The Automaticity of Everyday Life

THE AUTOMATICITY OF EVERYDAY LIFE
Advances in Social Cognition,

Volume X
This page intentionally left blank
THE AUTOMATICITY OF EVERYDAY LIFE
Advances in Social Cognition,

Volume X
Edited by
ROBERT S. WYER, Jr.

University of Illinois, Urbana-Champaign
Lead Article by
John A. Bargh
\}J ~~I~~~~~i?G~XP Press

NEW YORK AND LONDON
First Published 1997 by Lawrence Erlbaum Associates, Inc.
Published 2014 by Psychology Press

711 Third Avenue, New York, NY 10017
and by Psychology Press

27 Church Road, Hove, East Sussex, BN3 2FA
Psychology Press is an imprint of the Taylor & Francis Group,

an informa business
Copyright © 1997 by Lawrence Erlbaum Associates, Inc.
All rights reserved. No part of this book may be reprinted or

reproduced or utilised in any form or by any electronic,
mechanical, or other means, now known or hereafter invented,
including photocopying and recording, or in any information
storage or retrieval system, without permission in writing from
the publishers.
Trademark notice: Product or corporate names may be

trademarks or registered trademarks, and are used only for
identification and explanation without intent to infringe.
ICover design by Kathryn Houghtaling I

Library of Congress Cataloging,in,Publication Data
The Automaticity of Everyday Life

Advances in Social Cognition, Volume X
ISBN 13: 978-0-805-81699-0 (hbk)

ISBN 13: 978-0-805-81700-3 (pbk)
Contents
Preface vii
1 The Automaticity of Everyday Life

John A . Bargh 1
2 Environments and Unconscious Processes

Mahzarin R. Banaji, Irene V Blair, and Jack Glaser 63
3 Consciousness, Free Choice, and Automaticity

Roy E . Baumeister and Kristin L . Sommer 75
4 Some Thoughts Extending Bargh's Argument

Leonard Berkowitz 83
5 Associations to Automaticity
Charles S. Carver 95
6 Minding Our Emotions:

O n the Role of Automatic, Unconscious Affect
Gerald Clore and Timothy Ketelaar 105
7 Ifs and Thens in Cultural Psychology

Dov Cohen 121
8 Automaticity and Social Behavior:

A Model, a Marriage, and a Merger
Wendi L . Gardner and John T. Cacioppo 133
9 Rendering Accessible Information Relevant:

The Applicability of Everyday Life
Curtis D . Hardin and Alexander J. Rothman 143
V
vi Contents
10 The Automaticity of Academic Life:

Unconscious Applications of an Implicit Theory
Gordon D. Logan 157
11 Was the Cognitive Revolution Just a Detour

on the Road to Behaviorism? O n the Need
to Reconcile Situational Control and Personal Control
Walter Mischel 181
12 Preconscious Automaticity
in a Modular Connectionist System
Eliot R. Smith 187
13 The Vicissitudes of Social Behavior and Mental Life

Thomas K. Srull 203
14 Automatic but Conscious:

That is How We Act Most of the Time
Joseph Tzelgov 217
15 Reply to the Commentaries

John A . Bargh 231
Author Index 247
Subject Index 255

Preface
This is the tenth volume of the Advances in Social Cognition series. From its
inception, the purpose of the series has been to present and evaluate new
theoretical advances i n all areas of social cognition and information processing.
A n entire volume is devoted to each theory, allowing the theory to be evaluated
from a variety of perspectives and permitting its implications for a wide range of
issues to be examined.
T h e series reflects two major characteristics of social cognition: the h i g h level
of activity i n the field and the interstitial nature of the work. E a c h volume
contains a target chapter that is timely i n its application, n o v e l i n its approach,
and precise i n its explication. T h e target chapter is then followed by a set of
c o m p a n i o n chapters that examine the theoretical and empirical issues that the
target chapter has raised. These latter chapters are written by authors w i t h
diverse theoretical orientations, representing different disciplines w i t h i n psy-
chology and, i n some cases, entirely different disciplines. Target authors are then
given the opportunity to respond to the comments and criticisms of their work,
and to examine the ideas conveyed i n the c o m p a n i o n chapters i n light of their
o w n . T h e dialogue created by this format is both unusual a n d , we believe,
extremely beneficial to the field.
Theory and research i n the area of social cognition has traditionally focused
almost exclusively o n intentional, goal-directed information processing. In c o n -
trast, m u c h of the cognitive activity that underlies judgments and behavioral
decisions is likely to occur automatically, and often without awareness. A l t h o u g h
this possibility has sometimes been acknowledged by social cognition researchers,
the specific nature of the cognitive activity involved, and the way it influences
overt behavior, have seldom been specified i n detail.
For this reason, the work of John Bargh is of particular importance. Bargh's
conceptual and empirical contributions to an understanding of automaticity, w h i c h
span two decades, have long been recognized. In the present volume, he develops
a general theoretical formulation of automatic information processing that concep-
tually integrates the extensive research he has done, and discusses the implications
of the theory for comprehension, attitudinal judgments, and overt behavior. T h e
result is an exceptionally provocative contribution to theory and research i n social
information processing that is likely to have a profound influence o n our general
understanding of social phenomena.
vii
viii Preface
T h e potential importance of Bargh's chapter is matched by that of the compan-

ion articles. These articles, written by prominent researchers whose interests range
from cognitive science to cross-cultural psychology, not only help to refine and
extend Bargh's conceptualization but make important contributions i n their own
right. T h e issues they explore include: the interactive influence of individual and
environmental factors o n behavior; the interplay of conscious and nonconsious
processes; determinants of affect, emotion and aggression; biological and cultural
influences o n automatic processes; accuracy i n perceiving the sources of influence;
memory and resource allocation theories of judgment and performance; and the
implications of connectionist models for automaticity. A s a result, the volume as a
whole makes a valuable contribution to research and theory not only i n social
cognition, but i n many other areas as well.
In addition to the authors themselves, we want to acknowledge the invaluable
assistance of Lawrence Erlbaum Associates. Their continued support and encour-
agement of the Advances in Social Cognition series, and their commitment to the
publication of a high quality set of volumes, is deeply gratifying. It is a genuine
pleasure to work with them.
— Robert S. Wyer, Jr.

Chapter 7
The Automaticity of Everyday Life
John A. Bargh
New York University
MANIFESTO
If we are to use the methods of science in the field of human affairs, we must assume
that behavior is lawful and determined. We must expect to discover that what a man
does is the result of specifiable conditions and that once these conditions have been
discovered, we can anticipate and to some extent determine his actions. This
possibility is offensive to many people. It is opposed to a tradition of long standing
which regards man as a free agent, whose behavior is the product, not of specifiable
antecedent conditions, but of spontaneous inner changes of course.... If we cannot
show what is responsible for a man's behavior, we say that he himself is responsible for
it. The precursors of physical science once followed the same practice, but the wind
is no longer blown by Aeolus, nor is the rain cast down by Jupiter Pluvius.
(Skinner, 1953, pp. 6-7, 283)
A s Skinner argued so pointedly, the more we know about the situational causes
of psychological phenomena, the less need we have for postulating internal c o n -
scious mediating processes to explain those phenomena. N o w , as the purview of
social psychology is precisely to discover those situational causes of thinking, feeling,
and acting i n the real or implied presence of other people (e.g., Ross & Nisbett,
1991), it is hard to escape the forecast that as knowledge progresses regarding
psychological phenomena, there will be less of a role played by free will or conscious
choice i n accounting for them. In other words, because of social psychology's natural
focus o n the situational determinants of thinking, feeling, and doing, it is inevitable
that social psychological phenomena will be found to be automatic i n nature. T h a t
trend has already begun (see Bargh, 1994; Greenwald & Banaji, 1995), and it can
do nothing but continue.
1
2 Bargh
O f course, Skinner (e.g., 1978) was incorrect i n his position that cognition played
no role i n the stimulus control of behavior. Even modern animal learning theorists
in the Skinnerian tradition (e.g., Rilling, 1992) concluded that as soon as experi-
mental stimuli become more complex and extended over time than the simple static
tones and lights used by Skinner, cognitive mechanisms—especially perception and
representation—are indispensable for prediction and control of the animal's behav-
ior. However, as Barsalou (1992) pointed out, the fact that cognitive processes can
mediate the effects of situational stimuli o n responses does not make those responses
any less determined by those stimuli:
Like behaviorists, most cognitive psychologists believe that the fundamental laws of
the physical world determine human behavior completely. Whereas behaviorists view
control as only existing in the environment, however, cognitive psychologists view it
as also existing in cognitive mechanism. ... The illusion of free will is simply one more
phenomenon in that cognitive psychologists must explain, (p. 91)
In what follows, I argue that much of everyday life—thinking, feeling, and

doing—is automatic i n that it is driven by current features of the environment (i.e.,
people, objects, behaviors of others, settings, roles, norms, etc.) as mediated by
automatic cognitive processing of those features, without any mediation by con-
scious choice or reflection.
The Essential Automaticity of Social Psychological

Accounts of Human Nature
Theoretical accounts i n social psychology have always had a reflexive or automatic
flavor, because they lay out the situational factors causing the average person to
think-feel-behave i n a certain way. Take the following classic examples. For
thinking, if your own outcomes will depend o n the person you are about to meet,
you will spend the extra cognitive effort to learn about h i m or her as an individual,
instead of casually placing him or her into a stock category (Erber & Fiske, 1984).
For feeling, if you are i n a state of arousal, you tend to interpret your emotional
experience i n terms of how others i n the situation are reacting (Schachter & Singer,
1962). For behaving, if you are told to do something by an authority figure, you tend
to do it even if it means lying to another person (Festinger & Carlsmith, 1959) or
delivering an electric shock to a person who may be having a heart attack i n an
adjacent room (Milgram, 1963), and if another person needs help you will help if
you are the only person around, but not if there are others i n the vicinity who could
help (Darley & Latené, 1968).
In these several examples of situational influences o n cognitive processing,
emotional experience, and social behavior, the relation between situational features
and the effect of interest can be stated i n if-then terms: G i v e n the presence or
occurrence of a particular set of situational features (e.g., a person or event), a
certain psychological, emotional, or behavioral effect will follow.
1. The Automaticity of Everyday Life 3
The search for specifiable if-then relations between situations and psychological
effects also characterizes research o n automatic cognitive processes. A n automatic
mental phenomenon occurs reflexively whenever certain triggering conditions are i n
place; when those conditions are present, the process runs autonomously, inde-
pendently of conscious guidance (Anderson, 1992; Bargh, 1989,1996). Thus, research
and theory i n both domains, social psychology and automaticity, have, at the core,
the specification of if-then relations between situational events and circumstances
on the one hand, and cognitive, emotional, and behavioral effects o n the other.
The nature of these necessary preconditions (the if side of the equation) can
vary. Some require only the presence of the triggering environmental event; it does
not matter where the current focus of conscious attention is, what the individual
was recently thinking, or what the individual's current intentions or goals are. In
other words, this form of automaticity is completely unconditional i n terms of a
prepared or receptively tuned cognitive state. These are preconscious automatic
processes (Bargh, 1989) and are the major focus of this chapter. They can be
contrasted with postconscious and goal-dependent forms of automaticity (Bargh,
1989; Bargh & Tota, 1988), which depend o n more than the mere presence of
environmental objects or events. Postconscious automaticity is commonly studied
through the experimental technique of priming. Priming prepares a mental process
so that it then occurs given the triggering environmental information—thus, i n
addition to the presence of those relevant environmental features, postconsciously
automatic processes do require recent use or activation and do not occur without
it. Goal-dependent automaticity has the precondition of the individual intending
to perform the mental function, but given this intention, the processing occurs
immediately and autonomously, without any further conscious guidance or delib-
eration (e.g., as i n a well-practiced cognitive procedure or perceptual-motor skill;
see Anderson, 1983; Newell & Rosenbloom, 1981; Smith, 1994).
W h a t it means for a psychological process to be automatic, therefore, is that it
happens when its set of preconditions are i n place without needing any conscious
choice to occur, or guidance from that point on. M y thesis is that because social
psychology, like automaticity theory and research, is also concerned with phenom-
ena that occur whenever certain situational features or factors are i n place, social
psychological phenomena are essentially automatic. W h i c h of the different varieties
of automaticity a given phenomenon corresponds to depends o n the nature of the
situational (including internal cognitive) preconditions. Some situations may pro-
voke effects without any conscious processing of information whatsoever, and to
make the strongest and most conservative case for the automaticity of everyday life,
I confine myself i n this chapter to evidence of such preconsciously automatic
phenomena. B u t other situations might have their if-then reflexive effects by
triggering a certain intent or goal i n the individual, resulting i n attentional infor-
mation processing of a certain kind (i.e., an automatic motivation activation; see
Bargh, 1990). If the situation activates the same goal i n nearly everyone so that it
is an effect that generalizes across individuals, and can be produced with random
assignment of experimental participants to conditions, the only preconditions for
the effect are those situational features.
4 Bargh
One might well dispute this conclusion by pointing out the importance o f
mediating conscious processes and choice for the situational effects i n the previous
research examples. In the case of the bystander intervention research, for example,
the feeling of being less personally responsible to help if others are present (i.e.,
diffusion of responsibility) is said to mediate the effect of the number of bystanders
on the probability of helping (Darley & Latane, 1968). But if these conscious processes
do mediate the situational effect, then they must themselves be tied to those situations
in an if-then relation for there to be any general effect of the situational variable.
This may add extra steps to the if-then causal sequence (i.e., i/other possible helpers,
then feeling of less personal responsibility and then conscious decision not to help
and then no help given). For the effect to occur with regularity across individuals, the
feeling of less responsibility and the decision not to help, and so on, are also automatic
reactions to the situational information across different individuals.
But where is the evidence for those presumed conscious process mediators of the
effect? I confess I did choose the bystander intervention example for a reason; the
researchers had no evidence of the theoretical mediator of diffusion of responsibility
but instead inferred it from the effect of number of bystanders (Darley & Latané,
1968). T h e behavioral measure was taken as an indicator of the presence of the
cognitive mediator, i n other words (see discussion by Zajonc, 1980).
Bystander intervention research is not unique i n this regard. Following a review
of those studies i n which measures were made of behavior and the cognitive
processes believed to mediate it, Bern (1972) concluded:
Increase a person's favorability toward a dull task, and he will work at it more
assiduously. Make him think he is angry, and he will act more aggressively. Change his
perception of hunger, thirst, or pain, and he should consume more or less food or
drink, or endure more or less aversive stimulation. Alter the attribution, according to
the theory, and "consistent" overt behavior will follow.
There seems to be only one snag: It appears not to be true. It is not that the behavioral
effects sometimes fail to occur as predicted; that kind of negative evidence rarely
embarrasses anyone. It is that they occur more easily, more strongly, more reliably, and
more persuasively than the attribution changes that are, theoretically, supposed to be
mediating them. (p. 50)
Bern continued o n to give several examples of studies i n which both behavioral

and attributional dependent measures were collected, and i n which the behavioral
measure (e.g., eyelid conditioning, learning performance, pain perception, approach-
ing a feared object) showed clear effects, whereas the measure of the supposed
mediating conscious reasoning process showed a weak or absent effect.
Regardless of whether one shares Bern's conclusions regarding the limited media-
tional role played by conscious thought processes, the burden of proof has been
(unfairly) o n models that argue conscious choice is not necessary for an effect. To
convince skeptics that effects happen outside of consciousness, or do not require
conscious processing to occur, researchers have been made to jump through methodo-
logical hoops to establish nonconsciousness beyond any reasonable doubt. It might be
a step forward for social psychology to adopt the same level of healthy skepticism
for models that include a role for conscious mediation. Where is the evidence that
the mediating process exists, and where is the evidence of its mediation of the
observed effects? T h e assumption of conscious mediation should be treated with
the same scientific scrutiny as the assumption of automaticity.
The Inevitability of Continued Findings

of Automaticity
In developing the argument for the importance of automaticity within all of social
psychology, I am contending that social psychology has traditionally focused o n
situational determinants of behavior, and even within models such as attribution
theory that do posit a mediating role for conscious processes as opposed to
situational forces alone, there is insufficient evidence to support the position that
conscious mediation of situational effects is the rule rather than the exception.
Wherever such conscious mediators have been proposed, subsequent research
evidence has always constricted their importance and scope.
Note that, as research i n areas of social cognition such as attribution, attitudes, and
stereotyping progressed since the 1960s, evidence increasingly pointed to the relative
automaticity of those phenomena rather than the other way around. Take the case of
attribution theory. W h a t were once described i n terms of deliberative and sophisticated
steps of conscious reasoning (e.g., Kelley, 1967) were found to be "top-of-the-head"
(Taylor & Fiske, 1978), heuristic-based (Hansen, 1980), spontaneous (Winter &
Uleman, 1984), and finally automatic (e.g., Gilbert, 1989) reactions to the behavior of
others. The mediating role of one's attitudes on one's behavior moved from being
described i n terms of a conscious and intentional retrieval of one's attitude from
memory, to a demonstration of automatic attitude activation and influence (Fazio,
1986). The impact of cognitive structures such as stereotypes (e.g., Devine, 1989) and
the self (Bargh & Tota, 1988; Strauman & Higgins, 1987) o n person perception and
emotional reactions were shown to occur without needing involvement of intentional,
conscious processing (see Bargh, 1994; Greenwald & Banaji, 1995 for reviews).
The role of conscious choice was diminished even i n the realm of selection of an
individual's current processing goal. Social cognition models of the 1980s, for instance,
recognized how the outcome of processing was different as a function of the individual's
purpose i n processing the information. Yet the "goal-box" i n these flow-chart models
was presented as an exogenous variable that directed processing, not as an entity that
itself was caused by other factors (see, e.g., Smith, 1984; Srull & Wyer, 1986; Wyer &
Srull, 1986). However, as researchers uncovered more of the mechanism inside this
black box of goal selection (Atkinson & Birch, 1970; Bargh, 1990; Chaiken, Liberman,
ckEagly, 1989; Chartrand & Bargh, 1996; Gollwitzer & Moskowitz, 1996; Karniol &
Ross, 1996; Martin & Tesser, 1989; Martindale, 1991; Pervin, 1989; Wyer & Srull,
1989), the role presumably played by free will or conscious choice again was dimin-
ished—at least the need decreased to invoke the conscious will as a final recourse as it
became a superfluous explanatory concept.
6 Bargh
So even for social psychological models of the presumed cognitive mediating

processes, as research has advanced, so the role of conscious processing has diminished.
We have detailed knowledge of the situational features that produce a given phenome-
non for most people—a specifiable if-then relation tantamount to an automatic process.
But we also have a host of social-cognitive mediating processes such as attributions,
trait categorizations, attitudes, stereotypes, and goals, and these mediators are shown
increasingly to be equally automatic, if-then reactions to specific situational features.
THE PRECONSCIOUS CREATION

OF THE PSYCHOLOGICAL SITUATION
There is historical precedent i n theory and recent research evidence that automat-
icity plays a pervasive role i n all aspects of everyday life. N o t just i n input processes
such as perceptual categorization and stereotyping, which have been the principal
venue of automaticity research i n social psychology (see review i n Bargh, 1994);
not just i n the conscious and intentional execution of perceptual and motor skills,
such as driving and typing (see Newell & Rosenbloom, 1981; Bargh, 1996) or social
judgment (e.g., Smith, 1989)—but i n evaluative and emotional reactions, activa-
tion and operation of goals and motivations, and i n social behavior itself.
Environmental events directly activate three interactive but distinct psychologi-
cal systems, corresponding to the historical trinity of thinking, feeling, and doing
(see Fig. 1.1). By direct activation is meant preconscious—the strongest form of
automaticity (Bargh, 1989). Preconscious processes require only the proximal
registration of the stimulus event to occur—the event must be detected by the
individual's sensory apparatus, i n other words. G i v e n the mere presence of that
triggering event, the process operates and runs to completion without conscious
intention or awareness.
Evaluative
System
Environmental Motivational
Behavior
Features System
Perceptual
System
FIG. 1.1 Parallel forms of preconscious analysis.

A n individual's cognitive, affective, and motivational reactions to an environ-

mental event combine to constitute the psychological situation for h i m or her
(Koffka, 1925; Lewin, 1935; Mischel, 1973). A s it is the psychological situation
rather than the objective situation that then serves as the basis for further conscious
responses to the situation, the preconscious creation of the psychological situation
sets the stage and tone for all that follows an environmental event. M y focus is o n
the ways i n which the psychological situation is created preconsciously and auto-
matically for the individual.
T h e automatic, nonconscious perceptual interpretation of social stimuli was
demonstrated by a considerable number of studies (e.g., Bargh & Pietromonaco,
1982; Devine, 1989; Higgins, Rholes, & Jones, 1977; Niedenthal, 1990; Srull &
Wyer, 1979). Social behaviors are usually ambiguous enough to support various
interpretations or trait categorizations (e.g., independent or unfriendly; brave or
reckless), and so the readiness or accessibility of the relevant trait categories i n
memory—either through recent priming or chronic use i n the past—becomes
critical as to how that behavior will be understood. Moreover, biased assumptions
are often made about individuals based o n their social group membership, because
stereotypes of those groups automatically become active to influence person per-
ception outside of intent or awareness (see review i n Bargh, 1994). T h e evidence
for preconscious evaluation is more recent and perhaps not yet as well k n o w n .
However, it exists i n the domains of social attitudes (e.g., Bargh, C h a i k e n , Raymond,
& Hymes, 1996; Fazio, Sanbonmatsu, Powell, & Kardes, 1986), face recognition
(Murphy & Zajonc, 1993), and the neural substrates of emotional reactions
(LeDoux, 1989).
Before moving to a discussion of these bodies of evidence and their import, it
might be useful to present an a priori case for why preconscious influences should
be expected to obtain i n motivation and behavior as i n perception and affect. First
of all, theorists as diverse as Lewin (1935), M i s c h e l (1973), and Berkowitz (1984)
all argued that the psychological situation is not restricted to perceptual and
cognitive reactions to an event. Thus, for example, when we say something stupid
to a friend and wish we could take it back right away, we not only have cognitive
reactions, but also immediate emotional, visceral, and behavioral ones. A l l of these
reactions must be represented i n the mind.
Goals and behavioral responses do not exist i n some mysterious ether, but
correspond to mental representations i n m u c h the same way as do attitudes and
perceptual structures (see Bargh, 1990; Wyer & Srull, 1989). A n d because they are
mental representations, the same principle of automatization that produces auto-
matic perceptual interpretations, for instance, should apply to them as well. T h a t
is, as Hebb (1948) described, the principle of contiguous activation: Two (or more)
representations that tend to be active at the same time develop associative links to
one another. So if an individual makes the same categorizations (e.g., loyal) of a
given act (e.g., giving help to a friend during an exam) consistently over time, then
that trait representation will eventually become active whenever that behavior
pattern occurs i n the environment (Smith, Branscombe, c k B o r m a n n , 1988; S m i t h
& Lerner, 1986). If an individual makes the same evaluation (liking or disliking) of
8 Bargh
a given object consistently over time, then that evaluation will eventually become
active automatically whenever that object is perceived (Fazio et al., 1986).
A n d , if an individual has the same goal and intention within a given social
situation repeatedly over time, then that goal representation, with its associated
plans to attain the goal (Miller, Gaianter, & Pribram, 1960; Wilensky, 1983), will
become active automatically whenever those situational features are present i n the
environment to activate the internal representation of that situation (Bargh, 1990;
Bargh & Gollwitzer, 1994). T h i s hypothetical automatization of goal repre-
1
sentations through the consistent pairing of a given situation with the same
intention is at the heart of the auto-motive model of goal-directed action, to be
discussed next.
The Interface of World and Mind

H o w we immediately understand the world from moment to moment serves as the
starting point for everything we think, feel, and do i n response. Preconscious
processing is that initial stage of cognition i n which the world makes contact with
our mind. It operates o n sensory input and reduces and transforms it into meaning-
ful objects and events. T h e mental representations activated during preconscious
input analysis are those that were chronically associated with the stimulus event i n
the past.
Neisser's (1967) original description of preattentive processing limited it to the
recognition of patterns and to figural synthesis, so that what is furnished immedi-
ately to our conscious awareness and purposes while walking down the street are
cars and people and buildings and trees, not a blizzard of wildly moving light and
angles. Neisser, following the earlier work o n perceptual microgenesis (Flavell &
Draguns, 1957; Werner, 1956; Werner & Kaplan, 1963), persuasively demonstrated
that a considerable amount of cognitive work had to occur prior to conscious
awareness of a stimulus i n order to produce the common objects we take for granted.
Nonetheless, our subjective, conscious experience starts with these objects, as we
are not aware of the preconscious transformations that furnished them to our
awareness (see also Lazarus, 1982).
This principle of preconscious processing extends beyond the construction of
simple object percepts to also create for our conscious awareness the givens of our
social life and world. I review research showing that preconscious processing of
social information occurs as Neisser argued it does for nonsocial information, that
1
Again, although the objective situational features are the triggers that activate the chronic goal, this
occurs via the internal representation of that situation (i.e., its chronic construal or appraisal),which
may vary from individual to individual. For instance, one persons perceived threat may be another
person's perceived opportunity. Goals are formed in response to the way in which the situation is appraised
or interpreted by the individual, so the goal becomes automatically associated with the situational
representation; but as both the feature-to-representation and the representation-to-goal associations are
automatic, the perceptual registration of the objective features automatically results in activation of the
goal.
understandings and meanings about people and the social situations they inhabit
are furnished by these processes immediately and involuntarily, without any aware-
ness of their operation. We experience the output of these preconscious analysis as
if these meanings and understandings were clearly present i n the objective world,
when i n fact they are not (see Jones & Nisbett, 1971).
Aspects of the Psychological Situation
These immediate reactions are not just concerned with the categorization or
comprehension of the object or event, however. Lewin (1943) considered the
immediate psychological situation for the individual to consist of "needs, motiva-
tion, mood, goals, anxiety, ideals" (p. 3 0 6 ) — t h a t is, the totality of his or her
immediate reactions to the objective situation. In harmony with Gestalt principles
(e.g., Koffka, 1925), Lewin stressed the importance of this set of internal reactions
or meanings, and not the objective situation, as the stimulus for the individual's
behavior.
Mischel (1973) further developed the notion of the psychological situation i n
his social-cognitive model of personality. H e noted that an individual can have all
sorts of immediate reactions to a person or event, not limited to cognitive or
perceptual ones, but including (a) expectancies for what was going to happen next
i n the situation, (b) subjective evaluations of what was happening, (c) emotional
reactions one has had i n that situation i n the past and, most importantly to the
present thesis, (d) the behavioral response patterns one has available within the
situation based o n one's past experience.
W h a t the present argument adds to Mischel's (1973) analysis is that precon-
scious processes largely create the immediate psychological situation. T h e precon-
scious determines perceptual interpretations of the other people's behavior,
evaluative reactions to these people based on their physical features as well as their
actions, and one's own motives and behavioral responses within the situation.
In other words, there are three basic forms of preconscious analysis of the
environment that together constitute the immediate psychological situation: per-
ceptual, evaluative, and motivational-actional (see Fig. 1.1). T h e remainder of this
chapter reviews the evidence that these three types of reactions occur precon-
sciously o n the mere presence of the triggering stimulus. I argue that these three
systems operate simultaneously, i n parallel, and communicate with each other, so
that the output of one system has consequences for the others. For the same
environmental event to be processed immediately i n terms of its evaluative,
motivational, and perceptual implications, these different processing systems must
operate o n the same input at the same time (i.e., i n parallel). It would make a good
deal of sense if they shared information and perhaps operated o n the same cognitive
representations. Evidence of the existence of these causal links is presented.
I also argue that the operating characteristics of the three systems are not
identical. Rather, the three systems are dissociable, and they correspond to separate
processing modules (see Fodor, 1983; Jacoby, 1991; Johnson, 1983; and Tooby &
10 Bargh
Cosmides, 1992; for similar modularity arguments; and within this same series see
the recent contributions of K l e i n & Loftus, 1993, and Carlston, 1994, for further
evidence of dissociations between social-cognitive processes). Evidence that the
three systems are dissociated (see D u n n & Kirsner, 1988) is also presented.
THE DEVELOPMENT OF THE PRECONSCIOUS
Civilization advances by extending the number of important operations which we can

perform without thinking about them. (Whitehead, 1911, p. 143)
A l t h o u g h Lord Whitehead's claim was made at the grand level of civilization, it

holds equally well for the humble, individual human. Theoretically, preconscious
processes, like all automatic processes (Shiffrin & Dumais, 1981) develop out of
one's frequent and consistent mental, emotional, motivational, and behavioral
reactions to a given set of environmental features. Initially these reactions are
effortful and require conscious attention and intention. O v e r time, however, the
need for intention and attention diminish, given that the same categories or
evaluations or goals are always selected i n response to those features. Preconscious
automaticity models the regularities i n one's reaction to an event, and eventually
subsumes them so that the conscious mind no longer has to make decisions and
understandings it always makes the same way anyway. If this were not the case,
noted M i l l e r et al. (1960), none of us would be capable of getting out of bed i n the
morning.
Closer to Whitehead's point, the delegation of these routine processes to the
preconscious frees up processing capacity for the novel, creative work that only
conscious processing c a n provide—the chess master who c a n look far ahead
because the calculations that burden his or her opponent's attentional capacity
are made for h i m or her nonconsciously, the tennis champion for w h o m the
decisions as to where to r u n and which type of shot the opponent will attempt are
made preconsciously, freeing h i m or her to surprise and perplex the opponent with
a novel bit of strategy. Thus, an individual advances i n the same fashion as does
civilization.
Preconscious Processes as Mental Servants
Computer programmers are now developing interfaces for personal computers that
behave very m u c h like these preconscious mental processes. S u c h interfaces are
k n o w n as agent programming (Negroponte, 1995). Your personal agent program
resides i n your computer and performs such tasks as sorting your electronic mail,
sifting through the newsgroups you regularly enter, and finding postings that you
might be interested i n , among other functions. More importantly, such agents are
capable of programming themselves, mapping what they do onto the routines and
regularities demonstrated by their human user. A t the Massachusetts Institute for

Technology's M e d i a Lab, all users have such an intelligent agent program that over
time starts to take over, or automate, tasks such as scheduling meetings, electronic
mail responses, selection of what net-news to read, and so o n . Nicholas Negroponte
(1995), the director of the M e d i a Lab, summarized the concept of agent program-
ming:
The best metaphor I can conceive of for a human-computer interface is that of a

well-trained English butler.... It has to be able to expand and contract signals as a
function of knowing me and my environment so intimately that I literally can be
redundant on most occasions.... The concept of "agent" embodied in humans helping
humans is often one where expertise is indeed mixed with knowledge of you. A good
travel agent blends knowledge about hotels and restaurants with knowledge about
you (which often is culled from what you thought about other hotels and restaurants).
A real estate agent builds a model of you from a succession of houses that fit your taste
with varying degrees of success.... What they all have in common is the ability to
model you. (pp. 151, 155)
T h e present conceptualization of preconscious processes is no different: They

are mental servants that take over from conscious effortful processing those choices
that hardly ever vary, and so apparently are not worth bothering capacity-expensive
conscious processes with.
If you decide that you like something or someone, and you consistently have that
reaction, eventually that positive or negative evaluation is made for you when you
encounter that person or object, even if you are not thinking about how you feel at
all. If you take a certain kind of behavior as kind or insensitive, and do so
consistently, your preconscious agent eventually makes this categorization for you,
without your conscious involvement or knowledge. If you choose the same goal
within a certain situation, and do so consistently over time, that intentional choice
is eventually made for you when you enter that situation, and you may well behave
i n line with that goal without choosing it consciously. Conscious involvement is
bypassed i n the streamlined preconscious link between the environmental informa-
tion and one's perception, evaluation, and behavior.
The Preconscious and Skill Acquisition
This streamlining occurs for the same reasons and by the same process as does the
proceduralization or compilation of knowledge structures (Anderson, 1983; Smith,
1984, 1994), and skill acquisition (Bargh, 1996; Newell & Rosenbloom, 1981;
Wegner & Bargh, 1997). Note that i n both skill acquisition and knowledge
compilation, what were formerly separate procedures or components of the skill
become assembled into a single unit or structure. W h e n learning to drive, for
example, each component, such as turning the wheel the proper amount, pushing
the accelerator or brake pedal with the right force, or visually checking the traffic
12 Bargh
pattern o n all sides, requires considerable conscious attention and also needs to be
instigated by an act of intention or will. But eventually these components become
assembled into a larger unit, called driving, that still requires an act of intention or
will to be started, but when it is operating, the individual components no longer
require conscious choice or activation to operate. W h a t used to be several separate
skills each requiring an act of will to be engaged now become one single skill
requiring only one act of will.
T h e important point is that the basic idea of preconscious thought, evaluation,
or motivation—that which does not require an act of conscious will or intention to
occur—is already implicit within the research literature o n proceduralization and
skill acquisition. A s skills are acquired or procedures compiled i n these models, what
originally required an act of will to occur (e.g., hitting the brakes when seeing a stop
sign) can occur without that act of will with repeated pairing of stimulus features
and the intention to engage i n that skill.
THE THREE FORMS OF PRECONSCIOUS ANALYSIS
We now turn to a description of the three forms or systems of preconscious analysis

of the environment. In each section, evidence supporting the existence of direct
environmental control over that form of psychological reaction is presented. N e x t ,
the relation between that system of analysis and social behavior is described along
with evidence i n support of the automaticity of that pathway. Following the
schematic outline of Fig. 1.1, the three routes by which environmental stimuli
automatically and nonconsciously produce social behavior are traced: via automatic
social perception (i.e., the perception-behavior link), automatic evaluation (i.e.,
approach-avoidance motivation), and finally automatic goal and motive activation
(i.e., auto-motivation). A s the claim is made that these are three separate processing
modules, evidence is also be presented indicating their dissociations.
PERCEPTION
Preconscious effects i n social perception—those effects that are not mediated by

conscious intention—were demonstrated i n both impression formation and i n
stereotyping. Preconscious perceptual processes were shown to influence one's
2
2
T h e automaticity of social perception has been the most widely researched of the three forms of
preconscious analysis discussed here. A substantial amount of evidence supports its existence and thus
it is the least controversial of the three forms. Because several thorough reviews of this evidence already
exist (Bargh, 1989, 1994; Brewer, 1988; Greenwald & Banaji, 1995; Higgins, 1989; Smith, 1994; Wyer
& Srull, 1989), it is not reviewed in as much detail here as is the evidence regarding the automatic
evaluation and automatic motivation.
categorization and understanding of a person's behavior, and consequently the

impressions formed of that person. Trait concepts such as honesty, intelligence, and
aggressiveness, after frequent use i n understanding behavior relevant to them (i.e.,
behavior whose features match those i n the stored trait representation), can
eventually become capable of preconscious automatic activation i n the presence of
this behavior. This means that the behavior is encoded and categorized i n terms of
that trait regardless of the current focus of conscious attention or the current
processing goal (Bargh & Pratto, 1986; Bargh & T h e i n , 1985; Higgins, King, &
M a v i n , 1982; Uleman, Newman, ckMoskowitz, 1996; Winter & U l e m a n , 1984).
Trait Categorization of Behavioral Information
W h e n the behavioral information is clearly diagnostic of a given trait—when the

information matches a single trait concept very well and few, if any, others—it
activates that trait concept regardless of current conscious purposes. This is shown
by the fact that Srull and Wyer (1979) were able to activate or prime the trait
concepts of kindness and hostility by presenting scrambled sentences to participants
describing trait-relevant behaviors (e.g., "he fell her kicked") i n the guise of a
language ability test. Even though the participants' goal was not that of using the
information to form an impression of anyone, but merely to unscramble the words
i n the sentence, their subsequent impressions of a target person were more i n line
with the trait related to those behaviors than were the impressions of nonprimed
participants. Winter and U l e m a n (1984) and subsequent research by U l e m a n and
colleagues (see N e w m a n & U l e m a n , 1989; U l e m a n et al., 1996) showed that
sentences describing social behavior are encoded i n terms of the trait clearly
exemplified by that behavior even though participants' task is just to memorize each
of the sentences. Gilbert and colleagues (e.g., Gilbert, 1989; Gilbert, Pelham, &
K r u l l , 1988) showed the consequences of this automatic behavior-to-trait encoding
effect for conscious attributional processes. W h e n the participant is prevented
through attentional overload manipulations from gathering or integrating other
information as to the cause of a behavior (e.g., that concerning possible situational
causes), all he or she is left with is the automatic, default trait encoding, and so
tends to attribute the behavior to a dispositional trait of the actor's.
Construct accessibility research (see Bargh, 1989; Higgins, 1989; Wyer & Srull,
1989 for reviews) shows that the same unintended, preconscious interpretation of
behaviors can occur when the behavior is less than clearly diagnostic of a given trait
category. W h e n the behavior is ambiguously relevant to more than one trait
construct (see Bruner, 1957), the trait construct that is the most accessible, or easily
activated, from among the set of those applicable or relevant to the behavior will
be used to interpret the behavior. This greater top-down influence of construct
accessibility is not felt or experienced by the individual. Instead, the behavior is
perceived as clearly diagnostic of that trait—even though other participants who
do not have the trait as easily accessible would interpret the behavior differently
14 Bargh
(Bargh & Pietromonaco, 1982; Higgins, Rholes, & Jones, 1977; Srull & Wyer,
1979). Thus, if the individual is perceptually ready (Bruner, 1957) to perceive a
given trait, as a result of its heightened accessibility i n memory, preconscious
perception can occur even when the behavioral evidence is not diagnostic.
This increased accessibility of trait constructs can come either from recent
use—experimentally manipulated i n the previous studies through priming tech-
niques i n which stimuli semantically related to the trait are presented i n an
unobtrusive manner—or from frequent use i n the past by the individual. Techni-
cally speaking, only the latter form of accessibility produces truly preconscious
perceptual effects, because there are no conditions for producing such interpreta-
tions except the presence of the relevant behavioral information i n the environ-
ment. Priming effects involve the additional condition that the trait construct i n
question be recently used (and so are better termed postconscious processes; Bargh,
1989)—however, once a construct has been primed or recently used, the interpre-
tive effects it produces while active are indistinguishable from chronic or precon-
scious effects (see Bargh, Bond, Lombardi, & Tota, 1986; Bargh, Lombardi, &
Higgins, 1988).
Stereotyping
In the same way, social group stereotypes were found to be preconsciously
activated by the presence of features of the stereotyped group (see review i n
Bargh, 1994). R a c i a l , ethnic, gender, and age-related features of an i n d i v i d u a l
serve as diagnostic cues to his or her social group membership, and if there is a
stored stereotype of assumptions and beliefs about the characteristics of m e m -
bers of this group, it may become automatically active o n just the mere presence
of the group member (see Brewer, 1988). A s w i t h all preconscious processes,
what determines whether the stereotype becomes automatically activated i n this
way is whether it was frequently and consistently active i n the past i n the
presence of relevant social group features.
Evidence of the preconscious nature of stereotype activation comes from studies
i n which either (a) the stereotype is shown to become active subconsciously
(Devine, 1989), (b) conscious processing of the target information is prevented
through an overload manipulation (Pratto & Bargh, 1991), or (c) participants are
processing the stereotype-relevant information for conscious purposes unrelated to
people entirely (Mills & Tyrrell, 1983). Mills and Tyrrell, for example, had partici-
pants memorize a list of words presented one at a time. Unbeknownst to partici-
pants, o n certain series of trials consecutive words were related to either the male
or the female stereotype. (This was the only way these series of words were related).
Following each series, a word was presented consistent with the opposite stereotype.
Results showed that participants recalled words presented o n these "switch" trials
better than words within the consecutive series. Without participants being aware,
the words i n the series activated either the male or the female stereotype, w h i c h
was able to process subsequent stereotype-consistent stimuli using less attentional
capacity—a general feature of automatic processing (see Bargh, 1982; Bargh &
T h e i n , 1985; Gilbert, 1989; Macrae, M i l n e , & Bodenhausen, 1993). However, o n
encountering a stimulus word inconsistent with that stereotype, greater attention
was required and thus, that word was better recalled later—as are unexpected
stimuli i n general (see Fiske, 1980; Hastie & Kumar, 1979).
Devine (1989) activated the stereotype of African-Americans held by W h i t e
U . S . residents through the same subliminal priming manipulation Bargh and
Pietromonaco (1982) used to prime a single trait construct. However, Devine
demonstrated that a stereotype and not just a single trait construct was precon-
sciously activated by using as subliminal primes stereotype-relevant words that
were not related to hostility, although hostility was k n o w n to be a component of
that stereotype. N e x t , participants read about a fictitious target person (race
unspecified) who behaved i n an ambiguously hostile manner, and those partici-
pants whose A f r i c a n - A m e r i c a n stereotype had been primed rated the target as
being more hostile. T h e use of a subliminal priming technique, and the fact that
the target person was not explicitly depicted as A f r i c a n - A m e r i c a n suggest that the
activation and use of the stereotype i n this experiment was preconscious. It was
activated nonconsciously and unintentionally by environmental features relevant
to the stereotype, and then operated to influence perception of the target without
participants being aware of this bias.
The Self
T h e self-concept, which, like stereotypes, comprises a collection of interrelated trait
concepts (among other features), was also shown to become active automatically
in the presence of self-relevant stimuli, and, therefore, to affect self-perception and
emotions (Bargh, 1982; Bargh & Tota, 1988; Higgins, 1987; Strauman & Higgins,
1987). For example, Bargh (1982) showed that trait concepts belonging to the
individual's self-concept became active when trait-related stimuli were presented
to the unattended ear i n a dichotic listening study. This automatic activation was
evidenced by greater distraction away from the participant's conscious task com-
pared to when nonself-relevant stimuli were presented to the unattended ear,
although participants showed no awareness of the contents of the unattended
channel. In a different paradigm, Strauman and Higgins (1987) found that different
physiological reactions occurred to words related to the participant's ideal-self (i.e.,
aspirations) and ought-self (i.e., obligations) concepts. Specifically, participants
who felt they had not lived up to their hopes or duties actually experienced dejection
and agitation, respectively, after exposure to words related to those aspects of the
self. This occurred even though participants were not thinking intentionally or
consciously about the self at the time.
In summary, the interpretation of social behavior, whether it be one's o w n or
that of another person, and assumptions and expectancies about others' behavior
based o n their physical characteristics (e.g., skin color, gender features, voice
accent), can all be generated preconsciously i n the mere presence of these physical
16 Bargh
and behavioral features i n the environment. T h e next section traces the connection
between this automatic social perceptual system and behavioral responses to that
environment.
THE PERCEPTION-BEHAVIOR INTERFACE
The Principle of Ideomotor Action
Based o n the great capacity of humans and other primates for imitative behavior
(and speech i n humans), many prominent scholars argued that there is a strong
associative connection between perceptual and behavioral representations of the
same act, such that the very act of perceiving another person's behavior creates a
tendency to behave that way oneself (e.g., Bandura, 1977; Hilgard, 1965; James,
1890; Koffka, 1925; Lashley, 1951; Piaget, 1946; see review i n Prinz, 1990). James
labeled it the principle of ideomotor action, that thinking (consciously) about an
action activates the tendency to engage i n it. Piaget noted that the link between
perception and behavior must be innate, as the capacity to imitate is present i n early
childhood. In mentally retarded or brain-damaged patients for whom other con-
scious intentional forms of action control are unavailable, echoic or other imitative
reactions to others are still present (Prinz, 1990).
T h e theoretical mechanism invoked by Berkowitz (1984) to account for how
violence portrayed i n the mass media increased the probability of aggression i n the
viewer was James' principle of ideomotor action. A c t i v a t i o n was said by Berkowitz
to spread i n memory from representations of the violent acts perceived i n the media
to other aggressive ideas of the viewer, and this spreading activation occurred
"automatically and without m u c h thinking" (p. 410). A n experiment by Carver,
Ganellen, Froming, and Chambers (1983) tested this ideomotor action model of
the effect of aggressive cues o n aggression. In a first study, allegedly unrelated to the
critical experiment, the concept of hostility was primed for some participants,
following the procedure of Srull and Wyer (1979). T h e n , i n what they believed to
be an unrelated second experiment, participants were told to give shocks to another
participant (who was actually a confederate and received no actual shocks) when-
ever he or she gave an incorrect answer to a question. Participants primed with
hostility-related words gave longer "shocks" to the confederate than did nonprimed
participants.
The Automatic Effect of Perception on Action
For our present concern with whether social behavior can be produced entirely
automatically (i.e., nonconsciously), a critical aspect of the studies reviewed by
Berkowitz (1984) i n favor of the ideomotor action hypothesis (including the
Berkowitz & LePage, 1967, and Carver et al., 1983, experiments) is that participants
always had the conscious and intentional goal (given to them via experimental
instructions) to aggress against the confederate, i n that they were instructed to

shock the confederate for making incorrect answers. Would participants have
behaved i n a hostile manner to any degree if they had not been instructed to do so,
even if the concept of aggression had been primed? To demonstrate the existence
of preconsciously determined social behavior via the perception-behavior link, it
is necessary to show that the effect does not require conscious involvement or
intention, but merely the triggering environmental event. A recent study by Bargh,
C h e n , and Burrows (1996) tested this prediction.
Behavioral Consequences of Trait Concept Activation. We primed par-

ticipants with words related to either rudeness (e.g., assertive, rude, interrupt,
disturb) or to politeness (e.g., patient, polite, respectful) or neither (in the control
condition) i n the guise of a language test, i n what they believed to be a first
experiment. We expected that these primes would activate the perceptual construct
of rudeness or politeness i n our participants, exactly the same assumption that
guided use of this priming technique i n studies of impression formation. However,
we expected that this activated construct would have behavioral effects for the
individual and not only perceptual effects, i n line with the ideomotor action or
common-coding hypothesis.
Participants were then instructed to come down the hall to find the experimenter
when they were finished, so that they could participate i n a second, unrelated
experiment. W h e n the participant came down the hall, the experimenter was
engaged i n conversation with another participant, who was actually a confederate
of the experimenter. O n seeing the participant, the experimenter surreptitiously
started a stopwatch, but continued to answer the questions of the confederate. T h e
experimenter and confederate continued conversation for up to 10 minutes or until
the participant interrupted.
O u r results showed that considerably more (67%) of the participants randomly
assigned to the "rude" priming condition interrupted than did the participants
primed with "patience" related words (16%). Subsequent impression ratings of the
experimenter showed no differential perception due to the priming manipulation
(e.g., as rude or polite) that might have mediated behavior (see Herr, 1986; Neuberg,
1988). We did not expect any such differences because we did not design the
experimenter's behavior to be ambiguous i n any way with regard to rudeness or
politeness; i n general all participants felt the experimenter was moderately rude.
Extensive debriefing of participants indicated that they had no awareness of the
influence of the priming task o n their behavior.
These results, along with those of Carver et al. (1983), indicate that the same
priming manipulations that were shown to be successful i n influencing social
perception i n previous studies also influence the participant's social behavior. Trait
construct priming has the simultaneous effect of causing the participant to be more
likely to perceive that trait i n another person (given that the other person behaves
i n a way applicable to the trait construct; see Higgins, 1989), and to behave that
way himself or herself if such behavior is appropriate to the circumstances.
18 Bargh
Behavioral Consequences of Stereotype Activation. A logical derivation

from this parallelism i n the perceptual and behavioral effects of contextual trait
construct priming can be made to the case of stereotypes. Stereotypes are collections
of traits, among other features, and as discussed previously, they—like trait con-
structs—were demonstrated to become automatically activated i n the course of
perceiving another person (Brewer, 1988; Devine, 1989; Macrae et al., 1993; Pratto
& Bargh, 1991).Therefore, if nonconscious trait construct activation produces
trait-consistent behavior as well as perceptual influences, automatic stereotype
activation should also make the individual more likely to behave i n ways consistent
with the content of that stereotype.
Experiment 2 of Bargh, C h e n , et al. (1996) investigated the behavioral conse-

quences of automatic stereotype activation, a prediction based on the idea of the
perception-behavior link. Previous research on the content of the elderly stereotype
(e.g., Brewer, D u l l , & L u i , 1981; Perdue & Gurtman, 1990), as well as our own
pretesting, showed that it contains the notion of slowness and physical weakness.
We primed some participants with those other elements of the elderly stereotype
(e.g., forgetful, Florida, bingo) i n a scrambled sentence test (see Srull & Wyer, 1979)
as part of an experiment on language ability. Other participants were presented with
priming stimuli unrelated to the elderly stereotype (e.g., awkward, California, apples).
In order to demonstrate that the elderly stereotype itself was activated, and not
just the focal concept of slowness, none of the elderly priming stimuli was related
to slowness or weakness. We expected that the activation of the perceptual
construct of slowness or weakness by virtue of its participation i n the elderly
stereotype would have behavioral effects for the participant. Thus, if the priming
stimuli affected subsequent walking speed, it would indicate a mediational role for
the elderly stereotype itself o n the perceiver's behavior, as opposed to an effect of a
single trait concept (this was Devine's, 1989, procedure for demonstrating that the
A f r i c a n - A m e r i c a n stereotype had been automatically activated).
After the language test was completed, participants were led to believe the
experiment was over. They were thanked for their participation, and left the room.
T h e critical dependent measure was how long it took them to walk down the
hallway up to a piece of carpet tape about 40 feet away, as measured surreptitiously
by a second experimenter (blind to the participant's priming condition) posing as
another participant waiting outside the experimental room. A s hypothesized,
participants primed with the stereotypic content took longer to walk down the hall
after leaving the experiment than did control participants. A subsequent replication
with an additional set of participants produced the identical result. In both studies,
the experimenter caught up with the participant after he or she had passed the taped
line and fully debriefed him or her. W h e n later probed by the experimenter, no
participant showed any awareness of an effect of the language test on his or her
subsequent behavior or energy level.
In order to test an alternative explanation for this result i n terms of a mood effect
of the elderly priming stimuli—that it might have caused a depressed or sad mood
i n participants, causing them to walk more slowly—an additional group of partici-
pants were either primed with the elderly or the neutral priming stimuli, and then
completed the Salovey and Singer (1989) mood measure. There was no evidence
that participants i n the elderly priming condition were sadder than participants i n
the neutral priming condition; if anything, participants i n the elderly priming
condition reported being i n a nonsignificantly more positive mood than did partici-
pants i n the neutral priming condition.
We conducted a third experiment for two purposes: first, to assess the generality
of the elderly stereotype findings to a different stereotype altogether, and second,
to prime the stereotype subliminally i n order to rule out demand effects or other
conscious choice processes as convincingly as possible. In this experiment (Bargh,
C h e n , et al., 1996, Experiment 3), faces of young adult male African-Americans or
of young adult male Whites were subliminally presented o n the computer screen.
Participants engaged i n a dot estimation task i n which they were to respond as
quickly as they could on each trial as to whether the number of colored dots o n the
screen was odd or even. Immediately before the presentation of a trial (screen of
colored dots), a prime face was presented very briefly (13 msec) and pattern masked.
There were 130 trials i n the odd-even task, which lasted about 12 minutes.
Pretesting showed that participants found this task to be tedious and not enjoyable.
According to Devine (1989) and earlier studies of the A f r i c a n - A m e r i c a n stereo-
type held by many W h i t e U.S. residents, hostility is stereo typically associated with
African-Americans. Thus, our dependent measure i n this study was the degree of
hostility shown by the participant to a mild provocation that followed the dot
estimation task. O u r hypothesis, based o n the perception-behavior link, was that
subliminal presentation of the African-American faces to W h i t e participants should
automatically activate the trait concept of hostility as part of the African-American
stereotype, and, as a consequence, these participants would be more likely themselves
to respond i n a hostile manner, relative to participants primed with faces of Whites.
Following the last dot task trial, the participant was thanked by the experimenter
and moved to another seat nearby, i n view of the screen. Suddenly the computer
flashed error messages and beeped i n alarm that the participant's data was appar-
ently going to be lost due to a disk error. T h e experimenter voiced concern and
alerted the participant to the problem, saying, " O h , no, it looks like you might have
to do that task over again."
W h i l e this was going on, a hidden video camera across the room was recording
the participant's facial as well as verbal reactions to this piece of news. We had two
judges blind to the experimental hypotheses rate each participant's reaction o n
scales related to hostility, and after the session was concluded, we also asked the
experimenter—who was blind as to the participant's priming condition—to also
rate that participant's reaction to the request that he or she redo the dot task (in
actuality, no participant had to redo the task, as soon thereafter another message
appeared stating that the data had, i n fact, been saved after all.)
Results showed that once again, the automatic activation of a stereotype produced
stereotype consistent behavior. O n the judges' and experimenter's ratings, partici-
pants in the African-American prime condition showed a significantly more hostile
reaction to the provocation than did participants i n the W h i t e prime condition.
20 Bargh
Implications
We believe these findings have far-reaching implications for the question of the
automaticity of social behavior, and for the nature of social interaction. T h e fact
that perceiving another person's behavior, emotions, and so o n can make it more
likely that we ourselves behave that way suggests a possible explanation of empathic
reactions to others (see also Hodges & Wegner, i n press). O f course, true empathy
would also depend on whether our categorization of the other's behavior matched
the person's own understanding of it. Depending on our own chronic and temporary
category accessibility, the degree to which the other person feels we understand h i m
or her and empathize could be quite different.
These findings have equally important implications for the self-fulfilling nature
of stereotypes (e.g., Snyder, Tanke, & Berscheid, 1977). For example, suppose the
automatic activation of one's stereotype for African-Americans causes us to have
an automatic (unintentional and outside of our awareness) hostile reaction to an
A f r i c a n - A m e r i c a n person—perhaps communicated i n nothing more than the look
o n our face. O u r o w n automatic "first strike" might provoke a hostile reaction from
this person. However, we would only be aware of the person's apparently unpro-
voked hostility to us, and so we would interpret it as further supporting evidence
for our stereotypic beliefs. A recent study by C h e n and Bargh (1997) provided
evidence supporting this hypothesis. Compared to a nonprimed control group,
participants who were primed with African-American faces caused their subsequent
interaction partners to behave with greater hostility, as rated both by blind judges
and (even more importantly) by the primed participants themselves.
EVALUATION
There are two main lines of evidence of preconscious evaluation. First, the emo-
tional content of facial expressions was found to be picked up outside of conscious
awareness and intent to influence perceptions of the target individual (Murphy &
Zajonc, 1993; Niedenthal, 1990; Niedenthal & Cantor, 1986). Second, attitudes
toward social and nonsocial objects alike become active without conscious reflec-
tion or purpose immediately after encountering the attitude object (Bargh,
Chaiken, Govender, & Pratto, 1992; Bargh, Chaiken, et al., 1996; Fazio et al.,
1986).
Nonconscious Effects of Emotional

Expressions on Evaluation
Several studies showed that people are capable of detecting the emotional expres-
sion of faces outside of awareness, and that this information influences their
evaluations of target stimuli they subsequently consciously encounter. Niedenthal
and Cantor (1986) showed that participants liked the same faces better if the pupils
were dilated rather than constricted i n the photograph, although the participants
had no awareness that this feature influenced them so. Niedenthal (1990) showed
that subliminally presented facial expressions influenced the degree to w h i c h a
subsequent cartoon face was perceived as happy or sad, and Baldwin, Carrell, and
Lopez (1990) affected their participants' self-evaluations by subliminally flashing
photographs of smiling or frowning authority figures. Edwards (1990) found that a
subliminally presented facial expression presented prior to an attitude object
induced the formation of attitudes i n line with the valence of the facial expression,
and M u r p h y and Zajonc (1993) found a similar effect of subliminal faces o n
evaluative judgments of novel stimuli (Chinese ideographs).
A r e preconscious evaluation effects restricted to the special case of facial
expressions? The research o n automatic attitude activation shows that the precon-
scious evaluation effect is, instead, extremely general across social and nonsocial
stimuli.
Automatic Attitude Activation
Original Studies. Fazio et al. (1986) hypothesized that some attitudes

("strong"ones) would be more likely than others to become active automatically i n
the mere presence of the attitude object i n the environment. They operationalized
the strength of an attitude i n terms of how long it took the participant to give his
or her evaluation of the object; the faster the participant could evaluate the object,
the stronger the attitude was presumed to be. The names of the attitude objects
corresponding to each participant's fastest and slowest evaluative responses i n an
initial attitude assessment task served as the priming stimuli i n the sequential
priming task of Neely (1977). In this paradigm, a priming stimulus is presented for
a brief time (ca. 250 milliseconds) followed by a target stimulus to w h i c h the
participant is asked to respond. This time interval is too brief to allow the participant
to develop conscious expectancies and to implement strategic processes regarding
the target event (this usually takes at least 500 ms; see Neely, 1977; Posner & Snyder,
1975). Consequently, effects of the priming stimulus o n processing of the target
stimulus can only occur if the prime was processed preconsciously and automatically.
T h e target stimuli were positive and negative adjectives, which participants were
to evaluate (by pressing a "good" or "bad" button) as quickly as possible. Consistent
with their hypothesis, Fazio et al. (1986) found that primes corresponding to the
participant's strong (fast) but not weak (slow) attitudes did influence the adjective
evaluation response times. T h a t is, when the attitude object prime and the target
adjective shared the same valence, those times were faster than when prime-target
valence mismatched.
Generality of the Effect. Bargh et al. (1992) sought to investigate the gener-
ality of this automatic evaluation effect by studying the midrange of the attitude
strength distribution as well as the extremes. Based o n normative data for each of
22 Bargh
the 92 attitude object stimuli employed i n the Fazio et al. (1986) research, we
selected sets of positive and negative attitude objects that spanned the middle range
of the attitude strength (evaluation latency) distribution, and included them i n a
replication experiment. We obtained the automatic attitude activation effect for
the participant's idiosyncratically selected strong (fast) but not his or her weak
(slow) attitude object primes. However, we also obtained the automaticity effect for
the preselected midrange set of primes. This result suggested that the automaticity
effect was quite general across attitude objects.
Is Evaluation Truly Preconscious ? Subsequent experiments investigated the

conditions needed to produce preconscious evaluation effects. In these experi-
ments, aspects of the automatic attitude paradigm that might have artificially
produced the effect were systematically eliminated. For example, having partici-
pants consciously evaluate each of the attitude objects before testing whether their
associated attitude is automatically activated could have produced the effect as a
result of temporary activation or priming (see Bargh et al., 1986). However, when
we placed a 2-day delay between the attitude assessment and automaticity (adjec-
tive evaluation) tasks, we did obtain the automaticity effect, but for the participant's
weakest, as well as strongest, attitudes (Bargh et al., 1992; C h a i k e n & Bargh, 1993).
In other words, placing a delay between the attitude assessment and automaticity
tasks caused the effect to be found more generally for all attitude objects studied.
A n o t h e r aspect of the paradigm that stood i n the way of concluding the effect
was preconscious (i.e., unintended) was that i n the test of automaticity, participants
were given the explicit instructions to evaluate the adjectives as good or bad. This
conscious evaluation goal may have operated on the attitude object primes as well
as the adjective targets as they were presented concurrently i n time; thus the effect
would be produced by intention and not be preconscious. To eliminate this problem,
three experiments by Bargh, Chaiken, et al. (1996) had participants pronounce the
targets as quickly as they could instead of evaluating them. T h e pronunciation task
was shown to be a sensitive paradigm for detecting automatic spreading activation
effects (Balota & Lorch, 1986). We found the automatic evaluation effect once
again, for the strongest as well as weakest attitudes (Experiment 1). Next, we
removed other evaluative aspects of the paradigm, such as the immediately prior
attitude object evaluation task (Experiment 2) and the clearly evaluative adjectives
(Experiment 3), substituting mildly positive and negative nouns (e.g., water, bean)
as target stimuli. We continued to obtain the preconscious evaluation effect,
showing that it does not require conscious intention.
In summary, the automatic evaluation effect occurs regardless of the extremity
or strength of the prior attitude toward the object, and under conditions i n which
all aspects of intentional evaluative processing were removed. If anything, as those
conscious strategic processing conditions are eliminated from the paradigm, the
effect shows itself more clearly and pervasively. (We return later to a consideration
of why removing conscious aspects from the paradigm might also remove the
moderating effect of attitude strength as well.) A l l stimuli are evaluated immediately
as good or bad, without the participant intending to evaluate, having recently

thought i n terms of evaluation, or being aware of doing so. Therefore, everything
one encounters is preconsciously screened and classified as either good or bad,
within a fraction of a second after encountering it.
The Dissociation of Preconscious Evaluative

and Cognitive Processing
T h e findings of Bargh, Chaiken, et al. (1996), i n which any positively evaluated
object (e.g., water) facilitates the pronunciation of any other positively evaluated
object (e.g., Friday) even though the prime and target share no other semantic
feature i n common, cannot be accounted for by extant cognitive models of semantic
memory. Such prime-target matches, i n which only a single global feature is shared,
should produce the weakest of all priming effects (e.g., Balota & Lorch, 1986;
Collins & Loftus, 1975). In semantic network models, the degree to w h i c h two
concepts should prime one another—the probability that activation will spread
from one to the other—is a function of the number of relatively unique features the
two representations share i n common. Features (e.g., red; Collins & Loftus, 1975)
shared by many representations are posited to produce the weakest priming effects
because so many representations share that feature that the amount of activation
spreading to any one of them is minimal. O u r automatic evaluation results, on the
other hand, showed the pervasiveness of activation spread solely as a result of a
single feature (i.e., good or bad) that is shared by more representations than most
any other feature. A s many theorists argued (e.g., K u h l , 1986; Zajonc, 1980), there
seems to be something special about affective processing, i n that it does not play by
the same rules as does " c o l d " cognition.
There is neuropsychological evidence of a separation between affect and cogni-

tion as well. Korsakoff's syndrome patients cannot remember any biographical
information about target persons after 20 days, yet nearly 80% prefer the one
previously described as "good" than the one described as "bad" (Johnson, K i m , &
Risse, 1985). In animal research by L e D o u x and colleagues (see L e D o u x , Iwata,
Cicchetti, ckReis, 1988; Ledoux, 1989), brain areas (thalamic relay nuclei) repre-
senting the sensory features of acoustic stimuli associated with pain were found to
develop subcortical synaptical connections directly to the amygdala (responsible
for emotional responses to stimuli), bypassing the sensory cortex. In other words,
direct associative connections may develop between the stimulus feature repre-
sentation and the affective response mechanism, bypassing the cognitive processing
mechanism entirely.
Similar findings of a dissociation between affect and cognition were reported by
M u r p h y and Zajonc (1993). They conducted several experiments to test their affect
primacy hypothesis—that affective information is processed immediately and n o n -
consciously by a separate mental system. In support of this hypothesis, subliminally
presented faces of positive versus negative emotional states were shown to influence
judgments of the valence of ambiguous stimuli (Chinese ideographs). However,
24 Bargh
nonaffective properties of subliminal stimuli, such as the size of the presented

polygons, did not affect judgments of the Chinese ideographs along those dimen-
sions (i.e., whether the ideograph referred to something large or small).
Evaluation as a Qualitatively Different Aspect of Semantic Meaning.

Osgood, Suci, and Tannenbaum (1957) wrote 40 years ago that evaluation was the
primary semantic feature (accounting for most of the variance i n semantic differ-
ential studies of the semantic space) because it was probably accessed first among
all semantic features and often i n the absence of any other feature being accessed.
Osgood and colleagues, i n other words, were proposing the same dissociation
between affective and nonaffective processing as i n the present proposal and its
predecessors (e.g., LeDoux, 1989; Zajonc, 1980). A s is discussed i n the next section,
Osgood (1953) gave further reasons for his position that evaluative meaning had a
unique status among semantic features.
A study by Bargh, Litt, Pratto, and Spielman (1989) supported the Osgood et
al. (1957) contention. Participants were presented with a series of trait words via a
tachistoscope at stimulus durations below their conscious threshold. Following the
paradigm of M a r c e l (1983), we asked participants o n each trial either about the
semantic meaning of the subliminally presented word—whether another word was
a synonym—or whether they thought the subliminal word was positive or negative
i n meaning. Participants correctly answered the evaluative question at a better than
chance level, but at the same time were unable to answer the semantic question at
a rate better than chance. They had access to the evaluative meaning of the stimuli
in the absence of any access to the nonevaluative or semantic meaning.
Bargh, Raymond, and C h a i k e n (1996) tested whether the automatic evaluation
effect held for other major axes of semantic space as it does for evaluation. T h a t is,
evaluation might be the major dimension of semantic meaning, but it is not the only
one: activity (active-passive) and potency (strong-weak) account for a consider-
able amount of the variance i n semantic differential ratings as well (Osgood et al.,
1957). It is a possibility that evaluation is not unique i n its automatic qualities, and
that if we matched and mismatched primes and targets o n activity or potency we
might also obtain the automatic priming effect. If we did, it would suggest that the
effect is not special or unique to the case of affective processing. Thus, it is critical
to test this alternative account.
We asked a large group of students to rate each of the standard set of 92 attitude
objects used i n the previous automatic attitude research as well as the adjective
targets o n the dimensions of active-passive and strong-weak. We were then able
to construct prime-target pairs i n which both were strong, both were weak, or one
was strong and the other was weak (in Study 1), and similarly i n Study 2 for the
active-passive dimension. A t the same time, within each prime-target combina-
tion, we constructed one half to be of the same valence and one half to be of opposite
valence (i.e., both good, both bad, good-bad or bad-good pairs). In this way, we
could test for the presence of the evaluative priming effect at the same time we
tested for activity or strength priming effects. In both studies, there was no sign of
any automatic priming effect for the other two major dimensions o f semantic
meaning, but i n both studies the automatic evaluation effect was replicated.
T h u s there does seem to be something special about the evaluative d i m e n s i o n ,
as the same effects would not hold for other key dimensions o f meaning t h a n
evaluation.
Strategic Conscious Processes Interfere with Detection of Preconscious

Evaluation Effects. The second basis for concluding that the affective system is
dissociated from the nonaffective, or cognitive, processing system is evidence that
the more the conscious or strategic cognitive aspects of the task are eliminated, the
stronger and more general the automatic evaluation effect becomes (see also
C h a i k e n & Bargh, 1993). A s noted earlier, Fazio et al. (1986) showed this effect
first by including a 1000 msec prime-target delay condition—time enough for
strategic conscious processing effects to occur—and eliminating the automatic
attitude effect. T h e n , i n our first experiment (Bargh et al., 1992) we replicated the
original Fazio et al. (1986) paradigm exactly, and did not find the automatic
evaluation effect for the participant's weakest attitudes. But when we separated the
attitude assessment task from the automaticity test, we found the effect for all
attitude objects, regardless of attitude strength (Chaiken & Bargh, 1993). W h e n
we further removed aspects that induced or might have induced a conscious
evaluative mindset i n some way, such as the participants' explicit instructions to
evaluate the targets as good or bad i n the automaticity task (replacing this task with
pronouncing the targets instead) and common nouns instead of adjectives—clearly
positive or negative i n meaning—as the targets to be pronounced, the effect became
stronger and more pervasive. Throughout this entire series of experiments, therefore,
as steps were taken to remove conscious processing from the paradigm, the automatic
evaluation effect—the preconscious evaluation of all stimuli, social and nonsocial,
delivered through words and via pictures—came through louder and clearer.
These results are supportive of Murphy and Zajonc's (1993) hypothesis that the
more that conscious processing aspects were removed from their paradigm, the
clearer the affective priming effects would be. They demonstrated this by presenting
the same emotional faces above and below the participant's conscious threshold,
and only when the stimuli were presented subliminally did they affect judgments of
the ideograph targets.
THE EVALUATION-BEHAVIOR INTERFACE

(VIA MOTIVATION)
G i v e n the immediacy and pervasiveness of the automatic evaluation effect, the

question is, why do we evaluate everything we encounter as good or bad within .25
sec or so after encountering it? To what other systems is the output of the evaluative
module connected?
26 Bargh
Evaluation as a Trigger of Approach

and Avoidance Reactions
Several theorists, beginning with Lewin, proposed a direct link between evaluation
and approach-avoidance motivation. In his 1931 paper (reprinted i n Lewin, 1935),
he proposed this link quite explicitly: "Positive valence of an object i n the field has
attached to it an attraction motive or goal within the psychological situation, and
negative valenced objects have avoidance motives attached to them" (p. 92). In
accounting for why evaluation accounted for the lion's share of semantic meaning
of a concept, Osgood (1953) argued that the meanings of "signs" or semantic
representations are associated with overt (i.e., motoric) instrumental or behavioral
responses to the object i n question. Specifically, he contended that semantic
representations are linked to evaluative reactions such as approaching or avoiding
the object, and that these behavioral dispositions were included i n the repre-
sentation, or meaning, of the sign.
Neurophysiological Evidence. Recent neurophysiological evidence is con-

sistent with Lewin's and Osgood's conjectures. LeDoux et al. (1988) found that
separate pathways were involved i n autonomic and behavioral responses to fear-con-
ditioned stimuli i n rats. Electrolytic or chemical damage to one pathway between the
amygdala and the lateral hypothalamus interfered with the autonomic nervous
system reaction (increased blood pressure) to the conditioned stimulus but not the
behavioral response (bodily "freezing"), whereas damage to the other midbrain
interfered with the freezing response but not the autonomic. Thus there seems to be
a direct connection between emotional and behavioral representations i n rats.
Lang, Bradley, and Cuthbert (1990) proposed the valence of a stimulus is a basic
dimension by which the brain deals with information, with either a positive
approach or a negative avoidance motivational system activated by the stimulus.
These two basic affective-motivational systems are an appetitive, or approach, one
based o n positive valence, and an avoidance one based o n negative valence.
A c t i v a t i o n of one or the other motivational system by like valenced stimuli are said
to produce action dispositions. A l l affects, according to their model, are "primitively
associated" (p. 377) with either approach (approach, attachment, consummatory)
or avoidance (escape, defense) motives.
Recent research by Lang et al. (1990) and Cacioppo, Priester, and Berntson
(1993) provide further support for the evaluation-motivation connection. In the
studies by Lang and colleagues, basic reflex behaviors such as the startle response
to a sudden loud noise, or the knee flexion reflex, are stronger and more probable
when the participant is i n a negative emotional state than a positive one. Positive
reflexes such as salivation i n response to a tasty food stimulus are enhanced if the
participant is i n a positive emotional state or set and attenuated if he or she is i n a
negative set. Thus one's emotional or evaluative state predisposes one to valence-
consistent, approach versus avoidant behaviors, at an automatic and unintended
level (i.e., reflexes).
Cacioppo et al. (1993) showed that approach and avoidance feelings induced by
arm flexion versus extension influence attitude formation. Participants liked stimuli
more when at the same time their arm was flexed (i.e., pulling towards them)
compared to when their arm was extended (i.e., pushing away). A g a i n , this effect
is not intended nor conscious, as participants were unaware of any connection
between the position of their arm and their evaluations of the stimuli.
The Automatic Evaluation Effect and Approach-Avoidance Behavior.

Therefore, according to Osgood (1953), LeDoux (1989), and Lang et al. (1990),
and consistent with the findings of Cacioppo et al. (1993), the automatic evaluation
response demonstrated by Bargh, Chaiken, et al. (1996) should be connected to
and cause approach and avoidance behavioral responses (see also Lewin, 1926,
1935). It turns out that a student of Osgood's at Illinois, Solarz (1960) already tested
this prediction in an ingenious experiment. Participants held a lever that they could
3
either push away or pull towards them. On each trial, a card with a word printed
on it was exposed on a device mounted above the lever, which started a timer. O n
one set of trials, participants were told to push the word away from them if it was
unpleasant in meaning, and to pull the word toward them if it was pleasant; on
other blocks of trials they were given the opposite instructions. Results supported
Osgood's hypothesized linkage between evaluation and motivation: Participants
were faster to push away the unpleasant than the pleasant words, and faster to pull
toward them the pleasant than the unpleasant words.
Chen and Bargh (1996) recently conducted two experiments in order to directly
test the potential connection between the preconscious evaluation effect and
approach and avoidance motivations. In our first study, we conceptually replicated
the Solarz (1960) experiment, having participants evaluate as good or bad each of
a series of 92 stimulus words taken from the Bargh et al. (1992) norms. In a first
block of trials, participants either pushed a lever as quickly as they could to indicate
they disliked the stimulus whose name appeared on the screen and pulled the lever
to indicate they liked it, or vice versa. In the second block these instructions were
reversed. Our findings replicated those of Solarz exactly. Participants were faster to
make liking judgments by pulling the lever than by pushing it, and were faster to
make disliking judgments by pushing the lever than by pulling it.
However, in this replication as well as the original Solarz study, participants had
the conscious goal of evaluating the stimuli, and the thesis of the present chapter
is that these evaluative and motivational effects will occur just on the presence of
the stimulus; that is, preconsciously, without the need for conscious involvement.
Thus it must be that the same effects would occur if participants did not have the
conscious goal of evaluation. Following the same logic, Bargh, Chaiken, et al. (1996)
showed that the automatic evaluation effect was preconscious, as it occurred even
when participants did not have any conscious goal or intent to evaluate the stimuli,
but merely pronounced them.
3
M y thanks to Jerry C l o r e for alerting me to this study.
28 Bargh
Therefore, i n Experiment 2 of C h e n and Bargh (1996), we informed participants

that we were interested i n reaction times, and instructed them to move the lever
as quickly as they could when a word came o n the screen, i n order to "knock it off."
T h e words—again the names of positive and negative stimuli from the Bargh et al.
(1992) norms—were presented at random time intervals to enhance the reaction
time cover story. O n one half of the trials, participants pushed the lever to remove
the word from the screen; o n the other half, they pulled the lever (the order i n which
they pushed or pulled was randomized for each participant). N o t h i n g i n the
experimental instructions mentioned evaluation of the stimuli. Nevertheless, o n
the trials i n which participants pushed the lever, they were faster when the stimulus
had a negative valence, and o n the trials i n which they pulled the lever, they were
faster when the stimulus had a positive valence.
These findings show that the preconscious evaluation response extends to the
activation of motoric response tendencies, a direct preconscious route from stimulus
to motivation and behavior i n line with the proposals of Osgood (1953), Lang et
al. (1990) and others. Because the preconscious effect demonstrated by Solarz
(1960) and C h e n and Bargh (1996) travels through the evaluative system to
predispose the individual to certain behavioral responses, it constitutes another
source of evidence of automatic motivation—one that is continually occurring for
all stimuli the individual encounters based solely o n their emotional valence.
GOALS AND MOTIVATIONS
Goals Are Strategies That Interact

With the Environment
O n e of the historical roots of modern automaticity research is research o n skill
acquisition (see Bargh, 1996). A n y skill, be it perceptual, motor, or cognitive,
requires less and less conscious attention the more frequently and consistently it
is engaged (e.g., Newell & Rosenbloom, 1981). Eventually, with enough practice
and use, it becomes capable of operating with no conscious attention at all. In
social psychology, Smith's research (e.g., 1994) has, documented the decreasing
need for conscious guidance of social judgments with increased experience i n
making them.
These skills are intentional processes. A l t h o u g h they are very efficient and
require very little attention, they still require an act of intention or will to occur.
Consider walking down the street. A s automatic as that skill might be for most of
us—allowing us to daydream, plan, chew gum, and do lots of other things without
having to monitor or guide our legs and feet—we still have to intend to walk
somewhere i n the first place. W i t h experience, these automated skills come to
operate autonomously, so that once initiated they interact with the complex
environment as automated strategies. We adjust our walking to the other people
and the vehicular traffic and the weather and the light conditions without trying
to; part of the skill of being able to walk is the intake of current information and
basing our responses o n it. But the intake of walking-relevant information and our
adaptation to it is just as much part of the automated walking skill as are lifting our
legs and placing our feet o n the pavement. Because it is automated, we are not
aware of how we are selecting and using and reacting to the environmental
information.
Vera and Simon (1993) called this aspect the functional transparency of the skill.
G i v e n sufficient experience i n the domain, the relevant information is represented
at a highly abstract functional level, and it is this level (e.g., "driving to work") of
which one is aware, not the concrete level of details (e.g., "going down Depot Street,
wait for the light, signal a left turn, left onto Madison . . . "). W h e n one is just
learning how to drive, one has to make every decision consciously, even as detailed
as when to let go of the steering wheel during a turn. W i t h practice, that decision
does not need to be made consciously, as it is subsumed or compiled (Anderson,
1983) as part of the "making a turn" unit. Nonetheless, at this level of skill, the driver
still has to decide consciously when to make the turn. Eventually, even that decision
becomes functionally transparent—no longer needing to be made consciously—un-
der the even more abstract goal of "following the road" or "driving to work."
Therefore, the goal that is operating here autonomously and without conscious
guidance is not a single, static behavioral response to a stimulus, but an automated
strategy or plan for interacting with the environment i n order to achieve a desired
goal. W h a t is active is a mental structure that not only interacts with environmental
information, it requires that information to operate just as a car requires gasoline.
W h e n we refer to a goal or motive being triggered preconsciously, i n other words,
it is a goal with associated plans to achieve an outcome. A s Vera and S i m o n (1993)
described them, "Plans are not specifications of fixed sequences of actions, but are
strategies that determine each successive action as a function of current information
about the situation" (p. 17).
It is clear from the skill acquisition literature that the goals an individual
frequently and consistently pursues i n a given situation are capable of operating
autonomously and without the need for conscious guidance. W h a t starts them i n
motion? It is the activation of the goal or intention, the "top node" i n the goal system
under w h i c h the substrategies and processes are subsumed.
The Auto-Motive Model

O n c e activated, functionally transparent or automated skills can interact with the
environment i n a sophisticated way, taking i n information relevant to the goal's
purposes, and directing appropriate responses based o n that information, without
the need for conscious involvement i n those responses. T h e auto-motive model of
goal-directed action (Bargh, 1990) added just one assumption to this idea: T h e
entry point or trigger that starts that goal into operation can itself be subsumed and
removed from conscious choice. It can do so if that conscious choice point itself
becomes routinely associated with a set of environmental features. Environmental
representation and choice point can be compiled together if the situational repre-
30 Bargh
sentation and the goal i n question are repeatedly active together (Hayes-Roth,
1977; Hebb, 1948).
Although to claim that one can engage i n these goal-directed actions without
consciously intending to do so is also to argue that often one does not have conscious
control over one's responses to the environment, there seems no a priori reason not to
extend the principle of functional transparency to the instigation of the goal itself. A s
long as the same principles that caused conscious choice to be subsumed and eliminated
from the originally separate components of that skill apply to the instigating choice itself,
that choice should likewise be capable of delegation to the environment.
Thus, the central hypothesis of the auto-motive model is that this goal or
intention, this complex strategy of interacting with the world, can be started i n
motion by environmental stimuli. Stimuli i n the environment can directly activate
a goal, w h i c h will then become operative and guide cognitive and behavioral
processes within that environment, without any need for conscious decision.
This position, that the goals and motives guiding behavior can operate noncon-
sciously, has precedent. Jung (1927) argued that people often engage i n routine and
regular patterns of behavior, the motive for which might not be accessible to
consciousness. However, the individual may nonetheless experience the behavior
as consciously chosen, for he or she would supply a conscious motive or "rationali-
zation" for it:
We have grown accustomed to scrutinizing our own actions and to seeking rational
explanations for them. But it is by no means certain that our explanations will hold
water, indeed it is highly unlikely.... As a result of our artificial rationalizations it may
seem to us that we were actuated not by instinct but by conscious motives. (p. 301)
T h e notion of unconsciously operating motives does not appear only i n the writings
of psychodynamic theorists, however. Gazzaniga (1985) noted the same phenome-
non i n split-brain or Korsakoff's patients, of behavior generated by unconscious
activation of a goal that is then given a conscious rationalization. If a message is
flashed to the right hemisphere of such a patient, such as to get up and leave the
room, the behavior will occur. But when stopped by the experimenter and asked why
he or she is leaving, the patient is likely to respond almost immediately with a plausible
reason, such as, "I needed to get a drink of water." Hypnotized people, who have
ceded control over their behavior to the hypnotist (see Hilgard, 1965), show the
same ability to quickly rationalize behavior they did not instigate themselves. Hilgard
(1977) gave the example of a participant who was given the command to walk around
on the floor o n her hands and knees after she woke up. She was awakened, and then
crawled around on the floor, saying, "I think I lost an earring down here."
According to the auto-motive model, because goals and motives must be
represented i n the mind just as are other knowledge structures, they should be
capable of becoming automatically associated with representations of those envi-
ronmental features they are consistently paired with, just as do other automatic
associations (e.g., Shiffrin & Schneider, 1977). Thus, if an individual nearly always
pursues the same goal within a given situation, that goal will come eventually to be
preconsciously activated within that situation, independently of the individual's

conscious purposes at that later time.
W h e n the auto-motive model was first presented (Bargh, 1990), it had the status
of a n untested hypothesis, but since then several relevant experiments were
conducted. Two general types of automatic goal effects were studied: cognitive or
information-processing goals, and social-behavioral goals.
Automatic Activation of Information-Processing Goals
Several studies support the idea that an activated goal c a n operate outside of
awareness and can therefore unconsciously influence processing. O n e set of such
experiments looked at the residual effects of activated goals. Participants are given
a certain processing goal or mind set via explicit experimental instructions i n a first
task, and then it is shown that this goal continues to operate i n a subsequent,
ostensibly unrelated task.
Deliberation Versus Implementation. Gollwitzer, Heckhausen, and Steller

(1990) performed the first study of this kind. T h e experimenter induced either a
deliberative or an implemental mindset i n the participant by instructing h i m or her
to think about a personal problem i n one of two ways: either i n terms of alternative
ways to solve it, or i n terms of specific steps they might take to solve it. N e x t , i n an
ostensibly unrelated second experiment, participants were given a story completion
task. T h e story was one of several fairy tales and the participant was given just the
first few sentences. O n e , for example, concerned a medieval king who was going
off to war but did not want to leave his daughter alone i n the castle unprotected.
Participants could complete the story any way they chose, but those who had been
given a deliberative mind set i n the first experiment wrote more about all the
possibilities the king was considering, whereas those given an implemental, action-
oriented mind set continued the story with what the king actually did i n order to
solve his problem. Apparently, the goal used i n the first experiment was still active
and hence operated o n relevant input i n the second study, without participants
being aware of or choosing this mode of processing.
Motives in Processing Persuasive Communications. C h a i k e n and col-

leagues (Chaiken, Giner-Sorolla, & C h e n , 1996; C h e n , Shechter, & C h a i k e n ,
1996) also used the unrelated-task paradigm to prime various processing goals. In
the first task, participants were given scenarios to read i n which the target person
was portrayed as being concerned with accurately understanding what was going
on, or with making a good first impression o n another person. This manipulation
was intended to activate either an accuracy or an impression-management motiva-
tion i n the participant. In the second, apparently unrelated experiment, participants
were given an attitude issue (e.g., gun control) and informed that they would be
discussing this issue with another participant (in reality there was no other partici-
32 Bargh
pant). T h e other participant was described as holding either a pro or c o n position

with respect to that issue.
Next, participants read an essay containing arguments o n both sides of the
controversy, during which they wrote down their reactions to the essay. T h e n , after
reading it, participants were asked to give their own attitude about the topic. T h e
stated attitudes of participants who had earlier read about a person trying to make
a good impression were more i n line with that of the participant they expected to
meet than were the positions of participants who had read about a person concerned
with accurate information processing. Furthermore, content analysis of the thought
protocols revealed that participants evaluated essay arguments supporting the other
participant's position more positively i n the impression-management condition
than i n the accuracy condition.
Thus, i n this study, as i n Gollwitzer et al. (1990), the conscious activation of a
cognitive processing goal or motivation i n one context increased the likelihood that
this goal, rather than other, relevant goals, was used i n processing subsequent
information. This occurred even though participants were not aware of and did not
intend this subsequent influence.
Cognitive Consistency Motivation. In research by Bator a n d C i a l d i n i

(1995; Cialdini, 1994), cognitive consistency goals were either primed or not primed
i n a first experiment. T h e n , the same participants took part i n an ostensibly
unrelated cognitive dissonance experiment. Participants were told they would be
interacting with another person, and then had them read an essay purportedly
written by that person. T h e content of this essay either communicated that the
other person really valued consistency i n beliefs and behavior, or did not. Next, i n
what was an allegedly unrelated experiment, participants were asked to write an
essay i n favor of their university implementing comprehensive examinations as a
requirement for graduation. This position was opposite to all participants' actual
personal positions o n the matter. Participants wrote this counterattitudinal essay
either under free choice or no choice conditions. Following completion of the essay
they were asked for their o w n positions o n the issue.
A c c o r d i n g to dissonance theory, participants writing counterattitudinal essays
under free choice conditions should become more favorable towards the issue than
participants who believed they had no choice i n writing the essay (Wicklund &
Brehm, 1976). In the Bator and Cialdini (1995) study, this effect was obtained only
for those participants whose consistency motivation had been primed i n the
ostensibly unrelated first experiment. Participants i n the no-prime condition held
the same final position on the comprehensive exam issue whether they had written
the essay under free choice or no choice conditions. The results supported Cialdini's
(1994) hypothesis that dissonance and other consistency effects were obtained i n
previous research because of the communication of subtle consistency cues to the
participant by the experimenter i n those paradigms, a situation that the consistency
priming manipulation simulated. For our purposes, the fact that cognitive consis-
tency motivations can be primed is encouraging support for the auto-motive model.
Preconscious Activation and Operation

of Social Information Processing Goals
To demonstrate the preconscious activation and subsequent nonconscious

operation of cognitive processing goals, however, the role of conscious intent
must first be eliminated. A t present, the evidence shows that a goal, recently
consciously chosen and pursued, has a lingering influence i n subsequent c o n -
texts i n w h i c h it is not consciously chosen. A s such, this is a postconscious effect
of recent experience (Bargh, 1989) that depends o n a conscious choice of the
goal i n order to occur. To demonstrate that the cognitive processing goal is
triggered directly and unconditionally by the environmental event w i t h no
intervening role played by intentional conscious processing whatsoever, one
must show that the goal can be primed passively or nonconsciously and still
produce its signature effects.
A pair of experiments by Chartrand and Bargh (1996) supports this prediction.
We replicated two previous research paradigms, both of which had shown different
processing outcomes (memory organization, judgments) depending o n the partici-
pant's processing goal when encountering the stimulus information. However,
whereas the previous studies gave participants one or the other processing goal via
explicit instructions, we primed the goals passively, and i n one case, subliminally.
Impression Versus Memory Coals. The first experiment replicated the clas-
sic study by Hamilton, Katz, and Leirer (1980). In their study, participants read a
series of behaviors with instructions either to form an impression of the actor or to
memorize the information. Participants had greater free recall of the target's
behaviors, and greater degree of organization of the material i n memory according
to trait category (sociable, intelligent, athletic, religious), when they had an impres-
sion formation objective than when they had a memory objective.
In our study, we did not give our participants any explicit instructions i n how to
process the information. Rather, we told them merely to read it as we would ask
them questions about it later. To prevent participants from spontaneously having
an impression formation goal, we presented only the behavioral predicates (as had
H a m i l t o n et al., 1980; e.g., "had a party for some friends last week"), without
informing them that the behaviors had been performed by a single individual. Before
exposing participants to the behaviors, however, we had them perform an ostensibly
unrelated "language experiment" i n which they were unobtrusively exposed to the
priming stimuli via the scrambled sentence test described earlier (Srull & Wyer,
1979). Embedded i n the 15 items of this test were words related either to the goal
of forming an impression of someone (e.g., opinion, personality, evaluate) or to the
goal of memorizing information (e.g., absorb, retain, remember).
O u r results replicated those of H a m i l t o n et al. (1980) exactly. T h a t is, partici-
pants whose impression formation goal was primed recalled significantly more of
the behaviors than did participants i n the memorization condition. Moreover, their
recall protocols showed significantly higher clustering according to trait category.
34 Bargh
On-Line Impression Formation. O u r second experiment was a replication of

Hastie and Kumar (1979) and related person memory studies that used their
paradigm (e.g., Bargh & T h e i n , 1985; Srull, 1981). Unlike those studies, however,
we did not give our participants any explicit goal to form an impression of the target
person, and we again presented only behavioral predicates. Instead, we subliminally
primed the impression formation goal for some participants, presenting impression
related stimuli outside of conscious awareness (i.e., parafoveally, very briefly, and
masked) i n the context of a speeded reaction time task (Bargh & Pietromonaco,
1982). Other participants were not exposed to such impression related stimuli during
the reaction time task. This task was described to participants as a separate experi-
ment from the critical task that followed.
T h e major prediction was that participants whose impression formation goal had
been subliminally activated would show evidence of on-line impression formation
(Bargh & T h e i n , 1985; Hastie & Park, 1986; Lichtenstein & Srull, 1987), that is,
impressions formed prior to being explicitly asked for their opinion of the target
person by the experimenter. We hypothesized that participants whose impression
goal had not been primed would not form an impression until asked for it by the
experimenter (Srull, 1981), and so they would not show evidence of on-line
impression formation effects. There are three signatures of on-line impression
formation. O n e is a direct influence of the information presented o n impression
judgments that is not mediated by the information the participant has just recalled
(prior to the impression ratings) o n a surprise free recall test. Separating the direct
from indirect influences can be done through path analytic techniques (Bargh &
T h e i n , 1985). A n o t h e r indication of on-line impressions would be judgments that
more greatly differentiated the target persons o n the trait dimension o n which they
varied (honesty-dishonesty), given that there were clear differences i n the degree
of honesty of the two targets. O n e half of the participants were presented with 12
honest and 6 dishonest (and 6 neutral) behaviors, and one half the participants
with 6 honest and 12 dishonest (and 6 neutral) behaviors (following Bargh & T h e i n ,
1985). T h e third signature of on-line impression formation for w h i c h we tested was
the emergence of a recall bias for the minority behavior type, which occurs only
after the participant forms an impression and then processes subsequent impres-
sion-incongruent information more elaborately, i n an attempt to integrate it with
that impression (Srull, Lichtenstein, & Rothbart, 1985).
After the behaviors were presented, all participants were given a surprise free-recall
test, being asked to write down all of the behaviors they could remember. Then, they
were told that all behaviors had been performed by the same person and were asked to
rate the person with respect to both honesty and other, unrelated traits. Having
participants give their impressions after just recalling the behaviors should increase the
degree of correspondence between the memory and impression rating measures.
However, if participants had formed and stored an impression on-line during informa-
tion acquisition (see Carlston, 1980), honesty ratings should be a direct function of the
proportion of honest (vs. dishonest) behaviors presented, independently of the propor-
tion of honest to dishonest behaviors the participant had just recalled.
O u r findings supported these predictions. O n l y for participants whose impres-

sion formation goal had been triggered subliminally were impression ratings a direct
and significant function of the information presented. Moreover, the impressions
formed by these participants were significantly more polarized along the honest-dis-
honest dimension than those of other participants, more accurately reflecting the
clear difference i n the honesty of the two target people.
Goals Operate the Same Whether Activated

Consciously or Nonconsciously
These results provide clear support for the hypothesis, generated from the auto-mo-
tive model, that goals operate o n relevant information and attain their purpose (e.g.,
impression formation), regardless of whether they were activated consciously or
nonconsciously. T h a t is, the source of the activation does not matter, nor does
whether the person intends to pursue the goal, or whether the person is aware of
having that goal at the time.
T h e auto-motive model argues that motivations such as accuracy, defense,
impression management, and consistency exist i n chronic form i n some individuals,
and that every person has chronic goals that are triggered automatically by envi-
ronmental stimuli. Priming is an excellent technique for experimentally manipulat-
ing automatic goal activation and operation. However, the results of priming studies
are intended to generalize beyond temporary contextual influences to chronic,
context-independent states. T h e effects of chronic accessibility mimic those of
priming or temporary accessibility, and the two forms combine additively, suggesting
a c o m m o n underlying mechanism (i.e., amount of activation; Bargh et al., 1986;
Bargh et al., 1988; Higgins, Bargh, & Lombardi, 1985). Thus, the auto-motive
model assumes that the findings of studies i n which goals are primed generalize to
cases i n which those goals exist for the individual i n a chronically accessible form,
so that absolutely nothing is needed for that goal to become active within the
situation to which it is associatively tied, not even recent prior use (as i n priming
studies). However, this is an assumption, and the arguments of the auto-motive
model would be strengthened if the same motivations were to be shown to exist i n
chronic form as individual differences.
In the C h a i k e n et al. (1996) and Cialdini (1994) research, chronic individual
differences i n impression-management motivation and consistency motivation,
respectively, were demonstrated. The same effects were obtained i n their paradigms
when the individual difference variable was substituted for the experimental prim-
ing manipulation, without the need of introducing a recent priming event. In a study
by C h e n , Shechter, and C h a i k e n (1996), high self-monitors showed a greater
tendency than did low self-monitors to have impression-management motivations
within persuasion situations and to adapt their o w n expressed attitudes to what they
believed to be the attitudes of their experimental partner. C i a l d i n i , Trost, and
Newsom (1995) demonstrated individual differences i n a "preference for consis-
tency" questionnaire that predicts responses i n the foot-in-the-door, balance, and
36 Bargh
dissonance experimental paradigms. T h e classic findings i n those paradigms were

obtained only for those participants with a chronic consistency motivation. Those
with no such chronic motivation did not show consistency effects within the
standard experimental situations.
Automatic Social Behavior
Bargh, Gollwitzer, and Barndollar (1996) examined a goal-conflict situation, that

between achievement and affiliation i n a classroom setting. M a n y of us have
experienced a conflict between wanting to raise our hands constantly to a teacher's
questions and wanting to avoid having the other students dislike us because we are
showing them up. O f t e n this conflict is resolved by "dumbing d o w n " and not
achieving at the highest possible level i n order to also have friends and be accepted
by our peers.
Priming Achievement Versus Affiliation. In the experiment, we primed par-

ticipants with words related to achievement (e.g., strive, success) or affiliation (e.g.,
friend, sociable) i n an initial word search puzzle. T h e n , the participant worked
together with another participant (actually a confederate) as a team to find as many
words o n each of a series of five additional word search puzzles. This confederate,
however, was very bad at the task and, as the experimental session progressed,
became more and more humiliated for not doing well. T h e participant was thus
placed i n a goal-conflict situation where he or she could achieve a high score, but
at the cost of hurting the confederate's feelings. Participants primed with achieve-
ment stimuli found significantly more words o n the puzzle than did participants
primed with affiliation stimuli, especially o n the early trials of the task. It was
expected that the priming effect might become overwhelmed by the affiliation
demands of the experimental situation itself, because the priming manipulation
simulates a situational effect and the actual situational features themselves are a
m u c h more powerful contextual effect. A l t h o u g h this effect held as a main effect
across all participants, it was m u c h more apparent for males than females. Debrief-
ing of participants revealed no awareness of the possible influence of the priming
manipulation o n their performance.
The second experiment determined whether priming manipulations interact with
chronic motivations i n the same way they do with chronic construct accessibility i n
impression research (Bargh et al., 1988; Higgins et al., 1985). In those social perception
studies that pitted priming effects against competing chronically accessible alternatives
for the same behavior (e.g., independent vs. aloof), priming effects determined impres-
sions for a short period but after that chronic tendencies dominated.
The Interaction of Primed and Chronic Goals. In our second experiment

we again made use of the achievement versus affiliation paradigm, but also prese-
lected participants for the study based o n their chronic achievement and affiliation
motivations. Achievement motivation was measured i n the standard way using the
Thematic Apperception Test (TAT; Murray, 1943), a projective device that has
been used to measure the achievement motive for many years ( M c C l e l l a n d , 1953;
Sorrentino & Higgins, 1986). A s part of a mass testing demonstration at the
beginning of the semester, potential participants were asked to tell what was going
on i n a picture (from the standard T A T ) of a young man looking out of an open
window. T h e questions asked of participants about the picture were the standard
ones such as " W h a t is going on i n the picture?" " W h a t will happen next?" " W h a t
is the person i n the picture thinking?" and we coded answers to these questions i n
terms of achievement-related themes following the scoring key of Heckhausen
(1990). To assess affiliation motivation, we administered the Jackson (1974) Per-
sonality Research Form (PRF), which contains an affiliation subscale. Participants
who were selected for the experiment had either a high achievement motive and a
low affiliation motive, or a low achievement motive and high affiliation motive.
W i t h i n these two groups of participants, one half were primed o n achievement and
the others were primed o n affiliation.
A g a i n , as i n our first experiment, for female participants there were no significant

effects for either the priming manipulation or due to the chronic achievement-af-
filiation differences. For males, however, results replicated our first study for the
early trials: Participants whose achievement goal had been primed performed at a
reliably higher level than did the other participants on the word search task. O n
the later trials, for males again, the temporary goal priming wore off, and now the
participant's chronic motivational tendencies took over. O n the later trials, for
males, chronically achievement-motivated participants scored higher than did the
chronically affiliation motivated participants.
W h y were the results stronger for males and weaker or nonexistent for females
in these studies? Actually, the sex difference we obtained is identical to that i n most
previous studies of achievement motivation effects using the T A T as a selection
device (see Horner, 1974, for a review). It is so common not to find these effects for
women, for instance, that even when the same data are available for female
participants they are not even analyzed (Reumen, A l w i n , & Veroff, 1984). Thus,
the lack of achievement motivation priming effects for females actually replicates
the classic achievement motivation literature, and strengthens our belief that we
are priming achievement motivation with our manipulation.
A second potential reason why there was no difference i n the achievement and
affiliation conditions for women, either i n primed or chronic form, could be that
women are socialized to fulfill their achievement motives through affiliation if that
is possible i n the situation (Higgins, 1991). Thus, our female participants may have
not experienced the achievement versus affiliation goal conflict we created as did
the male participants, attaining both goals by focusing o n the feelings of the
confederate.
In any case, as i n the Cialdini (1994) and C h a i k e n et al. (1996) research on
primed and chronic information-processing goals, the nonconscious effects of
primed achievement and affiliative goals also exist i n the real world i n chronic form.
This is important because priming is used as a stand-in within the auto-motive
38 Bargh
model for chronic motivational tendencies. The finding that primed and chronic
achievement and affiliation motives interact over time i n the same way as primed
and chronic trait constructs is crucial because it demonstrates that we are activating
nonconsciously with our achievement and affiliation priming manipulations the
same underlying variable as chronic individual differences i n achievement and
affiliation motivation (see Bargh et al., 1988). Thus, our results support the
auto-motive postulate that chronic motivational states can be triggered noncon-
sciously and then operate to affect behavior, i n this case, actual performance o n a
word search task.
Dissociation Evidence: Motivational Qualities

of Primed Goal States
We are claiming that the achievement and affiliation primes are activating motivational
states. W h y do we contend that the stereotype priming effect is due to a perception-be-
havior pathway, and the achievement priming effect is due to a motivation-behavior
pathway? A s both effects are produced by trait construct priming manipulations (in the
case of stereotype activation, the trait construct is primed indirectly via its stereotype
membership), why are different explanations invoked?
First of all, it is difficult to see how the same mechanism could have produced
the achievement priming effects and also the stereotype-behavior effects. Assume
to begin with that both effects are due to goal activation and not to perceptual
activation. If so, then what is the goal or motive i n the elderly stereotype study just
described (Bargh, C h e n , et al. 1996), i n which priming the elderly stereotype caused
participants to walk more slowly down the hall following the experiment? It is hard
to see how priming the elderly stereotype produces a motive i n the participant to
walk more slowly. It is similarly difficult to understand the A f r i c a n - A m e r i c a n study
results i n terms of a motive, although one could argue that there is a motivation to
act i n a hostile manner towards African-Americans as part of that stereotype. Still,
we know of no evidence or prior conjecture of such an automatic motivational
component to that (or any other) stereotype.
Alternatively, assume that both effects are due to automatic perceptual and not
motivational activation. That is, it could be argued that what was activated i n the
achievement-affiliation studies was not an achievement or affiliation motive but
the perceptual trait construct of achievement or affiliation, and this passive "cha-
meleon effect" is what caused the participants' tendency to score higher on the word
search task (or not). This seems more plausible. However, the fact that we attain
the same quality of effect with our achievement priming manipulation as with the
classic T A T measure of achievement motivation argues that we did indeed prime
a motivational state. A n d it is hard to reconcile the sex difference we obtained with
a purely perceptual account, because it is highly unlikely that men but not women
possess the perceptual trait construct of achievement. A s noted, a motivational
account has no difficulty with the sex difference, as achievement motivation effects
using the T A T as a selection device historically have been obtained for men but
not for women.
Still, i n order to make a stronger case that motives and not perceptual structures
are responsible for these behavioral effects, we conducted additional studies to test
for the presence of qualities associated with motivational states—qualities that are
not predicted by any purely cognitive account of our findings. These qualities are
(a) persistence o n a task i n the face of interruptions or obstacles (Lewin, 1926;
Ovsiankina, 1928; see also Heckhausen, 1990; W i c k l u n d & Gollwitzer, 1982), and
(b) an increase i n motivational tendency over time (Atkinson & Birch, 1970), as
opposed to the decrease i n activation strength over time predicted by all cognitive
accounts of priming (e.g., Higgins et al., 1985). 4
Persistence in the Face of Obstacles. Bargh, Gollwitzer, L e e - C h a i , and

Barndollar (1997) showed that achievement-primed participants show greater
persistence o n a task i n the face of an obstacle than do neutral-primed participants.
Some participants were primed with achievement-related stimuli, and the remain-
ing participants with neutral stimuli. Participants participated three at a time, with
partitions between their desk chairs so that they could not see each other. However,
all three participants faced the front of the room, where a hidden video camera
recorded them during the experimental session. After completing the priming task,
under the instructions that it was a separate "language ability" measure, participants
were given a rack of Scrabble letter tiles and told to find as many words with those
letters as they could i n the next 3 minutes, and write each down o n the piece of
paper provided. The experimenter then explained that she had to leave the room
to r u n another experiment, but that if she could not get back by the end of the 3
minutes, she would give the signal to "stop"over the room's intercom.
Participants were then told to begin, and the experimenter left the room. A t the
end of the 3 minutes, participants were told to stop. The dependent measure was
the proportion of participants who continued to work o n finding the words after
the signal to stop was given, as monitored by the experimenter via the hidden video
camera. The results were as predicted: 55% of the participants i n the achievement
priming condition persisted i n the task after being told to stop, whereas only 22%
of those i n the no priming condition did so.
Male and female participants alike showed this achievement priming effect. T h i s
supports the explanation for the earlier lack of effect for female participants i n terms
of the presence of affiliation opportunities i n the situation (see Higgins, 1991). In
4
A distinction needs to be made between the strength of a priming effect per se and the relative
influence over time of a decision or judgement that has been influenced by priming. I am referring to the
former, to the relative potential strength of a priming effect as time passes prior to its influences on
responses to the environment. Wyer and Srull (1989) documented (e.g., Srull & Wyer, 1980, 1983) that
the relative effect of a trait judgment that was influenced by priming may increase over time as the other
possible sources of influence (i.e., the behavioral information itself) are cleared from working memory.
There is a difference between predicting an increase over time in a primed constructs potential effect
prior to use, and an increase in the relative impact of a judgment influenced by a priming manipulation.
N o cognitive model of priming, spreading-activation (e.g., Higgins et al., 1985) or otherwise (e.g., Wyer
& Srull's, 1989, bin model) predicts an increase with time in the eventual effect of a priming event.
40 Bargh
the Bargh et al. (1997) paradigm, affiliation was not an option, and so it could not
be used as a route for women to express their achievement goal. Consequently,
achievement priming influenced their behavior i n the face of the stop-signal obstacle.
Increase in Motivational Tendency Over Time. In a second experiment,

another motivational quality was assessed: whether the priming effect increased over
time, as Atkinson and Birch (1970) argued was true of unfulfilled motivational states.
A purely cognitive explanation cannot predict an increase i n the priming effect itself,
as all accounts of perceptual and cognitive activation predict a decrease or decay i n
activation following the priming event (e.g., Higgins et al., 1985). In Wyer and Srull's
(1989) bin model, as time passes the probability increases that other relevant constructs
will displace the primed one on top of the bin (i.e., as the most accessible for use). Even
then, the bin model does not predict an increase i n the primed constructs accessibility
as there is nowhere to go but down from the top. Thus, it should be possible to
demonstrate this hypothesized dissociation i n the effect over time of the identical
priming manipulation on an impression formation versus a behavioral task.
Participants first performed a matrix word search task i n w h i c h achievement-re-

lated or neutral priming stimuli were presented. Next, one half of the participants
in each priming condition experienced a 5-minute delay before the dependent
measure was assessed, whereas the other half did not. Delay condition participants
drew their family tree i n as much detail as they could (this task did not satisfy any
primed achievement motive). Next, some participants read about a target person
who behaved i n an ambiguously achievement-oriented way (e.g., he had not
studied all semester and stayed up all night before the test to cram for it) and then
rated the target o n achievement-related trait dimensions, whereas others found as
many words as they could i n a set of Scrabble letter tiles.
T h e impression task results replicated previous findings: W i t h no postpriming
delay, achievement-primed participants rated the target person as more achieving
and striving than did participants i n the neutral priming condition. After a 5-minute
delay this difference disappeared, indicating that the priming effect o n perceptual
interpretation had dissipated. T h e behavioral task results showed exactly the
opposite effect. T h a t is, the achievement-primed participants performed better than
the neutral-primed participants when there was no delay, replicating the Bargh,
Gollwitzer, et al. (1996) findings, but this effect increased over time. The perform-
ance level of neutral condition participants remained the same i n the no-delay and
the delay conditions, but that of achievement-primed participants significantly
increased as a function of postpriming delay.
These results show a clear dissociation between the behavioral and judgmental
effects of priming over time, i n that the direction of the effect of delay is reversed
between the two dependent measures ( D u n n & Kirsner, 1988). T h e increase i n
performance as a function of achievement priming cannot be explained except by
recourse to motivational qualities of the primed state. O u r obtained effect of
achievement priming on behavior, i n other words, cannot be merely an effect of the
activation level of a perceptual representation.
O n e additional point to be made i n the wake of these results is that it is a goal

or strategy that is clearly being activated by our priming manipulation, and not a
specific behavioral tendency. If we were just priming a specific behavioral tendency,
as i n a simple S - R connection, it would be enacted right away. Instead, the activated
goal follows the principle of applicability (Higgins, 1989): A n accessible repre-
sentation does not operate o n its own, i n the absence of relevant input, but only i n
the presence of environmental information for which it is applicable. Notably, A c h
(1935), an early theorist of the will, defined intentional states i n a similar way.
A c c o r d i n g to A c h , it is usually not the case that one begins acting immediately on
the activation of a motivational tendency. Rather, one waits for the opportune
moment i n time; the occurrence of situational events that give one the chance to
attain the goal (see also Vera & Simon, 1993).
Automatic Goal Effects are Independent

of Current Conscious Purposes
A n o t h e r objection might be raised to our conclusion that automatically activated

goal states operated nonconsciously i n these studies. It might be that although goal
states were indeed primed, this merely made them more accessible to conscious
choice processes. A l t h o u g h participants were not aware of the source of this
accessibility (i.e., the relation between the priming event and the subsequent tasks),
they still could have consciously chosen to achieve, or affiliate, when put into the
task situation.
T h e evidence discussed thus far argues against this possibility, however. First of
all, our o w n debriefing of participants gave us no indication that they had
consciously chosen their various behavioral strategies. T h e Bargh et al. (1997)
experiments, w h i c h revealed increasing effects of primed goal states over time,
speak against the role of conscious choice as well. In these studies, the dependent
measure was not the choice of behavior among possible alternatives, as i n the
previous studies, but the presence of heightened goal desire and increasing effort
over time. It is difficult to see how these effects are somehow a matter of deliberate
choice.
It is nonetheless important to test this final and key proposal of the auto-motive
model: once goals are activated, they operate on any relevant input without
conscious intent or guidance. A s stated earlier, it does not matter for goal operation
whether the activating event was conscious and intentional or not, just like it does
not matter for construct accessibility effects i n social perception whether the
activating event was conscious and intentional or passive and unintentional (Bargh,
1992; Higgins, 1989). O n c e activated and set i n motion, goal representations
theoretically operate o n any relevant input, even those the person does not mean
to process i n that way. W i t h this final postulate, the auto-motive model is complete
i n specifying that the entire chain of events from environmental stimulus to goal
operation can occur nonconsciously, without the person needing to intend or be
aware of having that goal or pursuing it (see Fig. 1.2).
42 Bargh
1. Goals are mental representations

2. M e n t a l representations (e.g., stereotypes) are capable of becoming activated
preconsciously
3. .. Goals can be activated preconsciously
4. O n c e activated they operate outside of awareness to guide information
processing and behavior
5. T h e entire sequence from environmental event to cognitive process execu-
tion or behavior enactment is nonconscious
6. Automatically actived goals display qualities of motivational states.
FIG. 1.2. The Auto-Motive Model (summarized from Bargh, 1990).
T h e standard method for demonstrating that an effect is not due to conscious

intent is to show that it differs from the effect that would occur when that
unconscious influence is not operating (Jacoby, 1991; see review i n Bargh &
Barndollar, 1996). Therefore, i n order to show that activated goals operate i n the
absence of conscious guidance, we assessed whether an activated goal could
produce effects opposite to those found when participants intend it to be operative.
Unintended Operation of Intentional Processing Goals. Bargh and Green

(1996) presented participants with a videotape of a conversation between two men,
with participants told beforehand either that the tape concerned (a) a job interview
for a restaurant waiter position, (b) a job interview for an investigative crime
reporter position o n a city newspaper, or (c) a conversation between two acquain-
tances who had not seen each other for some time (this "acquaintance" condition
was intended as a control i n which no explicit evaluative goal was given). Pretesting
had shown that the qualities of a good reporter (e.g., tough, aggressive, dominant)
were believed to be the opposite of those that would make a good waiter (e.g.,
friendly, acquiescent). A l l participants saw the same conversation, which was
scripted to be ambiguous enough to fit any of the three cover stories.
Halfway into the conversation, a third male ("Mike") knocked, entered the
doorway of the room, and asked the interviewer (who had his back to the camera)
whether he was ready for lunch. T h e interviewer said he was sorry but he was too
busy at the moment to go to lunch, and maybe later or another time. A t this point,
the critical experimental manipulation occurred: In one condition, M i k e became
irritated and told the interviewer that he was also very busy that day and could not
wait. W h e n the interviewer persisted that he could not leave right then, M i k e said
that he could not wait, they would have to make it another time, and shut the door
hard behind h i m . In the other condition, M i k e became very apologetic for inter-
rupting and quite calmly said he would wait outside.
Immediately after the tape had finished, we informed participants that we were
actually interested i n their opinion of M i k e , the person who interrupted about the
lunch date, and asked participants to rate Mike's likability. We hypothesized that
even though our participants had no conscious intention to evaluate M i k e , as their
attention was focused o n the conversation between the other two men, they would
nonetheless do so i n line with the goal that was currently operating based o n the
experimental instructions. In other words, if they were evaluating the target person
in terms of his qualifications for being a waiter, they would evaluate M i k e using the
same processing goal without knowing it, and likewise if they were evaluating the
target person for a crime reporter position. In each case, their evaluation of M i k e
would be more positive if his behavior fit the qualities that were valued for that job
description, and more negative if his behavior did not fit those qualities. In the
control condition, i n which participants were not given the waiter or the reporter
processing goal, evaluations of M i k e should be i n line with how one would evaluate
another person i n general, based o n his or her behavior.
Specifically, we expected participants i n the control condition to like "polite
M i k e " more than "surly M i k e . " We expected this difference to be even more
pronounced i n the waiter-goal condition, given the value placed o n deference and
docility i n a waiter. A n d , our major prediction was that participants i n the reporter
condition would like surly M i k e better than polite M i k e because surly M i k e was a
better fit to the position of a crime reporter.
A s expected, participants i n the control condition did like the polite version of
M i k e better than the surly version. A l s o as predicted, this difference was stronger
i n the waiter-goal condition. Most importantly, participants i n the reporter-goal
condition, who were considering the interviewee (not Mike) for the crime reporter
position liked surly M i k e better than polite M i k e — e v e n though, judging from the
control condition results, those same participants would have formed the com-
pletely opposite evaluations had they not been assessing an entirely separate
individual for a crime reporter job.
Auxiliary trait ratings of M i k e showed that these effects o n liking were not due
to participants categorizing Mike's behavior differently based o n their particular
processing goal. For instance, i n the reporter condition, participants rated surly
M i k e just as stubborn and rude and disagreeable as did the other participants, and
waiter condition participants rated polite M i k e just as unadventurous and passive
as did the reporter participants. In other words, reporter-condition participants
liked surly M i k e better despite having accurately perceived h i m as behaving badly.
A n d if they had not been thinking about a third party's suitability for a particular
line of employment, their liking ratings of M i k e would have been very different.
A g a i n , these results are predicted by the auto-motive model. W h e n a goal is
operating, it operates on any and all available information for w h i c h it is applicable,
regardless of whether that is the source of information (e.g., person) the individual
intends it to process. Activated processing goals, i n other words, operate o n their
own, autonomously. Judgments are made as a result that are clearly counter to what
the individual would make if he or she intended to process that source of informa-
tion; for instance, our pretest participants who focused their attention o n M i k e
instead of the other two actors i n the tape clearly disliked surly M i k e and liked polite
M i k e . Real-world versions of this effect are not difficult to imagine. For example, a
person who works all day i n an environment that values certain traits (e.g., an
aggressive, competitive atmosphere) might well become attracted to a coworker
44 Bargh
because he or she possesses those qualities, with potentially disastrous results—be-

cause if you asked h i m or her off the job about the ideal mate, you would very likely
get a different description.
Simulating Actual Environment-Goal Links

In the research described thus far, the assumption was made that the experimental
situations correspond to real-world counterparts—specifically, to social environ-
ments containing features that are chronically associated with cognitive or behav-
ioral goals. However, these experimental manipulations actually activated the goal
i n question through presentation of stimuli directly relevant to the goal itself: words
like impression and evaluate for the impression goal, and succeed and strive for the
achievement goal, and so forth. A l t h o u g h these stimuli were successful i n activating
the corresponding goal concept and producing goal-directed behavior, they did not
correspond to situational features. W h a t is needed is a demonstration that environ-
mental stimulus features that are semantically unrelated to the goal i n question are
capable of automatically activating that goal.
Previous priming research uniformly relied on stimuli that are synonymous with
or directly relevant to the mental representation they are intended to activate:
words synonymous with kind to activate the concept of kindness (e.g., Srull & Wyer,
1979), or the word furniture to activate the names of kinds of furniture (Neely,
1977), and so o n . Early tests of spreading activation theory i n the 1970s (e.g., Collins
& Loftus, 1975; Lorch, 1982; Meyer & Schvaneveldt, 1971) did involve stimuli
that were not synonyms but were still close semantic associates (e.g., s u n - m o o n ,
doctor-nurse). But the auto-motive model posits that features of environments will
activate goals associated with them, and these features and goals need not share
any semantic features whatsoever. Thus, the situation party could activate the goal
withdraw, be unobtrusive i n an individual who has pursued that social strategy within
parties habitually i n the past (see Bargh, 1990). T h e environmental feature school
could activate the goal achieve or the goal affiliate depending o n w h i c h chronic goal
the individual possesses i n that situation; but there are no semantic features i n
common.
Power as a Situational Feature Linked to Goals. In considering ways of

testing this aspect of the model, we considered what kinds of environmental features
were likely to become associated with goal-states. A n d there is one such feature
that, more than any other, is associated with goals: situations i n w h i c h one has
power. By definition, power i n a situation is the ability to attain your o w n personal
goals (Cartwright, 1959; Russell, 1938; Thibaut & K e l l e y , 1959). Thus, the concept
of power is a likely candidate to become automatically linked with the individual's
goals, especially those he or she pursues when i n a position of relative power within
a given situation.
O n e social problem that is noteworthy for the role that situational power plays
i n it is sexual harassment, and, i n a different sense of "power," sexual aggression (see
Bargh & Raymond, 1995; Bargh, Raymond, Pryor, & Strack, 1995; Brewer, 1982;
Pryor, 1987; see also Kipnis, 1976). Frequently, cases of sexual harassment involve
power differentials such that the (almost always male) perpetrator has some form
of power over the important outcomes of the (almost always female) victim (see
Brewer, 1982; Fitzgerald, 1993), and uses that to coerce her into granting sexual
favors. W h a t made this issue especially intriguing as a potential application of the
auto-motive model is that i n the majority of cases, perpetrators do not realize or
understand that their behavior is harassment (Fitzgerald, 1993)—something the
Bob Packwood diaries illustrated all too clearly.
H o w could this be? Brewer (1982) cogently applied the actor-observer attribution
difference to this situation (Jones & Nisbett, 1971; see also Kipnis, 1976, o n the role
played by actor-observer perceptual differences i n the abuse of power more gener-
ally), noting that the relatively powerful perpetrator does not perceive his own power
within the situation. Rather, what he sees is the subordinate's friendliness, agreeabil-
ity, passivity, and so on. The subordinate, on the other hand, is well aware of the
power position of the boss and of his control over her outcomes. Thus, the boss may
attribute his behavior to those situational features (the smiling, agreeable subordi-
nate), whereas the subordinate may attribute it to features that are salient to her (the
boss and the implied threat to her if she does not go along with him).
T h e auto-motive model can be applied to this situation. Those who sexually
harass and aggress do so at least i n part because of an automatic association between
the concept of power and the goal of sexuality (Bargh et al., 1995). T h a t is, the goal
of sex is automatically associated with mental representations of situations i n which
the individual has power. If power features of the situation activate the sexuality
goal automatically, this goal will operate outside awareness to guide behavior, and
the individual will not be aware of this influence (i.e., the role that his relative power
played i n his behavior toward the woman). Rather, he will attribute his behavior to
those features of the situation he is aware of (her smile or compliments or deference;
Kipnis, 1976) and his activated sexuality goal may well cause h i m to interpret those
features i n sexualized ways (e.g., she is flirting with me; she is attracted to me).
The Automatic Power-Sex Association in Sexual Harassers. A key predic-

tion then is that men who are likely to sexually harass or aggress (or both) should
show evidence of this automatic link between power and sex, whereas men who are
not likely to harass or aggress should not. In addition, it should then be possible to
prime the goal of sexuality with stimuli semantically unrelated to sexuality, but
synonymous with the situational feature presumed to be tied to that goal—namely,
power related stimuli. M e n who are likely to sexually harass or aggress, therefore,
should show evidence of having their sexuality goal primed when they are primed
with power related stimuli—for example, by finding a woman more attractive than
otherwise; whereas other men should show no effect of power priming on their
attraction toward the woman.
T h e results of two experiments reported by Bargh et al. (1995) confirmed these
predictions. Participants i n both studies were preselected based on their responses
46 Bargh
to Pryor's (1987) Likelihood to Sexually Harass (LSH) scale and Malamuth's (1989)
Attractiveness of Sexual Aggression ( A S A ) scale. T h e L S H presents participants
with 10 scenarios i n which a male protagonist has some form of leverage over an
attractive woman, such as catching her taking money from the cash register where
they both work. For each scenario, participants are asked to give the probability
that they would propose not using that leverage i n return for sexual favors, if they
were sure that nothing bad would happen to them as a consequence. T h e A S A asks
participants to indicate how arousing and attractive are each of a wide variety of
sexual practices. T h e key items for our purposes were rape and otherwise using force
to have sex with a woman. Participants who either scored i n the highest or the
lowest quartiles o n these scales participated i n our studies.
In Study 1, participants pronounced a series of words as quickly as they could.
This pronunciation task was demonstrated to be a sensitive measure of automatic
mental associations (Balota & Lorch, 1986; Bargh, Chaiken, et al., 1996). O n each
trial, prior to the presentation of the target word to be pronounced, a prime word
appeared very briefly (90 msec), at a randomized location o n the screen that was
outside of the participant's foveal (roughly, conscious; see Bargh et al., 1986)
processing area, and was immediately masked by a string of letters. These proce-
dural steps combined to ensure that the prime words were presented subliminally
and that participants were not even aware that words were being presented at all.
Phenomenally what they experienced were flashes of light.
Primes and targets were related to the concepts of either power or sex or neither (the
control stimuli; the sex-related stimuli were only ambiguously related—such as bed and
motel—because of the likely distorting effect of embarrassment or surprise o n pronun-
ciation latencies for directly related words such as intercourse or sex). Thus, we could
assess the effect of power related primes versus neutral primes o n the speed of
pronouncing both sexually related stimuli and power related targets. Participants who
scored highly o n the L S H or A S A were significantly faster to pronounce the sexuality
related targets that were preceded by power related primes compared to control primes.
Thus, the results showed that there indeed was an automatic link between the concepts
of power and sex for these subjects, but not for others.
A second experiment of Bargh et al. (1995) tested whether the presence of power
cues i n a situation would automatically activate the goal of sexuality, causing the
operation of that goal within an interpersonal situation. This should also be true
only for those participants for whom the automatic link exists between power and
sex. Participants took part i n the experiment individually, along with a female
confederate posing as another participant. In what was purported to be an unrelated
first experiment o n language ability, both participant and confederate completed a
16-item word-fragment completion task. For one half of the participants, 6 of the
items were related to power (e.g., str—g, out-ri-y), and for the remaining partici-
pants none of the items contained power related words.
Next, participant and confederate worked separately, but at adjacent tables, o n
a task allegedly to do with understanding visual illusions. Standard visual illusions
were projected o n a wall, and i n each case, the participant and confederate were
asked to give an explanation of why the illusion occurred. Finally, the participant
and confederate were shown into separate rooms, and the participant was informed
that the experiment was actually about impression formation, specifically the kinds
of impressions people formed of those with whom they had only a minimal
interaction, such as between himself and the "other participant." H e was asked to
complete a questionnaire concerning his impression of her, being led to understand
she was doing the same concerning him i n the other room.
This questionnaire contained two key items, concerning how attractive the
participant found the confederate, and also his desire for future contact with her.
A s predicted, participants likely to sexually aggress found the confederate to be
more attractive when their concept of power had been primed than when it had
not been; the power priming manipulation had no effect on participants who were
not likely to sexually aggress. In short, men with a tendency or proclivity to sexually
aggress against women found the identical woman more attractive when their
concept of power had been surreptitiously activated than when it had not been. To
generalize to the workplace, the boss or supervisor who finds his subordinate
attractive might well not find her so if he had met her outside of the office, o n an
equal power footing.
There are obvious practical implications of automatic power-goal associations
for sexual harassment and aggression, and the misuse of power i n general (see
Bargh & Raymond, 1995; Kipnis, 1976), but these findings are of theoretical
import as well. They show that perceptions as well as behavior (see Pryor, 1987)
are indeed triggered nonconsciously by environmental features, and that i n d i v i d -
ual differences corresponding to chronic feature-goal associations do exist and
result i n different reactions to the same situation. These are important findings for
priming research i n general because they move priming effects out of the direct
activation of the mental representation by synonymous stimuli onto a level of
representation closer to the outside world. In other words, representations of
situations activated directly by relevant features are directly connected to second-
level representations of goals, so that the perception of the feature preconsciously
activates the goal.
Summary
These studies have several implications. First, behavioral and cognitive goals can
be directly activated by the environment without conscious choice or awareness of
the activation. Second, the goals, once activated, direct information-processing and
social behavior. T h i r d , the states activated by the priming manipulations i n these
studies have motivational qualities. Fourth, these states also exist i n chronic form
and there are individual differences i n these chronic motivations. Finally, the
activated goals operate autonomously, bypassing the need for any conscious selec-
tion or choice, but producing outcomes different from those that would occur if the
individual would choose if the goal were not primed. In short, every postulate of
the auto-motivation model (Bargh, 1990) was supported by these studies, demon-
strating that the entire sequence from environmental information to goal and
48 Bargh
motivation to judgment and action can and does occur automatically and uncon-
sciously.
MOVING FROM SOCIAL COGNITION TO SOCIAL IGNITION
T h e study of automaticity progressed dramatically i n the 1990s. N o longer are

researchers content to confine themselves to perceptual or judgmental phenomena;
not once was it recognized that everything psychological was fair game—that
anything could be primed.
What Have We Been Priming All These Years?

It is easy to fall into the trap of thinking that the only effect an experimental
manipulation is having is the one that is being measured. T h e recent evidence of
automatic evaluation, motivation, and behavior shows that i n about 1975, social
cognition labs began priming not just what they thought they were—perceptual
trait constructs—but behavioral tendencies and motivations and evaluations. A l l
of these systems, according to the present proposal, are engaged immediately and
i n parallel by an environmental event. We may choose one dependent variable to
catch certain effects of our priming manipulation, but this does not mean the other
effects are not occurring.
This should have been clear as long ago as the Carver et al. (1983) study, w h i c h
used the same hostile priming manipulation as did Srull and Wyer (1979) but
instead of influencing impression formation with it, influenced participants' own
behavior. T h a t was the clue that both the perceptual and the behavioral effects were
primed i n parallel—in fact, Carver et al. (1983) proposed the idea of the behavioral
schema to account for these simultaneous effects. In my view, we are only now
catching up with the implications of that study.
Associations Between Systems

Because the three types of effects described are occurring i n parallel, we need to
learn more about how they influence each other. It would be surprising if these
different reactions were not highly interassociated with each other. By the basic
principle of contiguous activation (Hebb, 1948), all of these reactions occurring i n
parallel should be richly intertwined. Thus the activation of the internal repre-
sentation of a social situation by those features i n the environment should set i n
motion immediate perceptual, affective, and behavioral responses, to the extent
those were regularly enacted i n the past.
Several forms of such interrelations were discussed: Perception is linked strongly
to behavioral tendencies and evaluation to behavior via approach-avoidance
motivation. Goals are linked to perception and to evaluation as well as to behavior:
Nonconscious activation of cognitive processing goals affects person perception i n

the Chartrand and Bargh (1996) studies, nonconscious activation of consistency
motivation affects evaluative processes such as attitude change (Bator & C i a l d i n i ,
1995), and nonconscious activation of impression management, defense, or accu-
racy motivation affects evaluation of persuasive messages (Chaiken et al., 1996).
Previous theorists (e.g., Fazio, C h e n , M c D o n e l , & Sherman, 1982; Fiske &
Pavelchak, 1986; Strauman & Higgins, 1987) described mechanisms by w h i c h the
act of perception can have automatic affective or evaluative consequences (see
review i n Spielman, Pratto, & Bargh, 1988): if the evaluation is stored w i t h i n the
perceptual category (Fiske & Pavelchak, 1986) or so strongly associated with the
object representation that it is activated i n the course of perceiving the object
(Fazio et al., 1982), or the affect is generated by a discrepancy between the trait
concept activated i n perception and the individual's stored standards for his or
her behavior (Strauman & Higgins, 1987). T h a t is another interconnection
between processing systems: Automatic perceptual processes have automatic
evaluative consequences.
Dissociations Between Systems
The claim is that these three preconscious processing modules are richly intercon-
nected, but at the same time they have different internal operating structures and
rules, so they are different, too. W h y is it necessary to propose separate, parallel
modes of preconscious processing of social information?
Because across the board of our proposed lines of research—evaluation,
perception, and a c t i o n — n o one general cognitive model can account for all of
our obtained results. Existing spreading activation models of semantic memory
cannot account for the pervasive and strong evaluative priming effect, w h i c h
occurs based o n the sharing of a single, c o m m o n feature (see Bargh, C h a i k e n ,
et al., 1996); or why the effect is stronger and more pervasive w h e n the role that
strategic cognitive processes play i n the paradigm is reduced. N o purely cogni-
tive m o d e l of priming effects predicts an increase i n strength of the achievement
goal priming effect over time, as the Bargh et al. (1997) experiment found for
the behavioral—but not the perceptual—task. Likewise, passive effects of
perception o n behavior, especially the elderly stereotype effects found by Bargh,
C h e n , et al. (1996), are difficult to explain i n terms of automatic m o t i v a t i o n .
A n d social-perceptual effects of priming o n impression formation are content-
specific and not globally evaluative or affective i n nature. If a positive or
negative trait construct is primed that is not applicable to the ambiguous target
behavior, there is no priming effect—a finding of the very first priming study
(Higgins et al., 1977) and replicated consistently thereafter (see Bargh et al.,
1986; Erdley & D ' A g o s t i n o , 1988; Higgins, 1989). T h u s , trait construct p r i m -
ing effects appear to be due to the perceptual system as they c a n n o t be
a c c o u n t e d for by the evaluative (immediate and global good vs. bad classifi-
cation) system.
50 Bargh
THE ROLE OF CONSCIOUSNESS IN AN AUTOMATIC WORLD
Automaticity pervades everyday life, playing an important role i n creating the

psychological situation from which subjective experience and subsequent conscious
and intentional processes originate. O u r perceptions, evaluations, and the goals we
pursue can and do come under environmental control. Because these perceptual
interpretations, likes and dislikes, and reasons for our behavior are not consciously
experienced, we make sense of them i n terms of those aspects of which we are
consciously aware, and our theories as to what would have caused us to feel or act
that way (Karniol & Ross, 1996; Nisbett & Wilson, 1977). For example, the sexual
harasser (with the automatic power-sex mental association) attributes his feelings
of attraction towards a woman to her physical features or her friendly (perceived as
flirtatious) behavior or both (Bargh et al., 1995). O f course, as this choice of
example illustrates, our understandings of what cause us to think, feel, and do are
i n large part after-the-fact rationalizations (Gazzaniga, 1985; Steele, 1988). A s
Gazzaniga argued, consciousness may exist i n order for us to make sense and a
coherent pattern out of all of it, so that one feels a sense of stability and control—a
quite adaptive feeling to have, judging by the consequences when it is not present
(e.g., Abramson, Seligman, Teasdale, & D'Agostino, 1985; Taylor, 1989).
I emphatically push the point that automatic, nonconscious processes pervade
all aspects of mental and social life, i n order to overcome what I consider dominant,
even implicit, assumptions to the contrary. But i n making the case, pace Skinner,
that even goal-directed, complex social behavior need not require conscious
cognitive choice processes, something must be said about the conditions under
w h i c h nonconscious control is believed to occur, and exactly how unnecessary I am
claiming consciousness to be.
The Assumption of Conscious Mediation:

A Legacy of the Serial Stage Model
Let us consider what has changed i n psychology since the 1960s, concomitantly
with the decreased role of conscious choice. The most fundamental change was a
movement away from serial stage of cognition based o n the computer metaphor
(e.g., A t k i n s o n & Shiffrin, 1968; Newell & Simon, 1972) to models i n w h i c h many
mental operations are carried out simultaneously, i n parallel (e.g., Hintzman, 1988;
Rumelhart & M c C l e l l a n d , 1986). In my opinion, it was the serial stage model i n
which conscious judgment and reasoning processes were assumed to follow percep-
tion and precede responses to the environment that caused us to overestimate the
mediational role of conscious processes. This meta-assumption put conscious
recognition and reasoning processes as a causally prior stage, almost as a roadblock
i n the way of affective reactions and behavioral responses.
Early cognitive models, i n other words, equated cognition with conscious cog-
nition (see Bowers, 1981; Lazarus, 1982), and we have been cleaning up after this
misconception ever since. It led to the assumption that conscious recognition was
a necessary precondition for affective reactions. Erdelyi (1974) showed that the
reason why perceptual defense findings as a concept ultimately failed to persuade
most psychologists i n the 1950s and 1960s was that no one could get around the
notion that the stimulus had to 'be perceived before it was perceived'; that i n order
for it to be defended against and shut out of consciousness, it had to first be
perceived to be k n o w n to be something to be defended against. It was implicitly
assumed that perceptual registration had to be conscious, so it was impossible to
understand how something could be consciously perceived before it had been
consciously perceived. Erdelyi almost single-handedly restored the good name of
the N e w Look by amassing conceptual and empirical objections to this assumption.
Zajonc's (1980) argument that affective reactions could be immediate and
independent of "cognitive" (i.e., conscious) information processing was counterin-
tuitive only because of the implicit belief i n the serial stage model. If different
psychological functions can operate on input at the same time, the hypothesis of
immediate affective reactions prior to or i n the absence of conscious recognition of
the stimulus appears much more plausible.
M y own implicit adherence to the stage model nearly led me to conclude that
the extent of direct automatic influences of the environment o n social cognition
was limited to perceptual interpretation and did not extend to making judgments
or behavioral decisions or other responses to the environment (Bargh, 1989, 1990).
T h e assumption I held was that these judgments and decisions had to precede and
determine any intentions the individual formed and any behavior he or she enacted.
It was only by playing devil's advocate as to how the direct effect of the environment
could possibly breach this apparent asymptote at the judgment and decision (i.e.,
goal-setting) stage that the hypothesis of automatic goal activation was formed
(Bargh, 1990). A g a i n , it was the metaview of serial processing stages that made the
notion that motivations could be directly activated by the current environmental
information difficult for me to see.
In parallel models such as the present one, there is no theoretical, a priori
requirement for conscious processes to mediate the perceptual, evaluative, or
behavioral effect, as there was i n the serial stage models of the 1960s that still
pervade, implicitly or explicitly, social cognition today. This is despite the fact that
since the 1960s, the research evidence has caused the explanatory power of
conscious mediational processes to dwindle dramatically. A s noted earlier, whereas
attributional models once posited sophisticated, "analysis of variance" reasoning
processes to be the rule (e.g., Kelley, 1967), we now know that much of attributional
judgment is spontaneous, unintended, and nonconscious (e.g., Gilbert, 1989;
N e w m a n & Uleman, 1989; Taylor & Fiske, 1978; U l e m a n et al., 1995). Whereas
evaluative judgments were once thought to be computed consciously based on a
consideration of recognized stimulus features (e.g., Anderson, 1974), Zajonc (1980)
argued, and research verified (e.g., Bargh, Chaiken, et al., 1996; L e D o u x , 1989;
M u r p h y & Zajonc, 1993; Niedenthal, 1990), that affective reactions can be prior
to, more immediate, and independent of even the most basic conscious processes
such as recognition of the stimulus. A n d now, as the research reviewed demon-
52 Bargh
strated, even intentions and goals, and the cognitions and behaviors that are carried
out i n pursuit of those goals, can become automated and bypass conscious choice
and guidance.
Is Consciousness Riding Into the Sunset?
In removing consciousness from its privileged place at the mediational center of

everything, by moving from a serial stage to a parallel process metatheory, one is
not claiming that there is no role or function for conscious processing. W h e n Galileo
removed the Earth from its privileged position at the center of the universe, the
Earth still existed, just with diminished importance. W h e n Darwin removed human
beings from their privileged position among living creatures, human beings still
existed, just with a diminished sense of importance. Consciousness still exists as we
move from a serial to a parallel model of mind. In fact, by getting rid of its overstated
position i n the middle of serial models, we may end with a clearer sense of its role
and purpose.
For one thing, although conscious processing can no longer be viewed as

necessary for behavior and judgments and evaluations to be made i n a given
situation, it is of course necessary for the development of those preconscious
processes i n the first place. These had to be enacted or engaged i n effortfully and
consciously to begin with, and like any skill or mental process, only after consider-
able use could they recede into the preconscious (Vera & Simon, 1993). W i t h o u t
conscious processes to construct them, adapt them, modify them i n the face of trial
and error, and then engage i n them consistently and frequently over time, the
preconscious processes discussed i n this chapter would not exist. Moreover, as stated
at the outset, preconscious perceptual and evaluative processes provide the starting
point for conscious, subjective experience and decisions as to how to respond to
that subjective environment (Neisser, 1967). They were described as mental
servants that free up conscious capacity for nonroutine tasks.
This is less true of preconscious motivations, because the automatically activated
goal then takes control over the rest of the mind's machinery (see Wyer & Srull,
1989). But even i n the case of these automatic motivations, it is possible for a person
to become aware of his or her actions and, as i n the case of bad habits, attempt to
change those behavior patterns. This question of how automatic and conscious
motivations interact when i n conflict is one of practical as well as theoretical
importance, and we are now investigating parameters of this interaction.
But those who believe free will is not a scientific concept, and that as research
advances the contents of the black box of "conscious choice" will grow ever smaller
(e.g., Barsalou, 1992; Skinner, 1953), will likely object that these flexible and novel
conscious processes are nonetheless determined by situational and cognitive fac-
tors. W i t h enough knowledge, that is, we will be able to predict those apparently
"free" mental processes as well. Certainly the trend of research since 1980 is i n this
direction. So it may well be that there ultimately is no future role for conscious
processing i n accounts of the mind, i n the sense of free will and choice.
But there is another quality to what we call conscious processes that is unlikely
ever to be shown to be unnecessary, and that is its serial and inhibitory nature. M a n y
years ago, Lashley (1951) wrestled with the problem o f how the mind, i n w h i c h
thoughts, images, memories, and ideas were not bound to time and space, could
direct behavior i n the real world, where events happened one at a time. Kltiver
(1951), i n discussing Lashley's paper at the symposium i n which it was presented,
posed the problem quite succinctly:
As regards the relation of thinking to temporal organization, we are, it seems to me,

confronted with a certain dilemma. Ideas, concepts, and meanings themselves have no
reference to time and space, and, yet, the expression, formulation, and identification of
ideas are processes proceeding in time and occurring in space. (p. 136)
T h e difficulty is that the mind is exquisitely capable of moving around i n time,

and of doing many things at once, but the body cannot. T h e individual must live
and act i n a physical world i n which time is a dimension and in w h i c h events happen
i n order, not simultaneously. N o t coincidentally, conscious processing is serial i n
nature, with an inhibitory capability that prevents one from trying to do more than
one thing at a time (see Posner & Snyder, 1975; Shallice, 1972). T h e purpose of
consciousness, therefore, may be to connect a parallel mind to a serial world.
Ironically, then, moving from a serial to a parallel model of the m i n d may have
greatly decreased the causal importance of conscious processes i n everyday life, but
at the same time guaranteed that Skinner ultimately will lose his long argument
that consciousness is an epiphenomenon.
ACKNOWLEDGMENTS
Preparation of this chapter was supported i n part by Grant SBR-9409448 from the
National Science Foundation. Portions of the research described were presented in invited
addresses to the 1994 American Psychological Society convention in Washington, D . C . ,
and the 1995 American Psychological Association convention i n N e w York City. I
thank Bob Wyer for his insightful feedback, and Peter Gollwitzer, A d van Knippenberg
and Leonard Berkowitz for their comments and suggestions on a previous draft.
REFERENCES
Abramson, L.Y., Seligman, M . E . P., & Teasdale, J. (1978). Learned helplessness in humans: Critique
and reformulation. Journal of Abnormal Psychology, 87, 49-74.
A c h , N . (1935). Analyse des Willens. In E. Abderhalden (Ed.), Handbuch der biologishen Arbeitsmethoden
(Vol. 6, Part E). Berlin: Urban & Schwarzenberg.
Anderson, J. R. (1983). The architecture of cognition. Cambridge, M A : Harvard University Press.
Anderson, J. R. (1992). Automaticity and the A C T * theory. American Journal of Psychology, 105,
165-180.
Anderson, N . H . (1974). Cognitive algebra: Integration theory applied to social attribution. In L .
Berkowitz (Ed.), Advances in experimental social psychology (Vol. 7, pp. 1-101). New York: Academic
Press.
54 Bargh
Atkinson, J. W., & Birch, D . (1970). A dynamic theory of action. New York: Wiley.
Atkinson, R. C . , & Shifrrin, R. M . (1968). Human memory: A proposed system and its control processes.
In K . W. Spence & J. T Spence (Eds.), Advances in the psychology of learning and motivation (Vol. 2).
New York: Academic Press.
Baldwin, M . E., & Lopez, D . F. (1990). M y adviser and the Pope are watching me from the back of my
mind. Journal of Experimental Social Psychology, 26, 435-454-
Balota, D. A . , & Lorch, R. E , Jr. (1986). Depth of automatic spreading activation: Mediated priming
effects in pronunciation but not in lexical decision. Journal of Experimental Psychology: Learning,
Memory, and Cognition, 12, 336-345.
Bandura, A . (1977). Social learning theory. Englewood Cliffs, NJ: Prentice-Hall.
Bargh, J. A . (1982). Attention and automaticity in the processing of self-relevant information. Journal
of Personality and Social Psychology, 43, 425-436.
Bargh, J. A . (1989). Conditional automaticity: Varieties of automatic influence in social perception and
cognition. In J. S. Uleman & J. A . Bargh (Eds.), Unintended thought (pp. 3-51). New York: Guilford.
Bargh, J. A . (1990). Auto-motives: Preconscious determinants of thought and behavior. In E. T. Higgins
& R. M . Sorrentino (Eds.), Handbook of motivation and cognition (Vol. 2, pp. 93-130). New York:
Guilford.
Bargh, J. A . (1992). Being unaware of the stimulus versus unaware of its interpretation: W h y sublimi-
nality per se does not matter to social psychology. In R. Bornstein & T. Pittman (Eds.), Perception
without awareness (pp. 236-255). New York: Guilford.
Bargh, J. A . (1994). The four horsemen of automaticity: Awareness, intention, efficiency, and control
in social cognition. In R. S. Wyer & T . K . Srull (Eds.), Handbook of social cognition (2nd ed., pp. 1-40).
Hillsdale, NJ: Lawrence Erlbaum Associates.
Bargh, J. A . (1996). Principles of automaticity. In E. T. Higgins & A . Kruglanski (Eds.), Social psychology:
Handbook of basic principles (pp. 169-183). New York: Guilford.
Bargh, J. A . , & Barndollar, K. (1996). Automaticity in action: The unconscious as repository of chronic
goals and motives. In P M . Gollwitzer & J. A . Bargh (Eds.), The psychobgy of action (pp. 457-481).
New York: Guilford.
Bargh, J. A . , Bond, R. N . , Lombardi, W. J., & Tota, M . E. (1986). T h e additive nature of chronic and
temporary sources of construct accessibility. Journal of Personality and Social Psychology, 50, 869-878.
Bargh, J. A . , Chaiken, S., Govender, R., & Pratto, F. (1992). The generality of the automatic attitude
activation effect. Journal of Personality and Social Psychology, 62, 893-912.
Bargh, J. A . , Chaiken, S., Raymond, P, &. Hymes, C . (1996). T h e automatic evaluation effect:
Unconditionally automatic attitude activation with a pronunciation task. Journal of Experimental Social
Psychology, 32, 185-210.
Bargh, J. A . , Chen, M . , & Burrows, L. (1996). Automaticity of social behavior: Direct effects of trait
construct and stereotype activation on action. Journal of Personality and Social Psychology, 71, 230-244.
Bargh, J. A . , & Gollwitzer, P M . (1994). Environmental control of goal-directed action: Automatic and
strategic contingencies between situations and behavior. Nebraska Symposium on Motivation, 41,
71-124.
Bargh, J. A . , Gollwitzer, P M . , & Barndollar, K . (1996). Social ignition: I. The automatic activation of
motivational states. Manuscript submitted for publication, New York University.
Bargh, J. A . , Gollwitzer, P M . , Lee-Chai, A . , & Barndollar, K. (1997). Social ignition:II.The motivational
qualities of primed goal states. Manuscript in preparation, New York University.
Bargh, J. A . , & Green, M . (1996). Unintended consequences of intentional information processing. Manu-
script submitted for publication, New York University.
Bargh, J. A . , Litt, J., Pratto, F., & Spielman, L A . (1989). O n the preconscious evaluation of social
stimuli. In A . E Bennett & K. M . McConkey (Eds.), Cognition in individual and social contexts:
Proceedings of the X X V International Congress of Psychology. (Vol. 3, pp. 357-370). Amsterdam:
Elsevier/North-Holland.
Bargh, J. A . , Lombardi, W. J., & Higgins, E. T. (1988). Automaticity of Person x Situation effects on
impression formation: It's just a matter of time. Journal of Personality and Social Psychology, 55, 599-605.
Bargh, J. A . , & Pietromonaco, P. (1982). Automatic information processing and social perception: T h e
influence of trait information presented outside of conscious awareness on impression formation.
Journal of Personality and Social Psychology, 43, 437-449.
Bargh, J. A . , & Pratto, F. (1986). Individual construct accessibility and perceptual selection. Journal of
Experimental Social Psychology, 22, 293-311.
Bargh, J. A . , & Raymond, P. (1995). T h e naive misuse of power: Nonconscious sources of sexual
harassment. Journal of Social Issues, 51, 85-96.
Bargh, J. A . , Raymond, P., & Chaiken, S. (1996). The automatic evaluation effect: Does it hold for the other
major dimensions of semantic meaning as well? Unpublished manuscript, New York University.
Bargh, J. A . , Raymond, P., Pryor, J., & Strack, E. (1995). Attractiveness of the underling: A n automatic
power-sex association and its consequences for sexual harassment and aggression. Journal of Personality
and Social Psychology, 68, 768-781.
Bargh, J. A . , & Thein, R. D . (1985). Individual construct accessibility, person memory, and the
recall-judgment link: The case of information overload. Journal of Personality and Social Psychology, 49,
1129-1146.
Bargh, J. A . , & Tota, M . E. (1988). Context-dependent automatic processing in depression: Accessibility
of negative constructs with regard to self but not others. Journal of Personality and Social Psychology,
54, 925-939.
Barsalou, L. W. (1992). Cognitive psychology: An overview for cognitive scientists. Hillsdale, NJ: Lawrence
Erlbaum Associates.
Bator, R. J., & Cialdini, R. B. (1995). Priming a consistency motivation enhances cognitive dissonance effects.
Manuscript submitted for publication, Arizona State University
Bem, D. J. (1972). Self-perception theory. In L. Berkowitz (Ed.), Advances in experimental social psychology
(Vol. 6, pp. 1-62). New York: Academic Press.
Berkowitz, L. (1984). Some effects of thoughts on anti- and prosocial influences of media events: A
cognitive-neoassociation analysis. Psychological Bulletin, 95, 410-427.
Berkowitz, L., & LePage, A . (1967). Weapons as aggression-eliciting stimuli. Journal of Personality and
Social Psychology, 7, 202-207.
Bowers, K. S. (1981), Knowing more than we can say leads to saying more than we can know: O n being
implicitly informed. In D . Magnusson (Ed.), Toward a psychology of situations: An interactional
perspective. Hillsdale, NJ: Lawrence Erlbaum Associates.
Brewer, M . B. (1982). Further beyond nine to five: A n integration and future direction. Journal of Social
Issues, 38, 149-158.
Brewer, M . B. (1988). A dual process model of impression formation. In T. K. Srull & R. S. Wyer, Jr.
(Eds.), Advances in social cognition (Vol. 1, pp. 1-36). Hillsdale, NJ: Lawrence Erlbaum Associates.
Brewer, M . B., Dull, V , & Lui, L. (1981). Perceptions of the elderly: Stereotypes as prototypes. Journal
Bruner, J. S. (1957). Going beyond the information given. In H . E. Gruber, K. R. Hammond, & R. Jessor
(Eds.), Contemporary approaches to cognition (pp. 41-69). Cambridge, M A : Harvard University Press.
Cacioppo, J. T., Priester, J. R., & Bemtson, G . G . (1993). Rudimentary determinants of attitudes: II: A r m
flexion and extension have different effects on attitudes. Journal of Personality and Social Psychology,
65, 5-17.
Carlston, D . E. (1980). T h e recall and use of traits and events in social inference processes. Journal of
Experimental Social Psychology, 16, 303-329.
Carlston, D . E. (1994). Associated systems theory: A systematic approach to cognitive representations
of persons. In R. S. Wyer, Jr. (Ed.), Advances in social cognition (Vol. VII, pp. 1-78). Hillsdale, NJ:
Lawrence Erlbaum Associates.
Cartwright, D . (Ed.). (1959). Studies in social power. A n n Arbor, MI: Institute for Social Research.
Carver, C . S., Ganellen, R. J., Froming, W. J., & Chambers, W. (1983). Modeling: A n analysis in terms
of category accessibility. Journal of Experimental Social Psychology, 19, 403-421.
Chaiken, S., & Bargh, J. A . (1993). Occurrence versus moderation of the automatic attitude activation
effect: Reply to Fazio. Journal of Personality and Social Psychology, 64, 759-765.
56 Bargh
Chaiken, S., Giner-Sorolla, R., & Chen, S. (1996). Beyond accuracy: Defense and impression motives
in heuristic and systematic information processing. In P. M . Gollwitzer & J. A . Bargh (Eds.), The
psychology of action (pp. 553-578). New York: Guilford.
Chaiken, S., Liberman, A . , & Eagly, A . H . (1989). Heuristic and systematic information processing
within and beyond the persuasion context. In J. S. Uleman & J. A . Bargh (Eds.), Unintended thought
(pp. 212-252). New York: Guilford.
Chartrand. T. L., & Bargh, J. A . (1996). Automatic activation of impression formation and memorization
goals: Nonconscious goal priming reproduces effects of explicit task instructions. Journal of Personality
and Social Psychobgy, 71, 464-478.
Chen, M . , & Bargh, J. A . (1996). A n automatic effect of (all) attitudes on behavior: Preconscious approach
and avoidance responses to liked and disliked stimuli. Manuscript submitted for publication, New York
University.
Chen, M . , & Bargh, J. A . (1997). Nonconscious behavioral confirmation processes: T h e self-fulfilling
nature of automatically activated stereotypes. Journal of Experimental Social Psychology.
Chen, S., Shechter, D . , & Chaiken, S. (1996). Getting at the truth or getting along: Accuracy and
impression-motivated heuristic and systematic processing. Journal of Personality and Social Psychology,
71, 262-265.
Cialdini, R. B. (1994, October). The strain for consistency: A history, a measure, and a surprise. Plenary
address to the annual meeting of the Society for Experimental Social Psychology, Lake Tahoe, N V .
Cialdini, R. B., Trost, M . R., & Newsom, J. T. (1995). Preference for consistency: T h e development of
a valid measure and the discovery of surprising behavioral implications. Journal of Personality and Social
Psychobgy, 69,318-328.
Collins, A . M . , & Loftus, E. F. (1975). A spreading activation theory of semantic memory. Psychobgical
Review, 82, 407-428.
Darley, J. M . , & Latané, B. (1968). Bystander intervention in emergencies: Diffusion of responsibility.
Journal of Personality and Social Psychobgy, 8, 377-383.
Devine, P. G . (1989). Stereotypes and prejudice: Their automatic and controlled components. Journal
of Personality and Social Psychobgy, 56, 680-690.
Dunn, J. C . , & Kirsner, K. (1988). Discovering functionally independent mental processes: T h e principle
of reversed association. Psychobgical Review, 95, 91-101.
Edwards, K . (1990). T h e interplay of affect and cognition in attitude formation and change. Journal of
Personality and Social Psychobgy, 59, 202-216.
Erber, R., & Fiske, S. T. (1984). Outcome dependency and attention to inconsistent information. Journal
Erdelyi, M . H . (1974). A new look at the New Look: Perceptual defense and vigilance. Psychobgical
Review, 81, 1-25.
Erdley, C . A . , & D'Agostino,P.R. (1988). Cognitive and affective components of automatic priming
effects. Journal of Personality and Social Psychobgy, 54, 741-747.
Fazio, R. H . (1986). How do attitudes guide behavior? In R. M . Sorrentino & E. T. Higgins (Eds.),
Handbook of motivation and cognition (Vol. 1, pp. 204-243). New York: Guilford.
Fazio, R. H . , Chen, J., McDonel, E. C , & Sherman, S. J., (1982). Attitude accessibility, attitude-behavior
consistency, and the strength of the object-evaluation association. Journal of Experimental Social
Psychobgy, 18, 339-357.
Fazio, R. H . , Sanbonmatsu, D . M . , Powell, M . C . & Kardes, F. R. (1986). O n the automatic activation
of attitudes. Journal of Personality and Social Psychobgy, 50, 229-238.
Festinger, L., & Carlsmith, J. M . (1959). Cognitive consequences of forced compliance. Journal of
Abnormal and Social Psychobgy, 58, 203-210.
Fiske, S. T. (1980). Attention and weight in person perception: T h e impact of negative and extreme
behavior. Journal of Personality and Social Psychobgy, 38, 889-906.
Fiske, S. T., & Pavelchak, M . (1986). Category-based versus piecemeal-based affective responses:
Developments in schema-triggered affect. In R. M . Sorrentino & E. T. Higgins (Eds.), Handbook of
motivation and cognition (pp. 167-203). New York: Guilford.
Fitzgerald, L. F. (1993, February). The last great open secret: The sexual harassment of women in the workplace
and academia. Edited transcript of a Science and Public Policy Seminar presented by the Federation
of Behavioral, Psychological, and Cognitive Sciences, Washington, D C .
Flavell, J. H . , & Draguns, J. (1957). A microgenetic approach to perception and thought. Psychological
Bulletin, 54, 197-217.
Fodor, J. A . (1983). The modularity of mind. Cambridge, M A : Bradford Books.
Gazzaniga, M . (1985). The social bram. New York: Basic Books.
Gilbert, D . T. (1989). Thinking lightly about others: Automatic components of the social inference
process. In J. S. Uleman & J. A . Bargh (Eds.), Unintended thougfxt (pp. 189-211). New York: Guilford.
Gilbert, D. T., Pelham, B. W., & Krull, D. S. (1988). O n cognitive busyness: W h e n person perceivers
meet persons perceived, Journal of Personality and Social Psychology, 54, 733-739.
Gollwitzer, P. M . , Heckhausen, H . , & Steller, B. (1990). Deliberative and implemental mind-sets:
Cognitive tuning toward congruous thoughts and information. Journal of Personality and Social
Psychology, 59, 1119-1127.
Gollwitzer, P. M . , & Moskowitz, G . (1996). Goal effects on thought and behavior. In E. T. Higgins & A .
Kruglanski (Eds.), Social psychology: Handbook of basic principles (pp. 361-399). New York: Guilford.
Greenwald, A . G . , & Banaji, M . R. (1995). Implicit social cognition: Attitudes, self-esteem, and
stereotypes. Psychological Review, 102, 4-27.
Hamilton, D. L., Katz, L. B., & Leirer, V O . (1980). Organizational processes in impression formation.
In R. Hastie, T. M . Ostrom, E. B. Ebbesen, R. S. Wyer, D. L. Hamilton, & D. E. Carlston (Eds.), Person
memory: The cognitive basis of social perception (pp. 121-153). Hillsdale, NJ: Lawrence Erlbaum
Associates.
Hansen, R. D . (1980). Commonsense attribution. Journal of Personality and Social Psychology, 39,
996-1009.
Hastie, R., & Kumar, P. (1979). Person memory: Personality traits as organizing principles in memory for
behaviors. Journal of Personality and Social Psychology, 37, 25-38.
Hastie, R., & Park, B. (1986). The relationship between memory and judgment depends on whether
the judgment task is memory-based or on-line. Psychological Review, 93, 58-268.
Hayes-Roth, B. (1977). Evolution of cognitive structure and process. Psychological Review, 84, 260-278.
Hebb, D . O . (1948). Organization of behavior. New York: Wiley.
Heckhausen, H . (1990). Motivation and action. New York: Springer-Verlag.
Herr, P. M . (1986). Consequences of priming: Judgment and behavior. Journal of Personality and Social
Psychology, 51, 1106-1115.
Higgins, E. T. (1987). Self-discrepancy: A theory relating self and affect. Psychological Review, 94,
319-340.
Higgins, E. T. (1989). Knowledge accessibility and activation: Subjectivity and suffering from uncon-
scious sources. In J. S. Uleman & J. A . Bargh (Eds.), Unintended thought (pp. 75-123). New York:
Guilford.
Higgins, E. T. (1991). Development of self-regulatory and self-evaluative processes: Costs, benefits, and
tradeoffs. & L. A . Sroufe (Eds.), The Minnesota Symposia on Child Development: Self processes and
development (Vol. 23). Hillsdale, NJ: Lawrence Erlbaum Associates.
Higgins, E. T., Bargh, J. A . , & Lombardi, W. (1985). Nature of priming effects on categorization. Journal
of Experimental Psychology: Learning, memory, and cognition, 11, 59-69.
Higgins, E. T., King, G . A . , & Mavin, G . H . (1982). Individual construct accessibility and subjective
impressions and recall. Journal of Personality and Social Psychology, 43, 35-47.
Higgins, E. T., Rholes, W. S., & Jones, C . R. (1977). Category accessibility and impression formation.
Journal of Experimental Social Psychology, 13, 141-154.
Hilgard, E. R. (1965). Hypnotic susceptibility. New York: Harcourt Brace & World.
Hilgard, E. R. (1977). Divided consciousness. New York: Wiley.
Hintzman, D . L. (1988). Judgments of frequency and recognition memory in a multiple-trace memory
model. Psychological Review, 95, 528-551.
58 Bargh
Hodges, S., & Wegner, D . M . (in press). Automaticity and control of empathy. In W. J. Ickes (Ed.),
Empathic accuracy. New York: Guilford.
Horner, M . S. (1974). Fear of success in women. In J. W Atkinson & J. O . Raynor (Eds.), Motivation and
achievement (pp. 76-117). Washington: Winston.
Jackson, D. N . (1974). Personality Research FormManual. Port Huron, MI: Research Psychologists Press.
Jacoby, L. L. (1991). A process dissociation framework: Separating automatic from intentional uses of
memory. Journal of Memory and Language, 30, 513-541.
James, W. (1890). The principles of psychology. New York: Holt.
Johnson, M . K . (1983). A multiple entry, modular memory system. In G . H . Bower (Ed.), The psychology
of learning and motivation (Vol. 17, pp. 81-123). New York: Academic Press.
Johnson, M . K . , Kim, J. K . , & Risse, G . (1985). D o alcoholic Korsakoff's syndrome patients acquire
affective reactions? Journal of Experimental Psychology: Learning, Memory, and Cognition, 11, 22-36.
Jones, E. E., & Nisbett, R. E. (1971). T h e actor and the observer: Divergent perceptions of the causes
of behavior. In E. E. Jones, D . Kanouse, H . H . Kelley, R. E. Nisbett, S. Valins, & B. Weiner (Eds.),
Attribution: Perceiving the causes of behavior (pp. 79-94). Morristown, NJ: General Learning Press.
Jung, C . G . (1927). T h e structure of the psyche. In R. F. C . Hull (Trans.), Collected works of C. G . Jung
(Vol. 8, pp. 283-342). Princeton, NJ: Princeton University Press.
Karniol, R., & Ross, M . (1996). T h e motivational impact of temporal focus: Thinking about the future
and the past. Annual Review of Psychology, 47, 593-620.
Kelley, H . H . (1967). Attribution theory in social psychology. In D . Levine (Ed.), Nebraska Symposium
on Motivation (Vol. 15, pp. 192-241). Lincoln: University of Nebraska Press.
Kipnis, D. (1976). The powerholders. Chicago: University of Chicago Press.
Klüver, H . (1951). Discussion of Lashley's chapter. In L. A . Jeffries (Ed.), Cerebral mechanisms in behavior:
The Hixon Symposium (pp. 136). New York: Wiley.
Klein, S. B., & Loftus, J. (1993). T h e mental representation of trait and autobiographical knowledge
about the self. In T. K . Srull & R. S. Wyer, Jr. (Eds.), Advances in social cognition (Vol. 5, pp. 1-49).
Koffka, K . (1925). Die Grundlagen der psychischen Entwicklung. Osterwieck: Zickfeldt.
Kuhl, J. (1986). Motivation and information processing: A new look at decision making, dynamic change,
and action control. In R. M . Sorrentino & E. T. Higgins (Eds.), Handbook of motivation and cognition
(Vol. 1, pp. 404-434). New York: Guilford.
Lang, P J., Bradley, M . M . , & Cuthbert, B. N . (1990). Emotion, attention, and the startle reflex.
Psychobgical Review, 97, 377-395.
Lashley, K . S. (1951). T h e problem of serial order in behavior. In L. A . Jeffress (Ed.), Cerebral mechanisms
in behavior: The Hixon symposium (pp. 112-136). New York: Wiley.
Lazarus, R. S. (1982). Thoughts on the relations between emotion and cognition. American Psychologist,
37, 1019-1024.
LeDoux, J. E. (1989). Cognitive-emotional interactions in the brain. Cognition and Emotion, 3, 267-289.
LeDoux, J. E., Iwata, J., Reis, D. J. (1988). Different projections of the central amygdaloid nucleus
mediate autonomic and behavioral correlates of conditioned fear. Journal of Neuroscience, 8,
2517-2529.
Lewin, K . (1926). Vorsatz, Wille und Bedurfnis. Psychobgische Forschung, 7, 330-385.
Lewin, K . (1935). A dynamic theory of personality. New York: McGraw-Hill.
Lewin, K . (1943). Defining the 'field at a given time.' Psychobgical Review, 50, 292-310.
Lichtenstein, M . , & Srull, T. K . (1987). Processing objectives as a determinant of the relationship
between recall and judgment. Journal of Experimental Social Psychology, 23, 93-118.
Lorch, R. F., Jr. (1982). Priming and search processes in semantic memory: A test of three models of
spreading activation. Journal of Verbal Learning and Verbal Behavior, 21, 468-492.
Macrae, C . N . , Milne, A . B., & Bodenhausen, G . V. (1993). Stereotypes as energy-saving devices: A
peek inside the cognitive toolbox. Journal of Personality and Social Psychology, 66, 31-41.
Malamuth, N . M . (1989). T h e attraction to sexual aggression scale: Part O n e . Journal of Sex Research,
26, 26-49.
Marcel, A . J. (1983). Conscious and unconscious perception: Experiments on visual masking and word
recognition. Cognitive Psychology, 15, 197-237.
Martin, L. L., & Tesser, A . (1989). Toward a motivational and structural theory of ruminative thought.
In J. S. Uleman & J. A . Bargh (Eds.), Unintended thought (pp. 306-326). New York: Guilford.
Martindale, C . (1991). Cognitive psychology: A neural-network approach. Pacific Grove, C A : Brooks/Cole.
Meyer, D . E., & Schvaneveldt, R. W. (1971). Facilitation in recognizing pairs of words: Evidence of a
dependence between retrieval operations. Journal of Experimental Psychology, 90, 227-234.
McClelland, D. C . (1953). The achievement motive. New York: Appleton-Century-Crofts.
Milgram, S. (1963). Behavioral study of obedience. Journal of Abnormal and Social Psychology, 67,
371-378.
Miller, G . A . , Galanter, E., & Pribram, K. H . (1960). Plans and the structure of behavior. New York: Holt,
Rinehart & Winston.
Mills, C . J., & Tyrrell, D. J. (1983). Sex-sterotypic encoding and release from proactive interference.
Mischel, W. (1973). Toward a cognitive social learning reconceptualization of personality. Psychological
Review, 80, 252-283.
Murphy, S. T , & Zajonc, R. B. (1993). Affect, cognition, and awareness: Affective priming with optimal
and suboptimal stimulus exposures. Journal of Personality and Social Psychology, 64, 723-739.
Murray, H . (1943). The Thematic Apperception Test manual. Cambridge, M A : Harvard University Press.
Neely, J. H . (1977). Semantic priming and retrieval from lexical memory: Roles of inhibitionless spreading
activation and limited-capacity attention. Journal of Experimental Psychology: General, 106, 226-254.
Negroponte, N . (1995). Being digital. New York: Knopf.
Neisser, U . (1967). Cognitive psychology. New York: Appleton-Century-Crofts.
Neuberg, S. L. (1988). Behavioral implications of information presented outside of awareness: T h e effect
of subliminal presentation of trait information on behavior in the Prisoners Dilemma Game. Social
Cognition, 6, 207-230.
Newell, A . , & Rosenbloom, P. S. (1981). Mechanisms of skill acquisition and the law of practice. In J.
R. Anderson (Ed.), Cognitive skills and their acquisition (pp. 1-55). Hillsdale, NJ: Lawrence Erlbaum
Associates.
Newell, A . , & Simon, H . A . (1972). Human problem solving. Englewood Cliffs, NJ: Prentice-Hall.
Newman, L. S., & Uleman, J. S. (1989). Spontaneous trait inference. In J. S. Uleman & J. A . Bargh
(Eds.), Unintended thought (pp. 155-188). New York: Guilford.
Niedenthal, P M . (1990). Implicit perception of affective information. Journal of Experimental Social
Niedenthal, P M . , & Cantor, N . (1986). Affective responses as guides to category-based influences.
Motivation and Emotion, 10, 217-231.
Nisbett, R. E., & Wilson, T D . (1977). Telling more than we can know: Verbal reports on mental
processes. Psychological Review, 84, 231-259.
Osgood, C . E . (1953). Method and theory in experimental psychology. New York: Oxford University Press.
Osgood, C . E., Suci, G . S., & Tannenbaum, P H . (1957). The measurement of meaning. Urbana: University
of Illinois Press.
Ovsiankina, M . (1928). Die Wiederaufnahme unterbrochener Handlungen [The resumption of interrupted
goals]. Psychologische Forschung, 11, 302-379.
Perdue, C . W , & Gurtman, M . B. (1990). Evidence for the automaticity of ageism. Journal of Experimental
Pervin, L. A . (Ed.). (1989). Goal concepts in personality and social psychology. Hillsdale, NJ: Lawrence
Erlbaum Associates.
Piaget, J. (1946). La formation du symbole chez l'enfant. Paris: Delachaux & Niestlé.
Pittman, T. S., & D'Agostino, P R. (1985). Motivation and attribution: T h e effects of control deprivation
on subsequent information processing. InJ.H . Harvey & G . Weary (Eds.), Attribution: Basic and applied
issues (pp. 117-142). San Diego, C A : Academic Press.
60 Bargh
Posner, M . I., & Snyder, C . R. R. (1975). Attention and cognitive control. In R. L. Solso (Ed.), Information
processing and cognition: The Loyola symposium (pp. 55-65). Hillsdale, NJ: Lawrence Erlbaum Associ-
ates.
Pratto, F., & Bargh, J. A . (1991). Stereotyping based on apparently individuating information: Trait and
global components of sex stereotypes under attention overload. Journal of Experimental Social Psychol-
ogy, 27, 26-47.
Pryor, J. B. (1987). Sexual harassment proclivities in men. Sex Roles, 17(5-6), 269-290.
Prinz, W. (1990). A common coding approach to perception and action. In O . Neumann & W. Prinz
(Eds.), Relationships between perception and action (pp. 167-201). Heidelberg: Springer-Verlag.
Reuman, D . A . , Alwin, D . F., & Veroff, J. (1984). Assessing the validity of the achievement motive in
the presence of random measurement error. Journal of Personality and Social Psychobgy, 47, 1347-1362.
Rilling, M . (1992). A n ecological approach to stimulus control and tracking. In W. K . Honig & J. G .
Fetterman (Eds.), Cognitive aspects of stimulus control (pp. 347-366). Hillsdale, NJ: Lawrence Erlbaum
Associates.
Rumelhart, D. E., & McClelland, J. L. (1986). Parallel distributed processing. Cambridge, M A : M I T Press.
Russell, B. (1938). Power: A new social analysis. New York: Norton.
Salovey, P., & Singer, J. A . (1989). Mood congruency effects in recall of childhood versus recent
memories. Journal of Social Behavior and Personality, 4, 99-120.
Schachter, S., & Singer, J. L. (1962). Cognitive, social, and physiological determinants of emotional
state. Psychological Review, 69, 379-399.
Shallice, T. (1972). Dual functions of consciousness. Psychobgical Review, 79, 383-393.
Shiffrin, R. M . , & Dumais, S. T. (1981). T h e development of automatism. In J. R. Anderson (Ed.),
Cognitive skills and their acquisition (pp. 111-140). Hillsdale, NJ: Lawrence Erlbaum Associates.
Shiffrin, R. M . , &. Schneider, W. (1977). Controlled and automatic human information processing: II.
Perceptual learning, automatic attending, and a general theory. Psychobgical Review, 84, 127-190.
Skinner, B. F. (1953). Science and human behavior. New York: Free Press.
Skinner, B. F. (1978). W h y I am not a cognitive psychologist. In B. F. Skinner (Ed.), Reflections on
behaviorism and society (pp. 97-112). Englewood Cliffs, NJ: Prentice-Hall.
Smith, E. R. (1984). Model of social inference processes. Psychobgical Review, 91, 392-413.
Smith, E. R. (1989). Procedural efficiency: General and specific components and effects on social
judgment. Journal of Experimental Social Psychobgy, 25, 500-523.
Smith, E. R. (1994). Procedural knowledge and processing strategies in social cognition. In R. S. Wyer
& T. K . Srull (Eds), Handbook of social cognition (2nd ed., Vol. 1, pp. 99-152). Hillsdale, NJ: Lawrence
Erlbaum Associates.
Smith, E . R., Branscombe, N . R., & Bormann, C . (1988). Generality of the effects of practice on social
judgment tasks. Journal of Personality and Social Psychobgy, 54, 385-395.
Smith, E. R., & Lerner, M . (1986). Development of automatism of social judgments. Journal of Personality
Snyder, M . , Tanke, E. D . , & Berscheid, E. (1977). Social perception and interpersonal behavior: O n the
self-fulfilling nature of social stereotypes. Journal of Personality and Social Psychology, 35, 656-666.
Solarz, A . (1960). Latency of instrumental responses as a function of compatibility with the meaning of
eliciting verbal signs. Journal of Experimental Psychobgy, 59, 239-245.
Sorrentino, R. M . , & Higgins, E. T. (1986). Motivation and cognition: Warming up to synergism. In R.
M . Sorrentino & E. T. Higgins (Eds.), Handbook of motivation and cognition (Vol. 1 pp. 3 -19). New
York: Guilford.
Spielman, L. A . , Pratto, F., & Bargh, J. A . (1988). Automatic affect: A r e one's moods, attitudes,
evaluations, and emotions out of control? American Behavioral Scientist, 31, 96-311.
Srull, T. K . (1981). Person memory: Some tests of associative storage and retrieval models. Journal of
Experimental Psychobgy: Human Learning and Memory, 7, 440-463.
Srull, T. K . , Lichtenstein, M . , & Rothbart, M . (1985). Associated storage and retrieval processes in
person memory. Journal of Experimental Psychobgy: Learning, Memory, and Cognition, 11,316-345.
Srull, T. K., & Wyer, R. S., Jr. (1979). T h e role of category accessibility in the interpretation of information
about persons: Some determinants and implications. Journal of Personality and Social Psychology, 37,
1660-1672.
Srull, T. K . , & Wyer, R. S., Jr. (1980). Category accessibility and social perception: Some implications
for the study of person memory and interpersonal judgment. Journal of Personality and Social Psychology,
38, 841-856.
Srull, T. K., & Wyer, R. S., Jr. (1983). T h e role of control processes and structural constraints in models
of memory and social judgment. Journal of Experimental Social Psychology, 19, 497-521.
Srull, T. K . , & Wyer, R. S., Jr. (1986). T h e role of chronic and temporary goals in social information
processing. In R. M . Sorrentino & E. T. Higgins (Eds.), Handbook of motivation and cognition (pp.
503-549). New York: Guilford.
Steele, C . M . (1988). T h e psychology of self-affirmation: Sustaining the integrity of the self. Advances
in Experimental Social Psychology, 21, 261-302.
Strauman, T. J., & Higgins, E. T. (1987). Automatic activation of self-discrepancies and emotional
syndromes: W h e n cognitive structures influence affect. Journal of Personality and Social Psychology, 53,
1004-1014.
Taylor, S. E. (1989). Positive illusions. New York: Basic Books.
Taylor, S. E., & Fiske, S. T. (1978). Salience, attention, and attribution: Top of the head phenomena. In
L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 11, pp. 249-288). New York:
Academic Press.
Thibaut, J. W., & Kelley, H . H . (1959). The social psychology of groups. New York: Wiley.
Tooby, J., & Cosmides, L. (1992). T h e psychological foundations of culture. In J. H . Barkow, J. Tooby, &
L. Cosmides (Eds.), The adapted mind (pp. 19-136). New York: Oxford University Press.
Uleman, J. S., Newman, L. S., & Moskowitz, G . B. (1996). People as spontaneous interpreters: Evidence
and issues from spontaneous trait inference. In M . P. Zanna (Ed.), Advances in experimental social
psychology (Vol. 28, pp. 211-279). San Diego, C A : Academic Press.
Vera, A . H . , & Simon, H . A . (1993). Situated action: A symbolic interpretation. Cognitive Science, 17,
7-48.
Wegner, D . M . , & Bargh, J. A . (1997). Control and automaticity in social life. In D. Gilbert, S. Fiske, &
G . Lindzey (Eds.) Handbook of social psychology (4th ed.). Boston, M A : McGraw-Hill.
Werner, H . (1956). Microgenesis and aphasia. Journal of Abnormal and Social Psychology, 52, 347-353.
Werner, H . , & Kaplan, B. (1963). Symbol formation: An organismic-developmental approach to language and
the expression of thought. New York: Wiley.
Whitehead, A . N . (1911). A n introduction to mathematics. New York: Henry Holt.
Wicklund, R. A . , & Brehm, J. W. (1976). Perspectives on cognitive dissonance. Hillsdale, NJ: Lawrence
Erlbaum Associates.
Wicklund, R. A . , & Gollwitzer, P. M . (1982). Symbolic self-completion. Hillsdale, NJ: Lawrence Erlbaum
Associates.
Wilensky, R. (1983). Planning and understanding. Reading, M A : Addison-Wesley.
Winter, L., & Uleman, J. S. (1984). W h e n are social judgments made? Evidence for the spontaneousness
of trait inferences. Journal of Personality and Social Psychology, 47, 237-252.
Wyer, R. S., Jr., & Srull, T. K . (1986). Human cognition in its social context. Psychological Review, 93,
322-359.
Wyer, R. S., Jr., & Srull, T. K . (1989). Memory and cognition in its social context. Hillsdale, NJ: Lawrence
Erlbaum Associates.
Zajonc, R. B. (1980). Feeling and thinking: Preferences need no inferences. American Psychologist, 35,
151-175.
Zillman, D . , & Bryant, J. (1974). Effect of residual excitation on the emotional response to provocation
and delayed aggressive behavior. Journal of Personality and Social Psychology, 30, 782-791.
Chapter 2
Environments and
Unconscious Processes
Mahzarin R. Banaji
Yale University
Irene V. Blair
University of Colorado
Jack Glaser
Yale University
Even today, the strongest position i n psychology advocating the supremacy of

environments i n determining behavior remains that of B. E Skinner. H a l f a century
after the cognitive revolution and a full rejection of Skinner's antimentalism, his
bold optimism that human behavior is lawful and determined, that the sources of
predictive power lie i n the organism's environment, and that identifying them is the
only certain path to a technology of behavior is ironically inspirational to a social
psychologist working on fundamental questions regarding mental processes. John
Bargh is a product of late 20th century social psychology, a field that passed its
infancy with fortunate obliviousness of both the antimentalism of behaviorism and
the inattention to environments that characterizes the inward-looking stance of
modern cognitive psychology. From a historical point of view, it should occasion no
surprise that a person born of this tradition need not be burdened by shame or
conflict i n using a dead, anticognitive philosophy's insistence o n the power of
environments while speaking with ease about the power of automatic mental
processes.
In this target chapter, Bargh describes extensive programs of research on auto-
matic social processes, w h i c h when viewed as a collection, offer an impressive view
of how these processes operate i n everyday social life. O u r o w n position is compat-
ible with the one advocated i n the first chapter, and our comments will reaffirm and
add to selected issues. O u r main concern lies with the need for theories of the
63
64 Banaji, Blair, & Glaser
meaning and properties of transient and persisting environments and how they
produce their influence o n social processes (cognitive, evaluative, and behavioral).
We conclude that the research o n unconscious social processes reviewed by Bargh
not only provides new evidence about social perception, but also addresses deeper
questions about human nature. In our view, this research favors a new environ-
mental determinism i n understanding the causes of social behavior—one that is
necessarily informed by several decades of research o n social cognition.
From at least one perspective, the most important discoveries i n social psychol-
ogy are those that show the power of situational forces i n determining behavior,
with the two shining examples even 30 years later being experiments o n obedience
to authority (Milgram, 1963) and o n bystander nonintervention (Latané & Darley,
1968). These experiments (along with lesser known but equally impressive ones)
ought to be recognized as landmarks i n the history of science, for i n them we have
the very first experimental evidence for an unpopular view of human nature. In
contrast to the perspective from other fields, and certainly i n opposition to lay
thinking, these studies provided the first experimental demonstrations that humans
do not and more accurately, cannot, choose their actions as freely as they or their
observers expect. Rather, forces i n the situation, of which they may be little aware,
can have a determining influence on their actions, even those actions that have
immense consequences for the well-being and survival of themselves and their
fellow beings. The view of human nature revealed by these early experiments
continues to be a difficult one to endorse, perhaps especially by Western minds,
because it suggests that the will to freely choose a course of action may be illusory.
S u c h a view is additionally problematic because it pointedly raises the question of
whether reward for benevolent actions or retribution for heinous ones should
legitimately be assigned to the actor who performs them.
The profundity of these implications and the staying power of these demonstra-
tions i n our textbooks notwithstanding, it is the simple truth that these programs
of research did not propagate. After a few years' worth of laboratory and field
iterations of each basic finding, they ceased to inspire new work commensurate with
their impact or to produce advances on the scale of other theoretical orientations
i n psychology such as psychoanalysis, behaviorism, or information processing. W h y
was this the case? W h y were such stunning experimental discoveries not the basis
of a full-fledged and more influential perspective o n social behavior? There are
many explanations to offer, but one that the target chapter suggests to us is that
these accounts lacked grounding i n a theoretical system capable of explaining the
mechanisms that link environmental effects to social processes. A s Bargh's research
exemplifies, the availability of theories and methods to analyze automatic processes
offers a way out of some explanatory darkness.
We focus on two issues. First, we discuss the problem of accuracy, or more to the
point, inaccuracy i n perceiving the sources of influence o n judgment and behavior.
In particular, when causes are removed i n time or space from the effects they
produce, namely, when causal action occurs at a distance, the relationship between
the two may most naturally lie outside awareness. This point allows a connection
to be made between many classic findings i n social psychology showing inaccuracies
2. Environments and Unconscious Processes 65
i n assigning appropriate causes for behavior and the automatic processes that
underlie them. Second, we point out the value of construing the individual's
environment i n more microscopic terms to include vast numbers of potential
causes of thought, feeling, and action that may lie outside conscious awareness.
T h e target chapter offers many elegant examples of this, and we add some from
research o n the implicit and automatic use of knowledge and feelings about social
groups.
PERCEIVING ACTION AT A DISTANCE
Multiple strands of research i n social psychology have verified that perceiving the
cause of actions as emanating from the actor rather than the environment is a robust
human characteristic. This point was not only made i n the obedience and helping
research mentioned earlier, but more directly by research o n the attribution of
causality, now commonly referred to as the fundamental attribution error (Ross, 1977)
or the correspondence bias (Jones & Gerard, 1967). We use a physical metaphor here,
for it nicely suggests that this bias may be part of a more general human inability to
accurately perceive "action at a distance," with the term action referring to causal
action.
U n t i l Newton's discovery, scientists, like their lay colleagues, incorrectly believed
that color resided i n the colored object. Even 300 years after this discovery, it is only
through formal education and not intuition that we know, for example, that
"brownness" is not a "property" of skin and that "brownness" does not "reside i n "
the skin. Rather, as N e w t o n (1671) reported, "For as sound, i n a bell or musical
string or other sounding body, is nothing but a trembling motion, and i n the air
nothing but that motion propagated from the object, ... so colors i n the object are
nothing but a disposition to reflect this or that sort of ray more copiously than the
rest . . . " Writing to Oldenburg i n 1672, he described with great excitement the
experiments showing that light consists of rays of unequal "refrangibility," and
concluded, "These things being so, it can be no longer disputed, whether there be
colours i n the dark, nor whether they be the qualities of the objects we see . . . " (p.
179).
We now know that a complex interaction of light as well as properties of the
object itself determine color as it is ultimately perceived. The role of the object i n
"causing" us to perceive color is easy to grasp, whereas genius was needed to discover
that light, a source operating at a distance from the perceived object and with no
perceivable physical link to the object played the crucial role it did. T h e perception
of the causes of social behavior as residing i n the actor arise from a similar underlying
inability to see action at a distance. W h e n asked for an explanation of the cause of
X's behavior, the response is likely to involve properties of X rather than Y, if Y (an
animate or inanimate cause) issues an influence that is physically and psychologi-
cally invisible. A n d just as surely as with optics, a correct interpretation of the causes
of behavior must include both properties of the subject (which are intuitively
accessible) and properties of the environment (which are intuitively less accessible).
T h e reason for the relative difficulty of the latter i n both cases, optics as well as
social perception, is that causes lie i n places that are unfamiliar or distant and
perhaps not easily available to conscious cognition.
Examining the operation of automatic processes o n social behavior takes the bull
by its horns. There is clear recognition i n these newer accounts of social behavior
that sources of influence that may not be within the grasp of the actor may
determine perceptions and beliefs, preferences, and actions. A l t h o u g h this idea has
been a necessary part of much social psychological research, it is only with the
explicit study of processes that lie outside conscious awareness and control that the
full range of their impact can be determined. T h e unique emphasis that Bargh offers
i n the early section of the target chapter is that such sources of influence lie i n the
environment of the actor. To enable a fuller account of the cycle of interaction
between environment and mind, we must identify causative properties of the social
environment, generate meaningful taxonomies of them, and test the nature of their
influences on social thought, feeling and behavior. Such an approach allows more
fruitful encounters with sources of causal action that lie at a distance from the effects
they produce.
MICROENVIRONMENTS AND MICROBEHAVIORS
A l l psychological activity occurs i n some space, and we follow an old tradition i n

broadly referring to that space and its contents as environment, although our focus
will necessarily be restricted to socially meaningful ones. We introduce the term
microenvironments to capture a class of environmental influences that are pervasive
and influential even though they are not easily perceived or comprehended because
of their "smallness," and the term microbehaviors to capture the responses they
evoke. A t t e n t i o n to these features is new to social psychology, but is well illustrated
in Bargh's focus o n automatic social processes.
Yet again, an analogy from the physical sciences may be handy. We know that
knowledge of the physical world changed dramatically with the transition from
examining gross structures available to the naked eye to particle level structures
unavailable to the naked eye. Likewise, there lie potential layers of social psycho-
logical structures that may only be available by peering at levels that are below those
of consciously accessible cognition. Shifts i n the level of analysis i n any field are a
complex result of advances i n theory and the availability of methods and tools (for
example, the invention of the electron microscope). The shift i n social psychology
occurred most dramatically, as it did i n other fields, through the use of (micro)
computers i n research, which make it possible to create controlled, high-speed
representations of the environment and obtain stable, high-speed responses to the
environment. Entire layers of behavior previously unavailable and unrecognized as
even existing are becoming tractable and reliably reproducible, especially those
requiring stimulus presentation outside conscious awareness and measurement
without the respondent's awareness or control. Investigations such as the ones

1
captured by Bargh's research show the gains resulting when attending to the
microscopic features of the environment and measuring its influence at the level of
multiple single judgments or microbehaviors.
The implications of such a focus are not trivial. We use a comment made by a
colleague, a developmental psychologist, to illustrate the point. Pointing to his
2-year-old daughter's preference for feminine objects such as a purse, he expressed
surprise that she liked feminine things even though her parents had never encour-
aged such choices. T h e example was generated by h i m to convey the idea that such
choices and preferences cannot therefore be said to be learned or acquired, but
rather rooted i n a more inherent preference of females for feminine objects and
conversely of males for masculine objects. T h e colleague is a fellow of respectable
intelligence, so the question is really one for us social psychologists: W h y have we
failed to communicate a theory of the ways i n which environments produce their
influence so that a contemporary psychologist, let alone a layperson, can be properly
informed about the mechanisms by which environments can influence behavior?
We think that for too long social psychology remained at the level of gross
descriptions of environments. Such a level is not inappropriate, and it gave us many
of the findings of which we are proud, such as the effects of direct threat by authority
figures, the influence of the sheer numbers of others, and so on. It is simply that
environments at levels that are far too microscopic to be visible can and do influence
behavior and being unaware of them can lead to causal errors of the sort captured
by our colleague's statement. A t t e n t i o n to microenvironments means attending to
the subtle and ongoing influences that shape preferences and desires, knowledge
and beliefs, motives toward or away from other social objects. Their influences, can
be powerful because they are not available to conscious awareness. T h e lack of
access to conscious awareness can be the basis of faulty theories of self and others.
T h e remarkable findings i n social cognition over the past 20 years have revealed
with much greater explanatory force than previously available the manner i n which
errors i n social perception not only occur, but are protected from correction. If the
influence of microenvironments is not detected, explanations for the actual cause
may proceed unhindered. A s experiments by Lewicki and H i l l (1987) showed,
learning the association between a physical feature such as the shape of a face and
a social attribute can occur with a single exposure and without awareness, show
generalization to other similarly structured faces, and reveal incorrect explanations
o n the part of subjects regarding the cause of their judgment.
a l t h o u g h new technologies allow such processes to be captured and recorded in an unprecedented

manner, we offer two caveats. First, the study of automatic social processes, as Bargh describes, has several
facets, some of which are best captured by the type of high-speed presentation and data collection
available through computerized techniques. However, other aspects of unconscious social behavior, ones
we referred to as implicit social cognition (Greenwald & Banaji, 1995) can be studied in a variety of ways,
not the least of which are simple paper and pencil measures, nonverbal physiological and behavioral
measures, and so on. Second, reducing phenomena from one level of analysis to a lower level is not a
mark of preference for the lower level. Rather, the assumption is that understandings across levels should
be logically consistent.
Social psychologists are not alone i n having ignored microenvironments. In

other areas of psychology, similar gross characterizations of environment abound.
T h e best example is perhaps the continuing assumption that environments are more
similar for children sharing the same family than those that are not, and this
thinking has been the basis of a large and well-established literature on intelligence
i n w h i c h children with varying genetic concordance within the same family are
compared with children raised i n different families. T h e notion that two individuals
may share the same gross environment (e.g., family) but not the same microenvi-
ronments (e.g., variations i n treatment within family), and that similarity i n such
microenvironments may be a powerful predictor of behavior remains a foreign
notion. However, the thesis and evidence i n the target chapter show just how
microenvironments can provide levels of analysis that were previously denied and
a level of prediction that may eventually be superior. Here, we are i n full agreement
with Bargh's optimism about the greater potential predictive power offered by
understanding environments and situations. We add that such evidence will emerge
from studying automatic social processes because these processes allow examination
of microenvironments and microbehaviors. There is some resistance to this idea,
even among those who are quick to acknowledge the importance of environmental
triggers more generally. For example, Jones (1990) wavered i n his conviction
regarding the influence of what we would call microenvironments: "Perhaps it is
the case that such hidden determinants are actually quite rare, that most of the
time our actions follow directly from our perceptions of the situation" (p. 117).
ACTION AT A DISTANCE IN SOCIAL MICROENVIRONMENTS:

EXAMPLES FROM STEREOTYPING AND PREJUDICE
In the context of Bargh's work on the automaticity of everyday life, there are
numerous reasons to focus attention on the phenomena of stereotyping and
prejudice. First, and most self-servingly, they are useful illustrations of the notion
of action at a distance, introduced earlier to capture the difficulty i n perceiving
causes that are physically and psychologically removed from their effects. Further-
more, there is special relevance of stereotyping and prejudice to the automaticity
of everyday life. We assume that the title of the target chapter was not an accidental
variation of Freud's (1901/1965) book, The Psychopathology of Everyday Life. Freud's
intention i n that book was to extend the principles of psychoanalysis from rare forms
of psychopathology to everyday ones, and the focus on stereotyping and prejudice
provides a similar extension i n modern social psychology. Such beliefs and attitudes
are no longer believed to be present merely i n a special class of individuals who
consciously affirm stereotypes and prejudices, but i n the everyday actions, beliefs,
and preferences of ordinary people. Finally, a focus on stereotyping and prejudice
provides a way to look at the consequences of automatic social perception i n a
domain that has implications for interpersonal and intergroup relations, a social
problem confronting every society.
People are universally influenced by sociocultural norms that engender stereo-

typing of and prejudice toward members of social groups. Often, such norms operate
invisibly, partly because causal action occurs at a distance and because the triggers
may be socially microscopic, shaping social cognition without awareness and
acknowledgment. Social knowledge structures form through the operation of
perfectly ordinary processes of attention, perception, and memory, and there is
m u c h research that we do not review showing the contents of stereotypes and
prejudices and the processes by which they operate. From our own recent research
and related work of others, a new understanding of the role of automatic processes
i n stereotyping and prejudice has emerged. Here, we discuss a few of the studies
that were not considered i n the target article to highlight their implications for the
automaticity of everyday life.
To illustrate the automaticity of social perception and beliefs, Bargh mentions
research o n stereotyping, focusing heavily o n Devine's (1989) experiments o n
automatic stereotyping and its relation to controlled expressions of prejudice.
A l t h o u g h this work is influential and relevant, it might better serve as a point of
departure for discussions of implicit and automatic stereotyping. There has been 2
considerable research o n automatic and implicit stereotyping and prejudice since

1989 that serves to both elucidate and complicate the issues.
We present selective research i n three sections to illustrate (a) general demon-
strations of implicit and automatic stereotyping and prejudice, (b) qualifications of
implicit and automatic stereotyping and prejudice, and (c) dissociations between
explicit and implicit or automatic and controlled stereotyping and prejudice.
Demonstrations of Implicit-Automatic
Stereotyping and Prejudice
Several demonstrations of the automatic activation and application of beliefs and
attitudes about social groups have appeared i n recent years that convincingly
establish the existence of automaticity i n this domain of everyday life. Banaji and
Greenwald (1995) showed that social category (gender) is implicitly used i n
judgments of fame, such that familiar male names are more likely judged to be
famous than equally familiar female names. This research went further i n locating
the source of the implicit bias i n the strictness of the criterion that subjects used i n
judgment—for equally familiarized male and female names, subjects set a lower
criterion for judging male than female fame. Banaji, H a r d i n , and R o t h m a n (1993)
likewise showed that prior exposure to stereotype content (sentences about depend-
2
T h e r e are many nuances in terminology that serve both to clarify and complicate the processes that
were referred to as conscious-unconscious, direct-indirect, explicit-implicit, and controlled-automatic.
We choose to use the label implicit to refer to research whose main purpose is to understand effects that
are produced when the source of influence on behavior lies outside subjects' conscious awareness, and
may only occur if the cause is thus hidden from awareness. We choose to use the label automatic to refer
to those effects that more naturally fall into Bargh's category of responses over which the subject may
have little control (even if there is awareness regarding the source of influence on behavior).
ence or aggressiveness) moderated the well-known category accessibility effect such

that only targets whose social category fit the previously activated stereotype (i.e.,
female targets i n the case of dependence priming and male targets i n the case of
aggressiveness priming) were judged more harshly.
W h a t is remarkable is the smallness of the familiarizing experience an environment
must offer (in this case, passing exposure with a name or stereotype knowledge) to show
an effect on judgment. Such findings give support to Bargh's claim i n the title of the
target chapter that automaticity is a pervasive feature of everyday life, and is consistent
with proposals made by those who study unconscious forms of memory regarding the
pervasiveness of implicit memory (Jacoby & Kelley, 1987). Additionally, studies of this
type show the problem with perceiving action at a distance. We continue with the
appropriation of Skinner (1971) to point out the subtle power of environments:
... the role of the environment is by no means clear. The history of the theory of
evolution illustrates the problem. Before the nineteenth century, the environment
was thought of simply as a passive setting in which many different kinds of organisms
were born, reproduced themselves, and died. N o one saw that the environment was
responsible for the fact that there were many different kinds (and that fact, signifi-
cantly enough, was attributed to a creative Mind). The trouble was that the environ-
ment acts in an inconspicuous way: it does not push or pull, it selects. For thousands
of years in the history of human thought the process of natural selection went unseen
in spite of its extraordinary importance. When it was eventually discovered, it became,
of course, the key to evolutionary theory.
The effect of environment on behavior remained obscure for an even longer time. We
can see what organisms do to the world around them, as they take from it what they
need and ward off its dangers, but it is much harder to see what the world does to
them. (p. 14)
Implicit stereotyping effects of the sort described fall into the category labeled
by Bargh as postconscious. Such effects, he says, "depend o n more than the mere
presentation of environmental objects or events ... postconsciously automatic
processes do require recent use or activation and do not occur without it." (chap.
1, p. 3). However, research also supports Bargh's main focus of interest i n the target
chapter, namely preconscious automatic processes. This form of automaticity "is
completely unconditional i n terms of a prepared or receptively tuned cognitive
state" (p. 3). Early work by Gaertner and M c L a u g h l i n (1983) and Dovidio, Evans,
and Tyler (1986) set the stage for later studies that more conclusively demonstrated
the automatic activation of social category knowledge i n information whose primary
meaning may and more importantly, may not denote the social category. Thus,
Banaji and H a r d i n (1996) showed that words like mother and father, w h i c h denote
gender, but also words like nurse and mechanic, which connote gender, facilitate the
subsequent speeded judgment of gender congruent male and female pronouns. Blair
and Banaji (1996a) further expanded the set of primes to include gender stereotypi-
cal traits (e.g., emotional, aggressive) and nontrait attributes (e.g., laundry, cigar) and
showed facilitation on name judgment (e.g., Jane, John). However, more complex
relationships between preconscious and postconscious effects may exist than are
currently recognized. Automatic effects of the sort we have reported (Banaji &
H a r d i n , 1996), which appear at first glance to be preconscious (in that they are not
conditional o n cognitive preparedness) may turn out not to be so. Blair and Banaji
(1996a), for example, showed that such automatic effects are susceptible to prepar-
edness i n the form of expecting to be confronted with counterstereotypes.
Studies such as these point to the power of social category knowledge i n
automatic judgment. Just as the denotative meaning of a word is automatically
activated o n presentation, as shown by the vast amount of research o n semantic
priming (Neely, 1991; Ratcliff & M c K o o n , 1988), and just as the evaluative
component of information is automatically activated on encountering an attitude
object (Fazio, Sanbonmatsu, Powell, & Kardes, 1986; Bargh, C h a i k e n , Govender,
& Pratto, 1992), the social category meaning of ordinary information whose primary
(denotative) meaning does not refer to social categories (e.g., veteran, ballet,
basketball colonial) is automatically activated o n exposure. A s Blair and Banaji
(1996a) noted, these findings are "disturbing because such processes reveal the
potential to perpetuate prejudice and discrimination independent of more control-
led and intentional forms of stereotyping ... because people may be either unaware
of the automatic influences o n their behavior or believe that they have adequately
adjusted for those influences, they may misattribute their (stereotypic) response to
more obvious or seemingly justifiable causes, such as attributes of the target" (p.
26). The importance of these findings is underscored by other findings that do not
show the automatic effects of seemingly plausible variables of automatic influence
such as word potency (see Bargh, chap. 1).
Moderators of Implicit-Automatic
Stereotyping and Prejudice Effects
Perhaps the most interesting feature of recent research on automatic social category
effects is its complexity. A l t h o u g h unconscious effects may be pervasive they are
neither unpredictable, a point Bargh makes about this entire category of effects, nor
inevitable, as our data show. In each program of research, we demonstrated
conditions under which implicit or automatic effects may or may not occur, and it
is these interaction effects that provide an understanding of just how environments
activate and provide the basis for application of social category knowledge. In the
studies that tap what Bargh calls postconscious effects, we showed that stereotyping
is crucially dependent o n activation or fluency triggered by the environment. In the
fame judgment experiments, subjects without prior exposure to names did not show
differential use of the criterion to judge male versus female fame (Banaji &
Greenwald, 1995). Likewise, Banaji et al. (1993) showed that i n the absence of
environmental triggers of abstract stereotypic knowledge, subjects did not judge a
male and female target to vary along stereotypic dimensions. In both cases, some
specific form of activation was necessary to produce the effect. However, the
potency of the stimulus required may be quite mild, and the ease with which such
triggers are available i n everyday environments leads us back to the point made i n
the previous section regarding the pervasiveness of the everyday microenviron-

ments that are ripe for producing social category effects.
In the preconscious effects of social category knowledge, too, qualifications of
the automatic activation of stereotypes are evident. Blair and Banaji (1996a)
showed that consciously imposed expectancies or intentions can moderate the
occurrence of automatic stereotype priming, especially when cognitive resources
are available to do so. Variations i n these factors (intention, availability of cognitive
resources) can produce anywhere from a reduction of the automatic stereotype
priming effect to a complete reversal of it. Environments can provide many levels
of influence o n intentions and cognitive resources. Direct and even coercive
strategies may be used to both encourage and suppress the use of social category
knowledge. But, along the lines suggested by Skinner, that environments select
courses of action, we expect that environmental triggers that encourage and reduce
the use of social category knowledge may occur without the conscious operation of
intentions and goals. N e w evidence showing that environments may select coun-
terstereotypic information leading to reduced automatic stereotype priming is
available i n Blair and Banaji (1996b).
Dissociations Between Automatic

and Controlled Processes
A m o n g the provocative findings reported i n Devine's (1989) report, one that
caught the imagination of many social psychologists was the finding that variation
i n explicitly expressed prejudice did not predict implicit stereotyping. T h e finding
has both theoretical and practical implications, and here we focus on the theoretical
aspects. In the research performed since that study was published, there were several
reports of similar findings. In our own research, we showed that subjects' explicit
gender stereotypes do not predict the extent of the false fame bias (Banaji &
Greenwald, 1995), and that attitudes toward language reform and gender egalitari-
anism not predict the automatic activation of gender stereotypes (Banaji & Hardin,
1996). These findings, as Greenwald and Banaji (1995) discussed, may parallel
findings i n research on memory showing the dissociation between explicit and
implicit forms. Such findings inevitably lead to discussions of the "separateness" of
conscious and unconscious systems, with even the term systems connoting a
fundamental segregation of these modes of thought. There is reason to be cautious
i n endorsing separate systems, i n spite of the early evidence showing dissociations
between explicit and implicit modes. First, as with other seeming dissociations i n
social psychology (e.g., that attitude and behavior were not related), more appro-
priate comparisons between explicit and implicit measures may reveal greater
concordance across measures (see, e.g., Fazio, Jackson, D u n t o n , & Williams, 1995).
A s with research o n implicit and explicit memory, the debate will need to be
more focused on the properties of the new measures that are being developed to
capture automatic and implicit processes and revisions of older measures of con-
trolled and explicit processes. Bargh's claims of separate evaluative, cognitive, and
motivational systems will need greater precision i n definition and more convincing
empirical evidence that it is indeed meaningful to speak of three separate systems.

In particular, the proposal for a separate motivational system, i n part because it has
received the least empirical attention, needs greater scrutiny. A t present, the effects
reported as support for it may more parsimoniously be accommodated within the
cognitive system.
CONCLUSION
Freedom and dignity ... are the possessions of the autonomous man of traditional
theory, and they are essential to practices in which a person is held responsible for his
conduct and given credit for his achievements. A scientific analysis shifts both the
responsibility and the achievement to the environment.
(Skinner, 1971, p. 22-23)
Causal action at a distance is difficult to perceive and identify. However, attention

to automatic social processes allows theoretical mechanisms to be specified that
show the link between features of the environment and internal mental processes.
Microlevel social environments reveal entirely new layers of social processes for
study, and here, attention to automatic social processes provide unprecedented
theoretical advantages i n understanding social behavior, i n part due to the meth-
odological and technological advances that accompany it. Bargh has provided social
psychology with some of the best examples of these advances.
O u r own work focuses on how knowledge about social groups and feelings toward
them can play an implicit and automatic role i n judgments of individual members.
Because the causes of such judgments and behavior reside at some remove from
conscious awareness and control, they can lead perceivers to be blind to their use of
such knowledge and targets to be blind to such knowledge being used in their favor or
against them (Banaji & Greenwald, 1994). Skinner was entirely wrong i n equating
explanations involving mental processes with explanations using divine intervention,
and he was also wrong i n transferring all achievement and responsibility from the
individual actor to the environment. We now know that complex interactions between
actors and their environments, when understood, can explain when and how much of
achievement and responsibility emanates from one and the other. It is an exciting
moment i n social psychology to be able to examine the role of fundamental transducers
of social action, the social groups of which we must be members. However distant their
action and microscopic their influence, they play a ubiquitous role i n the magnitude of
the responsibilities we have and the ease with which we procure our achievements.
ACKNOWLEDGMENTS
The writing of this chapter was supported by National Science Foundation Grant
SBR-9422241. We are grateful to R. Bhaskar, Nilanjana Dasgupta, Richard Hackman,
Curtis H a r d i n , Kristi Lemm, and Robert Wyer for comments o n a previous draft.
REFERENCES
Banaji, M . R., & Greenwald, A . G . (1994). Implicit stereotyping and prejudice. In M . Zanna & J. M .
Olson (Eds.), The psychology of prejudice: The Ontario symposium Vol 7 (pp. 55-76). Hillsdale, NJ:
Banaji, M . R., & Greenwald, A . G . (1995). Implicit stereotyping in judgments of fame. Journal of
Personality and Social Psychology, 68, 181-198.
Banaji, M . R., & Hardin, C . (1996). Automatic gender stereotyping. Psychological Science, 7, 136-141.
Banaji, M . R., Hardin, C . , & Rothman, A . (1993). Implicit stereotyping in person judgment. Journal of
Bargh, J. A . , Chaiken, S., Govender, R., & Pratto, F. (1992). T h e generality of the automatic attitude
Blair, I. V , & Banaji, M . R. (1996a). Automatic and controlled processes in stereotype priming. Journal
Blair, I. V., & Banaji, M . R. (1996b). The effect of exposure to counterstereotypes on automatic stereotyping.
Unpublished manuscript, University of Colorado.
Dovidio, J. F., Evans, N . , & Tyler, R. B. (1986). Racial stereotypes: T h e contents of their cognitive
representations. Journal of Experimental Social Psychology, 22, 22-37.
Fazio, R. H . , Jackson, J. R., Dunton, B. C . , & Williams, C . J. (1995). Variability in automatic activation
as an unobtrusive measure of racial attitudes: A bona fide pipeline? Journal of Personality and Social
Psychobgy, 69, 1013-1027.
Fazio, R. H . , Sanbonmatsu, D. M . , Powell, M . C . , & Kardes, F. R. (1986). O n the automatic activation
of attitudes. Journal of Personality and Social Psychobgy, 50, 229-238
Freud, S. (1965). The psychopathobgy of everyday life. New York: Norton.
Gaertner, S. L., & McLaughlin, J. P. (1983). Racial stereotypes: Associations and ascriptions of positive
and negative characteristics. Social Psychobgy Quarterly, 46, 23-30.
stereotypes. Psychobgical Review, 102, 4-27.
Jacoby, L. L., & Kelley, C. M . (1987). Unconscious influences of memory for a prior event. Personality
and Social Psychobgy Bulletin, 13, 314-336.
Jones, E. E. (1990). Interpersonal perception. New York: Freeman.
Jones, E . E., & Gerard, H . B. (1967). Foundations of social psychobgy. New York: Wiley.
Latane, B., & Darley, J. M . (1968). Group inhibition of bystander intervention in emergencies. Journal
Lewicki, P., & Hill, T. (1987). Unconscious processes as explanations of behavior in cognitive, personality,
and social psychology. Personality and Social Psychobgy Bulletin, 13, 355-362.
Milgram, S. (1963). Behavioral study of obedience. Journal of Abnormal and Social Psychobgy, 67,
371-378.
Neely, J. H . (1991). Semantic priming effects in visual word recognition: A selective review of current
findings and theories. In D. Besner & G . Humphreys (Eds.), Basic processes in reading: Visual word
recognition (pp. 264-336). Hillsdale, NJ: Lawrence Erlbaum Associates.
Newton, I. (1672). Letter to Henry Oldenburg, February 6, 1672. In I. B. Cohen & R. S. Westfall (Eds.),
Newton (pp. 171-181). New York: Norton.
Ratcliff", R., & McKoon, G . (1988). A retrieval theory of priming in memory. Psychobgical Review, 95,
385-408.
Ross, L . D . (1977). T h e intuitive psychologist and his shortcomings: Distortions in the attribution
process. In L. Berkowitz (Ed.), Advances in experimental social psychobgy (Vol. 10, pp. 176-221). New
York: Academic Press.
Skinner, B. F. (1971). Beyond freedom and dignity. New York: BantanvVintage.
Chapter 3
Consciousness, Free Choice,

and Automaticity
Roy E Baumeister
Kristin L. Sommer
Case Western Reserve University
T h e increased understanding of automatic processes fundamentally influenced and

altered social psychology's view of human nature since 1985. In John Bargh's target
chapter, he shows why he is a leader i n illuminating these processes. H i s elegant
reasoning and innovative experiments shed considerable light o n how motivational
and cognitive processes alter people's behavior with often little or no conscious
awareness that they are being affected.
A l t h o u g h we have no quarrel with Bargh's procedures, data, or specific inter-
pretations of research findings, we do wish to question one overarching theme of
his work. Parts of his chapter, particularly the beginning and end, suggest that the
understanding of automatic processes may eventually take over psychology to the
extent that conscious processes and deliberative choice become outdated, super-
fluous concepts. In his words, "it may well be that there ultimately is no future for
conscious processing i n accounts of the mind, i n the sense of free will and choice"
(chap. 1, p. 52). In our view, such a conclusion requires a drastic leap of faith that
goes far beyond what the data warrant. Beyond that, we want to propose a different
understanding of the role of conscious processes i n human behavior. Bargh may
have trouble finding evidence of the effects of consciousness because he is looking
in the wrong place.
Specifically, we propose that the role of consciousness is to override automatic,
habitual, or standard responses o n the infrequent occasions when such intervention
is needed. Consciousness thus undermines the lawful, predictable nature of human
behavior and produces a situation of relative indeterminacy. S u c h an approach
allows us to treat Bargh's contributions as vital keys for achieving a new, expanded
view of human nature and mental functioning—but nonetheless a slightly different
view than the one he suggests i n his chapter.
75
76 Baumeister & Sommer
WHAT CONSCIOUSNESS DOES NOT DO
O n e important predecessor of Bargh's approach to consciousness was Julian Jaynes.

Jaynes devoted much of his life to the study of consciousness, and i n his main work
(1976), he concluded from a literature review that consciousness was not necessary
for thought, perception, judgment, learning, memory, language, concepts, and
behavior. A careful interdisciplinary study of ancient history and literature led h i m
to the further conclusion that the members of early civilizations were not conscious,
at least not i n the modern sense of the term. They were certainly awake and
processing information, but their behavior mainly followed patterns that Bargh
would label automatic and ballistic. W h e n a choice point was reached, the stress
of choosing would cause the person to experience an auditory hallucination, w h i c h
people interpreted as being the voice of a god, and which told the person what to do.
Early literature such as the Iliad, for example, did not depict human characters
as conscious beings with inner lives or thought processes. The human characters of
the Iliad were never described as deciding anything, nor do they ponder, scheme,
regret, worry, or the like. A t each decision point, a god stepped i n and told the
character what to do. A s Jaynes pointed out, a literal reading of the Iliad indicates
that the gods directed the Trojan War and were the source of all volition. T h e
human characters simply did their bidding, not unlike puppets.
In contrast, the Odyssey depicted a very different mentality, and Jaynes sub-
scribed to the view that it was originally written several centuries after the Iliad.
Odysseus did have an inner life. H e made plenty of his own decisions, often after
inner processes of planning and pondering. H e was not depicted as a puppet or
servant of the gods.
For present purposes, the point is that Jaynes' argument dovetails with Bargh's
analysis of the superfluity of consciousness. Prior to 2000 B C , according to Jaynes,
there were cities, empires, and large civilizations, complete with complex social
environments, writing, economies, technological innovation, and authority struc-
tures (in which gods played a prominent role, if one takes their writings literally).
Yet Jaynes wrote that none of the people were conscious. Hence, almost the
complete range of human social behavior is possible without consciousness. "Social
psychological phenomena are essentially automatic," writes Bargh (chap. 1, p. 3),
whereas "conscious choice is not necessary for an effect" (chap. 1 , p. 4).
Bargh, if anything, understates the case here. Consciousness is not only unnec-
essary for many psychological effects, sometimes it is even counterproductive. O n e
of the first demonstrations of how consciousness interferes with the lawful progress
of behavior was by Hefferline, Keenan, and Harford (1959). They employed a
conditioning paradigm i n which human subjects were reinforced for a subtle muscle
movement i n the hand. Subjects who were unaware they were being conditioned
at all showed the fastest learning curves. Those who were told i n a vague way that
they were being conditioned showed somewhat slower learning. Those who were
told specifically to try to learn the particular response that was being reinforced
showed the slowest learning. Thus, the standard principles of behavior (in this case,
3. Consciousness, Free Choice, Automaticity 77
learning by reinforcement) operated most smoothly and effectively i n the absence

of consciousness. T h e more conscious people were of what was going on, the less
predictable and lawful was their behavior.
CONSCIOUSNESS AS OVERRIDE MECHANISM
T h e implication of Hefferline et al.'s (1959) findings is that consciousness interferes

with the lawful, predictable patterns of behavior and prevents behavior from
conforming automatically to standard principles. This implication suggests a func-
tional role for consciousness quite different than the one Bargh attacks, namely as
an essential mediator of human behavior. A s researchers from Hefferline to Jaynes
to Bargh showed, consciousness is not an essential mediator of human behavior
because behavior can occur i n elaborate, lawful, and predictable patterns without
consciousness. Instead, we suggest that the function of consciousness is precisely
the opposite: It overrides those lawful and predictable patterns.
Bargh may be quite correct i n suggesting that a great deal, even the vast majority,
of human behavior occurs i n automatic fashion with little or no need for conscious-
ness to mediate. Instead of asserting, however, that consciousness is wholly irrele-
vant and that all behavior will eventually be shown to be automatic, we suggest that
consciousness does guide a small but very important minority of human behavior.
To put it another way, we (unlike some other advocates of consciousness) readily
concede Bargh's assertion that a great deal of human behavior can dispense with
consciousness because it flows smoothly along automatic pathways, guided by
habits, past experience, established goals and motivations, and situational cues.
Consciousness does, however, intervene occasionally to take behavior out of those
rutted pathways. It can break a link i n the causal chain that normally leads straight
from motivations and activating cues to behavioral responses.
W h e n a hungry person sees food, the behavioral outcome is likely to be eating.
Consciousness is hardly necessary to mediate that response, which is found i n all
manner of other animal species whose claims of consciousness are tenuous. T h e
conscious human being can resist that temptation, based perhaps o n a dieting
resolution and a calculation of the day's caloric intake. Such conscious overrides
may account for only a small part of the total variance among all the encounters
between hungry people and available food i n human history, but the part is
nonetheless important.
T h e capacity to override automatic or well-learned response patterns would be
highly adaptive. A s Bargh (1982) noted, automatic response patterns tend to be
inflexible. Similarly, Jaynes (1976) suggested that compromise, adaptation to
changing circumstances, and deception were quite difficult before human beings
became conscious. Baumeister, Heatherton, and Tice (1994) found that the ability
to override one's unwanted responses is linked to health, happiness, and success
i n a variety of spheres. Consciousness can greatly increase the h u m a n being's
ability to survive and flourish by overriding established, automatic response
patterns when they are not optimal.
In particular, consciousness is especially valuable for coping with a highly

complex social reality and for dealing with novel, rapidly changing situations. A s
long as life remains fixed amid stable relationships and a fixed routine, automatic
response patterns may be quite sufficient to get by, and consciousness is not needed.
But i n a complex social world where relationships wax and wane or are replaced,
where interactions with strangers or other new partners are important and frequent,
where rapidly changing circumstances require frequent departures from routines
and familiar procedures, and where norms and expectations are subject to change,
consciousness will be invaluable because the automatic response patterns cannot
keep up.
Conscious overrides are especially important when changing circumstances alter
the desirability of certain responses. A given type of response may initially be the
optimal one, and so the person may learn to respond that way automatically. T h e n
social norms change, and the response is no longer desirable or acceptable. C o n -
sciousness then becomes quite useful and valuable to override that automatic
response.
Sexual harassment provides a useful example. Bargh's target article describes
some of his own work showing that some men have automatic associations between
power and sex, so that they may automatically feel sexual or romantic attraction
toward female subordinates. T h e behavioral response of making romantic advances
toward such women may then follow i n an automatic or natural fashion. In some
social settings i n the history of the world, the right to make such advances and enjoy
their outcomes was considered to be one of the prerogatives of power, and so it
would make sense for men to learn to act that way. Recent developments i n our
own society contradicts any such right and insist that subordinates must not be
subjected to such pressures. Powerful men (and women) must therefore curtail the
tendency to make sexual advances toward subordinates i n order to avoid social
stigma, occupational sanctions, and possible legal penalties.
Recently i n the news, a European film director's brief Hollywood career was
marred by a sexual harassment complaint. Apparently, he had remarked to a female
assistant that she had lovely breasts. "In Europe, that is still considered a compli-
ment," he sniffed o n his way back to the O l d World. W h e n changing norms can
convert a compliment into a legally punishable insult, it is not safe to rely solely o n
automatic response patterns.
Combating prejudice provides another example. Stereotypes and the derogation
of outgroups are found all over the world, but modern U.S. culture has promoted
the desirability of overcoming prejudice as a positive, powerful value. Bargh's target
chapter cites research by Devine (1989) suggesting that prejudicial responses are
often automatic i n everyone, and so prejudice is defeated by a conscious override
of such responses. To us, this suggests precisely the function of consciousness we
are espousing: It overrides and alters responses that may be automatic but have
become undesirable.
It is conceivable that with sufficient practice the nonprejudiced response may
become the automatic one. Bargh's chapter insightfully proposes that conscious
responses can serve a vital role i n enacting responses that have not become
automatic but that can gradually become automatic as the person learns and
overlearns them. We agree. Still, Bargh's analysis presents consciousness as essen-
tially preceding automaticity, whereas we think that insight should be augmented
with the complementary one i n which automaticity is already i n place and con-
sciousness comes along later to alter and override.
A more grandiose formulation of our view would be to say that consciousness is
a state of relative indeterminacy. Consciousness creates a hole or gap i n the
deterministic web of causal relationships that shapes human behavior, so to speak.
U n d e r familiar, comfortable circumstances, certain causes lead smoothly to certain
behavioral responses. Consciousness can disrupt and alter those connections,
thereby disengaging behavior from its usual causes.
There are two different ways i n which such a mechanism could be beneficial to
the human organism. O n e is simply that indeterminacy reduces predictability.
Throughout m u c h of history, human beings have been at the mercy of various
predators, including other human beings. If all consciousness does is disengage
standard causal responses so as to introduce an element of randomness into human
behavior, that would have value i n terms of foiling predators who may be stalking
one and trying to anticipate one's movements. Randomness may also foster crea-
tivity, allowing novel responses to emerge that can be reinforced and learned if they
produce good outcomes.
The other way i n which indeterminacy could benefit the human being is that it
can allow behavior to follow from wise, prudent choices that take the full range of
opportunities and threats, including novel contingencies, into account. By defini-
tion, automatic behavior is not based o n a thoughtful analysis of all the subtle
possibilities that exist i n a given situation. Such an analysis can only guide behavior,
however, if the automatic response is prevented (or at least held i n check until one
can decide whether it is the best response). In other words, consciousness can stop
people from responding like rats i n Skinner boxes—and thereby allow people to
take advantage of the power of human reasoning when selecting the best or most
desirable response.
A g a i n , we do not think that the majority of human behavior conforms to this
model involving conscious choice and reasoned analysis. Most behavior may indeed
be automatic. But consciousness can have powerful, valuable, even life-saving
effects even if it overrides the automatic response patterns only once i n a great
while.
THE FALSE DICHOTOMY OF ALL OR NOTHING
There is a rich intellectual tradition that framed the debate about consciousness i n
the terms i n which Bargh addresses it, and i n a sense our quarrel is more with that
tradition than with Bargh's own stand i n it. A c c o r d i n g to this tradition, human
behavior is either entirely dependent o n conscious choice and free will, or it is
entirely a lawful, predictable result of firm causal processes and principles i n a
mechanistic, deterministic fashion. We think this is a false dichotomy and that the
truth is somewhere i n between those extreme positions.
Sartre (1943/1956), for example, argued passionately that all human behavior
is inevitably free, and that people could always have acted differently than they did.
Yet his arguments invariably depended o n consciousness. Thus, i n a famous exam-
ple, he described a man who grows tired while walking and sits down, saying that
he cannot walk another step. Sartre said that probably the man could indeed have
walked another step, and he could also have stopped sooner rather than at that
exact point. We agree that he could probably have stopped elsewhere; but that does
not mean that his stopping there was a conscious product of free will. Instead, it
seems more likely that the automatic processing of inner cues pertaining to fatigue
and other factors prompted h i m to stop when the cues reached a certain criterion,
and so he automatically stopped there. To do otherwise would have required
consciousness to override that automatic decision process and insist on, say, walking
the additional distance to the next campsite. But i n Sartre's example, consciousness
did not override the automatic response, and the man sat down right there.
In the context of the all-or-nothing tradition, Bargh's reasoning is perfectly
sound. Thus, the traditional terms of the debate stipulate that either all behavior
is free and conscious, or all is automatic and determined. Bargh shows that some
behavior is automatic and able to occur without conscious mediation, which
disproves the one possibility (i.e., that everything is conscious). H e is therefore left
with the other conclusion, namely that consciousness is irrelevant, and he antici-
pates that psychology may eventually find that all social behavior is automatic. H e
notes correctly that the trend i n recent research findings was to curtail the sphere
of conscious choice and expand the sphere of automatic response, and this trend
does point toward an eventual future i n which the conscious sphere disappears
entirely and automaticity reigns supreme.
If one rejects the all-or-nothing terms of the debate, however, as we suggest, then
the extreme conclusions do not follow from the available evidence. It is no longer
safe to show that some behavior is automatic and then conclude that all behavior
is always automatic (a conclusion that is however correct and reasonable under the
all-or-nothing rule). We are quite ready to concede that some behavior is automatic,
and we suspect that most of it is. But not all.
People do have the conscious experience of making free choices o n a fairly
regular basis, and so to argue that all such impressions are mistaken would require
fairly powerful and extensive evidence—or must remain a leap of faith. N o research
findings justify such a leap; indeed, one could say that the research evidence
contradicts the behavior is fully determined by prior causes. After all, many
thousands of psychological studies relentlessly failed to achieve the deterministic
ideal of 100% prediction of human behavior, even i n the controlled and circum-
scribed sphere of laboratory experimentation.
Frankly, we cannot understand how any psychologist can remain a strict deter-
minist after reading the journals, unless motivated by such an extreme blind faith
as to remain impervious to evidence. T h e causality of psychological research
findings is almost always probabilistic, not deterministic, and probabilistic causation
entails some degree of indeterminacy. If both everyday direct human experience

and the great mass of research findings point to the existence of some margin of
indeterminacy i n human behavior, then it deserves to be acknowledged i n psycho-
logical theories.
In this comment, therefore, we suggest consciousness as a major source of this
indeterminacy. Automaticity is indeed adaptive, and Bargh's findings certainly
persuade us that the sphere of automatic response is even larger than previously
thought. They do not, however, persuade us that sphere contains a l l human
behavior. A small but very important portion of human behavior depends o n
conscious, free choice, particularly when consciousness overrides automatic re-
sponse patterns that happen to be undesirable or unsuitable due to novel or
changing circumstances.
REFERENCES
Bargh, J. A . (1982). Attention and automaticity in the processing of self-relevant information. Journal
of personality and Social Psychology, 43, 425—436.
Baumeister, R. F., Heatherton, T F., & Tice, D . M . (1994). Losing control: How and why people fail at
self-regulation. San Diego, C A : Academic Press.
Devine, P G . (1989). Stereotypes and prejudice: Their automatic and controlled components. Journal
Hefferline, R. F., Keenan, B., & Harford, R. A . (1959). Escape and avoidance conditioning in human
subjects without their observation of the response. Science, 130, 1338-1339.
Jaynes, J. (1976). The origin of consciousness in the breakdown of the bicameral mind. Reading, M A :
Addison-Wesley.
Sartre, J.-R (1956). Being and nothingness (H. E. Barnes, Trans.). Secaucus, NJ: Citadel Press. (Original
work published 1943).
Chapter 4
Some Thoughts Extending

Bargh's Argument
Leonard Berkowitz
University of Wisconsin at Madison
Bargh's paper is very important, for me at least, and I believe also for social
psychology and the social sciences. Indeed, I am so taken with his line of reasoning
that I would extend Bargh's argument even further and i n a variety of directions.
M y own theoretical bias should be acknowledged from the start: Bargh's thesis
is very m u c h i n accord with my research and theorizing regarding aggression. For
more than 30 years, I repeatedly argued that many assaultive actions are, to a
considerable degree, impulsive (i.e., automatic) responses to certain features i n the
immediate situation. M y research concentrated o n situational influences (in keep-
ing with social psychology's traditional focus, as Bargh points out), but also
emphasized, i n fundamental agreement with Bargh, that the strength or target of
an attack, and sometimes even the intention to aggress, are governed largely "by
current features of the environment" and are "not mediated by conscious choice or
reflection" (p. 2). This formulation was a version of what Bargh terms an auto-motive
model (chap. 1, p. 29) i n that it suggested (see Berkowitz, 1993a) that environmental
stimuli c a n activate an aggressive goal at times "without any need for conscious
decision-making" (chap. 1, p. 30).
O f course, Bargh does more than repeat and extend my own line of thought. H i s
formulation of automatic processes is both more sophisticated than the conception
I employed i n most of my writings and also more ambitious than anything I
attempted. Whereas much of my thinking, especially i n the 1960s and 1970s, was
guided by the H u l l - S p e n c e behavior-theoretic perspective, Bargh's discussion of
automatic processes is more up-to-date and more differentiated, as well as more
precise, than was mine. M y analysis of the Wisconsin aggression experiments
undoubtedly would have been sharpened if I had noted more explicitly that the
impulsive aggression displayed i n many of these studies was a case of what Bargh
calls goal-dependent automaticity. T h e participants typically were required to
83
84 Berkowitz
punish the available target to some degree, and the situational stimuli being
investigated usually governed only the intensity of this punishment. M y focus was
also m u c h narrower than Bargh's i n that I was concerned almost entirely with
affective reactions (e.g., Berkowitz, 1993a, 1993b), whereas Bargh audaciously
maintains that a variety of socially relevant thoughts, feelings, and actions are fairly
automatic i n nature and independent of conscious decision-making, whether these
reactions are affectively charged. I am sympathetic to his general position.
IMPLICATIONS FOR EXPERIMENTAL SOCIAL PSYCHOLOGY
I especially share Bargh's conception of the nature of social psychology. M o r e than

anything else, as he notes, this discipline seeks to "discover [the] situational causes
of thinking, feeling, and acting i n the real or implied presence of other people"
(chap. 1, p. 1). A s obvious as this concern with situational influences undoubtedly
is to the great majority of experimentally oriented social psychologists, i n my view
it is not sufficiently appreciated by those who are less taken with laboratory research,
whether they are lay persons, social scientists, or even other psychologists. N o t fully
recognizing the degree to which people can be affected by features of the setting,
they regard experimental social psychological findings as trivial (or even wrong)
and give them scant attention.
A good example can be found i n the widespread indifference to the results
obtained i n scores of experiments investigating the short-term effects of television
and film violence. A l t h o u g h this research was remarkably consistent i n demonstrat-
ing that the sight of other people fighting often increases the chances that the
viewers will be aggressive themselves for a limited time (see Berkowitz, 1984,1993a;
Wood, Wong, & Chachere, 1991), the general public and many social scientists
appear reluctant to accept these findings. For them, if the mass media have any
influence at all o n antisocial conduct, it is through the lessons they teach m u c h
more than through their transient effects o n what audience members might do and
think soon after they encounter the violent scene.
This apparent indifference to experimental social psychological research is not
simply a matter of inattention and lack of knowledge. M o s t of our cousins i n the
other social sciences assume as a matter of course that human behavior results from
conscious decision-making. Guided by this overriding supposition, they find it
difficult to believe that human feelings and actions can be controlled automatically
by situational details, that is, "without any conscious processing of information"
(chap. 1, p. 3) and conscious decision-making. T h e most popular accounts of
criminal conduct i n sociology and criminology are a good case i n point. These
analyses hold that people consider the costs and benefits that might arise from each
of the various actions they could undertake at the time, and then choose the
alternative they think is likely to produce the best possible outcome o n that occasion
at the lowest possible cost.
4. Extending Bargh's Argument 85
T h e W i l s o n and Herrnstein (1985) conception of criminality, i n their important

book Crime and Human Nature, is illustrative. A l t h o u g h W i l s o n and Herrnstein
assumed that people's decisions are rooted i n their earlier classical and instrumental
conditioning, they contended that criminal conduct is rational from the offenders'
point of view; i n virtually all instances the criminals think the rewards their illegal
actions will bring them exceed the costs that might be incurred. These writers also
stated that many criminally inclined persons are especially apt to discount the
possible negative consequences that could arise i n the relatively distant future.
Crimes of violence, for W i l s o n and Herrnstein, were basically similar i n important
respects to other illegal behaviors, and they tended to dismiss the notion of
impulsive actions. A violent assault impelled by intense rage, they claimed, "is no
more irresistible than cheating on one's income tax; it could have been suppressed
by a greater or more certain penalty" (p. 56).
Proponents of this type of analysis obviously do not subscribe to Bargh's line of
thought. They also fail to recognize that an impulse to action arises automatically i n
a variety of emotional states. A growing number of researchers (e.g., Frijda, 1986;
Roseman, Wiest, & Swartz, 1994; Tomkins, 1962) maintain that the arousal of some
emotions, such as fear and anger, activates a fairly specific motor program. Thus, we
have an urge to flee when we are afraid and presumably have the motor impulses to
attack someone when we are angry.
Getting closer to Bargh's general thesis, some of the opposition to laboratory
experimentation, even by psychologists, may stem from an essentially similar reluctance
to believe that people's actions are often fairly automatic responses to features of the
surrounding situation. Critics often contend that experiments give rise to an erroneous
picture of human nature. They say that laboratory studies frequently portray people
falsely as puppets who are unthinkingly affected by situational stimuli and not as they
truly are, rationalistic decision makers acting on the basis of their conscious under-
standing of the setting (see Berkowitz & Donnerstein, 1982). Rejecting the possibility
of automatic influences, they attribute any laboratory demonstrations of such effects to
artifacts supposedly inherent in experiments and not to human qualities.
I heartily agree with Bargh's general position. Persons interested i n gaining a truly
adequate understanding of the complexities of human conduct should at least adopt
a healthy skepticism toward the assumption that conscious processes are necessarily
involved i n all human behavior.
SOME IMPLICATIONS AND QUESTIONS
Bargh's stimulating chapter also has implications for social psychological theoriz-
ing (although I am not at all sure Bargh would agree with me, here). Paraphrasing
Bern's (1972) critique of dissonance and attribution theorizing, Bargh notes that
it is not at all uncommon i n social psychology to infer the operation of conscious
processes without having any evidence that these processes did indeed occur. I
wonder if the same basic point could not also be made about other kinds of
theorizing i n contemporary social psychology. Consider the now popular notion of
86 Berkowitz
heuristic processing. It is generally assumed that when people engage i n this relatively
effortless and not-especially-thoughtful information processing, they employ a
highly accessible "rule of thumb" i n judging or interpreting external occurrences
(see C h a i k e n , 1980). If they are feeling good at the time, let's say because they have
just finished eating a delicious meal, they might arrive at a favorable judgment of
an appeal made to them soon afterwards because, the idea goes, they processed the
message only heuristically. A n d so, if we follow Schwarz and Bless (1991), they
might use a "how do I feel about it?" heuristic. Presumably attributing their pleasant
mood to the received message (Schwarz & Clore, 1983), they supposedly think that
if they're feeling good, the appeal made to them must be good.
Bargh (and Bern) objects to the inferences made about conscious processes i n
traditional dissonance and attributional theorizing because these inferences were
unparsimonious and lacked empirical support. The same kind of objection could be
leveled against the notion of heuristic processing i n interpretations such as the one
just summarized. Social psychologists often uncritically employ this concept because
it is consistent with their general theoretical perspective and not because its usage
is clearly empirically warranted. Just as the early attribution experiments had little
direct evidence that people's causal analyses are based on the complex reasoning
conjectured by Kelley (one of Bargh's points), so do social psychological studies
employing the idea of heuristic processing rarely present direct evidence that the
participants actually thought of and used the assumed rules of thumb.
In some instances the observed findings can be understood i n other terms. For
example, years ago, when associationism was more dominant i n psychological
theorizing, researchers typically interpreted the persuasive or judgmental effects of
affective experiences using associationistic constructs (e.g., Griffitt, 1970; Razran,
1938). A n d so, i n the illustrative case just mentioned, they would say the subjects
accepted the persuasive message after they had eaten the delicious meal because
the resulting pleasant affect had generalized to the communicator or the message.
This affect generalization interpretation may be a more parsimonious explanation
of the attitude change than the explanation resting on untested inferences about
mediating heuristics (whether the conjectured rules of thumb are conscious or not).
This question about the parsimony of some kinds of contemporary social
psychological theorizing brings up a minor matter i n Bargh's discussion that puzzles
me. This has to with the distinction between preconscious and postconscious
automaticity. Preconscious processes theoretically "require only the proximal reg-
istration of the stimulus event to occur" (chap. 1, p. 6), although i n this "registra-
t i o n " the external stimulus is said to be transformed preconsciously so that it loses
its objective quality and becomes the event as understood (chap. 1, p. 6). In the
case of postconscious processes, on the other hand, a mental process apparently is
only "prepared" and supposedly does not occur unless other, "triggering environ-
mental information" is also present (chap. 1, p. 3). Bargh held that this latter type
of processing "is commonly studied" employing priming procedures (chap. 1, p. 3).
This observation, as well as other statements i n the chapter, seems to say that
priming effects are postconscious.
Perhaps because I am not sufficiently familiar with the priming literature and
with Bargh's other writings, I am confused as to why priming effects are said to be
the result of postconscious processing, whereas the principle of ideomotor action
and the operation of the auto-motive model are discussed as instances of precon-
scious processing. H i s observation that "priming is an excellent technique for
experimentally manipulating automatic goal activation and operation" (chap. 1,
p. 35), and his statement that "once a construct has been primed or recently used"
it may have some effects that "are not distinguishable from . . . preconscious effects"
(chap. 1, p. 12) add to my uncertainty.
Other considerations also lead me to question his seeming characterization of
priming effects as postconscious effects i n the target chapter. To repeat what was
said just before, his conception (widely shared by cognitivists) proposes that the
typical demonstration of a priming effect arises i n two steps: First, the initial
registration of the external event heightens the accessibility of a particular concept
or category (prepares this concept for use), and then, second, other triggering
information presumably then puts the prepared concept into overt use (to apply
one of Bargh's terms here, this is a kind of serial stage model). Is it not possible that
this postulation of a two-step process is unparsimonious i n some cases, and that i n
these instances the ideas produced by the external priming event might activate
semantically related motor impulses and ideas directly, independently of the sup-
posed triggering information? Bargh's conception of an auto-motive process and
the principle of ideomotor action suggest that this kind of direct effect occurs. If so,
the distinction between preconscious and postconscious effects, w h i c h is not
sharply drawn i n this chapter, becomes blurred.
T h e findings i n the second Bargh and C h e n experiment reported i n the target
chapter also suggest the kind of direct influence I have i n mind. In this study (chap.
1, pp. 18-19) the participants primed with words related to the stereotype of elderly
people tended to walk more slowly soon afterwards than did their counterparts who
had been exposed to other kinds of words. C o u l d the thought of elderly persons and
(presumably) their typical manner of walking have done more than increase the
accessibility of a faltering gait, but had, instead, affected the participants' motor
responses directly?
BRINGING IN OTHER IDEAS AND FINDINGS
Bargh presents a truly impressive body of research findings i n support of his line of
reasoning. Quite a few other ideas and research results i n the psychological
literature are also relevant, and i n the remainder of my commentary I call attention
to these other matters that Bargh (and others) might want to consider i n future
investigations.
Evidence From Emotions Research

Research i n the emotions domain is especially supportive of Bargh's basic argument.
Because of space limitations, I mention only a few investigations of the many that
88 Berkowitz
could be cited. Studies employing the Velten mood induction procedure testify to
how depression-related and elation-related thoughts can produce moods i n accord-
ance with these ideas and also affectively consistent, subtle behavioral changes i n
eye-contact patterns, and even i n hand movements and facial muscular activity
(see Berkowitz & Troccoli, 1986). Other experiments indicate that bodily move-
ments characteristic of particular emotional states can influence memory (Laird,
Cuniff, Sheehan, Shulman, & Strum, 1989), cognitions (Keltner, Ellsworth, &
Edwards, 1993), and even task-related behaviors (Riskind & Gotay, 1982) i n
keeping with these states.
I (e.g., Berkowitz, 1990, 1993a, 1993b) adopted a cognitive-neoassociationistic
approach i n interpreting results such as these, basically proposing, i n accord with
Bower's (1981) analysis of mood effects o n memory, that emotional states c a n
profitably be regarded as associative networks. T h e activation of any part of the
network—for example, by the performance of an emotion-related bodily move-
ment—tends to activate other network components (such as cognitions, memories,
and expressive-motor reactions) i n proportion to the degree of association among
these network parts. From this perspective, I suggest that Bargh would do well to
give more attention to associative processes than he now does. I say more about
this shortly.
Aggression Research
Situational Features Priming Aggressive Reactions: The "Weapons Effect."

Experimental investigations of aggressive behavior (in addition to the one study by
Carver et al., cited by Bargh) provide still more evidence of automatic reactions.
Some of these experiments have to do with the weapons effect first identified by
Berkowitz and LePage (1967). A s an example of what Bargh termed goal-dependent
automaticity, angered participants i n this early study assaulted their tormentor more
severely when guns were nearby than when neutral objects were present. T h e sight
of the weapons had apparently primed aggression-facilitating cognitions and motor
tendencies, thereby increasing their aggressive intentions.
A l t h o u g h there were some failures to replicate the original finding, other
experiments i n the U n i t e d States and i n other countries corroborated the Berk-
owitz and LePage results (see Berkowitz, 1993a). Going beyond goal-dependent
automaticity, some of these studies demonstrated that the mere sight of guns can
evoke aggressive reactions from nonangered persons as long as their inhibitions
against aggression are relatively weak (see Berkowitz, 1993a). In these instances,
the priming seems to have an auto-motive influence (as Bargh would put it),
producing an aggressive urge.
Zuzul (see Berkowitz, 1993a) recently obtained evidence along these lines i n an
interesting experiment with children i n Croatia. After the youngsters were exposed
either to a real gun or a toy gun or to no weapon, an adult spoke to the children,
making some remarks that were either positive or negative or neutral to guns. T h e
children's free play soon afterwards was affected by these variations. They were
especially likely to fight and hit each other (realistically rather than playfully) after
they were exposed to the sight of weapons, but m u c h more so if the adult was not
critical of guns than if he or she disapproved of these weapons. T h e gun-produced
priming effect was stronger if the youngsters were frustrated immediately before
their free play period than if they were not so thwarted, but even the nonfrustrated
children showed indications of a weapons effect.
Other Aggression-Facilitating Associations: Target Characteristics. In my

analyses of impulsive aggression a generation and more ago (e.g., Berkowitz, 1964),
I often emphasized that the strength of the exhibited attack could vary with the
available target's characteristics. Simply put, my argument was that people who
were associated with aggression for one reason or another would automatically elicit
aggressive reactions from others, but especially from those having aggressive incli-
nations.
A number of experiments conducted i n my laboratory during the 1960s and
1970s yielded results supporting this formulation, but I here cite only two studies
investigating the influence of observed aggression. In the earliest of these (summa-
rized i n Berkowitz, 1964) carried out at a time that boxing was an officially sponsored
collegiate sport, one of the two experimenters posing as a graduate student intro-
duced himself either as interested i n films or as a member of the University boxing
team. This first experimenter then insulted the subject i n one half of the cases and
treated the other participants i n another manner. Immediately afterwards, the
second, neutral experimenter showed the participant a brief movie that portrayed
either the travels of M a r c o Polo or a prize fight, and asked the subject to evaluate
the film. T h e naive student was also asked to provide confidential ratings of each
of the two experimenters, supposedly i n connection with a survey assessing student
opinions of experiments.
Figure 4.1 summarizes the results, using as our hostility measure the difference
between the evaluation of the first experimenter when he had been insulting as
compared to when he had acted i n a neutral fashion. A s the chart indicates, the
provocateur was rated somewhat more unfavorably after the aggressive film than
after the travel movie, but he received the most unfriendly evaluation if he was also
identified as a college boxer. This target's semantically linked connection with the
aggression-priming boxing film (and perhaps with aggression generally) apparently
intensified the aggressive inclinations he evoked with his insults.
Other studies extended this formulation by proposing that people previously
associated with rewarded aggression would tend to draw the strongest attacks from
those who were aggressively primed by seeing a violent movie. O n e of several
experiments testing this possibility was published by G e e n and myself (Geen &
Berkowitz, 1966). Each male participant was first introduced to the experimenter's
accomplice posing as a fellow subject, and was led to believe this individual's name
was either Bob Kelly or Bob Dunne or Bob Riley. In all cases, the accomplice then
provoked the participant by giving his performance on an assigned task an unfavor-
90 Berkowitz
DIUeremee from same bUI neulral E
10 8.6
8 10 10 10
6
o
Fight Movie Travel Movie
Nature of Movie Seen
_ P as Boxer ~ P as Film Student
Data adapted from Berkowitz, 1964.

(Scores are amount above ratings of same but neutral exptr in same condition).
FIG. 4.1. Hostility toward the provocateur as influenced by his associations with the fight
movie and/or aggression.
able evaluation. After this, the subject watched either the same prize fight scene used
in the previously mentioned study or a film of an exciting track race. The provocateur
named Kelly was semantically connected with the fight loser (who received a bad
beating) because that character in the prize fight movie was also called Kelly. When
the accomplice was introduced as Dunne, however, he was associated with the fight
victor in the aggressive film because the fight winner in the movie had the same
name. No movie character was called Riley. At the end of the film, the subject had
an opportunity to punish the accomplice with electric shocks, supposedly as his
judgment of how well the accomplice had done on his assigned task.
Figure 4.2 shows that the aggressive movie led to a significantly greater number
of shocks delivered to the provocateur target than did the neutral film only when
that person had the same name as the victim of the observed aggression. His
name-linked association with the individual the subjects saw being beaten up
strengthened the subjects' urge to hurt him.
These two observed violence experiments, as well as others, indicate that a truly
comprehensive account of the auto-motive effects of priming experiences would do
well to consider more than the priming itsel£ Such a formulation should also deal
more specifically with the nature and degree of association between the priming
material and significant features of the external situation.
Bargh does, of course, refer to associative influences. However, his consideration
of these influences is too limited in that it is confined to differences in associative
strength based on differences in frequency of conjunction. Associations can also
vary along other dimensions, such as on the basis of the stimuli's physical and
semantic similarity or even their degree of psychological relationship. Needless to
say, a substantial body of research is relevant here, but I confine myself to one
experiment concerned with Miller's (1948) classic stimulus generalization model of
internal conflict.
As was once well-known, Miller's model attempted to account for the target of
displaced aggression in terms of such concepts as the conflict between approach
and avoidance tendencies and stimulus generalization. Extending other research
findings, he postulated that both the strength of the tendency to perform a
goal-oriented response (in this case, to inflict injury) and also to avoid performing
the action (that is, to inhibit one's aggression because of the possibility of punish-
ment) increased the closer the organism came to the goal (here, the closer the
organism came to the perceived angering source). Assume we have a variety of
possible targets an aggressively disposed person might attack, and that these possible
targets vary in their degree of psychological association with the angering source.
The closer the aroused person comes to the provocateur, the stronger will be the
urge to aggress, but any inhibitions against aggression stemming from fear of
punishment will also be stronger. Because the avoidance gradient (inhibition) is
often steeper than the approach-attack gradient, when the angry person meets the
angering source the inhibitory tendency often completely suppresses the aggressive
inclination. Nevertheless, because the approach-attack gradient extends further
No. shocks ,I •• "
6 10
s 10 10
10
10
10
4
o
Kelly Dunne Riley
Tormentor's Name
_ Fight Movie ~ Track Movie
Modified from Berkowitz, 1993a, p.221

(data from 1966 Geen & Berkowitz expt)
FIG. 4.2. Number of shocks to the tormentor as influenced

by his association with fight movie characters.
92 Berkowitz
than the avoidance-inhibitory gradient, when the provoked person faces another
target, one having an intermediate degree of linkage to the angering source, the
approach-attack tendency could well be considerably stronger than the inhibitory
tendency so that this associated person may now receive a displaced open attack.
B o t h the instigatory and inhibitory tendencies are quite weak when the angered
person encounters someone psychologically far removed from the angering source,
and if any overt aggression takes place, it theoretically will be fairly mild.
One of the few experiments designed to test this analysis of displaced aggression
(Fitz, 1976) yielded supporting results. Because of its considerable relevance, I describe
this experiment in some detail. Each real subject, a male undergraduate, thought he
was participating along with three other male fellow students (actually the experi-
menter's confederates) in an investigation of the effects of evaluations o n creativity.
The subject was also led to believe that two of his partners were friends with similar
personalities. We call these two persons P and P's Friend. The third student was a
stranger to them all.
A s the experiment got under way, three experimental conditions were established:
a nonangered control group and two deliberately provoked conditions. In the latter two
cases, each subject's initial performance was deliberately belittled by P i n an insulting
manner. For the next phase, every subject was then told he was to evaluate the three
other students' "creative free associations" to stimulus words. These evaluations were
to be delivered i n the form of unpleasant noises sent to each "worker" the more intense
the noise the evaluator delivered, the more unfavorable was his assessment of the
worker's creativity. However, before the supposed evaluations began, the two angered
conditions were differentiated. One half of the provoked men were also informed that
there would be yet another phase of the study, i n which P (the provocateur) would
evaluate their work on a later task by giving them electric shocks (Anger-High Fear),
whereas the remaining subjects were started on their evaluations without being given
this latter information (Anger-Low Fear).
The subject then listened to each of the other workers' responses o n their task, with
the order i n which each worker was heard being systematically varied, and delivered a
noise blast as his evaluation of each response. Figure 4.3 reports the difference between
the mean intensity of the noise administered each target by the angered men and the mean
intensity of the noise delivered to each target by the nonangered (control) subjects.
T h e data for the angry but unafraid men reveal the generalization gradient
postulated by Miller's conflict model: T h e provocateur (P) received the most
unpleasant evaluation relative to what P was given by the controls, the target
associated with h i m (P's friend) received the next most unfavorable assessment,
and the increase i n noise intensity to the stranger was lower still. By contrast, the
angry men who were led to be fearful of P's retaliation show the aggression
displacement predicted by Miller: Their evaluation of both the stranger and P was
at about the same level as that delivered by the nonangered controls, but they were
m u c h harsher to the person associated with P, his friend. Where they apparently
tended to inhibit their aggression toward the potentially dangerous P, they evidently
were less reluctant to punish the man linked to h i m and were harsher to h i m than
to the stranger. This displaced aggression i n keeping with the theoretical model
hcrel •• In nol •• InlenlUy
14
10
12
10
10
8
10
6 10
2 10 10
o
Provocateur (P) P's Friend Stranger
Target Being Evaluated
_Low Fear ~ High Fear
Data modified from Fitz, 1976

(Amount above evaluation of same target by nonangered participants)
FIG. 4.3. Punitive evaluations of targets as a function

of fear level and association with provocateur.
indicates that the subjects' actions were influenced in a fairly automatic manner by
the stimuli in the situation.
CONCLUSION
My citation of these studies certainly should not be taken as a devaluation of

Bargh's chapter. It makes many valuable contributions. However, in my view,
Bargh's theoretical analysis should be broadened, especially by integrating relevant
theoretical conceptions and research from the past. Berkowitz and Devine (1995)
pointed out that the increasing recognition of automatic influences on social
phenomena by cognitive theorists means there is little necessary conflict between
the cognitive and associationistic perspectives. Bargh seems to acknowledge this,
but I believe his theoretical scheme would benefit from a greater incorporation of
associationistic ideas where appropriate.
REFERENCES
Bern, D. J. (1972). Self-perception theory. In L. Berkowitz (Ed.) Advances in experimental social psychology.
Berkowitz, L. (1964). Aggressive cues in aggressive behavior and hostility catharsis. Psychological Review,
71, 104-122.
94 Berkowitz
Berkowitz, L. (1984). Some effects of thoughts on anti- and prosocial influences of media events: A
cognitive-neoassociationistic analysis. Psychological Bulletin, 95, 410-427.
Berkowitz, L. (1990). O n the formation and regulation of anger and aggression: A cognitive-neoasso-
ciationistic analysis. American Psychologist, 45, 494-503.
Berkowitz, L. (1993a). Aggression: Its causes, consequences, and control. New York: McGraw-Hill.
Berkowitz, L. (1993b). Towards a general theory of anger and emotional aggression: Implications of the
cognitive-neoassociationistic perspective for the analysis of anger and other emotions. In R. S. Wyer,
Jr., & T. K . Srull (Eds.), Advances in social cognition: Perspectives on anger and emotion (Vol. VI, pp.
1-46). Hillsdale, NJ: Lawrence Erlbaum Associates.
Berkowitz, L., & Devine, P. (1995). Has social psychology always been cognitive? What is "cognitive"
anyhow? Personality and Social Psychobgy Bulletin, 21, 696-703.
Berkowitz, L., & Donnerstein, E . (1982). External validity is more than skin deep: Some answers to
criticisms of laboratory experiments. American Psychologist, 37, 245-257.
Berkowitz, L . , & LePage, A . (1967). Weapons as aggression eliciting stimuli. Journal of Personality and
Social Psychobgy, 7, 202-207.
Berkowitz, L., & Troccoli, B. T. (1986). A n examination of the assumptions in the demand characteristics
thesis: With special reference to the Velten mood induction procedure. Motivation and Emotion, 10,
339-351.
Bower, G . H . (1981). Mood and memory. American Psychologist, 36, 129-148.
Chaiken, S. (1980). Heuristic versus systematic information processing and the use of source versus
message cues in persuasion. Journal of Personality and Social Psychology, 39, 752-766.
Fitz, D . (1976). A renewed look at Miller's conflict theory of aggression displacement. Journal of
Frijda, N . H . (1986). The emotions. Cambridge, England: Cambridge University Press.
Geen, R. G . & Berkowitz, L. (1966). Name-mediated aggressive cue properties. Journal of Personality,
34, 456-465.
Griffitt, W. (1970). Environmental effects on interpersonal affective behavior: Ambient effective
temperature and attraction. Journal of Personality and Social Psychology, 15, 240-244.
Keltner, D . , Ellsworth, P.C., & Edwards, K. (1993). Beyond simple pessimism: Effects of sadness and
anger on social perception. Journal of Personality and Social Psychology, 64, 740-752.
Laird, J. D . , Cuniff, M . , Sheehan, K., Shulman, D . , & Strum, G . (1989). Emotion specific effects of facial
expressions on memory for life events. Journal of Social Behavior and Personality, 4, 87-98.
Miller, N . (1948). Theory and experiment relating psychoanalytic displacement to stimulus-response
generalization. Journal of Abnormal and Social Psychology, 43, 155-178.
Razran, G . (1938). Conditioning away social bias by the luncheon technique. Psychological Bulletin, 35,
693.
Riskind, J. H . , & Gotay, C . C . (1982). Physical posture: Could it have regulatory or feedback effects on
motivation and emotion? Motivation and Emotion, 6, 273-298.
Roseman, I. J., Wiest, C . , & Swartz, T. S. (1994). Phenomenology, behaviors, and goals differentiate
discrete emotions. Journal of Personality and Social Psychology, 67, 206-211.
Schwarz, N . , & Bless, H . (1991). Happy and mindless, but sad and smart? T h e impact of affective states
on analytic reasoning. In J. P. Forgas (Ed.), Emotion and social judgments (pp. 55-72). New York:
Pergamon.
Schwarz, N . , & Clore, G . L. (1983). Mood, misattribution, and judgments of well-being: Informative
and directive functions of affective states. Journal of Personality and Social Psychology, 45, 513-523.
Tomkins, S. S. (1962). Affect, imagery, and consciousness. New York: Springer.
Wilson, J. Q . , & Herrnstein, R. J. (1985). Crime and human nature. New York: Simon & Schuster.
Wood, W , Wong, F. Y., & Chachere, J. G . (1991). Effects of media violence on viewers' aggression in
unconstrained social interaction. Psychological Bulletin, 109, 371-383.
Chapter 5
Associations to Automaticity
Charles S. Carver
University of Miami
In his discussion of automaticity, John Bargh executes something of a tour de force.

H e describes compelling evidence of automaticity from a large number of studies i n
several domains of human experience, pertaining to cognition, affect, motivation,
and action. His target chapter is impressive i n its scope as well as its clarity.
O n e of his central themes—that contexts automatically cue behavioral quali-
ties w h i c h can slip easily into the stream of action—is an idea I have long seen as
plausible, o n both conceptual and phenomenological grounds. A s early as 1981
(Carver & Scheier, 1981) we argued (based heavily o n hints from cognitive
psychology—e.g., Price, 1974; Rosch, 1978) that there are links between informa-
tion i n memory that help people identify objects and settings, and other informa-
tion that tells people how to act toward those objects and i n those settings. We
made this argument partly because we believe that doing behavior requires some
k i n d of representation of the action to use as a guide. This " h o w - t o - d o - i t "
information would have to be stored i n memory. A straightforward inference is
that the how-to-do-it information must be linked i n some manner to the percep-
tual-cognitive information that bears on the when-to-do-it and where-to-do-it
questions.
O n the phenomenological side, there is no doubt that my o w n behavior is
repeatedly channeled by associations that pop up as the day goes by. It sometimes
seems as though my life is a continuing series of inadvertent side trips, triggered by
contextual cues I encounter while i n the process of doing something else (cf.
N o r m a n , 1981). A n important element Bargh adds to the picture is that some of
1
these cues make their way i n and act outside awareness. A l t h o u g h I believe Bargh
is right about that, I must admit to not being terribly happy about it. It's bad enough
bringing to mind a quote attributed to John Lennon to the effect that "life is what happens to you
when you're busy doing other things."
95
96 Carver
to be at the mercy of contextual cues, but it adds insult to injury to be told that the
process may often be automatic, and that afterward I am totally unaware of its
occurrence.
T h e target chapter raised several conscious associations i n my mind (and maybe
others that are still preconscious). I believe these associations reflect spreading
activation sparked by information i n the article (along with chronic partial activa-
tion of certain areas of my own mind), but it would not surprise me to learn that
Bargh had buried some subliminal cues, and that my associations have different
origins altogether. In any event, my associations are the subject of this chapter. For
the most part, I do not quarrel here with what Bargh asserts i n his target chapter.
Rather, I address questions that his assertions raise i n my mind regarding issues I'm
interested i n .
CONSCIOUSNESS, AUTOMATICITY, AND FEEDBACK
M u c h of our own discussion of the self-regulation of behavior (Carver & Scheier,

1981, 1990, i n press) has focused o n consciously mediated behavior. In part, that
was because we began our work with an interest i n the consequences of focusing
attention o n various aspects of the phenomenal field. In part we had probably fallen
into the trap that Bargh now points out—that is, assuming that consciousness is a
mediator of action more often than is actually true. It would have been easy for us
to assume this, because we studied people i n laboratories i n what usually were
unfamiliar situations. In order to do the tasks they'd been assigned, subjects had to
pay attention to their behavior. There was no opportunity for the behavior to
become automatic.
These studies of consciously mediated behavior led us to assert that such
behavior displays characteristics of feedback control. Is this assertion affected by
the demonstration that motivation and behavior can be influenced by preconscious
contextual cues? N o t much, i n my view. The phrase "feedback controlled" does not
equate to "consciously mediated." If people are often engaged i n the pursuit of
multiple goals (as I believe they are), it's natural to assume that some of these goals
are outside consciousness at any given moment. Nevertheless, it also seems natural
(to me, anyway) to assume that regulation with respect to these goals, whether they
are i n awareness, involves the use of feedback information. W h e n feedback indi-
cates movement away from a desired goal, the result can be feelings of distress (cf.
Simon, 1967). If the goal i n question is presently outside awareness, the feelings of
distress may be vague and their source hard to p i n down. B u t the fact that the
feelings can arise at all suggests that feedback is being monitored outside awareness.
A s an example, consider the man who has the following goals: (a) developing
his career, (b) being supportive of his wife's career goals, and (c) raising socially
responsible children. Since early childhood, he has also had a goal (d) of protecting
himself against rejection by not depending m u c h o n other people. H e doesn't realize
he has this last goal, because the process of pursuing it is by now very automatic.
This m a n is pursuing the first of his goals quite consciously; the next two are also
5. Associations to Automaticity 97
sometimes i n consciousness, although less often. Throughout, he continues to be

vigilant at a preconscious level for cues signalling a failure to maintain conformity
to the last goal. Because he doesn't consciously recognize it as a goal, however, he
doesn't clearly grasp why he feels uncomfortable i n certain kinds of social situations
i n w h i c h he must depend o n others. In my view, that discomfort arises because
information monitored outside awareness indicates a failure to maintain conformity
to the last goal. This, i n turn, implies to me a feedback loop.
In short, although the demonstration of preconscious influence o n behavior
raises many questions about how and under what circumstances context-cued
behavioral qualities can intrude o n ongoing action (cf. N o r m a n , 1981), I don't think
it challenges the basic notion that behavior is under feedback control.
WHAT IS CONSCIOUSNESS?
T h e target chapter caused me to think again about the nature of consciousness, a

question I hadn't thought about for a while. A metaphor I've tried out o n students
from time to time is to think of consciousness as the surface of a soup pot. A s heat
is brought to bear o n the bottom of the pot, bits of the soup heat and rise to the
surface. This process i n the pot is probably chaotic and dependent o n such questions
as how readily different substances absorb and lose heat, how large the various
pieces floating i n the soup are, how m u c h interference with movement is created
by being oversized, and so on. In an analogous way, as information is activated i n
memory (by whatever influences produce activation, including spreading activation
from other active nodes, residual activation from primes that directly activated the
information earlier), that information drifts higher i n the soup pot of the mind, like
a noodle or a vegetable. Whatever bits are at the top of the as soup correspond to
the bits of information i n the person's current conscious experience.
This metaphor suggests at least one further implication. It derives from the fact that
the pieces i n the heating soup are all continuously active to some degree, even if they
are nowhere near the top of the pot. They are still absorbing activation, still bumping
against each other; parts underneath are still supporting the parts of the soup that are
at the surface. The analogy suggest that the parts of the mind that are out of awareness
similarly remain engaged i n work, spreading activation amongst themselves, and i n
some cases serving to support the edifice that's made it to consciousness at the top of
the pot. In such a model, many different areas of partial activation compete continuously
for access to consciousness, but of necessity only some small fragment of these
competing elements can be i n consciousness as any given moment. This metaphor
seems consistent with some of the sense of the target chapter.
FREE WILL AND SELF-DETERMINATION
T h e processes described i n the target chapter suggest that patterns of thought,

feeling, motivation, and action are all responsive to cues that register only outside
98 Carver
awareness. W h a t does this say about the notion of free will? The rather strong
implication is that free will is illusory. Further, the illusion seems to stem from
subjective manifestations of cognitive mechanisms at work. T h a t is, when the
mental structures that specify a behavior are fully activated (thus conscious), the
subjective experience is an intention to do the action, or a willing of the action, or
a conscious belief that the action is the appropriate behavior at that moment. This
is the subjective sense of will. Consciousness seems to be necessary, though not
sufficient, for the sense of will.
But Bargh argued persuasively that the same quality of behavior can often be
induced by contextual cues which activate the same mental structures, although
activating them to lesser degrees. Because the mental structures are only partially
activated, the subjective manifestation of their activation never reaches conscious-
ness. In such cases, the sense of intent or will would be absent, even if the act
engaged i n is essentially the same as the act done "willfully." The two cases might
feel qualitatively different to the actor, but this feeling is illusory. T h e two cases
simply represent two places o n a continuum of the activation of the behavioral
schema, and thus a continuum of awareness of the behavior's emergence.
Free will has always been something of a problem for behavioral scientists. T h e
mere search for lawfulness i n behavior seems to reflect an implicit rejection of a
strong version of the free will position. If people had free will and exerted it very
often, any lawfulness would be fragmentary at best. Bargh seems to me to be saying
2
that there is lawfulness i n behavior, and that this lawfulness derives from differential
activation of information i n memory, regardless of how that activation comes to
exist. To me, that sounds like a rejection of the idea of free will.
CONSCIOUSNESS AND DEVELOPMENT OF AUTOMATICITY
Let me turn now to a consideration of consciousness. Bargh points to two properties

of consciousness that he implies are necessary to the organism's long-term survival.
First, consciousness somehow has the property of creating the very automaticity
that is the focus of Bargh's chapter. Consciousness is involved i n knowledge
compilation. W i t h o u t consciousness, automaticity cannot develop.
W h a t does it mean to say that conscious processing is necessary for the devel-
opment of automatic processes? Precisely what is taking place at the focus of
consciousness? Two ideas come to mind, which are not mutually exclusive (nor do
they originate with me). O n e of them is relatively simple. Specifically, an argument
can be made that consciousness as a process is involved i n encoding information i n
memory with enhanced redundancy (Carver & Scheier, 1981; Powers, 1980). A s
the person focuses attention o n something, the corresponding memory trace is
rendered stronger, more complete, or some such. Indeed, greater redundancy of
encoding may even be isomorphic to the subjective experience of awareness.
2
Some would reply that the large proportion of error variance in our findings raises questions as to
how lawful behavior really is, but I will ignore that part of the argument here.
Another approach to the function of consciousness stems from an examination of

which situations do and do not require attention (Norman & Shallice, 1986; Shallice,
1978). Although attention can obviously be directed to many aspects of experience, i n
the normal course of behavior attention is needed o n some occasions more than o n
others. Specifically, attention seems to be required at points i n action sequences where
decisions have to be made (Norman & Shallice, 1986; Shallice, 1978).
Consider more closely what's taking place at these decision points. These
decisions appear to involve determining what next act quality to engage i n , given
the current context. U n d e r what conditions do such decisions have to be made?
They must be made whenever memory doesn't provide the person with a sufficiently
consistent and reliable guide for what that next behavior should be. To put it
differently, such a decision must be made when the relevant memory stores do not
permit a default response. A c c o r d i n g to this view, attention is required any time a
behavior is needed and no default is i n place.
Two sets of cases would seem to fit this characterization. O n e of them is what
happens during the acquisition of new skills. In skill acquisition, attention is
recruited for compiling the knowledge that constitutes the skill. W h a t exactly does
knowledge compilation consist of? This term seems to refer partly to the process of
identifying categories of contexts i n which the person can be confident that a
particular next response is "appropriate" or "right." W h e n a context plus response
is stored i n memory with sufficient redundancy, that category of phenomena no
longer needs attention. Stated differently, the process of skill acquisition seems to
be one of determining whether some response quality can be used as a default i n a
given context. If a particular response proves to be appropriate often enough, the
decision to use that response becomes automatic and the context-act link begins
to fade from awareness because it is no longer being checked on.
T h e other group of situations where a behavior is needed and no default is i n place
would consist of all domains where there never is enough certainty to generate true
default values. People acting i n these domains might not identify their behavior as
"trying to develop a skill," but the lack of a default response means that attention is
required, for precisely the same reason as underlies the process of skill acquisition.
A t t e n t i o n is needed i n order to decide what act quality to engage in, given the present
circumstances. In such cases, the person must think through the decision consciously.
A l t h o u g h the person who repeatedly thinks through a particular kind of decision
i n a given domain of life may not be intentionally trying to acquire a skill, a case
can be made that skill acquisition is precisely what is happening, intended or not.
O v e r repeated instances the decision becomes more automatic and receives less
attention, provided there is some degree of regularity about what decision is correct
or appropriate. M o s t circumstances i n life permit the evolution of default "leanings,"
if not truly default responses. D o these classes of phenomena then come to require
less attention over time and repetition? The answer probably depends o n how
consistently a particular action quality turns out to be right for the context.
Several questions arise from this discussion about how default behaviors come
to exist. Creation of a default response presumably depends o n the accumulation
of instances i n which the response was found to be appropriate to the situation. This
100 Carver
requires some degree of consistency among instances. But how m u c h consistency

is enough? If the response has been correct 80% of the time i n the past does it qualify
to become a default? Does it have to have been correct 90% of the time? There
must also be requirements of some sort regarding the total number of instances
recorded i n memory. A n act that was found appropriate 80% of the total of five
times it was done i n the past is likely to be chosen the next time the situation arises,
but almost certainly it won't yet qualify as a default. A n act that proved appropriate
80% of the 500 times it was done is surely closer to being taken as a default.
LINEARITY OF CONSCIOUSNESS
AND THE STRUCTURE OF BEHAVIOR
In his closing section, Bargh addresses a second function of consciousness. H e

describes this function as fitting a parallel mind to a serial world. That is, conscious-
ness seems to experience the world i n a linear fashion, and there is also a linearity
about acting i n the world. Events i n behavior follow each other, with relatively few
cases of doing multiple activities i n parallel. O n the other hand, there is a good basis
for asserting that the human nervous system works i n a massively parallel fashion.
Consciousness, then, provides an interface between these two realities (the entire
soup pot bubbles on, but only a few elements are o n the surface at any given moment.)
There is a partial resemblance between this idea and an observation we've made
elsewhere (Carver & Scheier, i n press). Over the years we've made considerable
heuristic use of a model of hierarchicality i n behavior, which was posed some time
ago by Powers (1973). In this model, there are very high order goal values (such as
the idealized overall sense of self or of a relationship) that serve as reference values
for feedback processes. A level downward are what Powers termed principles (e.g.,
honesty, thrift, expedience—virtually anything to which one could attach a trait
label), which serve as reference values for feedback processes that operate i n service
of those at the higher level.
A t the next level down there are programs of activity (essentially the same as
Schank & Abelson's, 1977, script concept), i n which behavior becomes more clearly
identifiable as behavior (such activities as going out to dinner, going to the grocery
store, mowing the lawn, etc.). This model assumes the functioning of feedback loops
to control these program-level activities as well. Programs are composed of se-
quences—acts executed all-at-a-piece—which, i n turn, are made up of lower level
qualities of behavior, all of which are presumed to involve feedback processes.
It has always seemed to me that there is something special and unusual about
program-level behavior i n this model. For one thing, people do seem to spend m u c h
of their waking lives with that level of abstraction functionally superordinate i n
guiding their actions. Occasionally, people think quite consciously about higher level
concerns, but for the most part our conscious experience is at the program level.
Behavior at this level also seems to differ from behavior at other levels. Behavior
at the program level has a digital or sequential quality, whereas processes at higher
and lower levels have more of an analog feel to them. M a n y years ago Miller,
Galanter, and Pribram (1960) noted this difference regarding lower order processes.
They wrote that consciously planful activity (which seems to correspond to what
I'm calling the program level of control) has a digital quality, whereas processing at
lower levels has a more analog quality (see also Greene, 1972). W h a t gives
program-level functioning its digital character is the sequential decision making
that takes place there.
M i l l e r et al. didn't speak to levels of abstraction higher than conscious planful-
ness. It's of interest, however, that control at these even higher levels appears to
reassume an analog form, with behavior varying i n a vaguely quantitative way rather
than as a series of acts. Consider principles. A l t h o u g h the sense of a given principle
is clear, the quality it specifies doesn't correspond to a particular act. Rather, the
quality might be reflected i n a multitude of potential behaviors. T h e psychological
sense of how well you've been living up to a given principle is emergent from bits
of many events. It seems to be relatively easy to bring this sense to mind, though,
and oddly enough it tends to emerge with an analog rather than digital feel.
T h e digital quality is something that seems noticeably different at the program
level, compared to what goes o n at levels either higher or lower. It's never been clear
to me, however, what to make of this difference. Indeed, it isn't entirely clear whether
this is really a difference between levels of abstractness i n a hierarchy, as I portray it
here, or whether it's really a difference between feedback systems (analog at all levels)
and a different kind of function (planner or prioritizer or some other function) that
is, for some other reason, tied to the program level. In any event, this observation
about the feel of control being different at various levels is consistent with Bargh's
reminder to us that something is needed to fit a parallel mind to a serial world.
IS CONSCIOUSNESS AN EPIPHENOMENON?
Bargh concludes his article with the assertion that identifying the function of
consciousness as connecting a parallel mind to a serial world guarantees the
ultimate failure of Skinner's argument that consciousness is epiphenomenal. I'm
not so sure about that. Even if we expand the discussion to include the other
proposed function of consciousness—the development of automaticity—I'm still
not sure the conclusion holds.
I discussed earlier the idea that consciousness is implicated i n the development
of automaticity by virtue of the fact that automaticity accrues from repeated decision
making, and that consciousness seems to be involved i n that decision making. Does
this imply that consciousness is not epiphenomenal? I don't see that it does.
Consciousness may be simply the experiential readout of a problem solving process
that facilitates (and may even be necessary to) the organism's long-term functioning.
But it isn't obvious to me why it's necessary to assume that it is more than just that—a
readout of information about something that's going o n inside. A reasonable question
to ask i n return, of course, is if it's only a subjective readout, why do we have it? I
don't have an answer to that one, but I don't know why we have an appendix either.
102 Carver
There's a similar problem with the other function Bargh ascribes to conscious-
ness. A l t h o u g h I am sympathetic to the idea that consciousness serves to link a
parallel (and perhaps analog) mind to a serial world, I'm not sure that function
implies Skinner was wrong. Physiological psychologists have long had to deal with
the problem that some of their measurement devices yield an analog output,
whereas their data management devices need a digital input. T h e solution is an
analog-to-digital ( A - t o - D ) convertor. This convertor connects analog and parallel
processess within the organism (signals from the nervous system) to a serial world
(the computer that has to deal with the information contained i n the signals). But
I think most of us would regard any consciousness that might be experienced by an
A - t o - D convertor to be epiphenomenal. A s much as I would like to believe that
consciousness is not an epiphenomenon, then, I'm not convinced that ascribing
this function to consciousness ensures that to be the case.
ACKNOWLEDGMENTS
Preparation of this chapter was facilitated by N C I grant C A 6 4 7 1 0 . Send correspon-

dence to Charles S. Carver, Department of Psychology, University of M i a m i , C o r a l
Gables, F L 33124-2070; Internet Ccarver@umiami.ir.miami.edu.
REFERENCES
Carver, C . S., & Scheier, M . F. (1981). Attention and self-regulation: A control-theory approach to human
behavior. New York: Springer-Verlag.
Carver, C . S., & Scheier, M . F. (1990). Principles of self-regulation: Action and emotion. In E. T. Higgins
& R. M . Sorrentino (Eds.), Handbook of motivation and cognition: Foundations of social behavior (Vol. 2,
pp. 3-52). New York: Guilford.
Carver, C . S., & Scheier, M . F. (in press). O n the self-regulation of behavior. New York: Cambridge
University Press.
Greene, R H . (1972). Problems of organization of motor systems. In R. Rosen & F. M . Snell (Eds.),
Progress in theoretical biology (Vol. 2, pp. 303-338). New York: Academic Press.
Miller, G . A . , Galanter, E., & Pribram, K. H . (1960). Plans and the structure of behavior. New York: Holt,
Rinehart & Winston.
Norman, D . A . (1981). Categorization of action slips. Psychological Review, 88, 1-15.
Norman, D. A . , & Shallice, T. (1986). Attention to action: Willed and automatic control of behavior.
In R. J. Davidson, G . E. Schwartz, & D . Shapiro (Eds.), Consciousness and self-regulation: Advances in
research and theory (Vol. 4, pp. 1-18). New York: Plenum.
Powers, W. T. (1973). Behavior: The control of perception. Chicago: Aldine.
Powers, W. T (1980). A systems approach to consciousness. Inj. M . Davidson & R. J. Davidson (Eds.),
Psychobiology of consciousness (pp. 217-242). New York: Plenum.
Price, R. H . (1974). T h e taxonomic classification of behaviors and situations and the problem of
behavior-environment congruence. Human Relations, 27, 567-585.
Rosch, E . (1978). Principles of categorization. In E. Rosch & B. B. Lloyd (Eds.), Cognition and
categorization. Hillsdale, NJ: Lawrence Erlbaum Associates.
Schank, R. C . , & Abelson, R. P. (1977). Scripts, plans, goals, and understanding. Hillsdale, NJ: Lawrence
Erlbaum Associates.
Shallice, T. (1978). T h e dominant action system: A n information-processing approach to consciousness.
In K. S. Pope & J. L. Singer (Eds.), The stream of consciousness: Scientific investigations into the flow of
human experience (pp. 117-157). New York: Wiley.
Simon, H . A . (1967). Motivational and emotional controls of cognition. Psychology Review, 74, 29-39.
Chapter 6
Minding Our Emotions: On the Role

of Automatic, Unconscious Affect
Gerald Clore
Timothy Ketelaar
University of Illinois
In his chapter, Bargh argues that affective reactions are automatic and u n c o n -
scious, and that they influence evaluative judgment without ever being experi-
enced. In contrast, others have suggested that affective influence depends o n
subjective experience (Schwarz & Clore, 1983), that emotions arise from cognitive
appraisals (Ortony, Clore, & Collins, 1988), and that emotions are never u n c o n -
scious (Clore, 1994). These views sound diametrically opposed, but we suggest
that they are not incompatible.
Bargh makes several claims w i t h respect to affect: (a) an evaluation module
exists i n w h i c h every stimulus is immediately and unconsciously evaluated as
good or bad; (b) evaluation is an unconscious event that precedes the rest of
processing (reflecting the claim that preferences need no inferences [Zajonc,
1980] or that emotion is not cognitively mediated, w h i c h discussed i n the
following section); and (c) affective reactions influence evaluations automat-
ically without mediation by consciousness or choice. It is also implied that this
is the n o r m , that it is c o m m o n rather than merely possible. T h i s last c l a i m is
discussed i n a later section.
AUTOMATICITY AND THE CAUSES OF EMOTION
Cognitive emotion theories are concerned with appraisals, the cognitive bases of
emotional reactions. Appraisals are evaluations of situations with respect to one's
personal concerns. A r e these appraisals the same or different from the automatic
and unconscious evaluations that Bargh talks about?
105
106 Clore & Ketelaar
Evaluations are Unconscious

Unconscious evaluations are not unconscious emotions. Emotions are evaluative
(affective) states of the organism, and evaluations are necessary but by no means
sufficient to generate emotion. In our usage, one cannot actually have unconscious
emotions. In the tradition of psychoanalysis, some choose to call emotions " u n c o n -
scious" when they involve ideas that are activated by unconscious means. However,
that usage is potentially confusing, unless one continually explains that the emotion
itself is not unconscious.
Bargh proposes that affect emerges from a nonconscious source. A l t h o u g h
cognitive approaches maintain that emotions result from cognitive appraisals,
nothing i n appraisal theory requires conscious processing. The idea that such
appraisals are generally unconscious helps explain the role of feelings i n emotion.
A l o n g with others (e.g., Frijda, 1986; Oatley & Johnson-Laird, 1987), we suggested
that emotions inform the person about the output of appraisal processes (e.g., Clore,
1992). But why, readers sometimes ask, does one need to be informed about one's
own cognitive analysis of a situation? If the appraisals were conscious, such
information might indeed be unnecessary. Appraisals are generally not conscious,
even though they are being made more or less constantly. A s a result, some
conscious, attention-grabbing output of the process is needed to influence process-
ing priorities when necessary (Simon, 1967). Conscious feelings that reflect uncon-
scious appraisals appear to play that role. Hence, cognitive emotion theory, too,
assumes that m u c h of the origin of emotion is nonconscious, a problematic position
only if one believes that cognition requires consciousness.
Evaluations are Automatic

Bargh proposes that, "all stimuli are evaluated immediately as good or bad, without
the participant intending to evaluate ... everything that one encounters is precon-
sciously screened and classified as either good or bad, within a fraction of a second
after encountering it" (chap. 1, p. 23). There is a great deal implicit i n these
proposals, including that most psychologists believe evaluative reactions are usually
consciously intended.
Bargh suggests that social psychologists generally believe that attributions re-
quire "deliberative and sophisticated steps of conscious reasoning" and that the
influence of attitude involves "conscious and intentional retrieval of one's attitudes
from memory" (chap. 1, p. 5). O u r o w n reading of the literature has not given us
that impression. Bargh repeatedly contrasts "conscious intention" with " u n c o n -
scious automaticity" (chap. 1, p. 3), but it seems to us that these do not exhaust
the alternatives. Causal attributions, for example, are frequently neither conscious
and intentional nor unconscious and automatic. Rather they are perceptual.
Heider (1958), from whose writings attribution theory originated, was a Gestalt
psychologist concerned with perceptual rather than intentional processes. W h e n
we see one billiard ball strike another and the second ball moves, we do not first
see the event and then make an attribution (either intentionally or automatically).
6. Minding Our Emotions 107
Rather, the causal attribution is part of the perception that these are billiard balls.
Causing each other to move is what billiard balls do. Thus, automatic versus
intentional may not be the only alternatives for how we might think of evaluative
and other psychological processes.
Bargh argues that our behaviors, perceptions, and evaluations are governed by
automatic processes i n which the "process operates and runs to completion without
conscious intention or awareness" (chap. 1, p. 6). In contrast to this description of
the automatic process of making evaluation, the examples involve the retrieval of
prior evaluations as part of object identification. In these examples, evaluations
occur because they are constituents of objects. A l l parts of a whole must be present
when the whole is present, by definition. It is not clear what is added by specifying
that such parts will occur automatically, as though there is some other unseen
process at work. For example, to a Chicago Bulls' fan, classifying someone else as a
Bulls fan may automatically bring with it a positive evaluation. It is not that one
first identifies h i m or her as a fan and then evaluates whether that is good or bad;
the evaluation is part of the categorization. Bargh studied the elderly stereotype, of
which slowness is a component, and Devine studied the A f r i c a n - A m e r i c a n stereo-
type, of which hostility is a component. Bargh suggests that the activation of slowness
and of hostility are automatic consequences of the activation of the elderly and
African-American stereotypes, respectively. But it might be more accurate to say
simply that these attributes are parts of those stereotypes. Thus, what Bargh sees as
evidence of automatic processes often occurs simply as part of object identification;
in other words, much of what is being studied concerns structure rather than process.
O n the other hand, even if all of the automatic components Bargh discusses were
of this type, their consequences are often surprising. For example, Bargh shows that
priming some of the elements of the elderly stereotype ends up slowing the walking
speed of subjects. Slowness was not specifically primed, but because the stereotype
as a whole was primed, slowness too was apparently activated. Because it was not
tied to an explicit elderly stimulus, the activated but unattached concept was
available to influence subjects' momentary implicit views of themselves.
But with respect to evaluation, Bargh is not specific about the process. Thus, it
may be useful to differentiate three different levels of evaluative processing:
1. Evaluation at the first level consists of accessing pre-existing evaluative

meaning as part of stimulus identification, as previously discussed.
2. A t the second level, a new evaluation is made of a new stimulus, but it is
determined simply and directly by its compatibility with one's tastes or
attitudes.
3. A t a third level, more extensive processing may be required as events and
actions are appraised with respect to their personal implications.
Evaluation as a Constituent of Meaning. A s Bargh points out, Osgood long

ago showed that evaluation is the primary dimension of meaning. T h a t is, all verbs,
nouns, and adjectives i n all languages can be sensibly organized i n terms of their
evaluative connotations (Osgood, Suci, & Tannenbaum, 1957). In other domains,
108 Gore & Ketelaar
it is also common to find that the first order of cognitive business for the individual
is evaluation. A glance at common terms used to describe people, such as A n d e r -
son's (1968) list of 555 trait terms scaled for likeableness, shows that they are all
frankly evaluative. A l s o , i n factor analyses of interpersonal ratings, the first factor
is almost always a good versus bad factor.
If one agrees that a primary dimension of meaning is evaluation, then the
assertion that the perception of any stimulus automatically involves evaluation
would seem noncontroversial. It reduces to saying simply that perception involves
meaning. If meaning involves evaluation, then must not perception involve evalu-
ation? In other words, part of what it means to apprehend anything is to place it
into one's network of concepts, to categorize it, to give it meaning. A s soon as one
identifies a stimulus (or even has an hypothesis about what it may be), then it is
likely to have an evaluation, as well.
We suggest, therefore, that many of the automatic evaluations Bargh refers to
are actually part of our interpretations of the stimuli we encounter. T h e rather
amazing prevalence of evaluative reactions, even to what would seem to be virtually
neutral stimuli, does provide notable evidence of most of the examples are not about
evaluating as a process, but about the centrality of evaluative meanings that are
already inherent i n the stimuli. For such stimuli, preferences do need inferences, i n
the sense that the evaluations are not produced by some automated mechanism but
require object identification processes.
Evaluation as a Function of Taste. In some of his experiments, Bargh em-

ploys stimuli that are novel. Nonsense syllables were used, and these appear to be
classified as good or bad automatically. Thus, although most of Bargh's examples
concern pre-evaluated concepts, he also gives data that truly do involve automatic
evaluations. In these experiments, nonsense syllables are presented that norma-
tively scaled for evaluation. Presumably more pronounceable nonwords are evalu-
ated more highly than less pronounceable nonwords. Similarly, perhaps one
nonsense syllable or geometric figure can be evaluated as more positive than another
because it satisfies some gestalt principle of goodness of form. These are examples
i n which evaluation of a stimulus is immediately given when it satisfies some
preformed criterion of preference.
Echoing Zajonc, one of the boldest of Bargh's claims is that evaluation and
emotion are separate systems. H e suggests, along with L e D o u x , that whereas the
cognitive system is concerned with the meaning of a stimulus, affective processing
is concerned with the implications of the stimulus for the self. Cognitive emotion
theorists, o n the other hand, have not made such a distinction. They view emotion
as a consequence of meaning analysis (Mandler, 1984), as a consequence of ordinary
cognitive processing that happens to include goals and concerns. Whether the idea
of two systems or of one system accomplishing two purposes is the more useful may
not be easily answered. But the more general compatibility between Bargh's
automatic evaluations and cognitive emotion theory is easily seen i n Bargh's words
(Bargh, C h a i k e n , Raymond, & H y m e s , 1996):
Evidence of a ubiquitous automatic and preconscious evaluation effect is in line with

models of emotion production that posit a primary stage of environmental appraisals
(e.g., Lazarus, 1982, 1991, pp. 152-170; Ortony, Clore, & Collins, 1988). The
automatic classification of events as good or bad may be a first step to an emotional
experience, as such appraisal models assume ... (p. 123)
T h e Ortony, Clore, and Collins model (1988) proposes three kinds of evalu-
ations, each resulting i n a different kind of affective reaction. W h i c h of these occurs
depends o n one's focus of attention. The possibilities include focusing o n events,
actions, or objects. T h e three affective reactions are being pleased about events,
approving of actions, and liking objects. O n e is pleased about the outcome of events
when events are appraised as desirable with respect to one's goals. O n e approves of
actions when they are appraised as praiseworthy with respect to applicable stand-
ards. O n e likes objects when they are appealing with respect to one's tastes (and
attitudes). So, three kinds of psychological structures (goals, standards, atti-
tudes-tastes) correspond to three domains of attention (events, actions, objects),
which result i n three kinds of affective reactions (being pleased, approving, liking).
Bargh finds the account of affective reactions to objects to be most compatible with
his notion of automatic evaluations (Bargh et al., 1996):
In this regard, we endorse the distinction made by Ortony et al (1988) between

different targets of evaluation or appraisal: whether an event, a person, or an object
is being evaluated. According to this view, events are appraised in terms of one's goals,
people are appraised according to personal standards of "praiseworthy" behavior, and
objects are evaluated as "appealing" or not. Perhaps the automatic evaluation effect
is most closely allied with this latter variety of appraisal. (p. 124)
T h e object-based (attraction) reactions include such emotions as liking, loving,

hating, and being disgusted. W h a t presumably appeals to Bargh about this portion
of emotion space is that Ortony et al. (1988) said:
The Attraction emotions ... are among the most salient experiences we have. A t the
same time they appear to be more immediate, more spontaneous, and less affected by
accessible cognitive processes than almost all of the other emotions. (p. 156)
Thus, although Ortony et al. gave a cognitive analysis i n which emotions depend
o n appraisals of events, actions, and objects, they would agree that such emotions
as disgust or liking may involve a direct readout of one's tastes. W h e n asked why
one likes something, it is perfectly acceptable to say, "I don't know why, it's just
disgusting," or, "I just like it."
Evaluation as a Function of Appraisal. T h e appraisals involved i n other

emotions, however, may not be as direct as those based o n taste (e.g., Lazarus, 1982;
1991). Hope, fear, disappointment, relief, sadness, frustration, and joy may all
involve appraisals of events with respect to the desirability of their outcomes.
Similarly, pride, shame, admiration, and reproach all involve appraisals of actions
with respect to their praise- or blameworthiness.
Still, these appraisals are seen, even by traditional emotion theorists, as auto-
matic. For example, the work of A r n o l d (1960) is important as an early statement
of the cognitive approach to emotion. Her concept of appraisal was particularly
influential. She proposed that people implicitly evaluate everything they encounter,
and that such evaluations occur immediately and automatically. Similarly, Clore,
Schwartz, and Conway (1994) concluded that, "Emotions result from ongoing,
automatic, but implicit, appraisals of situations with respect to whether they are
positive or negative for one's goals and concerns" (pp. 326-327).
Categorization Versus Calibration. Despite all this agreement, there is one

respect i n which Bargh's assertions do diverge from what is assumed by appraisal theory.
A basic aspect of appraisal theory is that the intensity of emotional responses depends
on appraisals of the degree to which an outcome is desirable, an action is praiseworthy,
or an object is appealing. These, i n turn, depend on the importance of relevant goals,
standards, and attitudes-tastes. In contrast, Bargh reports that the automatic evalu-
ation effect is not influenced by the strength of relevant attitudes. H e proposes that
evaluation is a separable stage involving categorization of the goodness-badness only.
Bargh's results suggest that emotional intensity, the quantitative aspect of affect,
must reflect later processing. A c c o r d i n g to his data, the initial stage is a simple
categorization of the stimulus as good or bad. Does that mean that there are two
appraisal processes, one that appraises a stimulus as good or bad and a later one
that asks how good or bad? This idea is not found i n appraisal theories, although
Ortony et al. also distinguish between the cognitive variables that govern elicitation
of emotions and those that govern the intensity of emotion. They are not always
different factors, but conceivably they may operate at different stages. In any case,
i n the next part of this chapter, we suggest that categorization and calibration may
indeed be separable. Some evidence suggests that priming, including Bargh's
nonconscious priming, affects interpretations or categorizations, whereas affect-as-
information processes (Schwarz & Clore, 1983) affect judgments or calibration.
Do We Need a Special Evaluation Module?

There are two possible ways of thinking about Bargh's results. Evaluative processes
include separable categorization and calibration stages, with the categorization stage
being very fast and the calibration stage taking longer. Alternatively, the subliminal
masking paradigm used by Bargh may make it appear that there are two stages when
actually there is only one. According to this view, one stage is all that is required.
Evaluative meaning is not processed before the descriptive features of the word, the
masking stimulus simply interferes with the storage of features that would make the
stimulus recognizable later. Thus, the masking procedure, by stopping the registration
of the stimulus makes it look like there is a separate process of evaluation.
Two findings that provide support for Bargh's special evaluation module idea are
the absence of similar effects for the potency and activity dimensions, and the
absence of fan effects i n evaluative priming. However, if evaluation is the most
important and reliable categorization people make, it should also be the first and
most easy to detect. In factor analyses of ratings using common descriptive terms,
the difference between the variance accounted for by the first, evaluative factor and
other more descriptive factors is usually large. Perhaps the failure to find related
effects for the potency and activity dimensions simply indicates that these are less
important and reliable. W h e n a stimulus is presented, subjects may immediately
build a model for the entire stimulus, but evaluative information appears to be
separate simply because it involves a dichotomous choice that is both primary and
reliable. It should be easy to detect and frequently found even if it were merely part
of general object identification processes rather than constituting a separate auto-
matic evaluation module.
Similar logic leads to the prediction that one would not see a fan effect i n evaluative
priming. Spreading activation models of general mood effects have not fared well
because the assumption that one should find fan effects makes it implausible that
similarly valenced conditions could all get activated at once. However, the logic of
subliminal presentation is precisely that only the most salient features, such as evaluative
meaning, should survive a masking stimulus that blocks out all other, more subtle object
identification dimensions, dimensions that are less reliable and salient. W h e n these
descriptive features are blocked out, the fan effect should be also eliminated. In any
case, these are ways of accounting for the results without assuming a special evaluation
module separate from general object identification processes.
O n the other hand, Bargh may be correct, and special, very fast processes devoted
to evaluation may exist. Evidence encouraging such a view comes from Cosmides'
(1989) research o n cheater detection. Whereas Bargh suggests that we engage i n
rapid and efficient evaluation of objects on the basis of liking, Cosmides's data
suggest especially efficient evaluation of actions on the basis of fairness. H e r model
explains content effects on a logical reasoning task known as the Wasson selection
task. W h e n the content of the logical problems naturally tap the domain of social
exchange, otherwise difficult problems are solved quite easily because individuals
are able to make use of specialized reasoning skills (e.g., cheater detection) that
evolved to apply to that domain. Cosmides showed that individuals can evaluate
such exchange problems surprisingly efficiently. O f course, that research is not about
brief exposure times or rapid evaluations. But it does suggest that we may have special
problem solving competence for problems involving approval or disapproval. If such
an approval-disapproval module exists, then Bargh could also be right about the
liking-disliking module that he proposes. Moreover, what we earlier called level two
evaluation (preferences without inferences) could conceivably exist for the liking or
disliking of objects, and for the branch of the Ortony, Clore, and Collins emotion
tree involving the approval or disapproval of actions.
Consciousness Serves Action in a Serial World

Bargh makes a compelling argument that the serial nature of conscious thought is
an outgrowth of the necessity of interfacing a parallel mind with a serial world i n
w h i c h one must act i n real space and time. In the literature on emotion, a
satisfactory account of the relationship between emotion and behavior has always
been elusive. Accounts that focus o n animal models tend to see emotion as having
direct behavioral consequences. Others (e.g., Scherer, 1984) have argued that
emotion evolved as protocognition, as a psychological waystation between stimulus
and response that afforded flexibility. Emotions can thus provide information and
motivation, without triggering obligatory behavior. Seeing emotion as playing a role
at the interface of unconscious, parallel processes and conscious, serial processes is
a step forward i n tying emotion to behavior. A s i n congressional decision-making
processes, there are presumably multiple mental subcommittees working i n parallel
to appraise things within their purview. But they can be considered by the body as
a whole only serially. In congress, various contingencies may arise that alter the
agenda and reset priorities, so that important matters can be accorded greater
consideration. T h e same is true i n the chambers of the mind, and a role for emotion
is to assist i n this agenda setting.
Summary
In this section, we discussed the relationship between Bargh's automatic evaluation
effect and the cognitive appraisals believed to underlie emotions. We concur with
the view that the evaluation of stimuli is a superordinate goal, and that evaluation
is a central feature of meaning. Moreover, we agree that psychology generally failed
to appreciate the centrality of this fact.
O n the other hand, Bargh's assertions about consciousness and social psychol-
ogy are a straw man. It was less clear to us, for example, that attribution theorists
have assumed conscious intentions to engage i n causal analyses or that attitude
theorists have assumed conscious intentions to access attitudes. We suggest that
the dichotomy between automaticity and intentionality may be overdrawn. A t t r i -
butions, for example, are often perceptual rather than either automatic or inten-
tional, and evaluation is often a constituent of meaning rather than the result of a
separable process. Bargh describes the automatic effects of which he speaks as
processes, and implies that evaluative processes operate apart from descriptive ones.
M o s t of the examples, however, concern existing evaluative meaning, so that the
evaluation is not automatic so m u c h as constitutive. A n evaluative component is
available i n these instances as a necessary part of having identified the object.
We suggested that it may be useful to distinguish three levels of evaluation. A t
the first level, prior evaluation becomes available simply as a consequence of object
identification, as previously described. A t the second level, evaluations may be
automatic and instantaneous to the extent that they are based o n taste or attitude.
For example, if one has never developed a taste for them, eating oysters may be
disgusting. T h a t reaction may not require cognitive elaboration. Conversely, pre-
ferred foods may fit one's tastes like a key fits into a lock. This is presumably the
k i n d of evaluation Bargh has i n mind. A t this level, it is perhaps true, as Zajonc
(1980) suggested that preferences need no inferences.
To explain evoluation at the third level, a brief review of the Ortony, Clore, and
Collins theory of emotion (Ortony et al., 1988) was presented. T h e theory proposes
three kinds of evaluation i n which events can be seen as desirable or undesirable,

actions as praiseworthy or blameworthy, and objects as appealing or unappealing.
Relating his work to this account, Bargh suggests that his automatic evaluations are
closest to those that concern the appeal of objects.
T h e third level concerns appraisals of the desirability of events and the praise-
worthiness of actions. Evaluations i n these domains are not based o n taste or
attitude, but o n goals (when evaluating the outcomes of events) and standards
(when evaluating the praiseworthiness of actions). T h e tendency to appraise these
stimuli is no less automatic than those involving taste, but here, preferences
generally do not precede inferences. Unlike the lock and key examples of taste,
inferences must be made. T h e inferences must be made even if unconscious, and
even i f the making of them is automatic. Moreover, the inferences must be made
even though they may be made ahead of time, so that a later emotional reaction
seems immediate (e.g., "Wouldn't it be wonderful-awful if my pregnancy test
tomorrow turns out positive?").
Evaluation at the third level is what emotion theorists mean by appraisal, a
process i n which preferences need inferences. Evaluation at the second level
(involving liking o n the basis of taste or attitude) includes the phenomena pointed
to by Zajonc and by Bargh i n which preferences need no inferences. Evaluation at
the first level concerns evaluative meaning. In these cases, an evaluation is neces-
sarily involved as part of the previously established meaning of a stimulus. Here,
preferences also need no inferences, because the process of forming impressions is
not involved, because evaluative meaning was already established and is only being
accessed. M o s t of Bargh's examples fall into this class. These establish the ubiquity
and centrality of evaluation, but are not, i n our view, examples of automatic
evaluative processes i n the proper sense.
A difference between appraisal theory and Bargh's formulations is that Bargh
concludes that automatic evaluations are categorical, and he and his colleagues find
that these are not influenced by attitude strength. A central tenet of appraisal theory,
however, is that the intensity of feeling is a function of goal, standard, or attitude
importance. There are a couple of possible implications. O n e possibility is that
evaluations at levels two and three are qualitatively different from each other. It does
seem more sensible, for example, to talk about goal importance than taste impor-
tance. A n o t h e r possibility is that evaluation is simply a multistage process i n which
categorization into good and bad is made before calibration of goodness and bad-
ness—in other words, that qualitative and quantitative evaluations are separable.
AUTOMATICITY AND THE CONSEQUENCES OF EMOTION
We examined some implications of Bargh's proposals for the process of generating

affective and evaluative reactions. But i n addition, Bargh suggests that affective
reactions, however generated, also influence judgment automatically and uncon-
sciously. In contrast, others have suggested that the influence of affect o n judgment
is mediated by the experience of affect (e.g., Schwarz & Clore, 1996). T h e goal of
this section is to ask whether these divergent views can be reconciled. We argue
that the automatic affect model and the affect-as-information model are comple-
mentary rather than conflicting.
Evaluative Reactions Without Affective Experience

Some writers maintained that affective influences o n judgment may occur without
any phenomenal experience of affect (e.g., Damasio, 1994; L e D o u x , 1987; Zajonc,
1980). We believe that results that have been reported are easily misconstrued. For
example, Murphy and Zajonc (1993) observed that subjects evaluated u n k n o w n
Chinese ideographs more positively when they were preceded by the subliminal
presentation of a smiling rather than a frowning face. In terms of the feelings-as-in-
formation framework, this would suggest that subjects misattributed feelings elicited
by the subliminal faces as reactions to the ideographs because the subliminal
presentation technique made a correct attribution unlikely. Consistent with this
explanation, the effect was obtained only when the faces were presented sublimi-
nally, not when they were presented supraliminally. Apparently, supraliminal pres-
entation allowed subjects to identify the actual source of their reactions, rendering
them nondiagnostic for evaluating the ideographs.
These results appear consistent with a feelings-as-information account, but an

extended replication of the study suggests that they really concern separate proc-
esses. W i n k i e l m a n , Zajonc, and Schwarz (1995) attempted to manipulate the
perceived diagnosticity of the presumed affective reactions through attribution
manipulations. However, the observed effect remained unchanged. Specifically,
they informed subjects that a smiling (or frowning) face would precede each
ideograph (Experiment 1), or they exposed subjects to music said to elicit positive
or negative feelings (Experiment 2). Neither misattribution manipulation resulted
i n augmentation or discounting effects. However, subjects did not report having
any subjective experience of affect i n the experiment. Such faces, although clear i n
meaning even when presented subliminally, apparently do not elicit affective
experience as required by the affect-as-information hypothesis.
Because subjects did not report being i n a mood or having any affective
experience i n these studies, opportunities to misattribute their feelings to an
irrelevant source could not eliminate the effect. T h e results, therefore, do not fall
i n the purview of the affect-as-information hypothesis. In the absence of experi-
enced feelings, these affective priming studies seem more relevant to the nonaffec-
tive priming literature than to studies of mood and emotion. They do not show the
influence of "felt" affective experience but of "unfelt" (i.e., nonconscious) evalu-
ation or meaning appraisal.
Two Stages of Judgment

A consensus is emerging i n social psychology around the idea that social judgments
require two stages (e.g., Brewer, 1988; Gilbert, 1989; Srull & Wyer, 1979; Trope,
1986). The first stage involves gross categorization, and the second involves
adjustments, corrections, extensions, and evaluations. T h e initial stage involves the

unfettered use of existing schemas, scripts, stereotypes, habits, feelings, expecta-
tions, and personal attributions. The second may involve attention to new or
inconsistent information, attributional discounting, observations of specific behav-
iors or situational factors, learning, and so on. These two stages are associated with
different processing styles, the first stage being more automatic or heuristic, and the
second more controlled or systematic. W i t h i n this kind of framework, the automatic
affective processes Bargh discusses are presumably most relevant to the first stage.
T h e experiential processes on which the affect-as-information explanation for
judgment effects is focused (Schwarz & Clore, 1983, 1988, 1996) may be more
relevant to the second stage.
Does mood influence the evaluation of stimuli by priming mood-consistent
cognitive categories or by providing experiential information about unconscious
appraisals? Explanations of mood effects based o n accessibility of concepts (e.g.,
Forgas & Bower, 1988; Isen, Shalker, Clark, & Karp, 1978) and explanations based
o n affect as information (Schwarz & Clore, 1983) make a variety of contrasting
predictions. For example, priming approaches assume that mood effects should occur
only when a stimulus is ambiguous, so that affect can bias interpretation, whereas
the affect-as-information hypothesis assumes that mood effects do not require
ambiguity, because mood influences evaluation and not interpretation (Schwarz,
Robbins, & Clore, 1985). In addition, priming approaches assume that primed
concepts influence judgment at encoding when interpretations are being made, while
the affect-as-information hypothesis predicts that mood effects occur at the time of
judgment when comparing one's feelings to the judgment alternatives (Clore &
W i l k i n , 1985). The differences i n these predictions suggest that the processes
assumed by each explanation may apply to different stages of processing. Assessment
of this possibility will have to await research specifically designed to test it.
Why Unconscious Ideas Are Powerful

There may be a tendency to see the power of unconscious, automatic processes as
a challenge to the validity of formulations based on implicit attributions (e.g.,
Schwarz & Clore, 1983). However, we argue (in agreement with Freud) that
unconscious ideas are powerful (and potentially problematic) simply and solely
because they are easily misattributed. A specific idea spoken by a particular person
at a given occasion i n a limited context may have a highly specific meaning. T h e
reason for its presence i n mind is unambiguous and highly specific, because along
with it i n memory is the time, place, and substantive context i n which it was uttered.
A s a result, its role i n mental life is correspondingly constrained; that is, there are
episodic constraints o n its semantic activation. Research indicates that when a word
is processed, multiple possible meanings are activated i n memory, but because of
the converging activation of its modifiers and surrounding context, one becomes
aware of the intended meaning alone. T h e particularity of the resulting meaning
allows one to engage i n directed rather than random thought. In contrast, it is
sometimes difficult to converse with a schizophrenic, because such constraints do

not seem to operate as reliably. Such persons may display unconstrained activation
patterns, including clang associations, i n which their thought is guided by the sound
of the word rather than its meaning.
In some research, words or concepts may be presented without a salient context.
They are sometimes presented subliminally, as an irrelevant stimulus i n a Stroop
task, or while one is distracted by a secondary task. In such cases, the concept may
become accessible through recency and frequency of exposure but without a
meaningful context and thus without salient or memorable episodic constraints. In
these cases, the pattern of activation does not summate with that of its modifiers
and context to yield a particular and limited meaning. In its more random disper-
sion, the activation may combine with that from rather distant associations so that
novel ideas are entertained or poetic associations are expressed. Similarly, as a result
of persistent psychological conflict, unfinished personal business, and so on, chronic
activation of a particular concept may result i n a "one-track m i n d . " In all of these
cases, one can think of the mental contamination from the primed concept as
occurring because the activated semantic meaning is unconstrained by the usual
episodic or contextual meaning that activates one particular instance.
Freud (1915) gave a highly similar interpretation for why unconscious ideas are
potentially problematic. For Freud, too, the problem was that activation (instinctual
energy) that is no longer tied to a particular memory is free to drive associated ideas
into consciousness. In his formulation, the key to obsessions, metaphor, poetry,
drama, fantasy, and creativity lies i n the displacement of activation made possible
by the fact that the constraining details of a particular episodic source were no
longer available i n memory. W h a t Freud believed to take place via repression, Bargh
produced through backward masking; perhaps not so very different. In any case, it
is humbling to see that the distance across the 100 years from Freud's (1894) essay
to Bargh's i n 1997 is not as great as one might have imagined.
Hot-Wiring the Auto-Motive System
Bargh refers to automaticity i n terms of social ignition and the auto-motive model,
w h i c h makes one wonder whether the strong focus on automaticity, important as
it is, really provides an adequate view of social cognition. There is a sense i n w h i c h
Bargh implies that these are the "real" social cognitive processes, whereas those on
w h i c h we have previously focused are window dressing. But a critic could argue that
these automatic evaluation effects show only what happens to the auto-motive
system under particular circumscribed conditions, analogous to starting a car with
a screwdriver and wire clippers (i.e., hot-wiring the ignition system). D o these effects
really reflect something the mind is designed to do? Automobiles, for example, are
not designed to be started with a screw driver and wire clippers, but we all know
that the design of a car allows such hot-wiring to happen. By analogy, one might
argue that Bargh accumulated impressive evidence that automatic evaluations can
and perhaps do take place, but not necessarily how this fact fits into the evolved
design of our minds. It seems reasonable that speed, efficiency, obligatoriness, and
so on, were important factors i n the evolutionary "design" of evaluation systems,
but it would be a mistake to take away from such experiments the conclusions that
such automatic processes were somehow more primary, more real, or more essen-
tially human than controlled processing. O n e does, after all, have to go to great
lengths to see such effects at all. N o r m a l functioning involves an interaction
between automatic and controlled processing so that behavior is usually a joint
function.
Automatic and Controlled Processing.

Everyday thought and action apparently result from the interaction of two different
kinds of information processing—automatic and controlled processing. T h e focus
of Bargh's chapter is to emphasize the pervasiveness and importance of automatic
processes and to suggest that progress was slowed by an overemphasis o n controlled
processes. Presumably everyday cognitive behavior requires both making accom-
modations to incoming information by forming new concepts (a controlled process)
and deploying old concepts i n order to assimilate new information (an automatic
process). Organisms must learn as they go, but it is obviously efficient to bring to
bear what is already k n o w n whenever possible. Each situation elicits a frame, w h i c h
allows experience i n them to make affective and cognitive sense.
Thus, i n a particular situation, any and all of one's habits, expectations, beliefs,
prejudices, and so o n are activated and waiting i n the wings. In this way, the
organism, like the proverbial boy scout, can "Be Prepared" (e.g., O h m a n , 1988).
Rather than having these habits and primed actions simply elicited directly by
environmental stimuli, as envisioned by S - R psychology, higher organisms have an
elaborate mental waystation i n which all kinds of processing, planning, counterfac-
tual reasoning, and deliberation can take place before the organism is committed
to action. We do not rely purely o n automatic impulse nor engage only i n reasoned
action. We compromise, going forward with some automatic thoughts, feelings, and
actions and inhibiting others. W h e n ready reactions are not sufficient, conscious
thought and deliberation may be triggered that corrects impressions, analyzes
situations, and makes decisions for action. Bargh shows us evidence of those
automatic components involving the deployment of knowledge, expectation, and
habit. Because many of these processes are frequently gated out, they are ordinarily
invisible to us, making Bargh's findings seem surprising. They do tell us, as Bargh
suggests, that m u c h more is going o n mentally than the conscious thoughts of w h i c h
we are aware.
Summary
In the second part of this commentary, we focused o n the implications that findings
of unconscious affective influence have for models that emphasize the role of
implicit attributions for affective feelings. For example, W i n k i e l m a n et al. (1995)
found that attributional manipulations did not alter the ability of subliminally
presented faces to influence the evaluation of neutral stimuli. However, because
the subliminal faces did not elicit affective experience, there was nothing to
misattribute. The effect may therefore be more relevant to the priming literature
than to the mood literature. Differing predictions from priming and affect-as-infor-
mation models suggest that they may apply to different stages of the judgment
process. It was suggested that priming effects may affect categorization, whereas
experienced affect may influence calibration or judgment.
Rather than casting doubt o n the role of attributional processes, we argued that
nonconscious evaluative influences illustrate their role. Presumably the mask used
in subliminal displays interferes with the storage of descriptive features of the
stimulus that would otherwise limit its impact. W i t h o u t the storage of such features,
the primed meaning is not bound to a particular memory and is free to affect any
associated stimulus. We pointed out that this view is quite similar to the model
proposed by Freud about the power of unconscious ideas.
We suggested that Bargh's work o n automatic and unconscious affective proc-
esses is fascinating and important. In addition, we cautioned that it would be an
error to assume that they are more basic and revealing than the conscious systematic
processes with which they are always integrated. Moreover, it would be misleading
to build a conception of affective processes solely o n the automatic model. Affect
is not automatic and unconscious whereas cognition is controlled and conscious.
B o t h processes are intertwined i n both domains.
ACKNOWLEDGMENTS
T h e authors wish to acknowledge N S F Grant S B R 93-11879 and N I M H Grant

M H 5 0 0 7 4 to Gerald Clore, and to the N I M H T 3 2 - M H 1 8 9 3 1 Postdoctoral Train-
ing Program i n Emotion Research for support of Timothy Ketelaar.
REFERENCES
Anderson, N . H . (1968). Likableness ratings of 555 personality trait words. Journal of Personality and
Arnold, M . B. (1960). Emotion and personality. New York: Columbia University Press.
Bargh, J. A . , Chaiken, S., Raymond, P., & Hymes, C . (1996). T h e automatic evaluation effect:
Unconditional automatic attitude activation with a pronunciation task. Journal of Experimental Social
Bower, G . H . (1981). Mood and memory. American Psychologist, 36, 129-148.
Brewer, M . B. (1988). A dual process model of impression formation. In T. K. Srull & R. S. Wyer (Eds.),
Advances in social cognition (Vol. 1, 1-36). Hillsdale, NJ: Lawrence Erlbaum Associates.
Clore, G . L . (1992). Cognitive phenomenology: Feelings and the construction of judgment. In L. L.
Martin & A . Tesser (Eds.), The construction of social judgments (pp. 133-163). Hillsdale, NJ: Lawrence
Erlbaum Associates.
Clore, G. L. (1994). Why emotions are never unconscious. In P. Ekman & R. J. Davidson (Eds.), The
nature of emotion: fundamental questions (pp. 285-290). New York: Oxford University Press.
Clore, G. L., Schwarz, N . , & Conway, M . (1994). Affective causes and consequences of social information
processing. In R. S. Wyer & T Srull (Eds.) The handbook of social cognition (2nd ed., pp. 323-417).
Clore, G. L., Schwarz, N . , & Kirsch, J. (1983, May). Generalized mood effects on evaluative judgments.
Paper presented at the Midwestern Psychological Association, Chicago, IL.
Clore, G. L. & Wilkin, N . (1985, May). Does emotional bias occur during encoding or judgment. Paper
presented at Midwestern Psychological Association, Chicago.
Cosmides, L. (1989). The logic of social exchange: Has natural selection shaped how humans reason?
Studies with the Wasson selection task. Cognition, 31, 187-276.
Damasio, A . R. (1994). Descartes' error: Emotion, reason and the human brain. New York: Grosset/Putnam.
Forgas, J. R, & Bower, G. H . (1988). Affect in social and personal judgments. In K. Fiedler & J. Forgas
(Eds.), Affect, cognition, and social behavior (pp. 183-207). Toronto: Hogrefe International.
Frijda, N . H . (1986). The emotions. New York: Cambridge University Press.
Freud, S. (1915). Instincts and their vicissitudes. Collected papers (Vol. 4, pp. 60-83), New York: Basic
Books.
Freud, S. (1984). The defence neuro-psychoses. Collected papers (Vol. 1). New York: Basic Books.
Gilbert, D. T (1989). Thinking lightly about others: Automatic components of social inference process.
In J. S. Uleman & J. A Bargh (Eds.), Unintended thought (pp. 189-211). New York: Guilford.
Heider, F. (1958). The psychology of interpersonal relations. Hillsdale, NJ: Lawrence Erlbaum Associates.
Isen, A . M . , Shalker, T. E., Clark, M . S., & Karp, L. (1978). Affect, accessibility of material in memory,
and behavior: A cognitive loop? Journal of Personality and Social Psychology, 36, 1-12.
Lazarus, R. S. (1982). Thoughts on the relation between emotion and cognition. American Psychologist,
37, 1019-1024.
Lazarus, R. S. (1991). Emotion and adaptation. New York: Oxford University Press.
LeDoux, D. (1987). Emotion. Handbook of physiology: The nervous system (Vol. 5, pp. 419-459).
Washington, D C : A P A .
Mandler, G. (1984). Mind and body. New York: Norton
Murphy, S. T , & Zajonc, R. B. (1993). Affect, cognition, and awareness: Priming with optimal and
suboptimal stimulus exposures. Journal of Personality and Social Psychology, 64, 723-739.
Oatley, K., & ohnson-Laird, R N . (1987). Towards a cognitive theory of the emotions. Cognition and
Emotion, 1, 29-50.
Ohman, A . (1988). Preattentive processes in the generation of emotions. In V. Hamilton, G . Bower, &
N . Frijda (Eds.), Cognitive science perspectives on emotion and motivation (pp. 127-144). Amsterdam:
Martinus Nijhoff.
Ortony, A . , Clore, G. L., & Collins, A . (1988). The cognitive structure of emotions. New York: Cambridge
University Press.
Osgood, C . E., Suci, G. J., & Tannenbaum, R H . (1957). The measurement of meaning. Urbana: University
of Illinois Press.
Scherer, K. R. (1984). O n the nature and function of emotion: A component process approach. In K.
R. Scherer & R Ekman (Eds.), Approaches to emotion (pp. 293-317). Hillsdale, NJ: Lawrence Erlbaum
Associates.
Schwarz, N . , & Clore, G. L. (1983). Mood, misattribution, and judgments of well-being: Informative
and directive functions of affective states. Journal of Personality and Social Psychology, 45, 513-523.
Schwarz, N . , & Clore, G. L. (1988). How do I feel about it? The informative function of mood. In K.
Fiedler & J. Forgas (Eds.), Affect, cognition, and social behavior (pp. 44-62). Toronto: Hogrefe.
Schwarz, N . & Clore, G. L. (1996). Feelings and phenomenal experiences. In E. T. Higgins & A .
Kruglanski (Eds.), Social psychology: A handbook of basic principles (pp. 433-465). New York: Guilford.
Schwarz, N . , Robbins, M . , & Clore, G. L. (1985, May). Explaining the effects of mood on social judgment.
Paper presented at the Midwestern Psychological Association, Chicago, IL.
Simon, H . (1967). Motivational and emotional controls of cognition. Psychological Review, 74, 29-39.
Srull, T. K . , & Wyer, R. S., Jr. (1979). T h e role of category accessibility in the interpretation of information
about persons: Some determinants and implications. Journal of Personality and Social Psychology, 37,
1660-1672.
Trope, Y. (1986). Identification and inferential processes in dispositional attribution. Psychological Review,
93, 239-257.
Winkielman, P., Zajonc, R. B. & Schwartz, N . (1995, May). Subliminal affective priming is impervious to
attributional manipulations. Paper presented at the meeting of the Midwestern Psychological Associa-
tion, Chicago, IL.
Zajonc, R. B. (1980). Feeling and thinking. Preferences need no inferences. American Psychologist, 35,
151-175.
C h a p t er 7
Ifs and Thens in Cultural Psychology
Dov Cohen
University of Illinois at Urbana-Champaign
Bargh describes automatic processing as following an if-then conditional: if (state

of the world), then (act a certain way). H e persuasively makes the case that m u c h
of our everyday social behavior is driven by this sort of automatic processing. It
occurs when certain conditions are i n place and then runs without needing any
conscious choice or guidance from that point o n (chap. 1, p. 2).
M a n y cultural psychologists would agree with these assessments; m u c h of what
is out there i n the world affects us, but bypasses consciousness altogether. Some
would argue it has to be that way, that culture makes us do things, and leaves our
conscious muttering "just because" when we are asked why.
This agreement between social cognition and cultural psychologists is important
o n two grounds: substantive and methodological. O n substantive grounds, it is
interesting to examine just how m u c h of the if-then conditional is enculturated.
A n d o n methodological grounds, the investigation of automaticity c a n be an
important tool for uncovering shared—yet hidden—cultural truths. In this chapter,
I discuss the mutually reinforcing effects of culture and automatic processes, the
uses of automaticity research i n examining cultural issues, and the implications of
automaticity for collective representations and behavior.
CULTURAL CONSTRUCTS—IFS AND THENS
A s Nisbett and W i l s o n (1977) and Nisbett and Ross (1980) argued, we are correct
i n describing the causes of our own and others' behavior to the extent that our
theories are correct. A s psychologists and as Westerners, however, we have few
theories of behavior that are truly "cultural."
O u r cultural rules are likely to be made up of unquestioned assumptions and
bedrock beliefs about the world. We think that we respond to the world as we do,
121
122 Cohen
not because of our personal construals of the world, but because of how the world
really is (Ross & Nisbett, 1991). A n d just as we fail to understand that our
perceptions are personal construals of reality, it would be surprising if we understood
how m u c h of our perception also reflects the cultural construal of reality. U n d e r -
standing these cultural construals is important because culture is important to both
parts of the if-then conditional—to understanding the way the world is (if) and to
acting o n it once we have understood it (then).
A s issues become more and more abstract, philosophical, or theological, it
becomes obvious that cultures can differ dramatically i n their understanding of
things. But even at some extremely basic levels, cultures can differ drastically i n the
way they take i n the world. Because they are so basic, perhaps the most striking
findings come from research o n the way culture affects our perceptions of the
physical and social world.
Perception of the Physical World

Perceptions that occur so naturally that they seem to be a product of our brain's
hardwiring can vary tremendously across cultures. It is striking, for example, that
the M u l l e r - L y e r illusion, which seems so natural to Westerners, is not so "natural"
among people who do not live i n "carpentered environments" where structures with
right angles predominate (Segall, Dasen, Berry, & Poortinga, 1990). Similarly, it is
striking that "illusions" calling o n us to use depth cues have vastly different effects
o n people who live i n dense forests than those who live i n wide-open vistas (Segall
et al., 1990). O n reflection, such effects make sense. But it is still remarkable when
we read the anthropological account of the B a M b u t i Pygmy who traveled out of the
forest—with its 100 yards of visibility—and watched apprehensively as the "beetles
and ants" miles away on the plain enlarged into buffalos as he drove closer (Turnbull,
1961).
Such findings are striking because we tend to be universalists when it comes to
perception. T h e problem is that the images o n our retinas may be similar, but how
we use our visual cues to turn these images into what we "see" depends a lot o n our
natural and human-made environments. A n d o n percepts that are m u c h less basic,
the amount of cultural construction is, of course, greater.
Perception of the Social World

T h e social world is more ambiguous than the physical. So it should not be surprising
that our social perception is subject to just as much or more preconscious processing
than is our "physical" perception. The meanings given to our world by preconscious
schémas and motivations are highly enculturated.
Bartlett's English subjects hear a N o r t h A m e r i c a n folktale about ghosts and the
supernatural and remember stories that are rationalized and devoid of such c o n -
tent—the ghosts i n one case turning from goblins into the family name of the Ghost
clan (Bartlett, 1950). Morris and Peng's U . S . subjects watch cartoons of fishes
7. Ifs, Thens, and Culture 123
swimming and see a lone fish acting o n internal whims, whereas their Chinese
subjects watch the same cartoon and see a group of fish exerting social influence
(Morris & Peng, 1994). Israeli Arabs watch an episode of Dallas as Sue Ellen leaves
her husband and moves i n with her ex-lover's family, and they see an episode i n
which she moves i n with her own family (Gates, 1995). We see different things as
we try to make our social worlds comprehensible.
Thens
H o w we act o n the world after we perceive it—the "then" part of the conditional—is
also likely to follow cultural scripts that can get triggered automatically. T h e
"self-evident" truth of what we must do i n a situation can be the product of a
preconscious that is highly acculturated. Things that seem like a natural stimu-
lus-response connection differ markedly across cultures. U p until recently, it was
the law i n Texas that a man who killed after finding his wife i n bed with another
man was to be acquitted, being i n such an inflamed state of passion that his behavior
was inevitable under the circumstances. T h e law took account of the "automatic,"
scripted nature of the reaction. Killing was deemed a natural and acceptable
response; mutilation (presumably because it involved more conscious reflection)
was not. T h e enculturated preconscious of a 1970s Texan was understood to have
an if-then allowing killing but not mutilation. T h e enculturated preconscious of a
resident of Massachusetts had neither.
In our o w n research, we found differences i n how likely northerners and

southerners of the United States are to operate with culture of honor scripts.
Specifically, we argued that southerners belong to a culture of honor i n which insults
and affronts must be answered—often violently—or else "face" is lost. In a series
of laboratory experiments, we invited southern and northern students at the
University of Michigan into the lab and provoked them (Cohen, Nisbett, Bowdle,
& Schwarz, i n press). A s subjects walked down a narrow hallway, they were bumped
into by a confederate of the experimenter and called an "asshole."
From the qualitatively different set of reactions, emotions, cognitions, and
physiological responses, it is clear that the meaning of the insult and implicit rules
for what to do about it are profoundly different for northerners and southerners.
Southerners responded to the insult by getting angry, whereas northerners were
more likely to be amused by it, as rated by observers. Southerners were more likely
to subsequently show more hostility and dominance i n their thoughts and actions,
whereas northerners were not. A n d southerners seemed to think the insult would
damage their reputation i n the eyes of onlookers, whereas northerners did not.
Perhaps most strikingly, southerners and northerners showed different physi-
ological responses to the insult. Using a cover story about monitoring blood sugar
levels as they performed mechanical aptitude tests, we took saliva samples from
northerners and southerners before and after being bumped. We measured the
samples for testosterone (a hormone associated with aggression and competition)
and Cortisol (a hormone associated with stress and arousal). Southerners showed
significant increases in both hormones; northerners showed increases i n neither.
124 Cohen
T h e differences between northerners and southerners o n many measures were

not just a matter of degree; the two groups were showing entirely different patterns
of responding. O n e might have expected some of our measures—for example, the
physiological responses or the more reflexive reactions to the b u m p — t o be more
influenced by a type of automatic, uncontrolled response. But for other responses
(e.g., subsequent impression management behaviors several minutes after the
bump), one might have expected more of a mindfulness to mediate the effect.
However, attempts to find some sort of conscious attentiveness o n the part of
northerners or southerners were disappointing. Debriefings with some subjects
resembled a monologue that could have been written for a John Wayne movie. In
many other cases, they produced a lot of confused blustering as subjects groped for
why they did what they did or why it was important or not important to act tough.
For some subjects, after a number of probing questions, it was obvious that the
bedrock answer was "just because." Their goals, actions, and definitions of mascu-
linity were activated by virtue of the situation they were i n . It was not necessarily
a matter of conscious deliberation; they were just following the cultural script.
Attempts to push the analysis to a more introspective level would have been
meaningless.
A s the sociologist John Reed (1981) wrote,
How do southerners learn that violence is acceptable in some circumstances but not
others? This aspect of culture, I suggest, is simply taken in like the others. Like the
words to 'Blessed Assurance,' the technique of the yo-yo, or the conviction that okra
is edible, it is absorbed, pretty much without reflection, in childhood ... [As a
schoolboy], if you were called out for some offense, you fought. I guess you could have
appealed to the teacher but that just—wasn't done. And that phrase speaks volumes.
(p. 13, italics added)
Bargh argues for the automaticity of everyday life, and I think many cultural
psychologists would agree. O u r behaviors are driven by processes, evaluations, and
interpretations that just seem to be automatic, uncontrollable results of the situ-
ations i n which we find ourselves. W h a t makes such processes cultural is that
everyone i n our group has a common, shared understanding and that with similar
environmental inputs, we get strikingly similar results across people of the same
culture. It is not a question of idiosyncratic meaning making. It is a question of
shared knowledge that we assume, that we believe everyone else assumes, and that
is so m u c h " i n the air" that it just appears to be the way the world is.
AUTOMATICITY AND METHODOLOGY

IN CROSS-CULTURAL RESEARCH
T h e study of automaticity also has methodological implications for how cultural

research can be conducted. Triandis and colleagues examined elements of cultural
truth by presenting statements to groups of people and seeing whether and how
quickly, they come to agreement i n their feelings about these statements. If a

statement produces some arbitrarily high level of consensus (for example, 90 %)
and is arrived at quickly (for example, under 60 sec,) Triandis argues that these
statements reflect broad cultural agreement (Triandis, 1996).
A similar test—at the preconscious instead of the verbal level—might also be
useful for cultural psychologists. To the extent that stimuli are eliciting the same
sort of preconscious, automatic reactions from members of the same culture (and
not from members of a different culture), we have a cultural feature that is relatively
deeply embedded i n the individual. (This is just to say that the cultural difference
is affecting us at a very basic level. It is not to say that the cultural difference we
identified is important or worth caring about. T h a t is another matter altogether.)
It is unclear how tight the link might be, but one could test whether or not there
is a positive connection between the richness and strength of a cultural "syndrome"
(Triandis, 1996) and how likely and to what extent that syndrome has "gone
automatic." O n e might expect the strength of the syndrome i n the public world of
collective representations and actions to be positively correlated with the likelihood
of becoming automatic i n the private world of the individual's mind.
The Problem of Verbal Reports
Methodologically, automaticity is also a nice tool for cultural psychologists because

of the problem of verbal reports. Two obvious problems with verbal reports are that
people may not be able to introspect accurately and even if they could, they would
often lie i n reporting their feelings. A s stereotyping researchers showed, the exami-
nation of automatic, preconscious processes is one way to get around these issues
(see, for example, Devine, 1989).
But there are also subtler problems than people not knowing their feelings or
simply lying about them. Because of the web of public and cultural representations
i n w h i c h we are all enmeshed, there are multiple levels to our beliefs and feelings.
A s Baron and Straus (1988) argued, "cultural contradictions" i n many of our
ideologies make it very difficult to verbalize our true beliefs. T h e ideologies of the
culture and subcultures to which we belong overlap and layer o n top of each other,
so that at the top, we may have broad platitudes that obscure deeper differences.
A t the level of verbal reports, we may have these shared truisms, reflecting things
that people say but do not necessarily believe, as they give answers off the top of
their heads. (Or even if they believe them at a symbolic level, they may not act o n
them at an instrumental level—see C o h e n , 1996). Methods that examine our
preconscious processes may have promise i n letting us go beyond these platitudes
to examine cultural values and behaviors that get automatically activated.
Certainly, we can learn something about a culture by examining public dialogue
and representations. But it is also informative to uncover elements of a culture that
may be more hidden i n people's preconscious. A g a i n , the work o n automaticity and
stereotyping can be a good model, allowing us to examine the cultural contradic-
tions of U . S . citizens' feelings about race, gender, freedom, and equality. In addition,
126 Cohen
hard-to-verbalize attitudes about oneself and one's connection to the group and
the social world provide other examples of cases i n which unconscious processing
may be as important to examine as top-of-the-head verbal reports. There will be
many cultural truths beyond those that are most acceptable to speak or even to
consciously think and we may need to go down to the preconscious to get at them.
Weak and "Wrong" Verbal Reports
A n interesting and open question is how much of the norms of our culture we can
articulate and be aware of. Bern's (1972) review (cited by Bargh) argued that
conscious thought plays a limited role i n many processes, with behavioral effects
being larger and more reliable than the cognitive effects that are supposed to
mediate them. "It is not that the behavioral effects sometimes fail to occur as
predicted; that kind of negative evidence rarely embarrasses anyone. It is that they
occur more easily, more strongly, more reliably, and more persuasively than the
attribution changes that are theoretically supposed to be mediating them" (p. 50).
O u r attempts to show that southerners, unlike northerners, hold to a culture of
honor stance produced a similar frustration. We collected data i n laboratory
experiments, field experiments, homicide records, archival records of behavior, and
attitude surveys (Cohen, 1996; C o h e n & Nisbett, 1994, 1996; C o h e n , Nisbett,
Bowdle, & Schwarz, i n press; Nisbett & C o h e n , 1996; Reaves & Nisbett, 1996). In
terms of magnitude, the weakest N o r t h - S o u t h differences we found were o n the
attitude surveys. T h e attitude items were consistent, but the differences were far
less spectacular than differences i n behaviors. People were acting out the culture of
honor m u c h more easily and readily than they were able to articulate the culture
of honor.
We are not alone i n our frustration. Verbal reports may not only be the weakest
indicators of cultural difference; as the verbal reports become more and more
abstract, they may also become more likely to go i n the "wrong" direction. Peng,
Nisbett and Wong (1996) collected data from U . S . and Chinese students o n their
values. They found such surprising results as Chinese students ranking inde-
pendence higher than U.S. students, whereas U.S. students ranked loyalty, respect
for tradition, and humility higher than did Chinese.
A s the questions became more concrete, asking how Chinese and U . S. subjects
would act i n specific scenarios, the cultural differences moved back i n the "correct"
direction (that is, o n the concrete scenarios, results were more i n harmony with
experts' judgments of cultural patterns and stances). It is apparently easier i n some
cases to know the right thing and to do the right thing than it is to say the right
thing. T h e scripts, the motivations, the expectations are all i n our heads. But
because they are either so overlearned (or were never explicitly taught i n the first
place), they may bypass conscious processing altogether. O u r verbal reports and
judgments are most clearly tied to conscious levels of processing, and so they may
never get connected with the cultural rules embedded i n our preconscious. Cultural
patterns may get their power precisely because they work through this more basic
level of processing and are not subject to conscious, rational analyses.
AUTOMATICITY AND CULTURAL SOCIALIZATION
Just how cultural patterns might get into the preconscious is also an open question.
A t some point, they may have been the product of explicit socialization. A s Bargh
writes about our preconscious processes, at some point they "had to be enacted or
engaged i n effortfully and consciously to begin with, and like any skill or mental
process, only after considerable use could they recede into the preconscious" (chap.
1, p. 52).
In other cases, however, the lessons of socialization may have been more
implicitly absorbed from the beginning. To some extent, one might guess that the
implicit versus explicit nature of socialization might be a function of how widely
shared the cultural stance is and how new it is. First, if there are competing models
for how to behave i n a culture, socialization may have to be more explicit. A model
may have to shout to be heard above the others. If a model, however, is widely
shared, socialization may be more implicit, the assumptions of a cultural stance
never having to be made apparent. If it is shared by enough people and suffuses
enough social relations, the cultural stance may be just i n the air. Second, i n a similar
way, old cultural stances may be embedded i n enough practices that they may not
have to be schooled into people as new ones might.
A g a i n , it would be interesting to explore the differences between cultural
processes that are explicitly taught and those that are implicitly absorbed. There
may be profoundly different consequences for learning, changing, overriding, and
for going automatic o n processes that have been taken i n i n different ways. Perhaps
the lessons we learn implicitly are most likely to be the ones that get embedded i n
our preconscious, automatically activated by the environment and giving it mean-
ing—implicitly absorbed and used i n our daily lives without m u c h awareness. It is
this lack of awareness that may make them most resistant to change or to being
overridden by conscious processes. A g a i n , the link between the strength and type
of the public socialization and the private representation within the individual's
mind can be explored.
AUTOMATICITY AND COLLECTIVE PROCESSES
Automaticity also has implications for culture that go beyond individual behavior.
It has implications for what Durkheim (1938) called collective representations or what
Sperber (1991) called public representations. Laws, social policies, institutional
arrangements, and communal myths can all be affected by the collective workings
of the automatic processes Bargh identifies.
In the case of law and policy, the claim would not be that a legislator automat-
ically reacts to a bill or issue and votes yes or no, as if sleepwalking. Rather, the
128 Cohen
claim is that culture puts the initial frame on issues, that this initial frame is very
difficult to change once set, and that this frame starts i n motion affective reactions
that can activate automatically (knee jerk reactions may be an appropriate meta-
phor). Loaded issues—those involving abortion, affirmative action, gun control,
and so on—are likely to trigger such responses. Thus, elites fight to control the
vocabulary and imagery of a debate so that they can impose the frame they want
(Kinder & Sanders, 1990); they seek to control the if part of the if-then condi-
tional—defining the world i n a certain way—because the desired behavioral
reaction can then flow much more easily.
A s an example, framing an issue such as affirmative action as either involving
racial justice or reverse discrimination is going to produce profoundly different
effects. Even incredibly subtle differences—such as framing just the opposition to
affirmative action as based either on opposition to a) reverse discrimination or b)
unfair advantage—has significant effects on the constellation of people's attitudes
(Kinder & Sanders, 1990). Slight changes i n the wording of survey questions can
produce large changes i n opinions about topics where one would guess that people's
attitudes had long ago crystallized: race, poverty, freedoms, and so o n (Iyengar,
1990; Kinder & Sanders, 1990; Schuman & Presser, 1981). If affective and cognitive
reactions can be triggered by such subtle framing changes i n survey questions, one
can imagine the effects that are triggered by the framings and overlearned associa-
tions that are part of one's culture. Culture primes us to think of issues i n certain
ways and prepares us to accept the frames that are put on issues by elites.
Automaticity and Collective Action

Automaticity may be especially important to consider when we go beyond individ-
ual behavior to consider communal actions or consensus activities. Collective
activities require the participation of multiple people i n shared meaning systems.
They require cooperation, compromise, and understanding—either implicitly or
explicitly. To the extent that we all share some very deep meaning system and to
the extent that we all have the same scripts and motivations activated by stimuli,
then unified, collective activity is made m u c h easier.
Automaticity of response by individuals would allow collective behavior to flow
smoothly and with little resistance through a social system. If we are not dealing
with the sorts of affective reactions and scripts that get activated unthinkingly,
however, then m u c h more overt negotiation will be needed to generate and sustain
cultural representations. These clashes of meaning-making and ideology can bring
issues to the fore and change the public dialogue and its results (see Sunstein, 1995).
Multiplier Effects and Automaticity

O n e might even expect to see snowballing or multiplier effects when many people
are working together and operating off the same automatic processes. Bartlett
(1950) noted this as he watched stories travel from person to person among Indian
and among English subjects. A s stories traveled from English person to English
person or from Indian to Indian, gradually the stories became "rationalized," ideas
became omitted, emphases changed. "Individualizing features" dropped out until
the stories took o n a sort of "group stamp or character" that made collective sense
to the different groups (p. 173; for other discussions of similar phenomena, see
C o h e n & Nisbett, 1996; Faludi, 1991, Gates, 1995). T h e development of communal
myths probably owes a lot to the standardization produced by automatic processes
that reflect the biases of a culture.
Analogously, the potential for unthinking action that automaticity creates can
also become important for collective behavior. T h e multiplier effect of collective
processes that have gone automatic may produce behavior m u c h more intense than
that produced by individual automatic processes. Writers such as George O r w e l l
(1950) argued that a sort of automaticity may be an essential ingredient i n political
conformity. H e wrote:
When one watches some tired hack on the platform mechanically repeating the
familiar phrases—bestial atrocities, iron heel, bloodstained tyranny, free peoples of
the world, stand shoulder to shoulder—one often has a curious feeling that one is not
watching a live human being but some kind of dummy: a feeling which suddenly
becomes stronger at moments when the light catches the speaker's spectacles and
turns them into blank discs which seem to have no eyes behind them. A n d this is not
altogether fanciful. A speaker who uses that kind of phraseology has gone some
distance toward turning himself into a machine. The appropriate noises are coming
out of his larynx, but his brain is not involved as it would be if he were choosing words
for himself. If the speech he is making is one that he is accustomed to make over and
over again, he may be almost unconscious of what he is saying, as one is when one
utters responses in church. A n d this reduced state of consciousness, if not indispen-
sable, is at any rate favorable to political conformity, (p. 87)
Even if one does not fully buy her argument, one is reminded of H a n n a h
Arendt's (1964) report o n N a z i war criminal A d o l p h E i c h m a n n . E i c h m a n n , she
wrote, "was genuinely incapable of uttering a single sentence that was not a cliche"
(p. 48); and that between the cliches, the "language rules" and euphemisms that
were used to obscure reality, "the longer one listened to [Eichmann], the more
obvious it became that his inability to speak was closely connected with an inability
to t h i n k " (p. 49).
It was not malicious intentions or even careerism that A r e n d t kept stressing, but
literally Eichmann's "sheer thoughtlessness" (p. 287) that allowed h i m to do his job.
T h e "cog i n the machine" is the appropriate analogy for Arendt's analysis not only
because it suggests the immenseness of the bureaucracy, but also because it suggests
the routine, mindless operation of thousands of people turned into mechanized
pieces. O n l y i n a society where masses of people had turned off their ability to think
could the world have been so turned upside down as it was i n Germany, according
to Arendt's argument. The point, then, is not that political systems go o n automatic
pilot, but that the automaticity of our everyday life has collective consequences
130 Cohen
when individuals act together, that this automaticity is of great aid i n producing
conformity, and—political theorists argue—that this automaticity may be created
by those i n power for such purposes.
S u c h considerations are a long way off from the concerns of the present chapter.
But it seems right that (a) the automaticity of our everyday life is affected by the
cultures we are a part of and that (b) the automaticity of our thoughts and actions
has implications for collective behaviors and representations, as well.
Cultural psychologists have borrowed heavily from cognitive psychology and
social cognition research (Markus & Kitayama, 1991). Automaticity may be an-
other topic that cultural psychologists could make use of i n exploring the ifs and
thens of cultural rules. Automaticity might be used as a conceptual tool i n under-
standing the "just because" nature of culture, as a methodological tool for exploring
issues of culture and cognitive processing, as a technique for uncovering aspects of
culture that are hard to verbalize, and as a potentially important concept connecting
the private and public representations of individuals and collectivities.
ACKNOWLEDGMENTS
Thanks are due to Bob Wyer for his advice and encouragement. Work o n this
chapter was supported i n part by grants from the Russell Sage Foundation and the
University of Illinois Research Board.
REFERENCES
Arendt, H . (1964). Eichmann in Jerusalem: A report on the banality of evil. New York: Penguin.
Baron, L , & Straus, M . A . (1988). Cultural and economic sources of homicide in the United States.
The Sociological Quarterly, 29, 371-390.
Bartlett, F. C . (1950). Remembering: A study in experimental and social psychology. Cambridge, England:
Cambridge University Press.
Bern, D. J. (1972). Self-perception theory. In L. Berkowitz (Ed.), Advances in experimental social psychology
(Vol. 6, pp. 1-62). New York: Academic Press.
Cohen, D. (1996). Law, social policy, and violence: The impact of regional cultures. Journal of Personality
and Social Psychology.
Cohen, D . , & Nisbett, R. E. (1994). Self-protection and the culture of honor: Explaining southern
violence. Personality and Social Psychology Bulletin, 20, 551-567.
Cohen, D . , & Nisbett, R. E. (1996). Field experiments examining the culture of honor: The role of institutions
in perpetuating norms about violence. Unpublished manuscript, University of Illinois.
Cohen, D., Nisbett, R. E., Bowdle, B. F., Schwarz, N . (in press). Insult, aggression, and the southern
culture of honor: A n "experimental ethnography." Journal of Personality and Social Psychobgy.
Devine, P G . (1989). Stereotypes and prejudice: Their automatic and controlled components. Journal
Durkheim, E. (1938). The rules of sociological method. New York: Free Press.
Faludi, S. (1991). Backlash: The undeclared war against American women. New York: Crown.
Gates, H . L. (1995, October 23). Thirteen ways of looking at a black man. The New Yorker, 121, 56-65.
Iyengar, S. (1990). Framing responsibility for political issues: The case of poverty. Political Behavior, 12,
19-40.
Kinder, D . R., & Sanders, L. M . (1990). Mimicking political debate with survey questions: the case of
white opinion on affirmative action for Blacks. Social Cognition, 8, 73-103.
Markus, H . R., & Kitayama, S. (1991). Culture and self: Implications for cognition, emotion, and
motivation. Psychological Review, 98, 224-253.
Morris, M . W., & Peng, K . (1994). Culture and cause: American and Chinese attributions for social and
physical events. Journal of Personality and Social Psychology, 67, 949-971.
Nisbett, R. E., & Cohen, D . (1996). Culture of honor: The psychology of violence in the South. Boulder:
Westview Press.
Nisbett, R. E., & Ross, L. (1980). Human inference: Strategies and shortcomings of social judgment.
Englewood Cliffs, NJ: Prentice-Hall.
Nisbett, R. E., & Wilson, T. D . (1977). Telling more than we can know: Verbal reports on mental
processes. Psychological Review, 8, 231-259.
Orwell, G . (1950). Shooting an elephant. New York: Harcourt Brace.
Peng, K., Nisbett, R. E., & Wong, N . Y. C . (1996). Validity problems of cross-cultural value comparison and
possible solutions. Unpublished manuscript, University of Michigan.
Reaves, A . L , & Nisbett, R. E. (1996). The cultural ecology of rural white homicide in the southern United
States. Unpublished manuscript, University of Michigan.
Reed, J. S. (1981). Below the Smith and Wesson line: Reflections on southern violence. In M . Black &
J. S. Reed (Eds.), Perspectives on the American South: An annual review of society, politics, and culture (pp.
9-27). New York: Cordon & Breach.
Ross, L., & Nisbett, R. E. (1991). The person and the situation: Perspectives of social psychology. New York:
McGraw-Hill.
Schuman, H . , & Presser, S. (1981). Questions and answers in attitude surveys. New York: Academic Press.
Segall, M . H . , Dasen, P R., Berry, J., & Poortinga, Y. (1990). Human behavior in global perspective. New
York: Pergamon.
Sperber, D . (1991). The epidemiology of beliefs. In C . Fraser (Ed.), Psychological studies of widespread
beliefs (pp. 25-44). Oxford, England; Oxford University Press.
Sunstein, C . R. (1995, December 25). True lies. The New Republic, 26, 37-41.
Triandis, H . C . (1996). The psychological measurement of cultural syndromes. American Psychologist,
51, 407-415.
Turnbull, C . M . (1961). Some observations regarding the experiences and behavior of BaMbuti Pygmies.
American Journal of Psychology, 74, 304-308.
Chapter 8
Automaticity and Social Behavior:

A Model, a Marriage, and a Merger
Wendi L. Gardner
Northwestern University
John X Cacioppo
Ohio State University
T h e field of social cognition arguably represents the most influential marriage to date
of any two disciplines i n psychology. A l t h o u g h little more than 10 years old, this
coupling provided us with a wealth of new constructs, paradigms, and perspectives
with which to explore the underpinnings of social behavior. Nowhere has this
marriage been more fruitful than i n the distinction between automatic and controlled
processing i n social behavior. The construct of automaticity leant new understanding
across a variety of domains—being powerfully applied to the exploration of stereo-
typing, attitudes, and attribution, to name but a few. I n this volume, John Bargh
presents a provocative argument that everyday social behavior is driven by automatic
processes rather than controlled or conscious choices. H e oudines three routes
(perceptual, evaluative, and motivational) through which aspects of the environment
can "automatically and nonconsciously produce social behavior" (chap. 1, p. 12).
Bargh deserves credit for the evidence presented, and for the synthesis of findings
across domains and disciplines that this chapter represents. We applaud Bargh's
challenge to social psychology to look beyond people's intuitive explanations of
their own behavior; indeed, much of our o w n work has also dealt with the
exploration of social processes that are hidden from verbal reports and overt actions
(Cacioppo, Gardner, & Berntson, 1996; Crites, Cacioppo, Gardner, & Berntson,
1995). Kudos aside, we do have a few quibbles with the current chapter, namely the
equating of social phenomena invariably with automatic phenomena, and we begin
our response addressing this point. T h e n , we explore an additional, complementary
perspective that includes biological as well as cognitive phenomena i n exploring the
psychological underpinnings of everyday life.
133
134 Gardner & Cacioppo
ARE SOCIAL PHENOMENA INVARIABLY AUTOMATIC?
Throughout the chapter, Bargh argues that social processes are implemented largely
by automatic rather than conscious or controlled processes. His argument seems to
rest o n two lines of reasoning. First, the study of social psychology and the study of
automaticity are presented as equivalent because they both share the specification
of if-then relations between the environment and behavior:
Thus, research and theory in both domains, social psychology and automaticity, have
at the core the specification of if-then relations between situational events and
circumstances on the one hand, and cognitive, emotional and behavioral effects on
the other... My thesis is that because social psychology, like automaticity theory and
research, is also concerned with phenomena that occur whenever certain situational
features or factors are in place, social psychological phenomena are essentially
automatic, (chap. 1, p. 3)
A l t h o u g h it is true that research programs concerned with both social psychology

and automaticity share the goal of specifying causal (if-then) relations between
events, this feature is common to most empirical sciences. Scientific endeavors
generally rest upon the assumption of lawful and determined relations between
events—otherwise empirical study would be pointless. Thus, we disagree with the
tenet that causality and automaticity are synonymous, as Bargh seems to argue i n
this quote:
If the situation activates the same goal in nearly everyone so that it is an effect that
generalizes across individuals, and can be produced with random assignment of
experimental participants to conditions, the only preconditions for the effect are those
situational features (chap. 1, p. 3).
If the defining feature of an automatic effect is that it can produce a statistically

significant difference between randomly assigned experimental groups, then em-
pirical social psychology would essentially be the study of automaticity. But if this
were the case, the value of the distinction between automatic and controlled
processes would be lost. After all, controlled and conscious responses may also be
related i n a causal fashion to preceding events. Consider the case of persuasion.
O n e robust finding i n the literature is that greater attitude change is found when
participants are exposed to strong rather than weak persuasive arguments and also
possess both the motivation and the ability to process these messages (Eagly &
Chaiken, 1993; Petty & Cacioppo, 1986). The amount of persuasion is thus causally
related to the quality of the message, but only through the mediation of effortful
and deliberative processing. W h e n participants are distracted (traditionally thought
to affect controlled rather than automatic processes) no difference between strong
and weak messages o n attitude change is found (Petty & Cacioppo, 1986). D u a l
process models of persuasion represent one instance i n social psychology i n which
the distinction between more and less effortful processing was fruitful. However,
8. Automaticity and Social Behavior 135
they also demonstrate causal relations between antecedents (message quality) and
consequences (attitude change) that are mediated by nonautomatic effects. We
suppose that Bargh could apply a similar argument for automatic processes medi-
ating the effects of message quality i n the elaborative or systematic condition as he
did for explaining conscious mediators of Latane and Darley's (1970) diffusion of
responsibility effects:
... if these conscious processes do mediate the situational effect, then they must
themselves be tied to those situations in an if-then relation.... For the effect to occur
with regularity across individuals, the feeling of less responsibility and the decision not
to help, and so on, are also automatic reactions to the situational information across
different individuals. (p. 6)
However, we argue that this diminishes the value of the automatic versus
controlled distinction. If the attitude change occurring i n response to peripheral
cues i n the distraction condition is automatic (because it requires neither intention
nor elaboration to occur), then to argue that the effect of message quality that
replaces the effect of peripheral cues i n the effortful processing condition is also
automatic because it can be determined i n a causal fashion obfuscates the distinc-
tion between the two routes of persuasion.
A n o t h e r example also suggests the value of retaining the distinction between
automatic and controlled processing. Individuals who differ i n need for cognition
were shown to exhibit chronic motivational differences i n the effortful processing
of information (Cacioppo & Petty, 1982; Cacioppo, Petty, Feinstein & Jarvis, 1996).
T h e auto-motive model suggests that these differences stem from an automatic
trigger between environmental cues and a goal, a suggestion that the extant
literature supports (see review by Cacioppo et al., 1996). H i g h need for cognition
individuals do not necessarily choose to process the information they are given
effortfully. Rather, the presentation of new information may automatically trigger
thoughtful processing as a strategy, because this strategy was used repeatedly by the
individual similar situations. Despite the probable automaticity of this initial trigger,
however, the subsequent processing and the outcomes of such processes (i.e.,
judgments concerning the information) are nonautomatic i n nature, although they
are determined, at least i n part, by differences i n information quality. O n c e again,
a definition for automatic processing that demands that the initial trigger between
the presentation of new information and the motivation to think i n high need for
cognition individuals be called automatic but also demands that the link between
these individuals' judgments and information quality (mediated by thoughtful,
effortful processing) be called automatic, is problematic.
Bargh's first argument for the equivalence of social phenomena and automatic
phenomena is largely a semantic one, with the conclusion depending o n the way i n
which automaticity is defined. O n e might argue about what the critical attributes
of automatic and controlled processes should be (Bargh, 1994; Posner & Snyder,
1975; Shiffrin & Schneider, 1977), and our point is not to favor one or another set
of such attributes. Instead, our point is that the presence of if-then relations or
determinism alone should not be a critical defining attribute of automatic versus
controlled processes. The distinction between automatic and controlled process i n

empirical inquiries is valuable i n social psychology—defining automaticity i n such
a way that makes all processes automatic does not seem to serve the field well.
Semantics play less of a role i n Bargh's second line of reasoning for the similarity
of social psychological phenomena and automatic phenomena. Bargh uses the
inevitability of continued findings of automaticity as evidence that social phenom-
ena are indeed automatic. Bargh points to research programs that progressively
showed more of an influence of automatic processes i n domains i n which thoughtful
processes used to be dominant (e.g., attribution, attitudes). H i s o w n work persua-
sively demonstrates the role of automatic processes even i n the case of strategic and
goal-driven processing.
W h a t the research programs reviewed i n Bargh's chapter demonstrate, however,
is that many social psychological phenomena can be driven by primarily automatic
processes, not that they are driven primarily by them i n everyday life. T h e task
remaining for social psychologists interested i n this area is to determine when these
processes are relatively more automatic or controlled, and how the way i n w h i c h
the processes are mediated change the outcomes i n terms of social behavior. This
specification of the antecedents and consequences of automatic and controlled
social processes was nicely demonstrated i n Devine's (1989) model of stereotype
activation and i n Gilbert's (1989) model of attributional processes. Both of these
models showed the sufficiency of automatic processes i n producing certain social
outcomes (e.g., racial prejudice, dispositional attributions), but also demonstrated
the correction of these outcomes when people possess the motivation and the ability
to engage i n controlled and deliberative processing.
There will always be a place i n social psychology for the study of controlled
processes, because they are often precursors of automatization i n the first place. Bargh
uses the wonderfully descriptive metaphor of automatic processes as mental servants,
likening the automatic activation of mental constructs to the behavior of a well-
trained English butler who knows precisely what you want and when you might want
it. Let us not forget who trained the butler. Jeeves may bring the Queen Mother tea
with cream (rather than with lemon) at precisely 3:15 (rather than 4:00), because
this is what she initially made the effort to ask for at the beginning of his employ.
Likewise, an automatic interpretation of "self-discipline" to explain the behavior of
a thin and stylish colleague who forgoes dessert (rather than interpreting the behavior
as a dislike of sweets) is a result of initially conscious deliberation of the motivations
of similar others i n similar situations. A s Bargh states:
... if an individual makes the same categorization ... of a given act ... consistently
over time ... if an individual makes the same evaluation ... of a given object
consistently over time ... if an individual has the same goal and intention within a
social situation repeatedly over time ... then that goal representation ... will become
active automatically ... (chap. 1, p. 7)
T h e precursors of automaticity i n social processes are often controlled processes

and conscious decisions; this is sufficient to ensure that the study of social psycho-
logical phenomena should not be reduced to the study of automatic processes.
A t the same time, Bargh's questioning of the prevailing assumption of conscious

mediation i n social behavior is clearly warranted. By pointing out areas of social
psychology i n which the discovery of automatic processes called into question the
omnipresence of controlled processes, he argues that the assumption that thought-
ful processes mediate social behavior "should be treated with the same scientific
scrutiny as the assumption of automaticity" (chap. 1, p. 5). We ardently agree—the
burden of proof should rest equally o n those who assume automaticity and those
who assume conscious deliberation. However, we argue that continued findings of
automatic processes occurring i n what was thought to be the domain of controlled
processes do not obliterate the value of the study of those controlled processes. After
all, as psychologists continue to break down any controlled process, additional
subprocesses will inevitably be identified—many of w h i c h will be automatic. T h i s
does not diminish the impact of the controlled process or the importance of
understanding the controlled process; it merely represents an additional level of
analysis, a complementary but not exclusive perspective. A s a concrete example,
consider eating behavior, or more specifically, the phenomena of dieting. A n
understanding of the psychology of dieting may be approached from a variety of
perspectives. O n e focus might be the identification of relevant neural processes,
such as the role of hypothalamic functioning i n the regulation of appetite. A n o t h e r
could be the exploration of automatic cognitive processes evoked i n response to
either environmental or internal hunger cues. Both of these represent powerful
determinants of dieting behavior; but knowledge at neither of these levels of analysis
diminishes the value of understanding the behavior at the level of psychological
experience—the investigation of the conscious, controlled, and effortful strategies
of the dieter.
T h e investigation of automatic processes provides a valuable perspective o n
social phenomena. Indeed, it is yet another example of how our field has been
vitalized by the marriage between social and cognitive perspectives. Bargh's review
of the evidence that social phenomena are largely automatic is one testament to
the value of multiple levels of analysis. H e clearly demonstrates how m u c h more of
social behavior can be understood when the unconscious and automatic cognitive
underpinnings of behavior are considered i n addition to the level of conscious
experience—we would like to encourage an even further expansion.
A MULTILEVEL PERSPECTIVE ON
THE PSYCHOLOGY OF EVERYDAY LIFE
Just as the understanding of automatic cognitive processes brought m u c h to bear

to the explanation of social phenomena, we believe that a multilevel social neuro-
science perspective (e.g., including an understanding of biological processes) would
likewise enrich the field. Bargh's work shows that research at no single level of
analysis may sufficiently describe social psychological phenomena—neither an
understanding of the conscious processes alone nor a sole dependence upon
unconscious processes captures the richness of social experience, nor predicts the
vagaries of social behavior. Similarly, a biological perspective i n isolation could never
hope to explain the complexities of everyday psychological phenomena. However,
as an addition to the understanding of both the unconscious automatic and
consciously controlled cognitive processes, consideration of the neural underpin-
nings of these processes may have m u c h to offer (cf. Cacioppo & Berntson, 1992).
Bargh opens the door to this type of multilevel integrative analysis when
describing the neurophysiological evidence i n support of the automaticity of evalu-
ation. Because of the fundamental and adaptive nature of the approach-avoidance
distinction across species, the ability to quickly evaluate stimuli would be predicted
to be hardwired into the biological system (Berntson, Boysen, & Cacioppo, 1993;
Lang, Bradley, & Cuthbert, 1990; Zajonc, 1980). Bargh reviews provocative re-
search that provides evidence this may be so and demonstrates how preconscious
evaluation may then impact behavior.
In explaining the possible operation of unconscious motivations, Bargh draws
support from research observing neuropsychological populations. For example,
when a split-brain patient complies with the command issued to the isolated (and
nonverbal) right hemisphere, the verbal left hemisphere will almost immediately
rationalize the action taken. Just as the field of social psychology gained from an
understanding of the fundamentals of cognitive psychology, many advances i n
cognitive psychology were spurred by research and theory i n neuroscience. T h e
understanding of unconscious perception, implicit memory, and similar cognitive
phenomena were significantly advanced since the 1970s by insight gained from the
neuropsychological literature. The observation of "Hindsight" i n functionally blind
patients, the startling procedural learning capacities of the amnesiac H . M . , and the
"unconscious" abilities of the isolated right hemisphere i n split-brain patients made
essential contributions to the field of cognition (Squire, 1987).
Likewise, a recognition of the capabilities and constraints of the neural systems
underlying both conscious and unconscious mentation can benefit theory i n social
cognition. A s an example, consider two metatheoretical underpinnings of the
model Bargh proposes: (a) the assumption of parallel rather than serial processing
and (b) independent but interacting processing modules. Bargh states:
My implicit adherence to the stage model nearly led me to conclude that the extent
of direct automatic influences of the environment on social cognition was limited to
perceptual interpretation ... it was the metaview of serial processing stages that made
the notion that motivations could be directly activated by the current environmental
information difficult for me to see. (chap. 1, p. 51)
Bargh further outlines the pervasiveness of the serial stage conceptualizations of

cognition that were proposed i n the 1960s i n the thought and theories of modern
social cognition researchers. H e offers a plausible reason for the continued adher-
ence to serial models: O u r own consciousness is serial i n nature and thus a serial
model of cognition is intuitively appealing. Bargh's chapter argues for going beyond
our intuitive understanding of the motivators of social behavior. O n e way i n which
we may be guided i n this task is through observation and appreciation of the

underlying neural system. The workings of the neural system may provide valuable
information i n addition, and sometimes i n opposition to the conclusions toward
which common sense deduction often urges us. For instance, the assumption of
serial processing may be based on our own conscious experience; that this experi-
ence is overlaid upon a nonserial set of processing is evident from research on the
neural underpinnings of this experience (Kolb & Whishaw, 1996; Thompson,
1993). Thus, models such as Bargh's, i n which many mental operations are carried
out simultaneously rather than i n serial may provide better explanation of psycho-
logical phenomena, and are more realistic from a purely neurophysiological point of
view.
The second metatheoretical view Bargh's model adheres to—that of inde-
pendent but interactive processing modules—is also supported by the neuroscien-
tific literature. Research by L e D o u x and colleagues (LeDoux, Iwata, C h i c h e t t i , &
Reiss, 1988) aptly demonstrated that the acquisition and representation of affective
memories can operate at multiple, interrelated levels within the brain. Conceptual
extensions of this work also made their way into social psychology. For instance,
Cacioppo, Marshall-Goodell, Tassinary, and Petty (1992) discussed how racial
attitudes can be acquired at higher and lower levels of the nervous system and how
each may operate (sometimes i n conflict) to produce both intentional and unin-
tentional stereotypic behavior. Finally, recent data implying that the brain areas
involved i n nonevaluative judgments point to overlapping but different neural
structures than those involved i n nonevaluative judgments (Cacioppo, Crites, &
Gardner, i n press) and the finding that discrete neural areas are implicated i n the
experience of positive and negative affect (George et al., 1995) provide further
illustrations of interacting but separate neural systems underlying socially relevant
processes. Thus, research i n neuroscience is consistent with the independent but
interactive processing modules Bargh suggests. Bargh draws o n the neuroscientific
literature (Gazzaniga, 1985; LeDoux, 1989) to support his model of simultaneously
active, separate, and interacting processing modules, and i n so doing takes the first
step toward the integrative analysis necessary to understand the psychology of
everyday life.
A social neuroscience perspective may also offer new challenges to the field of
social cognition. T h e neuropsychological example of commisurotomy (split-brains)
is a fascinating one because it provides further support for the notion that motives
can operate outside of conscious awareness, and because it questions the very
notion of what the field has termed "conscious" process. Is consciousness simply
the ability to report our experience? If so, then research with split-brain patients
illustrates that "unconscious" processes can also behave i n what appears to be a
controlled and deliberative fashion. For example, reports of "cross-cueing" i n split
brains exist, wherein the left hand (right hemisphere) will "act out" a word to which
it was exposed in a one-handed game of charades, attempting to communicate the
word to the left hemisphere. This behavior appears deliberative, despite the fact
that it is nonverbal. Currently, the social psychological literature equates " u n c o n -
scious" with "nonverbalizable." The inability of a research participant to report
stimuli, motives, or processes is the operational definition of unconsciousness

prevalent i n the field. T h e neuroscientific literature calls into question the suffi-
ciency of using the inability to verbally report experience as either a definition or
an operationalization of unconsciousness. Questions concerning the nature and
function of consciousness exemplify the type of research that must be answered by
integrating a variety of perspectives, from biological to social.
CONCLUSION
Bargh presents a provocative view of the power and omnipresence of automatic

processes i n social behavior. H e provides abundant evidence for the impact of
automaticity i n everyday life, and a model that aptly describes the ways i n which
automatic processes c a n mediate the situation-behavior relationship—even i n
domains such as goals and strategy selection, which intuitively seem conscious and
controlled. O n e of the few aspects we found with which to argue is the sentiment
that all social phenomena are essentially and invariably automatic phenomena.
Bargh argues forcefully for the pervasiveness of automaticity i n social psychology i n
order to overcome "dominant, even implicit, assumptions to the contrary" ( c h a p . l ,
p. 50). We wholeheartedly endorse the importance of existing and future research
i n this domain, but with the reminder that, however informative it proves to be, the
study o f automatic processes is a complement to existing perspectives, not a
substitute. W h a t Bargh's chapter does well is to exemplify the value of nontradi-
tional perspectives on social phenomena. The appeal of this type of synthesis is clear
i n the breadth of psychological effects Bargh is able to explain. A complete
understanding of the psychology of everyday life, of course, will require looking
across levels—from basic genetic and neural processes, to unconscious and auto-
matic cognition, to conscious and deliberate thought and action to contextual and
cultural determinants. By describing the contribution of automatic cognitive proc-
esses i n goal-directed action and other social phenomena, Bargh fires an early salvo
for a multilevel integrative analysis of social behavior. Indeed, research o n the
automaticity of social behavior might best represent the benefits of the marriage
between social and cognitive psychology. It would be regrettable if i n the enthusiasm
to include this level of analysis, other levels of analysis were ignored or defined to
be irrelevant. Just as the marriage of two people should neither obliterate the
separate identity of either one nor exclude interpersonal relationships with others,
the marriage between social and cognitive psychology should have synergistic
effects o n both fields while allowing mergers with other fields and levels of analysis.
REFERENCES
Bargh, J. A . (1994). The four horsemen of automaticity: Awareness, intention, efficiency, and control
in social cognition. In R. S. Wyer, Jr., & T. K . Srull (Ed.), Handbook of social cognition (2nd ed., pp.
8-13). Hillsdale, NJ: Lawrence Erlbaum Associates.
Berntson, G . G . , Boysen, S. T., & Cacioppo, J. T. (1993). Neurobehavioral organization and the cardinal
principle of evaluative bivalence. Annals of the New York Academy of Sciences, 702, 75-102.
Cacioppo, J. T., & Berntson, G . G . (1992). Social psychological contributions to the decade of the brain:
T h e doctrine of multilevel analysis. American Psychologist, 47, 1019-1028.
Cacioppo, J. T , Crites, S. L , Jr., & Gardner, W. L. (in press). Attitudes to the right: Evaluative processing
is associated with lateralized late positive event-related brain potentials. Personality and Social
Psychology Bulletin.
Cacioppo, J. T., Gardner, W. L , &. Bertson, G . G . (in press). Attitudes and evaluative space: Beyond
bipolar conceptualizations and measures. Personality and Social Psychology Review.
Cacioppo, J. T., Marshall-Goodell, B. S., Tassinary, L. G . , & Petty, R. E . (1992). Rudimentary determi-
nants of attitudes: Classical conditioning is more effective when prior knowledge about the attitude
stimulus is low than high. Journal of Experimental Social Psychology, 28, 207-233.
Cacioppo, J. T., & Petty, R. E. (1982). T h e need for cognition. Journal of Personality and Social Psychology,
42, 116-131.
Cacioppo, J. T., Petty, R. E., Feinstein, J., & Jarvis, B.(1996). Dispositional differences in cognitive
motivation: T h e life and times of individuals varying in need for cognition. Psychological Bulletin, 119,
197-253.
Crites, S. L., Jr., Cacioppo, J. T., Gardner, W. L., & Berntson, G . G . (1995). Bioelectrical echoes from
evaluative categorization: II. A late positive brain potential that varies as a function of attitude
registration rather than attitude report. Journal of Personality and Social Psychology, 68, 997-1013.
Eagly, A . H . , & Chaiken, S. (1993). The psychology of attitudes. Fort Worth: Harcourt Brace.
Gazzaniga, M . (1985). The social brain. New York: Basic Books.
George, M . S., Ketter, T. A . , Parekh, P. I., Horwitz, B., Herscovitch, P., & Post, R. M . (1995). Brain
activity during transient sadness and happiness in healthy women. American Journal of Psychiatry, 152,
341-351.
Gilbert, D . T. (1989). Thinking lightly about others: Automatic components of the social inference
processes. In J. S. Uleman & J . A . Bargh (Eds.), Unintended thought (pp. 189-211). New York: Guilford.
Kolb, B., & Whishaw, I. Q . (1996). Fundamentals of human neuropsychology. New York: Freeman.
Lang, P. J., Bradley, M . M . , & Cuthbert, B. N . (1990). Emotion, attention, and the startle reflex.
Psychological Review, 97, 377-395.
Latané, J, B., & Darley, J. M . (1970). The unresponsive bystander: Why doesn't he help? New York:
Appleton-Century-Crofts.
LeDoux, J. E. (1989). Cognitive-emotional interactions in the brain. Cognition and Emotion, 3, 267-289.
LeDoux, J. E., Iwata J . , Chichetti, P., &Reiss, D.J. (1988). Different projections of the central amygdaloid
nucleus mediate autonomic and behavioral correlates of conditioned fear. Journal of Neuroscience, 8,
2517-2529.
Petty, R. E., & Cacioppo, J. T. (1986). Communication and persuasion: Central and peripheral routes to
attitude change. New York: Springer-Verlag.
Posner, M . I., & Snyder, C . R. R. (1975). Attention and cognitive control. In R. L. Solso (Ed.), Information
processing and cognition: The Loyola symposium (pp. 55-85). Hillsdale, NJ: Lawrence Erlbaum Associ-
ates.
Shiffrin, R., & Schneider, W. (1977). Controlled and automatic human information processing: II.
Perceptual learning, automatic attending, and a general theory. Psychological Review, 84, 127-190.
Squire, L. (1987). Memory and the brain. New York: Oxford University Press.
Thompson, R. (1993). The brain: A neuroscience primer. New York: Freeman.
Zajonc, R. B. (1980). Feeling and thinking: Preferences need no inferences. American Psychologist, 35,
151-175.
Chapter 9
Rendering Accessible Information

Relevant: The Applicability
of Everyday Life
Curtis D. Hardin
University of California at Los Angeles
Alexander J. Rothman
University of Minnesota
The proposition that the situation can shape an individual's thoughts, feelings and
actions is so familiar as to be a social psychological truism (e.g., Ross & Nisbett,
1991). However, in demonstrating the power of the situation to "automatically"
affect judgment and behavior, John Bargh extends this thesis by challenging
prevailing assumptions that social behavior is necessarily predicated on conscious
choice. In this chapter we focus our attention on a question that follows directly
from the issues set out in the target article: Given the rich array of information
accessible to the individual at any moment, which elements of information will
actually guide judgment and behavior?
By way of example, consider the following situation. You sit on a New York City
subway, bombarded with a kaleidoscope of sights, sounds, and smells. Alone and in
combination, the range of people and objects in the immediate environment elicit
a changing array of thoughts and feelings. Your eyes settle on the person across the
aisle. What determines your specific impression? The assumption that thoughts and
feelings influence judgment as a function of their relative accessibility is noncon-
troversial (e.g., Uleman & Bargh, 1989), and is the focal mechanism in Bargh's
argument for the automaticity of everyday life. But given the diversity of stimuli in
the environment, there are a multitude of thoughts and feelings accessible at any
point in time. What determines the degree to which your impression is guided by
the music blaring from a nearby "boombox", the homeless person sleeping in the
corner of the car, the antisemitic graffiti scrawled across the wall, or the magazine
article in your lap? In short, what accounts for the selective and predictable manner
143
144 Hardin & Rothman
in which accessible information is used in judgment? The answer that has received
the most support to date is that information is used to the degree that it is accessible
(for a review of relevant models see Higgins, 1996). However, social cognition
researchers also recognized that the influence of accessible information is predicated
on its relevance or "applicability" to the task at hand (e.g., Higgins, 1990, 1996;
Wyer &Srull, 1989). In the target chapter, Bargh notes, for example, "an accessible
representation does not operate on its own, in the absence of relevant input, but
only in the presence of environmental information for which it is applicable" (chap.
1, p. 41, italics added).
Despite the consensus that applicability is essential to delineating the course of
information processing, little was done other than to define the term applicability
and specify the point at which it is assessed in the information-processing sequence.
In this chapter, we argue for the importance of identifying the factors that regulate
the use of accessible information. Our broad aim is to suggest that understanding
the automaticity of everyday life requires an understanding of what defines the
applicability of accessible information to the task at hand. Moreover, we propose a
direction from which an understanding of applicability might be pursued—one we
believe is consistent with the essential thrust of Bargh's approach.
STRUCTURING THE AUTOMATICITY OF EVERYDAY LIFE
As Bargh and colleagues demonstrated, the extent to which the situation can shape
human judgment and action is dramatic. Actions that to all accounts are products
of conscious reflection (e.g., interrupting someone, judging someone hostile) are,
in fact, regulated by situational information affordances and their ability to elicit
automatic responses. Because the power of the situation rests in large part on its
ability to render particular sets of thoughts, feelings, and goals accessible, research
has focused on providing evidence that judgment and behavior are influenced by
what information is most accessible (e.g., Higgins & Bargh, 1987). For example, the
observation that, under particular conditions, people wait longer before interrupt-
ing an experimenter provides evidence that the concept of politeness versus
rudeness was successfully rendered more accessible by a recent priming manipula-
tion (Bargh & Chen, 1995). Such findings reveal that exposure to specific sets of
stimuli may elicit the "preconscious creation of the psychological situation" (Bargh,
chap. 1, p. 7). Traditionally, experiments manipulated the degree to which informa-
tion is accessible and then observed its effect in situations specifically designed to
facilitate its use. Thus, the conditions that determine whether information, once
rendered accessible, is used have yet to be well elucidated. Recall that an early
demonstration of the effect of accessible information on judgment also revealed the
specificity with which these effects are obtained (Higgins, Rholes, & Jones, 1977).
Although manipulating the accessibility of information applicable to the target's
behavior (e.g., reckless, adventurous), influenced judgment, manipulating the
accessibility of inapplicable information (e.g., listless) had no such effect. Given the
9. Applicability 145
facility with which information can be rendered accessible independently of its

applicability to judgment, identifying what makes information applicable is neces-
sary to understanding mental life.
What makes information applicable? Most theorists define applicability in terms
of the "fit" or "overlap" of features between the information represented in memory
and the particular object, task, or judgment under consideration, yet none specify
the principles by which features overlap. Interestingly, the absence of formalized
attention to the question of applicability may have led to the tacit assumption that
the relevance or applicability of information to a particular task can be intuitively
discerned (cf. Higgins, 1996). In practice, if the applicability of information is not
defined by the intuitions of researchers, then it is defined by the intuitions of pilot
subjects who make explicit judgments of "relevance" during the development of
experimental materials. Even when studies manipulate applicability (e.g., Higgins
et al., 1977; Sedikedis, 1990), their methods reveal the tacit assumption that
applicability is essentially knowable at a conscious level to experimenters, subjects,
or both. Although we agree that this sometimes may be the case, it is unlikely to be
the general rule (e.g., Nisbett & Wilson, 1977). We think it is likely that when
applicability is fully understood, it will be defined by processes that often operate
outside of subjective awareness or deliberative control, consistent with Bargh's
primary thesis.
Congruent with Bargh's framing of automaticity, we propose that applicability
may be usefully approached in learning terms. Specifically, we define information
as applicable to the extent it was utilized in similar circumstances for similar
purposes. From this perspective, applicability reflects the contingent relationship
between specific information and the specific uses to which it is put, which develop
out of the habitual application of information in particular situations over time
(Rothman & Hardin, 1997; see also, Lewicki, 1985; Smith, 1990). Applying
information to judgment is conceived as a behavioral action that can be learned
with surprising specificity and maintained over extended periods of time (Smith,
1990; Smith, Stewart, &Buttram, 1992). The similarity between this operationali-
zation and Bargh's specification of accessibility reflects the assumption that appli-
cability mediates both the initial activation and the subsequent application of
information in the judgment process. For example, the observation that a particular
behavior (e.g., pounding the table) tends to elicit a specific evaluative response
reflects a contingent relationship that developed over time between the eliciting
cue and a particular response to that object (Smith & Zarate, 1992). Similarly, the
extent to which one habitually thinks that giving to panhandlers is generous rather
than patronizing defines generous as applicable to the act and patronizing as inappli-
cable, even under circumstances in which patronizing might be at least as accessible
as generous (e.g., Erdley & D'Agostino, 1988).1 The virtue of operationalizing
applicability in this manner is that it suggests a process by which features of
1
Information that is applicable to judgment does not always result in assimilation effects on judgment
(Schwarz& Bless, 1992).
information become linked in terms of their pragmatic utility. This, in turn, may
afford more complete predictions of when accessible information is actually used.
Thus, we postulate that learned patterns of information use may regulate which
accessible contents are applied to a particular task at hand.
The concepts of accessibility and applicability are closely interwoven. In situations
where the same stimulus serves as both the prime (i.e., renders specific information
accessible) and the target of judgment, accessibility and applicability are extremely
difficult to distinguish. For example, if a colleague automatically and preconsciously
elicits feelings of envy and these feelings are used to evaluate him or her, then the
applicability that mediated their accessibility may be the same as that which mediated
their use. However, even when the same object serves as both the eliciting prime and
the target of judgment, information from other sources may influence judgment.
Most research concerning the role of accessibility in social judgment focused on
situations where information rendered accessible in ostensibly unrelated contexts
influences judgment or behavior. When information rendered accessible by one
aspect of the situation influences a response to a different aspect of the situation,
moments of applicability—as distinguished from accessibility—can be identified at
two places in the judgment process. First, applicability is likely involved in deter-
mining the association patterns that affect accessibility (e.g., the degree to which
hearing the phrase break a leg brings the concept aggressive to mind). Second,
applicability is likely involved in the relationship between information currently
accessible and the use to which it can subsequently be put (e.g., the degree to which
you have previously considered football an aggressive game). Applicability is involved
both at the stage of information activation and the stage of information use.
From this perspective, applicability does not require a conscious assessment,
consistent with Bargh's basic thesis. To the extent that conscious attention is
brought to bear on the judgmental process, it may occur only when there is sufficient
reason to question whether information should be used. Thus, judgments of
perceived usability (Higgins, 1996) or appropriateness (Strack, Martin, & Schwarz,
1988) are relegated to the role of moderators in the application of information to
judgment. People do not rely on information that is highly applicable to judgment
when features of the situation undermine its perceived informational value. Tradi-
tionally, threats to the diagnosticity of accessible information focused on the
perceived source of that information (e.g., Schwarz & Clore, 1983; Zanna &
Cooper, 1974). Because people are frequently aware of the thoughts that come to
mind before the particular source of those thoughts can be identified Oohnson,
Hashtroudi, & Lindsay, 1993), they may act based on the contents currently
accessible in the absence of any information about the source, operating under the
assumption that information, if applicable, is pertinent to the judgement at hand
(cf. Gilbert, 1991). Because the accurate identification of a particular source is more
likely when information is processed in a detailed manner, people may consciously
assess the informational value of accessible information only when features of the
judgment task either heighten the salience of the alternative source (e.g., Schwarz
& Clore, 1983) or increase the motivation to systematically process information
(e.g., Thompson, Roman, Moskowitz, Chaiken, & Bargh, 1994). Although we
assume that the process by which the applicability of information to judgment is

determined operates outside of conscious awareness, the actual use of information
applicable to the task at hand may be marked by a fluency o r — i n Vera and S i m o n s
(1993) term—afunctional transparency, and the phenomenological experience of
perceiving or using information fluently may be consciously recognized. In fact,
research suggests that under some conditions this experience can indicate the
validity or accuracy of a judgment (e.g., Begg, Armour, & Kerr, 1985; Hasher,
Goldstein, & Toppino, 1977).
APPLICABILITY IN PERSON
JUDGMENT AND STEREOTYPING
A l t h o u g h , as Bargh suggests, stereotype use is defined i n part by the accessibility of

the stereotype i n question, many stereotypes—such as those involving gender or
ethnicity—are accessible to the point of operating automatically for virtually
everyone (e.g., Devine, 1989). Yet, even those stereotypes that are chronically
accessible within a specified culture or group are not applied to social targets
indiscriminately. By definition, stereotyping involves the selective attribution of
specific traits or characteristics to particular social groups or their members (e.g.,
Ashmore & D e l Boca, 1981; H a m i l t o n & Sherman, 1994), epitomizing the impor-
tance of applicability i n constraining information processing.
Hence, stereotypes are broadly consensual rules of applicability that reflect
chronic (societal) tendencies to associate particular kinds of information with
particular social categories. A t the individual level of analysis, stereotypes reflect
particular learning histories of regularly practiced, group-based judgments. T h e
applicability of information to judgment reflects the degree to which an individual
has associated an attribute with a particular group (i.e., a use rate) rather than the
degree to which that attribute is objectively associated with a particular group (i.e.,
a base rate). Hence, the contingent learning perspective is orthogonal to the
question of whether a particular stereotype has a "kernel of truth."
Target Variables in Applicability

From the contingent use perspective of applicability, stereotyping is revealed i n
differences i n the accessibility of stereotype-related information and, more impor-
tantly, by the degree to which accessible information is discriminately applied to
judgments of social targets. We found that highly accessible information related to
gender stereotypes is applied selectively as a function of the gender of the target
(Banaji, Hardin, & Rothman, 1993). In a series of experiments, participants judged
the protagonist of a brief story after having recently unscrambled sentences that
described either aggressive, dependent, or neutral behaviors. Participants judged a
person who performed slightly aggressive behaviors as more aggressive after being
primed for aggressiveness if his name was Donald, but not when her name was
Donna. Despite the fact that aggression was highly accessible, it was not applied to
a female target, even though the male and female target had performed the identical
set of behaviors. In a parallel experiment, highly accessible dependence-related
information led participants to judge a person who performed slightly dependent
behaviors as more dependent, but only if the person was female. Again, the use of
accessible information in judgment was delimited by its applicability to the target's
social category. Similar evidence that highly accessible information is applied
selectively as a function of (gender) social category is the demonstration that
familiarity is falsely misattributed as fame more for male than female names (Banaji
& Greenwald, 1995). These findings suggest that in reflecting patterns of contin-
gent information use, stereotypes and other person categories selectively define
information as applicable to the judgment at hand.
Given the specificity with which highly accessible stereotyped information is

used in judgment, some findings Bargh reports are intriguing because they suggest
that applicability of information may become generalized (e.g., Bargh, Chaiken,
Govender, & Pratto, 1992). For example, Bargh and Chen (1995) found that
participants primed with Black versus White faces subsequently acted more hos-
tilely toward the experimenter. Bargh frames the effect in terms of a perception-be-
havior link, in which Black faces activate an African-American stereotype, which
activates the concept hostility, which in turn facilitates hostile behavior. This
analysis conforms to a spreading activation model of information processing, but
current evidence from the cognitive literature suggests that activation does not
spread very far, if at all (cf. McNamara, 1992; Ratcliff & McKoon, 1988; Shelton
& Martin, 1992). To the extent spreading activation models fail to find direct
support, such findings would seem to complicate the identification of rules of
applicability. The generality of Bargh and Chen's effect would be particularly
striking if the experimenter were not African-American, as it would suggest that
activating a stereotype can even influence judgments of people unrelated to the
stereotype (cf. Fazio, Jackson, Dunton, & Williams, 1995). Similar questions have
been raised by Bargh (1994) concerning Devine's (1989) finding that subliminal
exposure to words associated with an African-American stereotype increased
hostility judgments of a race-unspecified target.
However, most evidence is consistent with the proposition that applicability can
be defined by the degree to which concepts were directly associated in the past. For
example, in demonstrations of automatic gender priming effects, effect size varied
directly with apparent applicability (Banaji & Hardin, 1996; Hardin & Banaji, in
press). Across trials of word pairs presented serially in rapid succession, prime words
and target words varied according to gender. Automatic gender priming was
revealed by shorter judgment latencies when the prime-target relationship was
gender congruent than incongruent. The role of applicability was revealed by
findings that priming effects were larger under conditions in which (a) the primes
were gendered by definition (e.g., Mr, Mrs, host, hostess) versus gendered by
stereotypical association (e.g., nurse, doctor, secretary, mechanic), and (b) the judg-
merits were made on common pronouns (she, he) versus first names (Suzy, Johnny).
This pattern of results is consistent with presumable patterns of stimulus association
across participants' learning histories. For example, it is likely that each of the
priming stimulus words (e.g., secretary, mechanic) was contingently associated with
each of the target pronouns (e.g., he, she) many more times than was any given first
name (e.g., Marie, Miguel).
To this point, the applicability effects we discussed were found in experiments
examining whether a single type of accessible information was used in judgment.
What happens when two types of information, relevant to judgment, are simulta-
neously accessible? In a series of experiments, we employed a paradigm in which
two types of information associated with the availability heuristic were made
accessible prior to judgment (Rothman & Hardin, 1997). Participants recalled
either three or six behaviors relevant to the subsequent target of judgment.
Although the amount of judgment relevant information recalled was greater in
the six than three behavior condition, participants found that it was subjectively
easier to recall three than six behaviors (see also Schwarz et al., 1991). Hence,
subjective ease of retrieval (availability heuristic) and the amount of information
retrieved were both accessible for judgment, but were cast in methodological
opposition. We discovered that in outgroup judgment, people based their evalu-
ation on the subjective ease with which information came to mind, but in ingroup
judgment people based their evaluation on the amount of information that was
recalled. These results suggest that even under conditions in which two principally
independent pieces of information are simultaneously accessible (i.e., amount of
information and the ease with which it came to mind), perceivers will selectively
rely on one type versus another as a function of its applicability to the target of
judgment.
Why should ease be applicable to judgments of an outgroup, whereas the amount
of accessible information be applicable to judgments of an ingroup? We propose that
the relative use of accessible information is guided by chronic information-to-situ-
ation contingencies that reflect the prior use of experiential versus declarative
information. Specifically, outgroup judgment has long been observed to be espe-
cially feeling-based. For example, Asch (1952) suggested that the subjective feelings
of acceptance versus rejection are the defining feature of outgroup attitudes. More
recent work in social cognition identified similar relationships between subjective
feeling states and outgroup judgments (e.g., Dovidio & Gaertner, 1993; Esses,
Haddock, & Zanna, 1993). The repeated reliance on feeling-based information in
evaluating outgroups may have defined experiential information as particularly
applicable to judgments of outgroups. In contrast, because ingroup judgment is
characterized more by attention to declarative content (e.g., Linville, Fischer, &
Salovey, 1989), information such as the number of behaviors that come to mind
may be more applicable to judgment. The identification of chronic situation-to-in-
formation associations involving the contingent use of subjective feeling states and
declarative information in judgment is analogous to Bargh's demonstrations that
relatively abstract goals can become chronically invoked for use in particular
situations through the regular instantiation of situation-to-goal associations. Such
research also suggests that applicability can operate at levels that are extremely
unlikely to be mediated by conscious perceptions of utility or relevance.
Person Variables in Applicability
If applicability arises out of actual patterns of information use, then different

learning histories should result in different patterns of information use. O f course,
the construct accessibility literature demonstrates that what information is chroni-
cally accessible may differ from person to person (Higgins & Bargh, 1987), and
further shows that these differences are likely to have resulted from the repeated
use of that information over time (Bargh, Bond, Lombardi, & Tota, 1986). However,
we suggest that individuals differ not only in what information is (chronically)
accessible, but also in the functions that accessible information can serve.
Effects of the contingent use of accessible information may be revealed by
findings that social category membership regulates personal experience, particularly
when differences in information use are attributable to people's social group
membership. For example, Wyer, Bodenhausen, and Gorman (1985) demonstrated
that women and men used the identical accessible information in different ways
when evaluating a rape incident. Previously activating thoughts about close,
personal relationships caused men to attribute more responsibility to the rape
victim, but caused women to attribute more responsibility to the rapist. Rendering
accessible thoughts about women as sex objects decreased men's beliefs in the
victim's credibility but increased women's beliefs in the victim's credibility. The
different inferences drawn as a function of participants' gender category may reflect
the different perspectives afforded by their respective group memberships.
Research suggests that White and African-American participants have very
different implicit responses to Black and White faces that are consistent with their
presumable habits of responding to White versus African-Americans (Fazio et al.,
1995). For White participants, responses to positive adjectives were faster after
White than Black faces, and responses to negative adjectives were faster after Black
than White faces. In contrast, for African-American participants, responses to
positive adjectives were faster after Black than White faces, and responses to
negative adjectives were faster after White than Black faces. These effects are
consistent with the hypothesis that group membership affects the relevance of
accessible information to the target of judgment. In this example, although Afri-
can-Americans are well aware of prevailing negative stereotypes, they are likely to
have had much less practice than Anglo-Americans in contingently applying
negative versus positive descriptors to African-Americans (see also Zarate, Bonilla,
& Luevano, 1995).
Although differences in group membership can correspond to differences in
information use, this may not always occur. According to the contingent use
perspective, differences in stereotyping between groups should not be observed to
the extent that particular stereotyped associations are equally practiced across
groups. The incidental use of gender-based stereotypes by women and men appears
to be one such case. Across three very different experimental paradigms (Banaji et
al., 1993; Banaji & Greenwald, 1995; Banaji & Hardin, 1996), both men and
women equivalently used accessible information stereotypically in judgments of
male and female targets (see also Stangor, Lynch, Duan, & Glass, 1992). These
findings suggest that men and women are both highly practiced in the contingent
application of information as a function of the gender social category.
Individual differences in judgment suggest that applicability arises out of habits
of information use, and Bargh's research offers a number of relevant examples. The
research on the automatic but contingent link between power and sex is a case in
point (Bargh, Raymond, Pryor, & Strack, 1995). Most germane is evidence that
although power and sex are linked for those men who score highly on either Pryor's
(1987) Likelihood to Sexually Harass scale (LSH) or Malamuth's (1989a, 1989b)
Attraction to Sexual Aggression scale (ASA), only scores on the A S A correlated
with male participants' perceptions of a female confederate after they were primed
with power related information. Assuming that the L S H and the A S A scales tap
different ways in which power and sex may be associated—consistent with their
moderate intercorrelations (Bargh et al., 1995)—these findings fit nicely with the
learning approach to applicability. The L S H scale focuses on situations in which
men are willing to use power over women to obtain sexual favors, whereas the A S A
scale focuses on the degree to which men are aroused by sexual acts that involve
power. In comparing how these scales operationalize the association between power
and sex, the L S H appears to focus on a man's willingness to use power to obtain
sex, whereas the A S A identifies a tendency to find power sexually appealing. If so,
to the extent that power elicits sexual feelings (in need of an applicable outlet),
individual differences in A S A should predict responses to the (applicable) female
confederate. O n the other hand, the power-sex link assessed by the LSH may better
predict individual differences in sexist behaviors under conditions that more easily
allow men to feel that they have power over the female confederate (e.g , Rudman
& Borgida, 1995), a situation that may be more applicable to behavioral habits
tapped by the LSH.
(Un)learning Applicability
Changes in applicability may reveal the extent to which personal learning histories
define the relevance of information to judgment. To the degree that applicability
develops through the repeated use of information, applicability effects should get
stronger over time. Operationalizing applicability in learning terms, as Bargh defines
automaticity, has several developmental implications for the use of stereotypes and
other categories. Not only should evidence accrue that the use of these categories
is learned, but also that it may be unlearned. Although there are currently more
empirical examples of the former than the latter, in principle they are two sides of
the same coin.
The development of category use in judgment is observed both across the life
span as well as within individuals as a function of new instantiations of information
use. For example, research suggests that although the use of race and gender
categories develops rapidly throughout childhood, very young children are less
adept than older children at applying the categories in social judgment (e.g., Bigler
& Liben, 1993; Fagot & Leinbach, 1989; Ruble & Stangor, 1986). Moreover, a
recent review by Fyock and Stangor (1994) suggests that (a) adults have more
strongly developed expectations about social groups than do children, with expec-
tancy-congruent effects larger for adults than children, and (b) more practiced
expectations, such as gender and ethnicity, produce larger expectancy confirmation
effects than do less practiced expectations, such as personality dimensions (see also
Andersen, Klatsky, & Murray, 1990).
The observation that it is extremely difficult to avoid using overlearned catego-
ries is consistent with much of Bargh's own work on chronic accessibility. Moreover,
evidence consistent with this proposition in the domain of stereotyping is not only
anecdotally familiar but also experimentally corroborated (e.g., Stangor et al.,
1992). For example, Nelson, Biernat, and Manis (1990) showed that the influence
of expectations about men versus women's height are particularly difficult to
overcome. Participants judged males taller than females even under conditions in
which (a) task instructions exhorted them to avoid gender-based inferences, and
(b) they were told accurately that the female and male targets in the judgment
sample were on average equally tall. The only manipulation that significantly
reduced participants' reliance on stereotyped expectations involved applicability.
Reliance on stereotyped expectations was attenuated when women were described
in masculine terms and men were described in feminine terms.
Development over the life span is mediated by the actual use of information by
people in particular circumstances, and such effects have also been identified (e.g.,
Lewicki, 1985; Smith, 1990). For example, Lewicki, Hill, and Sasaki (1989) showed
that unconsciously perceived associations can be learned and may continue to
influence subsequent information processing even after the objective association
no longer exists. Smith and Zarate (1990) found that prior exposure to group-level
information as compared to individuating information facilitated subsequent pro-
totype-based processing of group-relevant targets. Prior categorization of a social
target in terms of outgroup membership similarly facilitated the application of
group-based information in a subsequent similar judgment situation (Zarate &
Smith, 1990). Under some conditions, the perceived diagnosticity of learned
contingencies may even preclude people from paying attention to other equally
relevant sources of information (Sanbonmatsu, Akimoto, & Gibson, 1994).
People are known to be quite sensitive to the degree of association among stimuli,
including dimensions of particular relevance to psychologists. For example, Ford
and Stangor (1992) varied the "diagnosticity" of particular traits in stereotype
formation about artificial groups. Using the A N O V A F-ratio as a metaphor for
social judgment, they found that holding variability constant, the larger the mean
difference between two groups on a trait, the more the trait was used as a basis for
group stereotyping. Congruently, holding mean differences constant, the less vari-
ability on a trait distinguishing two groups, the more the trait was used as a basis
for stereotyping (see also Park & Hastie, 1987). Findings reported by Hilton and
Fein (1989) make explicit the connection between such findings and the contingent
learning approach to applicability. They found that participants generally ignored
clearly irrelevant individuating information, whereas they sometimes neglected
relevant categorical information in the presence of irrelevant information that was
frequently (but not always appropriately) applied in similar judgments.
One hopeful implication of this body of research is that, given that the applica-
bility of information develops from its repeated use in particular situations, the
regular instantiation of new parameters that constrain its application should render
even very accessible information inapplicable to judgment. For example, if aspects
of a working environment successfully constrain sexually harassing behavior, the
thoughts a sexist man finds accessible by the presence of a female colleague may,
over time, become less and less applicable to his judgment and behavior in the
workplace. By the same token, however, the specificity with which these contingen-
cies operate may be such that in situations not regularly invoking these particular
constraints (e.g., interactions at the coffee shop), sexually harassing behavior will
be no less frequently practiced. Such an analysis is consistent with the observation
that people express seemingly different "personalities" across situations. The prem-
ise that the applicability of information to judgment guides prejudicial behavior may
also be relevant to the observation that prejudice toward outgroups continues to
reflect people's feelings about a group despite meaningful changes in their beliefs
about that group (Dovidio & Gaertner, 1993). Providing people with new informa-
tion about an outgroup can alter the accessibility of particular beliefs, but these new
beliefs may have a limited effect on judgment as long as people consider their
phenomenological response to an outgroup to be more applicable.
CONCLUSION
We argue that although the work of Bargh and his colleagues provides a substantial
advance in understanding when and how information is utilized in judgment, the
focus emphasized the role of accessibility with little attention paid to issues con-
cerning relevance and applicability. In outlining the manner in which applicability
can shape information use, we hope to encourage future empirical and theoretical
developments in this area. Moreover, we believe that by placing applicability in a
learning perspective it can be easily integrated into the theoretical perspective
outlined in the target chapter.
REFERENCES
Andersen, S. M . , Klatsky, R. L., & Murray, J. (1990). Traits and social stereotypes: Efficiency differences
in social information processing. Journal of Personality and Social Psychobgy, 59, 192-201.
Asch, S. (1952). Social psychology. Englewood Cliffs, NJ: Prentice-Hall.
Ashmore, R. D., & Del Boca, E K. (1981). Conceptual approaches to stereotypes and stereotyping. In
D. Hamilton (Ed.), Cognitive processes in stereotyping and intergroup behavior (pp. 1-35). Hillsdale, NJ:
Banaji, M . R., & Greenwald, A . G . (1995). Implicit gender stereotyping in judgments offame.Journal
Banaji, M . R., & Hardin, C . D. (1996). Automatic stereotyping. Psychological Science, 3, 136-141.
Banaji, M . R., Hardin, C . D., & Rothman, A . J . (1993). Implicit stereotyping in person judgment. Journal
Bargh, J. A . (1994). T h e Four Horsemen of automaticity: Intention, efficiency, awareness, and control
as separate issues in social cognition. In R.S. Wyer & T. K. Srull (Eds.), Handbook of social cognition
(2nd ed., Vol. 1, pp. 1-40). Hillsdale, NJ: Lawrence Erlbaum Associates.
Bargh, J. A . , Bond, R. N . , Lombardi, W. J., & Tota, M . E. (1986). T h e additive nature of chronic and
temporary sources of construct accessibility. Journal of Personality and Social Psychology, 62, 869-878.
Bargh, J. A . , Chaiken, S., Govender, R., & Pratto, F. (1992). T h e generality of the automatic attitude
Bargh, J. A . , & Chen, M . (1995). The chameleon effect: Automatic social perception produces automatic social
behavior. Unpublished manuscript, New York University.
Bargh, J. A . , Raymond, R, Pryor, J. B., &Strack, F. (1995). Attractiveness of the underling: A n automatic
power-sex association and its consequences for sexual harassment and aggression .Journal of Personality
Begg, I., Armour, V , & Kerr, T. (1985). O n believing what we remember. Canadian Journal of Behavioral
Science, 17, 199-214.
Bigler, R., & Liben, L. (1993). A cognitive-developmental approach to racial stereotyping and recon-
structive memory in Euro-American children. Child Development, 64, 1507-1518.
Devine, E G . (1989). Stereotypes and prejudice: Their automatic and controlled components. Journal
Dovidio, J. E , & Gaertner, S. L. (1993). Stereotypes and evaluative intergroup bias. In D. M . Mackie &
D. L. Hamilton (Eds.), Affect, cognition, and stereotyping: Interactive processes in group perception, (pp.
167-193). San Diego, C A : Academic Press.
Erdley, C . A . , & D'Agostino, P. R. (1988). Cognitive and affective components of automatic priming
effects. Journal of Personality and Social Psychobgy, 54, 741-747.
Esses, V. M . , Haddock, G . , & Zanna, M . R (1993). Values, stereotypes, and emotions as determinants
of intergroup attitudes. In D. M . Mackie & D. L. Hamilton (Eds.), Affect, cognition, and stereotyping:
Interactive processes in group perception (pp. 137-166). San EHego, C A : Academic Press.
Fagot, B., & Leinbach, M . (1989). T h e young child's gender schema: Environmental input, internal
organization. Child Devebpment, 60, 663-672.
Fazio, R. H . , Jackson, J. R., Dunton, B. C , & Williams, C . J. (1995). Variability in automatic activation
as an unobtrusive measure of racial attitudes: A bona fide pipeline? Journal of Personality and Social
Psychobgy, 68, 768-781.
Ford, T. E., & Stangor, C . (1992). The role of diagnosticity in stereotype formation: Perceiving group
means and variances. Journal of Personality and Social Psychology, 63, 356-367.
Fyock, J., & Stangor, C . (1994). The role of memory biases in stereotype maintenance. British Journal of
Gilbert, D. T (1991). How mental systems believe. American Psychobgist, 46, 107-119.
Hamilton, D., & Sherman, J. (1994). Stereotypes. In R. S. Wyer & T K. Srull (Eds.), Handbook of social
cognition (2nd ed., Vol. 2, pp. 1-103). Hillsdale, NJ: Lawrence Erlbaum Associates.
Hardin, C . D., & Banaji, M . R. (in press). Gender in language and thought. Social Cognition.
Hasher, L., Goldstein, D., & Toppino, T (1977). Frequency and the conference of referential validity.
Journal of Verbal Learning and Verbal Behavior, 16, 107-112.
Higgins, E. T. (1990). Personality, social psychology, and person-situation relations: Standards and
knowledge activation as a common language. In L. A . Pervin (Ed.), Handbook of personality: Theory
and research (pp. 301-338). New York: Guilford.
Higgins, E. T. ( 1996). Knowledge activation: Accessibility, applicability, and salience. In E. T. Higgins
& A . W. Kruglanski (Eds.), Social psychobgy: Handbook of basic principles, (pp. 133-168). New York:
Guilford.
9. A p p l i c a b i l i t y 155
Higgins, E. T , & Bargh, J. A . (1987). Social cognition and social perception. Annual Review of Psychobgy,
38, 369-425.
Higgins, E. T , Rholes, W. S., & Jones, C . R. (1977). Category accessibility and impression formation.
Journal of Experimental Social Psychobgy, 13, 141-154.
Hilton, J. L., & Fein, S. (1989). T h e role of typical diagnosticity in stereotype-based judgments. Journal
Johnson, M . K., Hashtroudi, S., & Lindsay, D. S. (1993). Source monitoring. Psychobgical Bulletin, 114,
3-28.
Lewicki, P. (1985). Nonconscious biasing effects of single instances on subsequent judgments. Journal of
Lewicki, R, Hill, T , & Sasaki, I. (1989). Self-perpetuating development of encoding biases. Journal of
Experimental Psychobgy: General, 118, 323-337.
Linville, R, Fischer, G . W , &Salovey, R (1989). Perceived distributions of the characteristics of in-group
and out-group members: Empirical evidence and a computer simulation. Journal of Personality and
Malamuth, N . M . (1989a). T h e attraction to sexual aggression scale: Part One. Journal of Sex Research,
26, 26-49.
Malamuth, N . M . (1989b). The attraction to sexual aggression scale: Part Two. Journal of Sex Research,
26, 324-354.
McNamara, T. R (1992). Theories of priming: I. Associative distance and lag. Journal of Experimental
Psychobgy: Learning Memory, and Cognition, 18, 1173-1190.
Nelson, T. E., Biernat, M . R., & Manis, M . (1990). Everyday base rates (sex stereotypes): Potent and
resilient. Journal of Personality and Social Psychobgy, 59, 664-675.
Nisbett, R. E., & Wilson, T D. (1977). Telling more than we can know: Verbal reports on mental
processes. Psychobgical Review, 84, 231-259.
Park, B., & Hastie, R. (1987). Perception of variability in category development: Instance versus
abstraction based stereotypes. Journal of Personality and Social Psychobgy, 53, 621-635.
Pryor, J. B. (1987). Sexual harassment proclivities in men. Sex Roles, 17, 269-290.
Ratcliff, R., & McKoon, G . (1988). A retrieval theory of priming in memory. Psychobgical Review, 95,
385-408.
Ross, L., & Nisbett, R. (1991). The person and the situation: Perspectives in social psychobgy. New York:
McGraw-Hill.
Rothman, A . J., & Hardin, C . D. (1997). Differential use of the availability heuristic in intergroup
judgment. Personality and Social Psychobgy Bulletin, 23, 123-138.
Ruble, D., & Stangor, C . (1986). Stalking the elusive schema: Insights from developmental and
social-psychological analyses of gender schemas. Social Cognition, 4, 227-264.
Rudman, L. A . , & Borgida, E. (1995). The afterglow of construct accessibility: T h e behavioral
consequences of priming men to view women as sexual object. Journal of Experimental Social Psychobgy,
31,493-517.
Sanbonmatsu, D. M . , Akimoto, A . A . , & Gibson, B. D. (1994). Stereotype-based blocking in social
explanation. Personality and Social Psychology Bulletin, 20, 71-81.
Schwarz, N . , & Bless, H . (1992). Constructing reality and its alternatives: A n inclusion/exclusion model
of assimilation and contrast effects in social judgment. In L. L. Martin & A . Tesser (Eds.), The
construction of social judgments (pp. 217-245). Hillsdale, NJ: Lawrence Erlbaum Associates.
Schwarz, N . , Bless, H . , Strack, E , Klumpp, G . , Rittenauer-Schatka, H . , & Simons, A . (1991). Ease of
retrieval as information: Another look at the availability heuristic. Journal of Personality and Social
Psychobgy, 61, 195-202.
Schwarz, N . , & Clore, G . L. (1983). Mood, misattribution, and judgments of well-being: Informative
and directive functions of affective states. Journal of Personality and Social Psychobgy, 45, 513-523.
Sedikedis, C . (1990). Effects of fortuitously activated constructs versus activated communication goals
on person impressions. Journal of Personality and Social Psychobgy, 58, 397-408.
Shelton, J., & Martin, R. C . (1992). How semantic is automatic semantic priming? Journal of Experimental
Psychology: Learning, Memory, and Cognition, 18, 1191-1210.
Smith, E. R. (1990). Content and process specificity in the effects of prior experiences. In T. K. Srull &
R. S. Wyer (Eds.), Advances in social cognition (Vol. 3, pp. 1-60). Hillsdale, NJ: Lawrence Erlbaum
Associates.
Smith, E. R., Stewart, T. L., & Buttram, R. T. (1992). Inferring a trait from a behavior has long-term,
highly specific effects. Journal of Personality and Social Psychology, 62, 753-759.
Smith, E. R., & Zarate, M . A . (1990). Exemplar and prototype use in social categorization. Social
Cognition, 8, 243-262.
Smith, E. R., & Zarate, M . A . (1992). Exemplar-based model of social judgment. Psychological Review,
99, 3-21.
Stangor, C , Lynch, L., Duan, C , & Glass, B. (1992). Categorization of individuals on the basis of multiple
social features. Journal of Personality and Social Psychology, 62, 207-281.
Strack, E , Martin, L. L., & Schwarz, N . (1988). Priming and communication: Social determinants of
information use in judgments of life satisfaction. European Journal of Social Psychology, 18, 429-442.
Thompson, E. R, Roman, R. J., Moskowitz, G . B., Chaiken, S., & Bargh, J. A . (1994). Accuracy
motivation attenuates covert priming effects: T h e systematic reprocessing of social information.
Uleman, J. S., &. Bargh, J. A . (1989). Unintended thought. New York: Guilford.
7-48.
Wyer, R. S., Bodenhausen, G . V , & Gorman, T. F. (1985). Cognitive mediators in reactions to rape.
Wyer, R. S., & Srull, T. K. (1989). Memory and cognition in its social context. Hillsdale, NJ: Lawrence
Erlbaum Associates.
Zanna, M . P., & Cooper, J. (1974). Dissonance and the pill: A n attribution approach to studying the
arousal properties of dissonance. Journal of Personality and Social Psychology, 29, 703-709.
Zarate, M . A . , Bonilla, S., & Luevano, M . (1995). Ethnic influences on exemplar retrieval and
stereotyping. Social Cognition, 13, 145-162.
Zarate, M . A . , & Smith, E. R. (1990). Person categorization and stereotyping. Social Cognition, 8,
161-185.
Chapter 10
The Automaticity of Academic Life:

Unconscious Applications
of an Implicit Theory
Gordon D. Logan
University of Illinois
In the target chapter, Bargh reports a number of fascinating instances of automatic

processing in everyday life, in which people unconsciously apply previously acti-
vated cognitive structures to routine behaviors, like walking, talking, and reacting
to people. My thesis in this chapter is that the phenomena that Bargh and others
observe in everyday life also occur in academic life. In fact, one could construe much
of the current work on automaticity in the social cognition literature as an example
of unconscious application of a previously activated cognitive structure. The
cognitive structure is the resource theory of automaticity. Bargh, himself, does not
endorse this theory consciously or otherwise in his target article or in his other
writings (e.g., Bargh, 1989, 1992; Bargh & Pietromonaco, 1982), but many of his
colleagues do (e.g., Gilbert, Pelham, &Krull, 1988; Wegner, 1994). My commentary
is addressed more to them than to Bargh.
Students of social cognition are not alone in applying this cognitive structure
unconsciously. Resource theory is the cornerstone of many approaches to automat-
icity outside the attention literature. Prominent examples appear in the literature
on memory (e.g., Jacoby, 1991), comprehension (Just & Carpenter, 1992), devel-
opment (Bjorklund & Harnisfeger, 1990; Kail & Salthouse, 1994), and aging (e.g.,
Craik & McDowd, 1987; Salthouse, 1990). My comments are addressed to these
literatures as well.
I, myself, was a victim of the phenomenon for many years. I became enamored
with resource theory in graduate school—Kahneman's (1973) Attention and effort
was an epiphany for me that led to my dissertation, my first research grant, and my
first publications (e.g., Logan, 1978, 1979). For the following decade, I thought
"automatically" in terms of resource theory throughout my academic life. In
157
158 Logan
seminars, in the library, at my writing desk, and in my lab, resource theory was my
first reaction to problems of attention and automaticity. Resource theoretic analyses
came to mind easily and they seemed right. There was no reason to think of
alternatives. Then, for reasons that escape me now, I became aware of my automat-
ism and I wondered if I could think about attention and automaticity without
thinking of resources. Imagine no resources. It was hard at first, but it became easier
with practice. My instance theory of automaticity (Logan, 1988) was the first fruit
of that labor, and it convinced me that I could think productively about attention
and automaticity without thinking in terms of resource theory. My current work
continues the practice, and resource theory no longer pervades my thinking. Free
at last, I wonder what other habits of thought I have automatized and what influence
they have on my academic life.
My purpose is to expose the cognitive structure to conscious awareness by
describing the resource theory of automaticity and articulating the problems that
led to its demise in the attention literature a decade ago. My hope is that by making
the theory explicit, I can make researchers aware of its influence on their profes-
sional behavior in academic life so that they can counteract it. My goal is to
neutralize the influence of a bad theory on otherwise excellent research. Research
by Bargh and others on the effects of making people aware of unconscious influences
suggests that I may be more likely to produce a contrast effect than neutralization,
but the contrast effect may have a positive influence on the future development of
the field.
RESOURCE THEORY
History
Resource theory was a child of the 1960s. It arose as a reaction to single-channel
(Welford, 1952) and filter (Broadbent, 1958) theories of attention. It has historical
roots in psychoanalysis—Freud's libido was a source of mental energy (see Schwartz
& Schiller, 1971; also see the foreword in Kahneman, 1973)—but the major impetus
was to resolve an insoluble issue in the attention literature.
All theories of attention must explain the fact that human's capacity for
processing information is limited. Single-channel theory and the filter theories that
followed it proposed a limited-capacity channel that could only deal with one thing
at a time. In front of the limited-capacity channel, interfacing it with the sensory
world, were preattentive processes that processed information in parallel without
any limitations on capacity. Between the preattentive processes and the limited-ca-
pacity channel was a filter that chose among the outputs of the preattentive process,
selecting one to pass through the limited-capacity channel (see Fig. 10.1).
This view of attention raised important questions about the locus of selection:
Where in the chain of processing did the filter reside? What level of processing
could stimuli reach without attention? What is the highest level of processing that
is done preattentively? The first theories adopted an early selection perspective,
10. Academic Life 159
Preattentive Filter Limited-

Processes Capacity
Channel
FIG. 10.1. Schematic description of single-channel or filter theories of attention, including

preattentive processes, a selective filter, and a limited-capacity channel.
arguing that the filter followed low-level sensory processes, that stimuli outside the
focus of attention received only cursory analysis, that preattentive processes dealt
with raw sensory features and nothing deeper (e.g., Broadbent, 1958). Almost
immediately, competing theories were proposed that advocated late selection
(Deutsch & Deutsch, 1963; Norman, 1968). They argued that the filter followed
semantic processing, that stimuli could be identified and categorized without
attention, and that preattentive processes computed meaning.
Although the theories were clear, the evidence was mixed. Some experiments
showed evidence of early selection (e.g., Cherry, 1953) while others showed evi-
dence of late selection (e.g., Gray & Wedderburn, 1960). Still others showed
ambiguous evidence. In a classic experiment, Moray (1959) showed that subjects
detected their own names on an unattended channel 35% of the time. The
percentage of detections was too high for early selection, which predicts a 0%
detection rate, and too low for late selection, which predicts a 100% detection rate.1
Resource theory arose in reaction to researchers' frustration with the difficulty
of resolving the early- versus late-selection issue. Resource theory finessed questions
about the locus of selection by removing attention from the chain of processes that
go from stimulus to response. The chain of processing remained intact, but attention
was no longer a stage of processing (see Fig. 10.2). Instead, it was a source of "mental
energy" that activated processing stages. In principle, any process could be activated
by allocating mental energy, so attentional selection could occur at any locus, early
or late (see Kahneman, 1973; Moray, 1967; Posner & Boies, 1971). By the middle
of the 1970s, filter theory was dead and resource theory was the dominant paradigm
for attention research.
What Is Resource Theory?

Often, it is as important to know what a thing is not as to know what it is. Resource
theory is many things to many people, especially resource theories that are implicit
and influence researchers' thinking unconsciously. Within the attention literature,
the theory became explicit, first as it rose and then as it died, and it is worth
recapitulating the explicit assumptions of resource theory here.
Moray's (1959) data were interpreted initially as evidence against early selection (which is appro-
priate) and evidence for late selection (which is not appropriate). It was not until the 1980s that
researchers noted that the 35% detection rate was too low for late selection (Kahneman & Treisman,
1984).
160 Logan
Capacity
Pool
Allocation
Policy
FIG. 10.2. Schematic description of a single-capacity resource theory. Note that the
attention mechanisms are not part of the chain of processes that extend from stimulus to
response. The limited pool of processing capacity serves the same explanatory function as
the limited-capacity channel in single-channel and filter theories. The allocation policy serves
the same (selective) function as the filter in single-channel and filter theories.
Capacity is Limited and Fixed. Resource theory assumes that the capacity
for processing information is limited. This is a common assumption in implicit
theories as well as explicit ones, but it is not a very strong assumption and it is not
the only assumption underlying resource theory. It is not a strong assumption
because, from a formal or technical perspective, all that is meant by capacity
limitations is that the effectiveness with which one process is carried out depends
on the number of other processes that are simultaneously active (where "effective-
ness" is measured in terms of reaction time and accuracy). According to Townsend
and Ashby (1983), capacity is unlimited if the time it takes to complete an operation
is unaffected by the number of other operations; capacity is limited if the time it
takes to complete an operation increases with the number of simultaneous proc-
esses. The idea that capacity is limited is not the same as the idea that capacity is
fixed. Capacity can be limited without being fixed. By analogy, when I shop for
groceries, my budget is limited in that I don't want to spend too much money, but
(since graduate school) it is not fixed. I will spend more if there are special sales or
interesting items and less if there are not. Capacity can also be limited and fixed.
When I get a grant, my budget is limited and fixed. The auditors in the university
administration will not let me spend more than the fixed amount I was granted.
From this perspective, resource theory assumes that capacity is both limited and
fixed. The amount of capacity available (to an individual person) is constant across
situations, tasks, strategies, and so on. This is a very strong assumption that goes
far beyond what is needed to account for the empirical observation that capacity is
limited. One could account for the empirical observation by assuming that capacity
is limited but not fixed, but that is not the approach that resource theorists take.
What do the data say? There is abundant evidence that capacity is limited but
virtually no evidence that capacity is actually fixed. Perhaps the reason for this sad
state of affairs is that it is much easier to demonstrate that capacity is limited than
to demonstrate that it is fixed. A l l that is required to demonstrate capacity
limitations is to show that the rate at which one thing is processed decreases as a function
of the number of other things that are processed simultaneously. Even this is not easy
because there are many demonstrations that an unlimited-capacity process can mimic
a limited-capacity process (Duncan, 1980; Townsend & Ashby, 1983).2 However, to
demonstrate that capacity is fixed, one needs to be able to measure the amount of
capacity expended in several different situations and show that the amounts add up to
a constant. For example, one might run a dual task experiment and vary the emphasis
placed on the two tasks. Subjects could be encouraged to try harder on Task A in one
condition and try harder on Task B in another. When experimenters try this manipu-
lation, performance is usually better on the emphasized task and worse on the de-em-
phasized task, but that only demonstrates capacity limitations, notfixedcapacity. One
would need to measure the amount of capacity expended on Task A and the amount
expended on Task B in the two situations and show that the amounts add up to the
same constant value in the two conditions. To my knowledge, no one has ever done
that. It would require a formal mathematical model, and no such model has been
developed to the extent that it could be used in such a test. Thus, there is no empirical
evidence that capacity isfixed.One of the cornerstone assumptions of resource theory
is not grounded in empirical evidence.
Capacity Allocation is Craded. Resource theorists assume that capacity can

be allocated in a continuous, graded fashion, and that performance varies continu-
ously as a function of the amount of capacity allocated. Most often, this idea is
expressed by drawing performance-resource functions, which plot performance as
a function of the amount of capacity allocated (see e.g., Navon & Gopher, 1979;
Norman & Bobrow, 1975). The function is usually smooth and negatively acceler-
ated, expressing the idea that performance gets better as more capacity is allocated,
but the improvement in performance for a given increment in the amount of
capacity allocated is smaller the better performance is before the increment (see
Fig. 10.3).
2
M a n y researchers investigating visual and memory search interpret an increase in reaction time and
error rate with the number of items in the display or the number of items in the memory set as evidence
of limited-capacity processing. This is not necessarily true. Limited-capacity processing predicts an
increase in reaction time and error rate, but unlimited-capacity processing can predict the same increase.
If the comparisons between the display items and members of the memory set are completely inde-
pendent, then the probability that all of the comparisons will be correct is the product of the probabilities
that each one will be correct. If the individual probabilities are equal, then the probability of a correct
response is p N , where p is the probability that an individual comparison is correct and N is the number
of comparisons. This value clearly decreases with the number of items in the display and memory set.
Reaction time predictions rely on a similar argument. T h e time taken for all of the comparisons to
complete will increase with the number of comparisons even if the individual comparison times are equal,
on average. If there is random variation in the comparison times (and that is a reasonable assumption),
then the time for all of the comparisons to complete is equal to the maximum of the individual comparison
times, and the expected value of the maximum increases with the number of comparisons (Townsend
& Ashby, 1983). T h e more comparisons there are, the greater the chance that at least one of them will
take an unusually long time to finish, and that is the reason for the increase.
162 Logan
Performance
Resource Allocation
FIG. 10. 3. A hypothetical performance-resource function depicting the idea that perform-
ance improves continuously as more resources are allocated.
Performance-resource functions are seductive. They express the assumption of

graded allocation very clearly, with an aura of quantitative science. The problem is
that no one has ever seen one; they cannot be observed directly. We can observe
performance, but we cannot observe resources or capacity directly. One research
strategy has been to plot performance operating characteristics or attention oper-
ating characteristics, in which performance on one task (in a dual task situation) is
plotted against performance on another (see Fig. 10.4; see also, e.g., Sperling &
Melchner, 1978). In principle, one could work backwards from the performance
operating characteristic to the performance resource functions. However, that was
never done formally, and many empirical performance operating characteristics are
difficult to interpret (see e.g., Kantowitz & Weldon, 1985). Thus, there is no
empirical evidence for the assumption that capacity allocation is graded. Another
cornerstone assumption of resource theory is not grounded in empirical evidence.
Capacity Can be Divided. Resource theories, implicit and explicit, assume

that capacity can be divided among concurrent processes so that concurrent
processes run in parallel. Capacity taken from one task can be given to another;
performance on the one task gets worse (smoothly) as the other improves. Again,
there is no evidence that capacity can be allocated in parallel. Formal analyses
showed that parallel and serial processes are extremely difficult to distinguish (see,
e.g., Townsend, 1990). Townsend and Ashby (1983) provided formal proofs that,
in many cases, parallel and serial models mimic each other perfectly so that it is
impossible to tell them apart.
From a less formal perspective, processes that look parallel can be mimicked by
a serial process that alternates rapidly between two tasks. Multiuser computers, for
example, appear to serve several users in parallel—at least from the perspective of
the user—whereas in fact, they alternate between users, usually at a rate that is too
fast for the users to detect. Part of the problem in distinguishing serial from parallel
processing is estimating the maximum rate at which serial processes can switch. If
we knew that, we could rule out serial processing in some cases. For example, if we
knew that the minimum switching time was 100 ms and we found that subjects were
engaged in two tasks during one 100-ms interval, we could reject serial processing,
arguing that there was not enough time for attention to switch between tasks.
Unfortunately, minimum switching times (or maximum switching rates) have been
very difficult to determine. Estimates of the time taken to switch attention vary by
at least an order of magnitude, from 20-40 ms (Treisman & Gelade, 1980) to 500
ms (Posner & Boies, 1971; Sperling & Reeves, 1978), depending on the task and
the method of measurement. Thus, it is extremely difficult to have faith in the
resource theory assumption that capacity can be allocated in parallel to two
processes or two tasks. Yet another cornerstone assumption of resource theory is
not grounded in empirical evidence.
Capacity is Unitary. The assumption that capacity is unitary was explicit in

early resource theories (Kahneman, 1973; Moray, 1967; Posner (St Boies, 1971). I
think it is implicit in most of the unconscious applications of resource theory in the
modern literature. Kahneman (1973) argued that performance was limited by the
availability of central processing capacity, which he assumed was (nearly) fixed (he
thought the amount of capacity available increased with arousal), allocated in a
graded fashion, and allocated in parallel to concurrent activities. Like Spearman's
g factor in intelligence, Kahneman's central processing capacity was involved, to a
greater or lesser degree, in every task the person performed. The same source of
mental energy was responsible for each and every capacity limitation a person
evidenced. Kahneman (1973) and others distinguished between interference due
to limitations on central processing capacity and "structural" interference, due to
Performance on Task B
Performance on Task A
FIG. 10.4. A performance operating characteristic, which plots performance on one task
against performance on another. Performance operating characteristics illustrate tradeoffs in
performance between tasks—Task A gets worse as Task B gets better. Performance operating
characteristics could be derived from performance resource functions, if performance
resource functions were observable.
164 Logan
overloading peripheral input and output systems, but structural interference was
usually viewed as obvious and uninteresting. One hand cannot be in two places at
one time. That (structural) limitation is physical, not psychological.
The idea that capacity is limited does not imply that capacity is unitary. From a
formal (descriptive) perspective, capacity is limited whenever the time taken to
complete a process increases with the number of concurrent processes. There may
be many reasons for that increase and different reasons may account for it in
different situations. There need be nothing in common to different cases of capacity
limitation. Thus, the resource-theoretic idea that capacity is unitary is a very strong
assumption.
A great deal of empirical evidence indicates that the unitary capacity assumption
is false. This evidence appeared in the 1970s. There were many experiments that
converged on the conclusion (for reviews, see Navon & Gopher, 1979; Wickens,
1980). One prominent example is a paper by Treisman and Davies (1973) titled
"Divided attention to ear and eye." They had subjects detect two simultaneous
targets in three different conditions: visual-visual, auditory-auditory, and audi-
tory-visual. In the first two conditions, the targets appeared in the same sensory
system (i.e., both to the eyes or both to the ears); in the third, one target appeared
in the eyes and one appeared in the ears. Treisman and Davies calibrated the
auditory and visual tasks so they were equal in difficulty. The question was whether
dual-task performance would be affected by the distribution of stimuli across sensory
modalities. If unitary capacity theory were right, the distribution should not matter.
The two stimuli should tax the same central processing capacity no matter how
they were put into the system. The data showed much more interference in the
within-modality conditions (visual-visual and auditory-auditory) than in the be-
tween-modality condition (auditory-visual), soundly rejecting the unitary capacity
account.
By the end of the 1970s, unitary capacity theories were dead, but resource
theories were alive, if not well. Single-resource theories were replaced by multiple-
resource theories (Navon & Gopher, 1979; Wickens, 1980), which proposed that
several different resources limited performance. Wickens (1984), for example,
argued that there were different resources for different processing stages (input,
output, and central processing), different resources for different modalities of input
(visual and auditory), different resources for different codes (spatial and verbal),
and different resources for different responses (vocal and manual). Multiple re-
source theories were true resource theories, in that they assumed that capacity was
fixed, that resources could be allocated continuously, and that resources could be
allocated in parallel. The main difference between them and unitary capacity
theories was in the number of different resource types that limited performance:
many versus one.
Multiple-resource theories inherited many of the problems of single-resource (or
unitary capacity) theories. There was still no evidence that any of the resources
were fixed, no evidence that any resource could be allocated in a graded fashion,
and no evidence that any resource could be allocated in parallel. The only evidence,
which was accepted with a great deal of consensus in the attention literature, was
that a unitary capacity or single-resource theory could not account for the data. For
many researchers, multiple-resource theory was a step backwards. It complicated
predictions and seemed incapable of falsification. It could accommodate any pattern
of results: Two tasks would interfere with each other if they shared the same
resources but they would not interfere if they used different resources. Trade-offs
between tasks might not be perfect (i.e., capacity might not sum to a constant even
if it could be measured) because the tasks might share some resources but not others.
The shared resources would trade off but the different resources would not.
Moreover, multiple-resource theories were largely theories of performance in dual-
task and divided-attention situations, and no longer theories of attention. Re-
searchers interested in attention began to look for alternative theories.3
The Death of Resource Theory

By the middle of the 1980s, resource theory was dead. Some researchers actively
attacked it, others found more attractive accounts of the phenomena it addressed,
and still others shifted their attention to other topics where resource theory
provided no insights. Allport (1980, 1989) was a highly vocal critic, primarily
attacking the unitary capacity view. One of the most stringent attacks came from
Navon (1984), who was a student of Kahneman's and one of the original propo-
nents of multiple-resource theory (Navon & Gopher, 1979). Navon noted that the
assumptions underlying resource theory were largely ungrounded and argued that
the idea of resources added little explanatory power to the machinery that was
necessary to account for phenomena in the first place. For example, typists appear
to shift from reliance on visual feedback to kinesthetic feedback as they acquire skill
(West, 1967). Resource theory would "explain" this as a shift from visual resources
to kinesthetic resources, but the idea of resources adds very little to the explanation.
It says nothing about why, when, and how typists shift their strategy or what visual
and kinesthetic strategies might be. The idea that visual and kinesthetic resources
are fixed, allocated continuously, and allocated in parallel plays no essential role in
the account.
For many researchers, resource theory failed to fulfill its promise. Resource theory
was attractive in the early days because it promised a quantitative account of
performance in dual-task and divided attention situations. It suggested that capac-
ity could be measured, that the fixed sum of the different measures, which represents
the total amount of capacity available, would become an important constant, like
Miller's (1956) magical number seven plus-or-minus two, and that trade-offs
between tasks could be quantified and predicted in advance. A decade and a half
later, none of these promises were fulfilled. Resource theory was bankrupt.
3
Strangely, the new theories were much like the early filter and single-channel theories that were
replaced by resource theory. Researchers proposed theories that distinguished preattentive processes from
subsequent attentional processes, much like the earlier theories (see e.g., Treisman & Gelade, 1980),
though they were focused more specifically on the details of specific tasks, like visual search.
166 Logan
In the early 1980s, theorists started conceiving of alternatives to resource theory.

T h e idea of structural interference, denigrated i n single-capacity theories, became
internalized i n multiple-resource theories (remember Wickens' input and output
resources). Later theorists used it as a cornerstone assumption, dropping the
association with resources. Dual-task interference and the costs of divided attention
were a function of the similarity between the tasks (e.g., Hirst & Kalmar, 1987;
Kinsbourne & Hicks, 1978).
Other researchers explored the idea that cross-talk or outcome conflict between
tasks was responsible for dual-task interference and divided-attention costs (e.g.,
N a v o n & Miller, 1987). Cross-talk occurred when people had trouble separating
the stimuli relevant to two tasks or two information sources. Outcome conflict
occurred when the outputs from one channel were confusable with the outputs from
the other. N a v o n and Miller, for example, showed that subjects had much more
difficulty dividing attention between two sources when the distractors o n one
channel were potential targets on the other. They had subjects monitor one channel
for boys' names and found that performance was much worse when boys' names
were distractors on the other channel. Resource theory could not account for this
result without assuming that resources were very specific (so that one resource was
dedicated to processing boys' names and another was devoted to processing city
names) and very numerous (so that one resource was dedicated to each concept i n
a person's semantic memory). Some people say that theories are overthrown by new
theories and not by data. The ideas of cross-talk and similarity provided new
theoretical alternatives to replace the dying resource theory.
Still other researchers were concerned that the poor control over timing i n the
continuous dual-task procedures of the 1970s may have produced results that
looked like parallel processing when i n fact subjects were alternating between tasks.
Pashler (1984) began an important series of experiments with the double stimula-
tion or psychological refractory period paradigm, which allowed precise control over
timing. H e showed very convincingly that a structural bottleneck, like the limited-
capacity channel of early filter theories, provided a better account than resource
theory. Bottleneck theories and resource theories made specific, contrasting predic-
tions about the pattern of performance, and Pashler's incisive experiments con-
firmed the bottleneck account and disconfirmed the resource theory account (see
e.g., Pashler, 1989).
W i t h i n the field of attention, the dominant paradigm shifted. Whereas re-
searchers i n the 1970s had been concerned primarily with dual-task performance
and performance i n divided-attention situations, researchers i n the 1980s became
interested i n visual search (Treisman & G e l a d e , 1980; Treisman & Schmidt, 1982),
negative priming (Neil, 1977; Tipper, 1985), and a variety of cuing tasks (Posner &
C o h e n , 1984). Researchers became interested i n grounding their theories i n neuro-
physiological data (e.g., Posner ^Petersen, 1990) or i n computational analyses (e.g.,
Bundesen, 1990; U l l m a n , 1984; Wolfe, Cave, & Franzel, 1989). T h e style of
theorizing changed. Global theories that provided general accounts of a broad range
of phenomena were replaced by more focused theories that provided detailed
accounts of specific phenomena. In many cases, resource theory was too general to
provide useful insights. Resource theory was a metaphor, and theorists were ready
to abandon the metaphor for more complete and more accurate descriptions.
The metaphor directed researchers' attention in the wrong direction. The key
idea behind resource theory is the idea of mental energy. Resources were the energy
sources that activated mental processes. Resource theory directed researchers'
attention toward the energy requirements of processing and away from the details
of the underlying computation. Post-resource-theory theorists were interested in
the details of the computation, wanting to specify the representations and processes
involved and the interaction between them. The mental energy involved was a
secondary concern. The key assumptions underlying resource theory—fixed capac-
ity, graded allocation, parallel allocation, and a unitary source—were not necessary
components of these theoretical accounts. Worse than dead, resource theory was
irrelevant.
RESOURCE THEORY AND AUTOMATICITY
History
Automatic processing was an important topic at the dawn of experimental psychol-
ogy more than 100 years ago. James (1890) discussed it extensively in his chapter
on habit. Solomons and Stein (1886) studied automatic writing, and Bryan and
Harter (1897, 1899) investigated the automatization of telegraphic skills. Research
on automaticity was suppressed by the behaviorist revolution early in this century
and it lay dormant until the mid-1970s. Around 1975, there was a renaissance of
research on automaticity, sparked by three seminal papers: LaBerge and Samuels
(1974), Posner and Snyder (1975), and Shiffrin and Schneider (1977; see also
Schneider & Shiffrin, 1977).
Preattentive Processing
The stage for the renaissance was set 20 years earlier when the first of the modern
attention theories was developed. Single-channel and filter theories assumed a
stage of preattentive processing that had many of the characteristics of automatic
processing. Preattentive processing was fast, obligatory, parallel, and effortless,
much like automatic processing. However, preattentive processing is not the same
as automatic processing (see Logan, 1992; Treisman, Vierra, & Hayes, 1992).
Preattentive processes are largely innate, whereas automatic processes are acquired
through learning. Researchers who tried to create preattentive processes through
learning, failed (Treisman et al., 1992). Preattentive processes occur prior to
attention, whereas automatic processes occur in parallel with it or follow it.
Preattentive processes provide the informational basis for attentional selection (i.e.,
they produce the perceptual objects attention chooses among), so it is necessary
that they precede attention. Preattentive processes are exclusively perceptual,
168 Logan
whereas automatic processes can be perceptual, conceptual, or motor. Thus,

preattentive processing is not the same as automatic processing. Nevertheless,
studies of preattentive processing set the stage for studies of automatic processes by
introducing researchers to the idea that some processing was done without atten-
tion and by creating the controversy over the locus of selection (early vs. late). That
controversy introduced researchers to the idea that relatively deep levels of proc-
essing could be accomplished without attention. The difficulty of resolving the issue
led to resource theories, as described previously.4
The Property-List Approach

The first accounts of automaticity after the renaissance took what I call a property
list approach, defining automatic processing in terms of a list of empirical properties
that distinguished it from attentional processing (LaBerge & Samuels, 1974),
strategic processing (Posner & Snyder, 1975), controlled processing (Shiffrin &
Schneider, 1977), or effortful processing (Hasher & Zacks, 1979). Automatic
processing was fast, effortless, obligatory, and unconscious, whereas attentional (or
strategic, controlled, or effortful) processing was slow, effortful, voluntary, and
conscious.
There were several problems with the property list approach. One serious
problem was that the property lists suggested that automatic processing was
dichotomous. Automatic processes had all of the properties on one side of the list
and attentional (strategic, controlled, or effortful) processes had all of the properties
on the other side of the list. Researchers were quick to point out that this was not
true empirically (see Kahneman & Henik, 1981; Paap & Ogden, 1981; Regan,
1981). The field reacted by construing automaticity as a continuum and arguing
that the degree of automaticity one attained was a function of learning. The
property lists characterized the ends of the continuum. Different properties could
be learned at different rates, so one property may be fulfilled before another at
4
M y discussion of preattentive processing invites confusion in readers more familiar with the social
cognition literature—especially Bargh's contributions—than with the attention literature. Bargh (1989,
1992) drew an important distinction between preconscious, postconscious, and goal-dependent auto-
matic processes, and his concept of preconscious automatic processing could be confused with the
attention literature concept of preattentive processing. They are not the same idea. Preattentive
processes are early sensory processes that operate on all inputs to produce perceptual objects that
subsequent attentional processes can choose among. Preattentive processes do not control responses or
influence behavior directly. Rather, their influence is mediated by attentional processes that take their
output as input. By contrast, Bargh's preconscious automatic processes are largely cognitive processes
that exert very strong and direct influences on behavior, as the literature reviewed in the target article
documents. Bargh argues that attention is not necessary to trigger preconscious automatic processes,
that preconscious automatic processing occurs whenever a relevant stimulus enters the perceptual
system. It seems to me that preconscious automatic processes might operate on the output of traditional
preattentive processes, just as attention does. Preconscious automatic processes do not require attention
to be triggered, but they do not necessarily precede attention as preattentive processes do.
intermediate stages of learning. A process might attain some degree of autonomy

(obligatoriness) before it became completely effortless (Logan, 1985).
In my view, the main problem with the property list approach was the lack of
justification for adding properties to the list. Different researchers listed different
properties and different numbers of properties, and there was no basis for reaching
consensus. Some researchers cited James (1890), but even James had no justifica-
tion for putting properties on his list. The solution to this problem is to propose a
theory of the processes that underlie performance and then deduce the properties
of automaticity from the theory of the underlying processes.
Resource Theory
Resource theory set the stage for the renaissance and served as the main theory of
the underlying processes in early approaches to automaticity. Resource theory
suggested that processes could vary in the amount of capacity they required, some
requiring a lot and some requiring a little. It was only a small step to interpret this
variation as the continuum of automaticity and propose that some processes—fully
automatic processes—required no capacity at all (see Logan, 1978). The contrast
between automatic processing and attentional (strategic, controlled, or effortful)
processing was in terms of the amount of resources required. Automatic processing
required none, whereas attentional (strategic, controlled, or effortful) processing
required a lot (see Fig. 10.5).
This view was important because it provided a justification for putting properties
on the list: Automatic processing was fast because it was not limited by the
availability of resources. Automatic processing was effortless because effort is
proportional to the amount of resources allocated, and automatic processes require
Capacity
Nonautomatic
Stimulus
Process
Automatic
Stimulus
Process
FIG. 10.5. The resource theory view of nonautomatic (top) and automatic (bottom) proc-
essing. Nonautomatic processes require two inputs to produce an output: An appropriate
stimulus and some amount of resources. Automatic processes require only one input—an
appropriate stimulus. Automatic processes d o not require resources and so cannot be
controlled by allocating resources.
170 Logan
no resources. Automatic processing was obligatory because voluntary control was

accomplished by allocating resources; a process that took no resources could not
be controlled in that fashion. And automatic processing was unconscious, because
people were only conscious of the things to which they allocated resources. Thus,
resource theory strengthened theories of automaticity by providing a theory of the
underlying processes. Automaticity theory returned the favor, strengthening re-
source theory by providing another phenomenon it could account for.
The Death of Resource Theories of Automaticity
Resource theories of automaticity were killed by the same things that killed resource
theories of attention. There was no evidence for the critical assumptions that
capacity was fixed, that it could be allocated in continuously, and that it could be
allocated in parallel. The alternatives to resource theory showed that resource
theory made few unique predictions, and in some prominent cases where it did, its
predictions were falsified (see Pashler, 1989).
In my view, resource theories of automaticity were on their death beds with the
advent of multiple-resource theory. Resource theories of automaticity were clearest
when they assumed a single resource or a unitary capacity. Multiple-resource
theories raised interpretative problems that no subsequent resource theory of
automaticity was able to solve. If automatic processing used less resources than
attentional (strategic, controlled, or effortful) processing, which resource did it use
less of? The answer was clear in single-resource theories—central processing
capacity—but it was no longer clear in multiple-resource theories. Visual tasks
might use fewer visual resources and auditory tasks might use fewer auditory
resources, but that would lead to the prediction that automatic visual tasks could
run in parallel with other visual tasks but not with other auditory tasks. A n
automatic and a nonautomatic visual task should produce the same amount of
interference when paired with a nonautomatic auditory task. Thus, automaticity
would not necessarily reduce dual-task interference.
Moreover, there was evidence that people shift strategies when acquiring skill,
which would be interpreted as shifting the resources they relied on. West (1967),
for example, showed that typists switched from reliance on visual feedback (visual
resources) to kinesthetic feedback (kinesthetic resources) as they acquired skill.
This suggests that subjects may rely more on some resources and less on others as
they acquire skill, and this contradicts the re source-theoretic assumption that
automatic processes use less resources. One way to deal with this problem would
be to propose that automatic processes use fewer resources in total than do
nonautomatic processes, but that raises a serious measurement problem no theorist
has addressed: How many units of one kind of resource are equivalent to one unit
of another kind of resource? Multiple-resource theory suggests that this problem
may be insoluble. Resources are distinguished from each other because they are
incommensurable. One cannot trade any number of units of one resource for a unit
of another.
The death of single-resource theory and the faltering ascendance of multiple -

resource theory left resource theories of automaticity knocking on heaven's door.
The death of resource theory removed the theory of the underlying processes that
gave coherence to resource theories of automaticity. There was no longer any
theoretical justification for claiming that automatic processing should have the
properties it does. Automaticity theory was back to the property list approach,
crassly empirical and theoretically bankrupt.
As resource theory lay dying, there was a kind of paradigm shift in studies of
automaticity, motivated by the idea that automaticity was a continuum. Researchers
became interested in the acquisition of automaticity. Shiffrin and Schneider (1977;
Schneider & Shiffrin, 1977) described the conditions necessary for automaticity to
be acquired, but they did not propose a theory of the acquisition itself. Several
theories proposed learning mechanisms that would produce the kinds of changes
associated with automaticity, taking a task with the properties of attentional
(strategic, controlled, or effortful) processing and changing it into a task with the
properties of automatic processing (see Anderson, 1982, 1987, 1992; Cohen,
Dunbar, & McClelland, 1990; Logan, 1988, 1990; MacKay, 1982; Newell &
Rosenbloom, 1981; Schneider, 1985). These theories raised a final challenge to
resource theories of automaticity: How is it possible that the resource requirements
of a process change with practice? What mechanism makes it possible for a process
to function in the same way but with fewer resource demands? To my knowledge,
this challenge was never answered. None of the current theories of automaticity
invoke resource theory as an explanatory construct. If resource theories of auto-
maticity were not dead, they were certainly abandoned, like an old Eskimo drifting
off to oblivion alone on an iceberg.
AUTOMATICITY AS MEMORY: A THEORETICAL ALTERNATIVE
The last decade has revealed a new approach to automaticity that does not rely on
the concept of resources, implicitly or explicitly. Many models were proposed, and
most of them provide formal theories of the acquisition of automaticity, imple-
mented as mathematical models or computer simulations. The models differ in
detail but share a common view that automaticity is a memory phenomenon.
Automatic performance is based on memory retrieval. Whereas novices must solve
problems with deliberate thought and conscious algorithms, skilled performers
simply retrieve past solutions from memory (see Fig. 10.6; see also Anderson, 1982,
1987, 1992; Cohen et al., 1990; Logan, 1988, 1990; MacKay, 1982; Newell &
Rosenbloom, 1981; Schneider, 1985).
This autormtkity-as-rnernory view provides a theoretical account of the process-
ing underlying automatic performance, and thereby provides justification for attrib-
uting some properties to automatic processing and others to nonautomatic
processing: Automatic processing is fast because memory retrieval is fast. O f course,
not all instances of memory retrieval are fast (try to recall your second-grade
172 Logan
Memory
FIG. 10.6. The automaticity-as-memory view of nonautomatic (top) and automatic (bottom)
processing. Nonautomatic processes involve the execution of a complex algorithm; auto-
matic processes involve memory retrieval.
teacher's name), but the ones underlying automatic performance are. Extensive
practice makes memory very strong and retrieval very rapid. According to automat-
icity-as-memory theories, people do not rely on memory retrieval until it is faster
than computing solutions with an algorithm (e.g., Logan, 1988).
Automatic processing is effortless because it involves only a single act of memory
retrieval that is triggered by stimulus presentation. By contrast, algorithmic com-
putation is difficult. The person must first think of a way to solve the problem and
then apply it. Often, the application involves several intermediate steps before a
final solution is attained.
Automatic processing is obligatory because memory retrieval is obligatory. Atten-
tion to a stimulus is sufficient to trigger retrieval of things associated with it, as
evidenced by the ubiquitous Stroop (1935) effect (for a review, see MacLeod, 1991).
By contrast, application of an algorithm requires several deliberate actions, from the
initial formulation of the problem to stepping through the algorithm that computes
the solution. Automatic processing may appear to be obligatory because it occurs so
quickly there is not much time to stop it. A small target is hard to hit with a "shot"
of inhibition. By contrast, application of an algorithm takes much more time and
presents a much larger target for control processes to inhibit (see also Zbrodoff &
Logan, 1986).
Automatic processing is unconscious because there are no intermediate steps to
present themselves to consciousness. The person may aware of the stimulus and aware
of the course of action that the stimulus retrieves from memory, but the act of memory
retrieval itself is not available to consciousness. We know the results, not the process
that produces them. By contrast, algorithmic processing involves many steps, each of
which presents a retrieval cue and retrieves something we can be aware of.
This is only one account of the properties of automaticity, mostly taken from my
own instance theory (Logan, 1988). Different theories focus on some properties more
than others, and their accounts differ in detail. My purpose here is to show that
automaticity-as-memory theories can account for the properties of automatic and
10. A c a d e m i c Life 173
nonautomatic processes i n a principled fashion without invoking the concept of

resources. Readers interested i n the details of the other accounts should consult the
original sources.
M o s t of the current theories of automaticity focus on the role of automatic
processing i n skilled performance, where people engage automatic processing
deliberately and automatic processing is compatible with their goals. M u c h of the
research i n the target chapter focuses on the unconscious, involuntary use of
automatic processing, and it may not be obvious how automaticity-as-memory
theories would account for those effects. The C o h e n et al. (1990) theory is a
counterexample, dealing with the Stroop effect, which is notoriously resistant to
voluntary control. I think that the other theories could be applied to the Stroop
effect and to unconscious, involuntary effects in a straightforward fashion by
examining their assumptions about memory retrieval.
Most automaticity-as-memory theories assume that memory retrieval is obliga-
tory, retrieving whatever is associated with the retrieval cues i n the current envi-
ronment. There is no guarantee that the things that are retrieved will be useful or
even compatible with the person's goals and desires. The results of retrieval are
useful i n skilled performance largely because skilled performance usually occurs i n
highly constrained environments i n which response requirements are consistent
from trial to trial and the retrieval cues are compatible with the performer's goals.
The less restrictive environments that Bargh and others study do not guarantee this
consistency, and so are open to pernicious effects of automatic processing as well as
beneficial ones. I believe that automaticity-as-memory theories could account for
the automaticities of everyday life i n a clear and elegant manner (without invoking
resource theory). Further research—and further theoretical development—will be
necessary to see whether my belief will bear fruit.
IMPLICATIONS: HOW TO COPE WITH THE LOSS
Resource theory is dead and resource theories of automaticity died along with it. In
the attention literature, at least, they were replaced by new theories of attention
and automaticity. Researchers should explore these new theories and see which
ones can be adapted to the problems they investigate. Most importantly, researchers
should be aware of the assumptions they are making about attention and automat-
icity. Unconscious processing may be acceptable i n everyday life, but it should not
be acceptable i n academic life. We should be aware of the assumptions on which
our theories are based and aware of the implications of those assumptions for our
research. In the remaining pages, I draw out a few of the implications of the death
of resource theory.
Don't Say It If You Don't Mean It

Researchers should not invoke capacity or resource theory unless they are willing
to stand behind the assumptions that underlie it. D o you have evidence that
174 Logan
capacity is fixed? D o you have evidence that capacity can be allocated in a graded
fashion? D o you have evidence that capacity can be allocated i n parallel? D o you
have evidence that a single resource limits performance? Researchers who are
unwilling to make all of these assumptions cannot endorse capacity or resource
theory. They should find some other way to express their ideas or to describe their
manipulations. It may be difficult at first, if their unconscious use of resource theory
is as strong a habit as mine was, but it will get easier with practice. Moreover,
searching for another way to describe manipulations of attention and automaticity
may produce surprising new results and theories.
Don't Claim That a Dual Task Consumes Resources
M a n y researchers i n social cognition and memory employ dual-task paradigms in

order to remove capacity from a task they are interested in. The implicit assumption
seems to be that the mere presence of a second task takes all of the capacity or all
of the resources away from the first one. I hope I convinced you that this assumption
is very controversial. The two tasks might not share any resources at all. Multiple -
resource theory suggests that many tasks do not overlap (very much) i n their
resource requirements, so pairing arbitrary tasks is unlikely to produce the desired
trade-off. T h e two tasks might share resources but the second task may not take all
of the resources away from the first. Capacity might be limited but not fixed, in
which case the first task will still have some capacity, albeit not the same amount
it would have i n single-task conditions. Capacity might be fixed but the second task
might not take all of the capacity available, leaving none for the other. A s a rule of
thumb, no task can be performed without any attention, so that above-chance
performance is evidence that the task received some attention.
The onus is o n the researcher to demonstrate the validity of the assumptions he
or she makes. If you assume that a second task takes all of the resources away from
the first, you must be prepared to demonstrate that capacity is fixed, that the two
tasks i n fact require the same resources, and the second task did consume all of the
resources. This is a tall order rarely met in any experiment.
Don't Claim That Automatic Processing

Is Capacity Free
The claim that automatic processes do not use resources is a theoretical statement,
not an empirical fact. A t a descriptive level, many automatic processes are subject
to capacity limitations, i n that the rate at which they execute depends o n the
number of other processes that are executed at the same time (Townsend & Ashby,
1983). A t a theoretical level, the claim that automatic processing is capacity free is
an endorsement of resource theories of attention and resource theories of automat-
icity. If you do not mean to endorse those theories and accept all the assumptions
they entail, do not make the claim. A t an empirical level, researchers should not
claim that automatic processes are capacity free unless they can demonstrate it
experimentally. They must have a way to measure capacity and show that the
measurement procedure, when applied to the automatic process, indicates that the
process does not use capacity. That, too, is a tall order rarely met in practice.
Don't Claim That Successful Performance

in Dual-Task Situations Means Automaticity
This implication is a corollary of the first: People can do two things at once for many
reasons besides automaticity. The second task may not be demanding enough, so
that plenty of capacity is left over for the first (assuming capacity theory is true).
The second task may demand different resources than the first, and so not interfere
with it (assuming multiple-resource theory is true). Subjects may actually be
alternating between tasks rather than concurrently performing them, in which case,
each task could receive the attention (or the amount of resources) it requires.
Learn to Live With Complexity
The main advantage of resource theory, especially the single-capacity version, was
that it was simple. It led to straightforward manipulations and straightforward
interpretations of results. Unfortunately, it appears to be wrong. That means that
the simple manipulations cannot have the effects they were intended to have and
the simple interpretations of the results were in error. For example, experiments that
manipulate cognitive load or cognitive "busyness" to "remove capacity" may not be
contrasting purely automatic and purely nonautomatic processes. They may not be
showing what happens when all resources are taken away from performance.
Instead, they may be showing what happens when a task is performed with fewer
resources than it usually requires, if one accepts resource theory and all that entails.
Or they may be showing that a task performed in a dual-task environment, subject
to cross-talk and competition for a structural bottleneck, is different from a task
performed alone. These descriptions are less elegant and perhaps less compelling
than the simpler interpretations, but they are more likely to be accurate and less
likely to lead future researchers astray.
Research on attention and automaticity went through a lot of changes in the
last 25 or 30 years. It is important to keep track of those changes, learning the lessons
attention researchers have to offer and adapting them to one's own research
projects. The theories may not appear simple and straightforward, especially when
they are new, but they may ultimately provide new insights into one's own research
problems, producing a clarity that was elusive before. Bargh's research in the target
article is a good example of the kind of clarity that can be achieved without invoking
the resource construct. Those who fail to read history are condemned to repeat it.
Those who fail to heed history and let unconscious processing pervade their
academic lives may be condemned to a fate that is even worse.
176 Logan
ACKNOWLEDGMENTS
This research was supported by Grant No. SBR 94-10406 from the National
Science Foundation. Correspondence may be addressed to Gordon D. Logan,
Department of Psychology, University of Illinois, 603 East Daniel Street, Cham-
paign IL 61820. Electronic mail may be addressed to glogan@s.psych.uiuc.edu.
REFERENCES
Allport, D. A . (1980). Attention and performance. In G . Claxton (Ed.), Cognitive psychobgy (pp.
112-153). London: Routledge & K e g a n Paul.
Allport, D. A . (1989). Visual attention. In M . I. Posner (Ed.), Foundations of cognitive science (pp.
631-682). Cambridge, M A : M I T Press.
Anderson, J. R. (1982). Acquisition of cognitive skill. Psychobgical Review, 89, 369-406.
Anderson, J. R. (1987). Skill acquisition: Compilation of weak-method problem solutions. Psychobgical
Review, 94, 192-210.
Anderson, J. R. (1992). Automaticity and the A C T * theory. American Journal of Psychobgy, 105,
165-180.
cognition. In J. S. Uleman & J. A . Bargh (Eds.), Unintended thought (pp. 3-51). New York: Guilford.
Bargh, J. A . (1990). Auto-motives: Preconscious determinants of thought and behavior. In E. T. Higgins
& R. M . Sorrentino (Eds.), Handbook of motivation and cognition (Vol. 2). New York: Guilford.
Bargh, J. A . (1992). T h e ecology of automaticity: Toward establishing the conditions needed to produce
automatic processing effects. American Journal of Psychology, 105, 181-199.
Bargh, J. A . , & Pietromonaco, R (1982). Automatic information processing and social perception: T h e
Journal of Personality and Social Psychobgy, 43, 437-449.
Bjorklund, D. F., & Harnisfeger, K. K. (1990). T h e resources construct in cognitive development: Diverse
evidence and a theory of inefficient inhibition. Devebpmental Review, 10, 48-71.
Broadbent, D. E. (1958). Perception and communication. London: Pergamon.
Bryan, W. L., & Harter, N . (1897). Studies in the physiology and psychology of telegraphic language.
Psychobgical Review, 4, 27-53.
Bryan, W. L., & Harter, N . (1899). Studies of the telegraphic language. T h e acquisition of a hierarchy
of habits. Psychobgical Review, 6, 345-375.
Bundesen, C . (1990). A theory of visual attention. Psychobgical Review, 97, 523-547.
Cherry, E. C . (1953). Some experiments on the recognition of speech with one and two ears. Journal of
the Acoustical Society of America, 25, 957-979.
Cohen, J. D., Dunbar, K., & McClelland, J. L. (1990). O n the control of automatic processes: A parallel
distributed processing account of the Stroop effect. Psychobgical Review, 97, 332-361.
Craik, E I. M . , & McDowd, J. M . (1987). Age differences in recall and recognition. Journal of Experimental
Psychobgy: Learning, Memory and Cognition, 13, 474-479.
Deutsch, J. A . , & Deutsch, D. (1963). Attention: Some theoretical considerations. Psychobgical Review,
70, 80-90.
Duncan, J. (1980). Demonstration of capacity limitation. Cognitive Psychology, 12, 75-96.
Gilbert, D. T , Pelham, B. W., & Krull, D. S. (1988). O n cognitive busyness: W h e n person perceivers
meet persons perceived. Journal of Personality and Social Psychobgy, 54, 733-740.
Gray, J. A . , & Wedderburn, A . A . I. (1960). Grouping strategies with simultaneous stimuli. Quarterly
Journal of Experimental Psychobgy, 12, 180-184.
Hasher, L., & Zacks, R. T. (1979). Automatic and effortful processes in memory. Journal of Experimental
Psychology: General, 108, 356-388.
Hirst, W., & Kalmar, D. (1987). Characterizing attentional resources. Journal of Experimental Psychology:
General, 116, 68-81.
Jacoby, L. L. (1991). A process dissociation framework: Separating automatic from intentional uses of
memory. Journal of Memory and Language, 30, 513-541.
James, W. (1890). Principles of psychology. New York: Holt.
Just, M . A . , & Carpenter, R A . (1992). A capacity theory of comprehension: Individual differences in
working memory. Psychological Review, 99, 122-149.
Kahneman, D. (1973). Attention and effort. Englewood Cliffs, NJ: Prentice-Hall.
Kahneman, D., & Henik, A . (1981). Perceptual organization and attention. In M . Kubovy & J.
Pomerantz (Eds.), Perceptual organization, (pp. 181-211). Hillsdale NJ: Lawrence Erlbaum Associates.
Kahneman, D., 6k Treisman, A . (1984). Changing views of attention and automaticity. In R. Paras-
uraman & D. R. Davies (Eds.), Varieties of attention, (pp. 29-61). San Diego, C A : Academic Press.
Kail, R., & Salthouse, T. A . (1994). Processing speed as a mental capacity. Acta Psychologica, 86, 199-225.
Kantowitz, B. H . , & Weldon, M . (1985). O n scaling performance operating characteristics. Human
Factors, 27, 531-547.
Kinsbourne, M . , & Hicks, R. (1978). Functional cerebral space. In J. Requin (Ed.), Attention and
Performance VII. (p. 345-362). Hillsdale, NJ: Lawrence Erlbaum Associates.
LaBerge, D., & Samuels, S. J. (1974). Toward a theory of automatic information processing in reading.
Cognitive Psychology, 6, 293-323.
Logan, G . D. (1978). Attention in character classification: Evidence for the automaticity of component
stages. Journal of Experimental Psychology: General, 107, 32-63.
Logan, G . D. (1979). O n the use of a concurrent memory load to measure attention and automaticity.
Journal of Experimental Psychology: Human Perception and Performance, 5, 189-207.
Logan, G . D. (1985). Skill and automaticity: Relations, implications, and future directions. Canadian
Journal of Psychology, 39, 367-386.
Logan, G . D. (1988). Toward an instance theory of automatization. Psychological Review, 95, 492-527.
Logan, G . D. (1990). Repetition priming and automaticity: Common underlying mechanisms? Cognitive
Logan, G . D. (1992). Attention and preattention in theories of automaticity. American Journal of
Psychology, 105,317-339.
MacKay, D. G . (1982). T h e problem of flexibility, fluency, and speed-accuracy trade-off in skilled
behavior. Psychological Review, 89, 483-506.
MacLeod, C . M . (1991). Half a century of research on the Stroop effect: A n integrative review.
Psychological Bulletin, 109, 163-203.
Miller, G . A . (1956). T h e magical number seven plus or minus two: Some limits on our capacity for
processing information. Psychological Review, 63, 81-97.
Moray, N . (1959). Attention in dichotic listening: Affective cues and the influence of instructions.
Quarterly Journal of Experimental Psychology, 11, 56-60.
Moray, N . (1967). Where is capacity limited? A survey and a model. Acta Psychologica, 27, 84-92.
Navon, D. (1984). Resources—A theoretical soup stone? Psychological Review, 91, 216-234.
Navon, D., & Gopher, D. (1979). O n the economy of the human processing system. Psychological Review,
86,214-255.
Navon, D., & Miller, J. (1987). Role of outcome conflict in dual-task interference. Journal of Experimental
Psychology: Human Perception and Performance, 13, 435-448.
Neill, W. T. (1977). Inhibitory and facilitatory processes in selective attention. Journal of Experimental
Psychology: Human Perception and Performance, 3, 444-450.
Newell, A . , & Rosenbloom, R S. (1981). Mechanisms of skill acquisition and the law of practice. In J.
R. Anderson (Ed.), Cognitive skills and their acquisition, (pp. 1-55). Hillsdale NJ: Lawrence Erlbaum
Associates.
Norman, D. A . (1968). Towards a theory of memory and attention. Psychological Review, 75, 303-312.
178 Logan
Norman, D. A . , & Bobrow, D. G . (1975). O n data-limited and resource-limited processes. Cognitive

Psychobgy, 7, 44-64.
Paap, K. R., & O g d e n , W . C . (1981). Letter encoding is an obligatory but capacity-demanding operation.
Journal of Experimental Psychobgy: Human Perception and Performance, 7, 518-527.
Pashler, H . (1984). Processing stages in overlapping tasks: Evidence for a central bottleneck. Journal of
Experimental Psychobgy: Human Perception and Performance, 10, 358-377.
Pashler, H . (1989). Dissociations and dependencies between speed and accuracy: Evidence for a
two-component theory of divided attention in simple tasks. Cognitive Psychobgy, 21, 469-514.
Posner, M . I., & Boies, S. J. (1971). Components of attention. Psychobgical Review, 78, 391-408.
Posner, M . I., & Cohen, Y. (1984). Components of visual orienting. In H . Bouma & D. G . Bouwhuis
(Eds.), Attention and performance X (p. 531-556). Hillsdale, NJ: Lawrence Erlbaum Associates.
Posner, M . I., 6k Snyder, C . R. R. (1975). Attention and cognitive control. In R. L. Solso (Ed.), Information
processing and cognition: The Loyola symposium, (pp. 55-85). Hillsdale NJ: Lawrence Erlbaum Associ-
ates.
Regan, J. E. (1981). Automaticity and learning: Effects of familiarity on naming letters. Journal of
Experimental Psychobgy: Human Perception and Performance, 7, 180-195.
Salthouse, T. A . (1990). Working memory as a processing resource in cognitive aging. Devebpmental
Review, 10, 101-124.
Schneider, W (1985). Toward a model of attention and the development of automatic processing. In M .
I. Posner 6k O . S. Marin (Eds.), Attention and Performance XI. (pp. 475-492). Hillsdale NJ: Lawrence
Erlbaum Associates.
Schneider, W & Shiffrin, R. M . (1977). Controlled and automatic human information processing: I.
Detection, search, and attention. Psychobgical Review, 84, 1-66.
Schwartz, F., & Schiller, R (1970). A psychoanalytic model ofattention and learning. Psychobgical Issues,
6, Monograph 23.
Shiffrin, R. M . 6k Schneider, W. (1977). Controlled and automatic human information processing: II.
Solomons, L. M . , & Stein, G . (1886). Normal motor automatism. Psychological Review, 3, 492-512.
Sperling, G . , & Melchner, M . J. (1978). The attention operating characteristic: Some examples from
visual search. Science, 202, 315-318.
Sperling, G . , & Reeves, A . (1980). Measuring the reaction time of an unobservable response: A shift of
visual attention. In R. S. Nickerson (Ed.), Attention and Performance Vlll (pp. 347-360). Hillsdale,
NJ: Lawrence Erlbaum Associates.
Stroop, J. R. (1935). Studies of interference in serial verbal reactions. Journal of Experimental Psychobgy,
18, 643-662.
Tipper, S. P (1985). T h e negative priming effect: Inhibitory priming by ignored objects. Quarterly Journal
of Experimental Psychobgy, 37A, 571-590.
Townsend, J. T. (1990). Serial versus parallel processing: Sometimes they look like Tweedledum and
Tweedledee but they can (and should) be distinguished. Psychobgical Science, 1, 46-54.
Townsend, J. T , & Ashby, F. G . (1983). Stochastic modeling ofelementary psychobgical processes. Cambridge,
England: Cambridge University Press.
Treisman, A . , & Davies, A.(1973). Divided attention to ear and eye. In S. Kornblum (Ed.), Attention
and performance IV (p. 10-30). New York: Academic Press.
Treisman, A . , & Gelade, G . (1980). A feature integration theory of attention. Cognitive Psychobgy, 12,
97-136.
Treisman, A . , & Schmidt, H . (1982). Illusory conjunctions in the perception of objects. Cognitive
Psychobgy, 14, 107-141.
Treisman, A . , Viera, A . , & Hayer, A . (1992). Automaticity and preattentive processing. American Journal
of Psychobgy, 105, 341-362.
Ullman, S. (1984). Visual routines. Cognkion, 18, 97-159.
Wegner, D. M . (1994). Ironic process of mental control. Psychobgical Review, 101, 34-52.
Welford, A . T (1952). T h e 'psychological refractory period' and the timing of high-speed performance:
A review and a theory. British Journal of Psychology, 43, 2-19.
West, L. J. (1967). Vision and kinesthesis in the acquisition of typewriting skill. Journal of Applied
Psychobgy, 51, 161-166.
Wickens, C . D. (1980). T h e structure of attentional resources. In R. S. Nickerson (Ed.), Attention and
performance Vlll (pp. 239-257). Hillsdale, NJ: Lawrence Erlbaum Associates.
Wickens, C . D. (1984). Processing resources in attention. In R. Parasuraman & R. Davies (Eds.), Varieties
of attention, (pp. 63-102). San Diego, C A : Academic Press.
Wolfe, J. M . , Cave, K. R., & Franzel, S. L. (1989). Guided search: A n alternative to the feature
integration model for visual search. Journal of Experimental Psychobgy: Human Perception and Perform-
ance, 15, 419-433.
Zbrodoff, N . J., & Logan, G . D. (1986). O n the autonomy of mental processes: A case study of arithmetic.
Journal of Experimental Psychobgy: General, 115, 118-130.
Chapter 7 7
Was the Cognitive Revolution Just a

Detour on the Road to Behaviorism?
On the Need to Reconcile Situational
Control and Personal Control
Walter Mischel
Columbia University
In chapter 1, John Bargh forcefully, often compellingly, argues that—with all due respect
to cognition and cognitive processes—it is the situational factors in the environment
that account for, and automatically drive, many, most, or virtually all of the complex
psychological phenomena of everyday life. The degree to which this hugely stimulating
article is controversial hinges on which of those quantitative qualifiers ("many?" "most?"
"all that are important?") Bargh really has in mind, and that is what remains most
provocative and unclear about the arguments he builds around thefindingshe surveys,
particularly his own dazzling results. He casts his thesis within the classic tradition and
the very definition of social psychology: The focus of the field is on the significance of
the social situation in the determination of social cognition, feeling, and action, with
the goal of demonstrating its remarkably strong and often subtle power.
Researchers in this vein have systematically compared this power of the situation
to that of the person as if the two were competing in a zero-sum game, in which
evidence for the power of one necessarily diminishes that of the other. We last saw
such a competition in the classic, fiercely controversial debate between social and
personality psychologists, which raged in the 1970s and still simmers (e.g., Ross &
Nisbett, 1991). Ultimately, it led to the belated recognition, inevitable in all such
controversies about "is my variable more important than your variable?" that, of
course, both are important and the task is to figure out how the interactions between
them work.
With that history in the background, Bargh now pits the power of the situation
and the if-then relations it automatically activates against the power of the person
181
182 Mischel
to exert purposeful, deliberate control and choice—modern terms for nothing less
than the concept of will. The competition he creates, stripped to its essentials, is
between automatic stimulus control on one side—the power of the situation to elicit
responses automatically—and, on the other side, purposeful mediated self-con-
trol—the power of the person to overcome its impact. Bargh squarely puts his bets
on the side of the external stimulus and automaticity.
THE IMPORTANCE OF THE SITUATION
From the start, Bargh provocatively sharpens the controversies that such a position
is sure to precipitate. He does so by purposefully aligning himself with Skinner, the
devil the cognitive revolution attempted to exorcise forever in the 1960s. Appar-
ently embracing the concept of the stimulus control of behavior, which has long
been unmentionable in polite cognitive company, Bargh's basic thesis is that:
... much of everyday life—thinking, feeling, and doing—is automatic in that it is

driven by current features of the environment... as mediated by automatic cognitive
processing of those features and not mediated by conscious choice or reflection, (chap.
1.P.2)
In this tour de force, Bargh is on an important mission that needs to be

undertaken periodically in order to challenge all the fancy but sometimes empty
cognitive mediating constructs that we psychologists so easily invent in order to
determine if they are just another more fashionable version of the "mental fictions"
that Skinner decried so long ago. If so, the field may again be in need of a thorough
shave from Occam's parsimonious razor, which Bargh so skillfully provides.
It is not that Bargh is impervious to cognition and to complex mediating
processes in the regulation of thought, feeling, and action. O n the contrary, he notes
that even in animal responses to complex stimuli, cognitive processes, such as
perception and representation, play a role. It is unclear, however, what kind of role
they play for him. Do the external stimulus and the cognitive mediators interact in
ways that potentially transform and change the meaning and impact of the situ-
ation? Or do the mediators passively transmit the stimulus until it elicits its
associated response in a straightforward if-then, S-R connection? In his words,
"Given the presence or occurrence of a particular set of situational features (e.g., a
person or event), a certain psychological or emotional or behavioral effect will
follow." (chap. 1, p. 2), and the specification of these relations then becomes the
core of social psychology.
In this, Bargh sounds like he is resurrecting, in cognitivized form, the concept of
the automatic stimulus control of behavior, and one wonders if he really means it.
But in a comprehensive analysis of social cognition, feeling, and action, the if-then
relationships that have to be considered include internal events and condi-
tions—the situations inside the head—such as the person's chronic affective states,
styles of encoding information, self-representations and expectations (e.g., about
one's own efficacy), goals, values, self-regulatory strategies, and action scripts, all
11. Reconciling Situational and Personal Control 183
of which are likely to interact with and change the impact of the external stimulus (e.g.,
Mischel, 1973, 1990; Mischel & Shodd, 1995; Mischel, Cantor, & Feldman, 1996). If
that is not the case, one wonders what the cognitive revolution was all about.
Intentionality and Goal-Directedness

Again, Bargh tries to take cognitive mediation into account, saying for example that
automatic processes depend on recent use or activation and, more important, on
"goal-dependent automaticity" which has "the precondition of the individual intending
to perform the mental function" (chap. 1, p. 3). But such talk of intentionality and
goal-dependence and psychological (rather than objective) situations sounds not even
remotely like the behavioristic concept of stimulus control and the Skinner "manifesto"
with which he sets the stage for his bold thesis. Thus, as one reads Bargh closely, it
becomes increasingly plain that he and the field moved a long way beyond the
automaticity and stimulus control of behaviorism, and that we need to reflect before
concluding that the cognitive revolution was mostly a detour in the progress of our field.
CONCURRENT EVIDENCE FOR THE POWER

OF THE SITUATION AND OF THE PERSON
Bargh's incisive arguments and strong evidence underline that one of the core conclu-
sions that needs to be drawnfrom50 years of research is that the situation of the moment
plays an enormously powerful role in the often automatic activation and regulation of
complex human social behavior. As an early spokesperson for the subtlety and ubiqui-
tous power of the situation in Personality and Assessment (Mischel, 1968), I find myself
applauding Bargh's convincing demonstration that, today's fashions not withstanding,
the significance of the situation in the regulation of human social behavior remains
formidable, even after three decades of cognitive revolution.
I became convinced myself of this point in my own work on the willingness and
ability of young children to delay gratification by continuing to wait for two pretzels
later as opposed to settling for one right now. We found that such a seemingly trivial
change in the situation as whether the pretzels remain exposed on the plate facing
the child, or are placed under it can change the average delay time from less than
1 minute to more than 10 (e.g., Mischel, Shoda, & Rodriguez, 1989). Whether the
young child finds delay of gratification excruciatingly difficult or easily achievable
hinges on the subtleties of the situation, and when these "ifs" are properly under-
stood and introduced, the "thens" that follow can become highly predictable. The
history of the field is of course full of such demonstrations.
Overcoming Stimulus Control Through Self-Regulation

The same set of studies also generated a second set of findings, however, that led
me to join the cognitive revolution (e.g., Mischel, 1973) and still keep me within
the party (Mischel & Shoda, 1995), and they also need to be considered in the
184 Mischel
evaluation of the role of the situation and Bargh's thesis. Most striking for me was
the finding that regardless of the objective stimulus facing the subject, it was its
mental representation that controlled the delay of gratification behavior. Namely,
when the mental representation focused on the "hot" consummatory features of
the stimulus, the frustration of continued delay of gratification became unbearable
for most children, and this was true even when the external stimulus facing the
subject was completely controlled. However, when the mental representation
focused on the "cool," informative cue properties, sustained, goal-directed delay of
gratification and "willpower" became manageable, again regardless of the external
stimulus in the situation (Mischel, Shoda, & Rodriguez, 1989; Mischel, 1996).
Thus, the way in which the child cognitively represented the stimuli during the
delay period profoundly transformed their impact: The power resides in the head,
not in the external stimulus.
If the crucial process is in the head, individuals should be able to influence it

through self-generated strategies to change the mental representations of the
reward objects, just as it is influenced by the strategies the situation suggests. In
fact, the same research program revealed impressively stable individual differences
in the ability to overcome stimulus control pressures in the purposeful pursuit of
long-term goals (e.g., Mischel, Shoda, 6k Peake, 1988; Shoda, Mischel, 6k Peake,
1990). For example, in laboratory situations in which individual differences in such
strategies were activated, those 4-year-old children who delayed longer became
more socially and cognitively competent young adults. They also achieved higher
levels of scholastic performance, as reflected in their S A T verbal and quantitative
scores (e.g. Shoda et al., 1990). Thus, behavior in the delay situation was a function
not only of the characteristics of the situation but also of the individuals in it. The
fact that these distinctively human efforts also require the strategic utilization and
support of situations undermines neither the role of the person nor of the situation,
but requires attention to their interplay (Mischel 6k Shoda, 1995; Shoda et al.,
1990). In the same vein, but with quite different methods and situations, the
implications of such interactions have been shown, for example, in the demonstra-
tion that chronic construct accessibility interacts with situational priming to affect
judgment and behavior, even when the stimulus information is extremely vague
(e.g., Higgins & Brendl, 1995).
THE ROLE OF INDIVIDUAL DIFFERENCES
Individual differences of course with regard to a wide range of internal cognitive-af-

fective mediators—from attentional styles and encoding strategies to available
constructs and action scripts—have been repeatedly demonstrated, and seem to
underlie the distinctive patterns of if-then relations that characterize the person's
social thought and behavior (e.g., Shoda, Mischel, 6k Wright, 1994). These indi-
vidual differences affect how features of the environment are selected, interpreted,
and processed and make the stimulus only one (major) element in a complex
interaction process that demands attention in a comprehensive analysis of the
11. Reconciling Situational and Personal Control 185
psychology of everyday life. In the course of such processing, the cognitive-affective

units within the persons processing system dynamically interact with each other as
well as with the information from the external stimulus features, potentially
transforming their impact and allowing the individual considerable power to
overcome (or at least modify) external stimulus control i n order, for example, to
persist i n the pursuit of important long-term goals (Mischel & Shoda, 1995;
Mischel, Cantor, & Feldman, 1996).
It takes an inordinately long stretch of the Skinnerian account of stimulus
control to fit such findings, as seen, for example, when Bargh tries to make the
"individual intending to perform the mental function" (chap. 1, p. 3) part of
stimulus control. Regardless of what one calls it (stimulus control? self-control?) or
just where it resides (outside, inside), the point seems by now undeniable: In
everyday life, as well as i n the laboratory, people can and do modify and transform
the power and impact of the stimuli they encounter and create, persisting i n pursuit
of long-term, difficult goals even in the face of potent barriers and temptations along
the route (e.g., Mischel et al., 1996). These phenomena (as when the habitual
smoker gives up tobacco, and the difficult new year's resolution to exercise is
actually executed) may be rare events, but it is their importance for being human,
not their frequency, that is at issue. Surely this is part of what psychologists must
explain in a comprehensive account of what is significant i n everyday life.
In summary, the prevalence and significance of the automaticity of if-then links
in everyday life that Bargh so elegantly demonstrates is not diminished (it may even
be enhanced) by the concurrent recognition of the field's other major conclusion:
Regardless of its frequency, humans do engage at least some of the time and under
some circumstances in self-regulatory behavior i n pursuit of their long-term goals
and values. In these moments, they manage to purposely modify, transform, and
even overcome the power of the immediate stimulus, interjecting their own personal
agendas between the external "if" and the observable external "then," in ways that
reveal their distinctive personality signatures (e.g., Mischel et al., 1996).
TIME TO REDEFINE SOCIAL PSYCHOLOGY?
Intermittently, Bargh reminds one of the consummate stage actor who connects
with the audience through subtle hints that signal implicit common understandings,
a shared knowledge base. The sophisticated reader will have no difficulty hearing
Bargh's asides, i n which he acknowledges that he knows we know he is quite
deliberately overstating his arguments, but for good reason. H e recalls, for example,
the many methodological hoops that he and his co-pioneers had to jump through
to convince skeptics that anything—not to say everything—happens outside of
conscious control. Having made that demonstration so incisively, Bargh now muses
(or seriously urges?) that we should adopt the same skepticism toward models in
which conscious mediation has a starring role. So be it. The point is fair, skepticism
should always be welcome in science, and most of life seems to run off automatically.
But evidence for the shrinking role of conscious mediation, and for the importance
186 Mischel
of automaticity and the situation, does not make situational determinism a more
adequate explanation today than it was when the cognitive revolution arose to
protest it. O f course Bargh knows that himself, and lets us know he does, and lets
us in o n a host of interesting insights in the process. T h e danger is that the casual
or unsophisticated reader can skim this rich contribution, remember its zealous tone
and opening Skinnerian manifesto, but miss the depth of its final wise conclusions.
Worse, it may perpetuate the classic definition and purpose of social psychology,
with which Bargh starts his essay, restricting it to the discovery of the "situational
causes of thinking, feeling, and acting i n the real or implied presence of other
people" (chap. 1, p. 1). But such a definition may encourage the field to demonstrate
(over and over) the power of the situation versus that of the person, framed as a
competition between two opposing entities, rather than address the dynamic
reciprocal interactions between these two codependents in which some of the most
interesting phenomena of social psychology are rooted (e.g., Higgins, 1990; Mischel
& Shoda, 1995).
REFERENCES
Higgins, E. T. (1990). Personality, social psychology, and person-situation relations: Standards and
knowledge activation as a common language. In L. A . Pervin (Ed.), Handbook of personality: Theory
and research (pp. 301-338). New York: Guilford.
Higgins, E. T , & Brendl, C . M . (1995). Accessibility and applicability: Some "activation rules"
influencing judgment. Journal of Experimental Social Psychology, 31,218-243.
Mischel, W. (1973). Toward a cognitive social learning reconceptualization of personality. Psychological
Review, 80, 252-283.
Mischel, W. (1990). Personality dispositions revisited and revised: A view after three decades. In L. A .
Pervin (Ed.), Handbook of personality: Theory and research (pp. 111-134). New York: Guilford.
Mischel, W. (1996). Personality and assessment. Mahwah, NJ: Lawrence Erlbaum Associates. (Original
work published 1968).
Mischel, W. (1996). From good intentions to willpower. In R M . Gollwitzer & J. A . Bargh (Eds.), The
psychology of action: Linking cognition and motivation to behavior, (pp. 197-218). New York: Guilford.
Mischel, W., Cantor, N . , & Feldman, S. (in press). Principles of self-regulation: T h e nature of willpower
and self-control. In E. T. Higgins & A . W. Kruglanski (Eds.), Social Psychology: Handbook of basic
principles (pp. 329-360). T X : New York: Guilford.
Mischel, W , & Shoda, Y. (1995). A cognitive-affective system theory of personality: Reconceptualizing
the situations, dispositions, dynamics, and invariance in personality structure. Psychological Review,
102, 246-268.
Mischel, W., Shoda, Y , & Peake, P (1988). The nature of adolescent competencies predicted by
preschool delay of gratification. Journal of Personality and Social Psychology, 54, 687-696.
Mischel, W., Shoda, Y , & Rodriguez, M . L. (1989). Delay of Gratification in children. Science, 244,
933-938.
Ross, L., & Nisbett, R. E. (1991). The person and the situation: Perspectives of social psychology. New York:
McGraw-Hill.
Shoda, Y , Mischel, W., & Peake, P K. (1990). Predicting adolescent cognitive and self-regulatory
competencies from preschool delay of gratification: Identifying diagnostic conditions. Developmental
Shoda, Y , Mischel, W., & Wright, J. C . (1994). Intraindividual stability in the organization and
patterning of behavior: Incorporating psychological situations into the idiographic analysis of person-
ality. Journal of Personality and Social Psychology, 67, 674-687.
Chapter 12
Preconscious Automaticity in a
Modular Connectionist System
Eliot R. Smith
Purdue University
John Bargh's target chapter provides an excellent summary and overview of many
lines of evidence for the importance of preconscious automaticity i n social thought,
feelings, and behavior. He correctly concludes that symbolic models of the sort that
were traditionally applied i n social psychology are inadequate to explain this
evidence. However, other models may hold more potential. T h e main part of this
chapter, following some discussion of definitional issues, outlines a type of theory
that, I believe, can accommodate the evidence Bargh presents. This theoretical
sketch rests o n the properties of connectionist networks rather than the symbolic
systems that have been the most familiar theories i n social cognition. It is only a
sketch, far from a well-developed theory. Yet, I hope it suggests potentially important
considerations as social psychologists begin exploring the connectionist models that
have been so influential i n other areas of psychology i n the last decade.
DEFINITION OF "AUTOMATIC"
T h e key term i n Bargh's article is automatic. It is unfortunate, then, that clear

definition of this term is marred by two types of confusion i n the chapter. First, Bargh
at least rhetorically contrasts automatically or situationally caused acts to "free will
or conscious choice" (chap. 1, p. 6). T h e introduction of the free will idea is
unnecessary and invites confusion. T h e notion of free will connotes a moral
dimension, the idea that an actor is responsible or can be attributed praise or blame
for an action. Yet, as others argued (e.g., Fiske, 1989), people can meaningfully be
held accountable for automatic as well as consciously chosen acts. T h e idea of free
will also implies that two people i n the same external situation may act differ-
187
188 Smith
ently—and despite Bargh's emphasis o n the reliable effects of social situations on

behavior, social psychologists are quite familiar with this meaning of free will as well,
We call it error variancel In most social psychological studies, it is far larger than the
variance due to situational manipulations. So defining automatic as a contrast to
free will does not work very well (see Franklin, 1995, for an excellent discussion of
free will).
Bargh's argument also seems to identify the idea of automaticity with the idea
of causality. In the beginning of Bargh's chapter, automatic seems to be defined as
"reliably manipulable." For example, "knowledge of situational features that pro-
duce a given phenomenon for most people [is] a specifiable if-then relation tanta-
mount to an automatic process" (chap. 1, p. 6). Furthermore Bargh writes that even
if a situation causes some behavior i n most people by eliciting a conscious intention,
the entire process can still be termed automatic.
This discussion seems to confuse the ideas of cause and automaticity. Automat-
icity has a well-accepted definition as a property of psychological processes (Bargh,
1994), whose indicators include lack of subjective awareness, insensitivity to
intention, and the like (as I discuss i n a moment). Research that manipulates
situational factors and measures psychological responses is well-suited to assessing
causation, but unless specific designs are used, cannot tell us anything about
characteristics of the mediating processes, including whether they are automatic.
Thus, if a situational factor caused a particular conscious thought or intention i n
most people, which i n turn led them to behave i n a certain way, Bargh would term
the process automatic, whereas I would argue that this is a demonstration of causal
potency—an entirely different matter. Space precludes further development of the
causality-automaticity distinction here, but readers are invited to refer to C o o k and
Campbell's (1979) seminal discussion.
So how should automaticity i n the sense targeted by this chapter— precon-
scious automaticity—be defined? O n e answer is implicit i n several of the empirical
demonstrations the chapter reviews. Preconscious automaticity implies lack of
mediation by consciously accessible thoughts or feelings. Operationally, therefore,
we can test this hypothesis by the usual means of testing mediation. A study
manipulates some situational factors, measures potential conscious mediators with
self-report techniques, and measures a final dependent variable (e.g., some behav-
ior). If appropriate data analyses show that the self-report variable does not
mediate the effect, that is consistent with the effect involving preconscious
automaticity.
A l t h o u g h this technique is valuable and obviously widely used, it also has
potential problems. It relies on subjects' cooperativeness and honesty for valid
self-reports. T h e researcher may be measuring the wrong potential mediator;
perhaps asking a different self-report question would reveal mediation by con-
sciously accessible thoughts. Also, careful conceptual definitions of automaticity
(see Bargh, 1994) emphasize more than inaccessibility to conscious awareness. For
one thing, automatic processes are supposed to be relatively impervious to manipu-
lations of conscious intention. Studies could be designed to test the automaticity
of a given process by manipulating the subject's conscious intentions (cf. Jacoby,
12. Modular Connectionist System 189
Toth, & Yonelinas, 1993). Automatic processes are also supposed to be insensitive
to the amount of available cognitive capacity. Tests could be designed using this
criterion to determine whether a given process is automatic (cf. Bargh & Tota,
1988). Future research should use a multiplicity of indications to draw conclusions
about the degree of automaticity of social psychological processes, but should avoid
assuming that any effect reliably elicited by a situational manipulation is automatic.
SKETCH OF A MODEL
Due to space constraints, I do not quibble with the individual studies Bargh
describes as evidence for pervasive automaticity of social psychological processes.
In addition, I agree with his major conceptual point, that current theories i n social
psychology are not well suited to explaining these phenomena. However, Bargh
states that i n his opinion "no one general cognitive model" (chap. 1, p. 49) can
account for all the results he discusses. I would like to take up the challenge implicit
in this statement, after dealing with one terminological issue. Bargh recognizes the
unfortunate ambiguity i n the term cognitive, which (a) can refer to any type of mental
processes, or (b) can be used i n a narrower sense, contrasted with affective or
motivational processes. Still, the term is used in both ways at various points i n the
chapter, and i n context it is unclear which meaning is part of the claim. If he means
(using the second definition) that no model without affective or motivational
components can account for all his evidence, this is true by definition. If he means
(using the first definition) that no model of mental processes using a single overall
set of operating principles can account for the evidence, I disagree. I intend to
outline a type of model I believe holds the potential to account for observations of
automaticity within social psychology. A s I do so, I comment on its relations to
Bargh's points.
Connectionist Network
With Distributed Representations
Space does not permit a general introduction to the properties of connectionist
models here. Introductions can be found i n several chapters i n Rumelhart,
M c C l e l l a n d , Asanoma, C r i c k , & E l m a n , e t a l . (1986) and M c C l e l l a n d , Rumelhart,
A s a n o m a , C r i c k , & Elman, et al. (1986), and i n C h u r c h l a n d and Sejnowski
(1992). Smith (1996) offered a brief overview for social psychologists. T h e most
significant properties of these models for present purposes are the following:
1. A l l representation and processing are performed by a set of simple and richly

interconnected units—idealized neurons. These units receive excitatory and inhibi-
tory inputs across weighted connections from other units, sum those inputs to
determine their own activation, and send the resulting output to other units.
190 Smith
2. A concept or object is represented by a pattern of activation distributed across

a set of processing units. Activation levels can change rapidly, and the current
activation pattern is identified with a transitory mental state.
3. Representations are not "stored" inertly until retrieved by a search process
and used. Instead, the same units are both representational vehicles and processing
mechanisms; a single mechanism, the flow of activation along connections between
units, accounts for both storage and processing of information. Retrieval is the
reinstatement of a previously processed activation pattern that is similar to the
current inputs.
4. Retrieval is made possible by a learning process that makes relatively enduring
changes in the connection weights as information is processed by the network.
Evolved, Modular System

Connectionist models have been developed with many diverse architectures. The
one I propose has a modular structure. Symbolic models of cognition, inspired by
the computer, tend to be unified models. These, including, for example, Wyer and
Srull (1989) and an earlier proposal of mine (Smith, 1984) are models in which a
serial, limited-capacity central processor operates in a uniform way on all types of
information. Some aspects of the human cognitive system (such as our ability to
reason using linguistically encoded thought) may work in this way. However, more
recent theories emphasize the modular nature of cognition (see citations in Bargh,
chap. 1). Humans evolved by adding brain structures and mental abilities on top of
those possessed by other primates, rather than by starting fresh with a new brain
design. Similarly, mammals added modules on top of those possessed by their
predecessors, and so on. Evolutionary older modules such as the brainstem and
limbic system are common across most vertebrates (Ornstein, 1986). This perspec-
tive implies, among other things, that we should not assume (as Bargh does; chap.
1) that all preconscious processes were originally conscious and became automatic
with practice. Some automatic processes must be assumed to be relatively "hard-
wired" by evolution.
Following Bargh's lead, let us particularly focus on modules that process sensory
inputs. The architecture I suggest (see Fig. 12.1) involves one or more modules that
receive input from sense organs and produce as output semantic representations of
"what's out there." We could call these perceptual modules. Other modules receive
their inputs from the perceptual modules. For example, evaluative modules scan the
semantic representations looking for patterns that can be identified as good or bad
(probably these are two separate modules rather than one with a bipolar output
signal; see Cacioppo & Berntson, 1994). Affective and motivational modules similarly
scan the output of perceptual interpretive modules, looking for patterns that
represent motivationally significant situations, such as the presence of food, danger,
novel stimuli, and so on. When they are activated by semantic patterns that they
are seeking, these modules trigger motor plans and ultimately overt behavior.
Motivational modules such as those that search for food should have their operation
regulated by internal signals representing the individual's state of hunger or satiety
Sensory inputs
Perceptual module:
computes semantic
representations
Affective module: Evaluative module:

computes affective computes evaluations
responses
Behavioral goal
module: represents
current goal
Motor systems
FIG. 12.1. Basic architecture of a modular connectionist system.

Note: Units within modules (not shown) are assumed to be interconnected with excitatory
and inhibitory links. Arrows between modules in the figure represent excitatory connections
from units in one module to units in another. For simplicity only one module of each type
(perceptual, evaluative, and affective) is shown; in actuality there may be several of each.
(Dorman & Gaudiano, 1995). We need not stay at this concrete level; the example
could equally well be an achievement motivation module that, when activated,
looks for abstract situational features conducive to achievement or comparisons
against standards of excellence.
Let me distinguish this modular architecture from two alternatives. First, Bargh
assumes that properties like good, bad, dangerous, interesting, and the like are
semantic features included in mental representations of objects, and argues that
they cannot explain observations of evaluative priming (Bargh, chap. 1). In my
thinking, they are not features in the semantic representation computed by a
perceptual module; they are computed separately, in evaluative or affective-moti-
vational modules. The difference is that a connectionist module computing evalu-
ation can recognize a number of unrelated patterns (see Churchland & Sejnowski,
1992) and can produce a common output for numerous such patterns. Therefore,
for instance, a module might recognize both flowers and puppies as good or pleasant
even if those patterns share no semantic features; we need not assume that all good
(or bad) things have anything in common.
Second, some theorists arguing for separate affective and cognitive systems (e.g.,
Zajonc, 1980) strayed dangerously close to treating the affective component as
mysterious, even ineffable (in contrast to the cognitive component, whose princi-
ples of operation are known at least in broad outline). This position risks becoming
unscientific by permitting no concrete predictions. Certainly Bargh does not mean
to fall into this camp, but by repeating the idea that affect "does not play by the
same rules" as does cognition (chap. 1, p. 23) he may unintentionally do so. My
proposal is definitely not of this sort. The individual connectionist units in all types
of modules follow the same operating principles (i.e., summing their inputs from
192 Smith
other units and sending their own resultant activation level over their outputs,
changing their connection weights according to a common learning rule). My
affective or motivational modules are no different in their internal operations from
any other modules; their differences are functional ones related to their specific role
in self-regulatory subsystems of the overall processing mechanism. Recurrent feed-
back connections among units within a module allow a given pattern, once elicited
by external inputs, to remain active for a time rather than dying out. This effect
may account for the observation (Bargh, chap. 1) that "motivational priming" can
be long-lasting or even have increasing effects over time, at least in the absence of
competing goals.
Pervasive Interactivity of Processing

The discussion so far may suggest a purely bottom-up system: Perceptual modules
look at sensory inputs and produce semantic representations of the environment.
Other modules look at those representations and produce their own outputs, but
being later in the information-processing chain, cannot influence the earlier per-
ceptual stages. However, a basic principle of neural systems as well as of the model
I sketch here is that processing is thoroughly interactive. This means that connec-
tions between modules are rarely one-way. If module A sends outputs to module B,
then B sends recurrent outputs back to A (Felleman & Van Essen, 1991; Movellan
& McClelland, 1995). From the perceptual level forward, the cognitive system
affects the information it processes.
Interactivity means that each module influences the activity of others and thus
modules cannot be put in a strict sequential order. Seemingly "later" modules, such
as the evaluative and motivational ones mentioned previously can influence the
way "earlier" perceptual modules settle on semantic interpretations of stimulus
input. Interactions between processing units between and within modules allow the
network as a whole to settle into a stable pattern that includes both a pattern in
the semantic module representing the meaning of the stimulus and a pattern in
evaluative modules representing its evaluation, and so on. Many connectionist
models embody this principle. The pioneering McClelland and Rumelhart (1981)
paper modeled letter and word recognition in an interactive system. If letters are
assumed to be recognized first and then put together into words as a later stage, it
becomes difficult to account for various empirical findings (such as people's better
ability to recognize letters in words than when presented in isolation). Instead,
word-level knowledge feeds back and influences letter recognition. Rueckl (1990)
and Seidenberg and McClelland (1989) proposed models of word pronunciation
and reading that involve separate but interacting modules for orthography, pronun-
ciation, and meaning.
In social psychology, I propose that affective and motivational modules influence
perception through similar interactive processes. Thus, even as a semantic inter-
pretation of a situation is being constructed, its motivational implications are also
being assessed—and feedback from motivational modules may influence the nature
of the final perceptual representation. For example, patterns representing motiva-
tionally favorable situations may be more stable and form on the basis of less
definitive input cues, compared to patterns representing less desirable situations.
Self-enhancing and other biases may be due to this sort of mechanism. Recent work
shows that having an accessible attitude toward an object makes that object auto-
matically grab attention (Roskos-Ewoldsen 6k Fazio, 1992). This might be explained
by recurrent connections back from an evaluative module to a semantic module. The
presence of a strong evaluative pattern elicited by the semantic pattern representing
the object may, through these recurrent connections, strengthen and stabilize the
semantic pattern in a way analogous to the effect of the semantic pattern for a word
on recognition of its orthographic pattern in Rueckl's (1990) model.
Context Sensitive Representations

A n important dynamic aspect of distributed representations is their flexibility and
context sensitivity. In traditional models, representations that are not currently
active are stored away inertly until accessed by a retrieval process. What is
eventually retrieved is exactly the same as what was stored. In connectionist models,
flows of activation through connection weights reconstruct a representation as a
distributed pattern of activation. In this process, any other current sources of
activation (e.g., patterns representing the person's mood, perceptually present
objects, current concerns, or goals) will also influence the "good person" repre-
sentation. For instance, thinking of a good person in the context of a crime reporter
might activate a representation that includes pushy and demanding, whereas in the
context of a waiter, the resulting representation might include features like gracious
and obsequious (see Bargh, chap. 1). Such thoroughgoing context sensitivity is an
inherent property of distributed representations (Clark, 1993).
Evidence in many areas of nonsocial (Barsalou, 1987) and social psychology
demonstrates the flexibility and context sensitivity of mental representations. For
example, Markus and Wurf (1987) advanced the notion of a working self concept,
the contextually relevant set of self-attributes that are currently active. According
to Wilson and Hodges (1992), attitudes are constructed on the spot in a flexible
and context-dependent manner rather than being retrieved from memory in
invariant form every time they are accessed. Stereotypes—although usually re-
garded as highly stable knowledge structures—are also sensitive to recent experi-
ences, a point that was strikingly illustrated by Bodenhausen, Schwarz, Bless, and
Wanke (1995). It seems likely that all types of cognitive representations will be
found to be flexibly reconstructed in a context-sensitive way rather than retrieved
from memory as they were stored—like items buried in a time capsule—as assumed
by many current symbolic theories in social psychology.
Within-Module Competition
Connections between modules are generally excitatory in nature. As noted earlier,
this principle rests on neurophysiological evidence and is part of most existing
modular connectionist models (Usher 6k McClelland, 1995). This makes sense
194 Smith
functionally i n that one wants to allow even subtle cues from one module to partially
activate a related representation i n another module, preparing it for full activation
to occur quickly if the cues become clearer or if many independent sources of
evidence accumulate.
However, connections within modules can be excitatory or inhibitory (Murre,
1992, Usher, & M c C l e l l a n d , 1995). Inhibitory connections allow for competition
among incompatible patterns, so that a given perceptual cue, for example, is not
simultaneously seen as part of two different objects. Furthermore, as Bargh notes,
motor plans must be carried out i n serial fashion for behavior to be organized and
effective. A l t h o u g h partial activation of many representations from multiple cues
is a reasonable approach for a perceptual system, simultaneous partial activation of
many different behaviors is only a recipe for disaster (Dorman & Gaudiano, 1995).
Therefore, at some level a winner-take-all competitive scheme, which can be
implemented by mutual inhibitory connections within a behavioral goal module, is
necessary.
Connections Modified by Learning

A connectionist network can learn a set of connection weights that permit it to
perform some useful task. The learning takes place through incremental changes
i n the weights as the network processes stimuli. Thus, repeatedly processing the
same stimulus i n the same way increases the efficiency and automaticity of that
specific process. Smith (1994; and Bargh, e.g., chap. 1) described implications of
automatization for social psychology.
Learning i n connectionist models can be of several different types (Churchland
& Sejnowski, 1992). First, learning can be "unsupervised" i n the sense that the
network simply observes its inputs and changes its connections based on regularities
i n the input. N o external "supervisor" is required to tell the network what the
"correct answer" is for each input. Unsupervised nets can, for example, detect sets
of features that covary across a number of input patterns. The process is analogous
to the statistical technique of factor analysis, which uncovers patterns of covariation
within a single set of variables (not divided into independent and dependent
variables). In statistical analysis, such patterns (i.e., factor scores) may serve as
inputs for further analysis. Likewise, i n a connectionist model, regularities detected
by unsupervised learning can serve as higher level input features to be further
processed by other networks.
Second, learning can be "supervised." Some source of information outside of a
given module indicates whether the module's responses are appropriate. For exam-
ple, consider a categorization task i n which a pattern of stimulus attributes is applied
to the input units of a network, and one of several possible output patterns becomes
active to represent the network's decision as to the category membership of the
stimulus. A network could be trained through supervised learning to perform such
a categorization task. Initially, all connection weights are given zero or random
values. A training pattern is presented to the input, and the network's output
observed. Using one of several specific procedures such as back-propagation, the
12. M o d u l a r C o n n e c t i o n i s t System 195
weights are then adjusted incrementally to reduce the discrepancy between the
network's output and the correct output provided by the supervisor (reflecting the
known category membership of the training stimulus). This process is repeated
many times with a given set of training stimuli. After enough training, the weights
usually stabilize at values that give adequate performance at categorizing the
training stimuli. T h e network can then be tested by presenting it with new stimuli
(not part of the training set) and observing how it categorizes them. The process is
analogous to the statistical technique of regression analysis, for the network learns
which input features to use in predicting the output category membership.
Supervised learning by definition requires assistance from outside the module
that is being trained, but may or may not require a supervisor outside the entire
connectionist system (e.g., an organism). Thus, a child may learn as his or her
parents correct his or her word usage or pronunciation (external supervision), or
may correct his or her own errors by trying to match behaviors to those others
perform. Built-in reward and punishment systems can also serve as supervisors,
setting connection weights so that organisms learn what behaviors produce good
outcomes, like nourishment and relatively novel stimuli, and avoid bad outcomes,
like pain. Franklin (1995) believed that such built-in values are essential guides for
the development of mind. Their existence means that a connectionist network need
not be treated as a tabula rasa that can learn just anything i n a suitable environment.
Instead, built-in (evolved) systems bias the connectionist modules' learning from
the beginning. A s Clark (1993) pointed out i n his excellent discussion, this
perspective gives us a broader view of the tired old nature-nurture controversy. We
need not assume that a given function is either learned or hardwired. Instead, we
could view a particular function as, say, 30% genetic. For example, specific details
(such as a liking for German chocolate cake) are learned, but biased and constrained
by built-in preferences (such as the general mammalian taste for sweets and fats).
Following the Principle of Accessibility

Connectionist learning, whether supervised or unsupervised, involves incremental
modification of the connection weights that takes place after the presentation of each
stimulus. Therefore, the recency and frequency with which a pattern was encoun-
tered during the learning process influence the ease with which it can be elicited by
a given set of cues. W h e n a stimulus is processed, the incremental weight changes
produced by learning make the current pattern and similar ones slightly easier to
reproduce in the future, at the expense of slightly distorting (and worsening perform-
ance on) unrelated patterns. In other words, the principle of accessibility is inherent
in the network's operation. In traditional theories within social psychology, accessi-
bility is not intrinsic to basic theoretical processes, but is explained by special ad-hoc
mechanisms, such as a storage battery containing time-varying amounts of charge
attached to each discrete representation (Higgins, 1989), or a top-down search of a
storage bin holding multiple copies of each representation (Wyer 6k Srull, 1989).
Connectionist models make the novel prediction that recent and frequent
exposure produce two distinct types of "accessibility" with different properties
196 Smith
(Wiles & Humphreys, 1993), one dependent on current unit activations and the
other on changes i n connection weights. First, a pattern of activation may persist i n
a module for a short time after a stimulus is processed, so that if the next pattern is
related to the first its processing may be facilitated (Masson, 1991). This type of
accessibility may underlie semantic priming, the observation that having just read
the word bread makes it easier for people to read butter. The activation patterns
representing bread and butter will overlap to a greater extent than do representations
of unrelated words; this is a property of the distributed representations produced by
typical connectionist learning rules (Clark, 1993). The connectionist account
predicts that this sort of priming should last only briefly and should be abolished by
one or two intervening unrelated words (which would create unrelated patterns of
activation).
Second, processing a stimulus leads to incremental changes i n the connection
weights i n a network. This change is long-lasting, and its effects diminish not with
time but with interference from unrelated patterns. M a n y people have an intuition
that the effects of weight changes caused by processing a stimulus on a single
occasion could not be demonstrable over days or even weeks, although "priming"
effects clearly can last that long (e.g., Smith, Stewart, & Buttram, 1992). However,
Wiles and Humphreys (1993) argued i n quantitative detail that this intuition is
misleading. If a particular stimulus is processed frequently over months and years,
the resulting systematic shifts i n connection weights will influence the individual's
processing characteristics for years, even a lifetime (a property termed chronic
accessibility i n the social literature).
A l t h o u g h the mechanisms are different, under some circumstances these two
forms of priming may have similar effects, such as increasing the probability that
people will assimilate an ambiguous stimulus to the primed category. Bargh argues
that the two forms depend on the same underlying mechanism (chap. 1, p. 35).
However, this conclusion can be questioned, for some evidence suggests that the
two types of accessibility can have somewhat different properties (Bargh, Lombardi,
& Higgins, 1988; Higgins, Bargh, & Lombardi, 1985; Smith & Branscombe, 1987).
Furthermore, more focused empirical tests of possible differences between two forms
of accessibility, hypothesized by this type of connectionist account but not by
existing models, would be of value.
Conscious Processing Layered on Top

Understandably i n an chapter devoted mainly to preconscious automatic process-
ing, Bargh does not give much attention to the role of conscious thought. His main
point i n this regard is the serial nature of conscious thought, which he suggests is
important i n connecting a mind incorporating multiple parallel processing struc-
tures to a "serial world" (chap. 1, p. 53). A s I noted earlier, the generation of
behavior does have to be serial. More precisely, winner-take-all competitive activa-
tion rules or other conflict-resolution mechanisms must operate so that behavior is
controlled i n an organized way rather than simultaneously by a multitude of partially
activated goals.
Strangely for a social psychologist, Bargh does not discuss another key charac-
teristic of conscious thought: its linguistic and therefore intrinsically social origins
and nature (see Levine, Resnick, & Higgins, 1993). Smolensky's (1988) vision of
connectionism elaborated o n this point. Smolensky held that people have two
separate processors. The top-levelconscious processor uses linguistically encoded and
culturally derived knowledge as its "program." This is the processor people use when
they follow explicit step-by-step instructions or engage i n conscious, effortful
reasoning. It is based on the same cognitive capacities that underlie public language
use, such as the ability to parse sentences into their components and to combine
words following grammatical rules. This system can recombine known linguistic
symbols into new patterns, and can quickly formulate and store symbolic expres-
sions representing newly learned knowledge. Ultimately, all these capacities must
rest on computations carried out by connectionist networks, which are assumed to
roughly characterize the way the brain works. For example, linguistic expressions
must be encoded as distributed patterns of activation and stored i n connectionist
memories (Smolensky, 1988). Numerous theorists are currently working o n con-
nectionist models of linguistic phenomena (Barnden, 1995; Clark, 1993; Elman,
1995; Shastri, 1995), although some earlier models i n this area were naive and
unrealistic (Pinker & Prince, 1988).
In contrast, i n Smolensky's (1988) model the intuitive processor is responsible for
most human behavior (and all animal behavior), including perception, skilled
motor behavior, and intuitive problem solving and pattern matching. T h i s proc-
essor does not rely o n language, but directly rests on properties of subsymbolic
connectionist networks. Learning in this system is slow, occurring only with
repeated experience. Processing i n this system can be described i n rational,
symbolic terms, but they will always be imprecise approximations.
Psychologists advanced many related dual-process models emphasizing the
distinction between controlled (conscious, systematic) and automatic (noncon-
scious, heuristic) processing (Sloman, 1996; S m i t h , 1994). Evidence for dual
processes includes, for example, the predictable effects of manipulations that
drain cognitive capacity (such as distraction) or increase or diminish m o t i v a t i o n
to process carefully; these manipulations seem to knock out conscious process-
ing and leave automatic processing relatively unaffected. Smolensky's approach
seems quite compatible with these models, although social psychologists often
incorporate important points that Smolensky failed to consider, such as the fact
that both cognitive capacity and motivation are typically required for people to
use the top-level conscious processor rather than the heuristically based i n t u i -
tive processor.
A n explicit model of the interrelationships of intuitive or preconscious and
conscious processing must incorporate accounts of the ways the two systems
interact. Bargh considers:
• Automatization (cf. Smith, 1994), or the decrease of resource demands as a

process is repeatedly carried out—at first under explicit conscious control but later
more and more automatically.
198 Smith
• Construction of reality, the idea that preconscious interpretive systems produce

our consciously accessible perceptual world as their output (Bargh, chap. 1). Any
conscious reasoning—in the absence of very special circumstances (cf. Martin, Seta,
6k Crelia, 1990)—takes automatically generated interpretations and feelings as a
starting point.
However, the systems interact in other ways:
• Although automatization of initially conscious processes certainly does occur,

in some domains intuitive processes may come first and explicit thought second.
For example, children seem to initially categorize animals by overall physical
appearance and only later learn to override these responses with fixed rules such as
that an animal is always of the same kind as its parents (Keil, 1989). Considerations
of development and evolution clearly indicate that use of the intuitive processor
can come first; its operations do not all reflect processes that were initially carried
out intentionally and consciously, although some do.
• In some circumstances, intuitive and reflective processes may run in parallel
and whichever finishes first (or reaches an acceptable level of confidence first)
controls the overt response. This was the basic assumption, for example, in Logan's
(1988) instance theory of automaticity.
• "Gut feelings" based on intuitive processes can influence our degree of moti-
vation to engage in conscious thought. For example, we may spend more time trying
to see flaws in an argument whose conclusion we do not like (for whatever intuitive,
nonrational reasons) than in an argument whose conclusion is congenial (cf. Gilbert,
1993).
• O n the basis of much evidence, McClelland, McNaughton, and O'Reilly (1995)
proposed a connectionist model with two separate memory systems: one analogous to
the neocortex, which changes connection weights only slowly and stores stably struc-
tured "schematic" general knowledge, and another analogous to the hippocampal
system, which rapidly learns new information and forms episodic traces that are
accessible to consciousness. New knowledge is transferred from the latter system into
the former in a slow "consolidation" process. One important implication of this model
for social psychology is a predicted difference in the type of information to which the
two systems attend. Schematic learning is chiefly concerned with regularities, so it
records primarily what is typical and expected. In contrast, rapidly formed episodic
memories should record the details of events that are novel and interesting: In other
words, this system should attend more to the unexpected and unpredicted. This
difference may well be reflected in the typicalfindingsof social psychological studies
that people attend to and recall mostly expectancy-inconsistent information when
forming a new impression, but mostly expectation-consistent information when work-
ing with a well-formed and solid expectation (Higgins & Bargh, 1987).
No doubt there are other forms of interaction between intuitive, preconscious,

automatic systems and reasoning that involves conscious awareness. A well-speci-
fied overall model of cognition must incorporate some account of the various
pathways and mutual influences, rather than minimizing the contribution of

conscious processes, as Bargh seems to do. In his recent review of evidence for two
forms of reasoning, Sloman (1996) wrote:
The common mode of operation of the two systems is clearly interactive. Together
they lend their different computational resources to the task at hand; they function
as two experts who are working cooperatively to compute sensible answers. One
system may be able to mimic the computation performed by the other, but only with
effort and inefficiency, and even then not necessarily reliably. The systems have
different goals and are specialists at different kinds of problems. But when a person is
given a problem, both systems may try to solve it, each may compute a response, and
those responses may not agree.... Because the systems cannot be distinguished by the
problem domains to which they apply, deciding which system is responsible for a given
response is not always easy. It may not even be possible, because both systems may
contribute to a particular response, (p. 6)
CONCLUSIONS
I believe that a model of the sort sketched here will display the types of automatic
processing that Bargh describes in his review. As sensory information enters the
system, encoded as distributed patterns of activation, perceptual modules begin to
compute semantic representations of "what's out there" based on the input and
prior experience encoded in the connection weights. But the computation is not
purely bottom-up or stimulus-driven; after a short time, activation spreads from the
perceptual to evaluative and affective modules, which begin to compute their own
representations—and through interactive feedback connections, to influence the
pattern of activity into which the "earlier" perceptual module settles. A l l this occurs
prior to conscious awareness; these modules are part of the system that construct
the individual's subjective experience.
In many ways, this chapter could be viewed as an extension and fuller specifica-
tion of Bargh's own suggested parallel processing model (his Fig. 1.1). However,
there are some important differences between my proposal and his, as I have already
noted in several cases.
• In contrast to Bargh, I doubt that all automatic processes were previously

consciously controlled.
• I do not consider good-bad to be among the semantic features included in
object representations.
• In connectionist theories, short-term priming and long-term "chronic acces-
sibility" are generated by distinct mechanisms (whose effects may often overlap),
rather than by common mechanisms.
• My proposal is for a system whose modules can be termed perceptual,
evaluative, and so on, based on their different functional roles in the overall system,
but which all rest on identical internal processes. Bargh states that no model with
uniform operating principles can account for his data, an argument that comes close
200 Smith
to saying that affective or evaluative systems are ineffable, playing by "different

rules" than the cognitive system (chap. 1).
• This connectionist model or its close relatives (e.g.,Chappell & Humphreys,
1994; McClelland et al., 1995; Seidenberg & McClelland, 1989) were demon-
strated to account for many other types of data Bargh does not address: context
sensitivity of concepts, implicit memory and priming, various types of interaction
between conscious and preconscious processing.
• This model can be made much more detailed and explicit, to the point where
computer simulations could be used to demonstrate its predictions. For example,
Smith and DeCoster (1996) demonstrated with explicit simulations that a single
module (roughly equivalent to the perceptual module in the multiple-module
system advanced here) can reproduce many phenomena that are the subject of
existing theories of person perception and stereotyping, as well as generate some
new predictions.
I said at the outset that this chapter would present a sketch of a model, and space
limitations hold me to that. However, the sketch presented here compares favorably
in the detail of its process assumptions to the parallel model advanced by Bargh in
the target chapter, as well as to other well-known models in social cognition. I look
forward to the time when modular, parallel models are developed—whether with
connectionist or more traditional symbolic assumptions—to the point where their
detailed predictions can actually be compared with empirical data.
ACKNOWLEDGMENTS
Preparation of this chapter was supported by a research grant (ROI MH46840) and
a Research Career Development Award (K02 MH01178) from the National
Institutes of Mental Health. Address correspondence to Eliot Smith, Department
of Psychological Sciences, Purdue University, W. Lafayette, IN 47907-1364, or
esmith@psych.purdue.edu.
REFERENCES
Bargh, J. A . (1994). T h e four horsemen of automaticity: Awareness, intention, efficiency, and control
in social cognition. In R. S. Wyer & T . K. Srull (Eds.), Handbook of social cognition (2nd ed., pp. 1-40).
Bargh, J. A . , Lombardi, W. J., & Higgins, E. T. (1988). Automaticity of chronically accessible constructs
in person X situation effects on person perception: It's just a matter of time. Journal of Personality and
Bargh, J. A . , & T o t a , M . E. (1988). Context-dependent automatic processing in depression: Accessibility
54, 925-939.
Barnden, J. A . (1995). Artificial intelligence and neural networks. In M . A . Arbib (Ed.), Handbook of
brain theory and neural networks (pp. 98-102). Cambridge, M A : M I T Press.
Barsalou, L. (1987). T h e instability of graded structure: Implications for the nature of concepts. In U .
Neisser (Ed.), Concepts and conceptual development (pp. 101-140). Cambridge, England: Cambridge
University Press.
Bodenhausen, G . V , Schwarz, N., Bless, H . , & Wanke, M . (1995). Effects of atypical exemplars on racial
beliefs: Enlightened racism or generalized appraisals? Journal of Experimental Social Psychology, 31,
48-63.
Cacioppo, J. T., & Berntson, G . G . (1994). Relationship between attitudes and evaluative space: A
critical review, with emphasis on the separability of positive and negative substrates. Psychological
Bulletin, 115, 401-423.
Chappell, M , & Humphreys, M . S. (1994). A n auto-associative neural network for sparse repre-
sentations: Analysis and application to models of recognition and cued recall. Psychological Review,
101, 103-128.
Churchland, P. S., & Sejnowski, T. J. (1992). T h e computational brain. Cambridge, M A : M I T Press.
Clark, A . (1993). Associative engines: Connectionism, concepts, and representational change. Cambridge,
M A : M I T Press.
Cook, T D., & Campbell, D. T (1979). Quasi-experimentation. Chicago: Rand McNally.
Dorman, C , & Gaudiano, P. (1995). Motivation. In M . A . Arbib (Ed.), Handhook of brain theory and
neural networks (pp. 591-594). Cambridge, M A : M I T Press.
Elman, J. L. (1995). Language processing. In M . A . Arbib (Ed.), Handbook of brain theory and neural
networks (pp. 508-513). Cambridge, M A : M I T Press.
Felleman, D. J., & Van Essen, D. C . (1991). Distributed hierarchical processing in the primate cerebral
cortex. Cerebral Cortex, 1, 1-47.
Fiske, S. T (1989). Examining the role of intent: Toward understanding its role in stereotyping and
prejudice. Inj. S. Uleman & J. A . Bargh (Eds.), Unintended thought: The limits of awareness, intention,
and control (pp. 253-283). New York: Guilford.
Franklin, S. (1995). Artificial minds. Cambridge, M A : M I T Press.
Gilbert, D. T. (1993). T h e assent of man: Mental representation and the control of belief. In D. M .
Wegner & J. W. Pennebaker (Eds.), Handbook of mental control (pp. 57-87). Englewood Cliffs, NJ:
Prentice-Hall.
Higgins, E. T. (1989). Knowledge accessibility and activation: Subjectivity and suffering from uncon-
scious sources. In J. S. Uleman & J. A . Bargh (Eds.), Unintended thought: The limits of awareness,
intention, and control (pp. 75-123). New York: Guilford.
Higgins, E. T , & Bargh, J. A . (1987). Social cognition and social perception. Annual Review of Psychology,
38, 369-426.
Higgins, E. T , Bargh, J. A . , & Lombardi, W. (1985). Nature of priming effect on categorization. Journal
of Experimental Psychology: Learning, Memory and Cognition, 11, 59-69.
Jacoby, L. L., Toth, J. R, & Yonelinas, A . R (1993). Separating conscious and unconscious influences of
memory: Measuring recollection. Journal of Experimental Psychology: General, 122, 139-154-
Keil, F. (1989). Concepts, kinds, and cognitive development. Cambridge, M A : M I T Press.
Levine, J. M . , Resnick, L. B., & Higgins, E. T. (1993). Social foundations of cognition. Annual Review of
Logan, G . D. (1988). Toward an instance theory of automatization. Psychobgical Review, 95, 492-527.
Markus, H . , & Wurf, E. (1987). The dynamic self-concept: A social psychological perspective. Annual
Review of Psychobgy, 38, 299-337.
Martin, L. L., Seta, J. J., & Crelia, R. A . (1990). Assimilation and contrast as a function of people's
willingness and ability to expend effort in forming an impression. Journal of Personality and Social
Psychobgy, 59, 38-49.
Masson, M . E. J. (1991). A distributed memory model of context effects in word identification. In D.
Besner & G . W. Humphreys (Eds.), Basic processes in reading: Visual word recognition (pp. 233-263).
McClelland, J. L., McNaughton, B. L., & O'Reilly, R. C . (1995). W h y there are complementary learning
systems in the hippocampus and neocortex: Insights from the successes and failures of connectionist
models of learning and memory. Psychobgical Review, 102, 419-457.
McClelland, J. L., & Rumelhart, D. E. (1981). A n interactive activation model of context effects in
letter perception: Part 1. A n account of basic findings. Psychobgical Review, 88, 375-407.
Smith
202
McClelland, J. L , Rumelhart, D. E., Asanuma, C , Crick, F. H . C , Elman, J. L., Hinton, G . E., Jordon,
M . I., Kawamoto, A . R , Munro, E W., Norman, D. A . , Rabin, D. E., Williams, R. J., & Zipser, D.
(Eds.). (1986). Parallel distributed processing (Vol. 2). Cambridge, M A : M I T Press.
Movellan, J. R., & McClelland, J. L. (1995). Stochastic interactive processing, channel separability, and
optimal perceptual inference: An examination of Morton's Law (Tech. Rep. N o . PDRCNS.95.4). Carnegie
Mellon University: Department of Psychology.
Murre, J. M . J. (1992). Learning and categorization in modular neural networks. Hillsdale, NJ: Lawrence
Erlbaum Associates.
Ornstein, R. (1986). Multimind. Boston: Houghton Mifflin.
Pinker, S., & Prince, A . (1988). O n language and connectionism: Analysis of a parallel distributed
processing model of language acquisition. In S. Pinker & J. Mehler (Eds.), Connections and symbols
(pp. 73-193). Cambridge, M A : M I T Press.
Roskos-Ewoldsen, D. R., & Fazio, R. H . (1992). O n the orienting value of attitudes: Attitude accessibility
as a determinant of an object's attraction of visual attention. Journal of Personality and Social Psychology,
63, 198-211.
Rueckl, J. G . (1990). Similarity effects in word and pseudoword repetition priming. Journal of Experimental
Psychology: Learning, Memory, and Cognition, 16, 374-391.
Rumelhart, D. E., McClelland, J. L., Rumelhart, D. E., Asanuma, C , Crick, F. H . C , Elman, J. L., Hinton,
G . E., Jordon, M . I., Kawamoto, A . H . , Munro, P. W., Norman, D. A . , Rabin, D. E., Williams, R. J.,
& Zipser, D. (Eds.).(1986). Parallel distributed processing (Vol. 1). Cambridge, M A : M I T Press.
Seidenberg, M . S., & McClelland, J. L. (1989). A distributed, developmental model of word recognition
and naming. Psychobgical Review, 96, 523-568.
Shastri, L. (1995). Structured connectionist models. In M . A . Arbib (Ed.), Handbook of brain theory and
neural networks (pp. 949-952). Cambridge, M A : M I T Press.
Sloman, S. A . (1996). T h e empirical case for two systems of reasoning. Psychobgical Bulletin, 119, 3-22.
Smith, E. R. (1984). Model of social inference processes. Psychobgical Review, 91, 392-413.
& T K. Srull (Eds.), Handbook of social cognition (2nd ed., Vol. 1, pp. 99-151). Hillsdale, NJ: Lawrence
Erlbaum Associates.
Smith, E. R. (1996). What do connectionism and social psychology offer each other? Journal of Personality
Smith, E. R., & Branscombe, N . R. (1987). Procedurally mediated social inferences: T h e case of category
accessibility effects. Journal of Experimental Social Psychobgy, 23, 361-382.
Smith, E. R., & DeCoster, J. (1996). Knowledge acquisition, accessibility, and use in person perception and
stereotyping: Simulation with a recurrent connectionist network. Unpublished manuscript, Purdue U n i -
versity.
Smith, E. R., Stewart, T. L., & Buttram, R. T. (1992). Inferring a trait from a behavior has long-term,
highly specific effects. Journal of Personality and Social Psychobgy, 62, 753-759.
Smolensky, R (1988). O n the proper treatment of connectionism. Behavioral and Brain Sciences, 11, 1-74.
Usher, M . , & McClelland, J. L. (1995). O n the time course of perceptual choice: A model based on principles
of neural computation. Unpublished manuscript, Carnegie Mellon University.
Wiles, J., & Humphreys, M . S. (1993). Using artificial neural nets to model implicit and explicit memory
test performance. In R Graf & M . E. J. Masson (Eds.), Implicit memory: New directions in cognition,
devebpment, and neuropsychobgy (pp. 141-165). Hillsdale, NJ: Lawrence Erlbaum Associates.
Wilson, T. D., & Hodges, S. D. (1992). Attitudes as temporary constructions. In L. L. Martin & A . Tesser
(Eds.), The construction of social judgments (pp. 37-65). Hillsdale, NJ: Lawrence Erlbaum Associates.
Wyer, R. S., & Srull, T K. (1989). Memory and cognition in its social context. Hillsdale, NJ: Lawrence
Erlbaum Associates.
Zajonc, R. B. (1980). Feeling and thinking: Preferences need no inferences. American Psychobgist, 35,
151-175.
Chapter 13
The Vicissitudes of Social Behavior

and Mental Life
Thomas K. Srull
University of Illinois at Urbana-Champaign
If is often said, or at least it has often been said to me, that social psychology has
been cognitive from the very beginning—cognitive i n its perspective, its subject
matter, its methodological orientation, and i n the sense that its dominant para-
digms have been experimental and designed to examine (primarily) mediating
cognitive processes i n social situations. A s is the case with any general statement
of this type, one could claim the statement is too grand, argue that it is an
oversimplification, or raise a variety of picayune points to challenge the statement
at the level of fine detail and analysis. I resist the temptation to develop an
"on-one-hand-but-on-the-other-hand" type of exposition here, for it would surely
take us too far afield.
Suffice it to say that, although such a statement is certainly an oversimplification,
it is still, i n essence, an accurate historical characterization of the field. T h e earliest
theoretical attempts to understand social facilitation (Triplet, 1897, 1898), social
loafing (Ringelman, 1913), group stereotyping (Zawadski, 1948), cooperation and
conflict (Deutsch & Collins, 1951), and several other phenomena often included
cognitive concepts that were invoked to explain the various empirical effects that
were observed. H o v l a n d and his colleagues from the Yale school of communication
and persuasion developed an approach that was, i n many respects, a precursor to
the general information-processing models that became prominent years later (see
e.g., H o v l a n d , Janis, ck Kelley, 1953; M c G u i r e , 1968, 1969).
It is interesting, however, that, despite this emphasis on cognition i n general, the
precise role of consciousness was rarely addressed. Even when social cognition came
to the forefront as a dominant and reasonably well-articulated metatheoretical
approach to social psychology, the specific role of consciousness always seemed to
be left for another day. For example, despite enormous empirical literatures on such
topics as self-monitoring (Snyder, 1987), self-consciousness (Buss, 1980), and
203
204 Srull
self-regulation (Bandura, 1986), the role of consciousness is rarely articulated,

which processes are assumed to be conscious or unconscious is rarely specified, and
so on.
Bargh emphasizes these points in a way that, I hope, will prod all social cognition
theorists to think about the precise role of consciousness more carefully. He is
certainly correct in suggesting that previous claims (and implicit assumptions)
about the role of mediating conscious cognitive processes have been accepted
without a great deal of criticism, and, in most cases, have never been pursued at
any level of detail. Although there are many individual research reports, and even
individual theories, in which articulating the precise role of consciousness was left
for another day, collectively, we never seemed to get around to it (for a thoughtful,
but very recent exception to this rule, see Greenwald & Banaji, 1995).
In contrast, I have been struck by the energy, subtlety, and exquisite level of
sophistication among researchers in artificial intelligence on just these issues (see,
e.g., the recent debate on the ontological nature of consciousness between Searle,
1995a, 1995b and Dennett, 1994, 1995). In fact, one recent review (Searle, 1995a)
discussed six entire books in artificial intelligence that are devoted to analyzing such
questions as how a particular conscious state arises, how a unique conscious state
can be identified, how neurobiological processes can give rise to particular psycho-
logical states, how conscious and unconscious processes interact, and myriad of
other, closely related issues (Crick, 1994; Dennett, 1994; Edelman, 1992, 1994;
Penrose, 1995; Rosenfield, 1994). It is ironic, although probably not surprising to
Bargh, that philosophers and neurobiologists have been more active participants in
these debates than those with traditional backgrounds in psychology. To the extent
that Bargh's arguments stimulate psychologists in general (and social cognition
theorists in particular) to pay closer attention to these issues, he will have accom-
plished something quite important.
If only parenthetically, I also point out that social cognition theorists have been
almost as silent on the topic of actual behavior. Although there are studies that
have measured a specific behavioral response that was theoretically derived, it is
fair to say that such studies are in a small minority. Also, social cognition theories
that specifically address how a psychological process (or psychological state) is
translated into a specific behavioral response are almost nonexistent. The fact that
Bargh implicitly calls our attention to this issue as well is another very useful
characteristic of his treatise.
APPRECIATING PROGRESS AND KEEPING PERSPECTIVE
I believe that several of Bargh's specific arguments are incomplete and, if I am

reading him correctly, others are misleading in what I consider to be important
respects. Therefore, I spend the bulk of my time attempting to offer some critical,
but constructive comments on his general themes. However, I think it is important
to note at the outset that these are minor criticisms that are made within the context
13. The Vicissitudes of Social Behavior and Mental Life 205
of a very impressive intellectual effort. If psychology in general is a very young discipline,

social cognition is still in its early infancy. There is a great deal we don't know, a subset
of which we don't even know how to study. It is doubtless necessary to remain constantly
vigilant and critical. However, it is also useful to appreciate the progress that is being
made, maintain an appropriate historical perspective, and develop a healthy respect for
past (obviously flawed and incomplete) theoretical efforts. I wish that Bargh had been
a bit more understanding of some of his progenitors.
Being an academic, and therefore one who is also more at ease challenging than
praising, I too dwell on the former. But Bargh's chapter has several strengths that
are so unusual and important that I think they are worth acknowledging explicitly.
First, both the theory and the exposition are unusually provocative, and I mean
this in the best sense of the word. Bargh has addressed several extremely compli-
cated issues and the scope of his analysis is impressively large. Moreover, he
consistently presents clear, strong arguments that are remarkably free of ambiguity
and hedging. Although this means that he is almost certain to be challenged on
matters both large and small, it also means that he has presented us with a clear,
complete, and bold theoretical statement. Bargh has never taken the easy way out
and should be applauded for his skill, thoroughness, and tenacity.
Bargh's model is not only broad, but also deep and integrative. Such integrative
models are not only extraordinarily difficult to construct, but are also extremely
important for further and more refined theoretical advancement. Berkowitz and
Devine (1989) made a related observation in what I consider to be one of the most
overlooked papers in the 10 years. They suggested that there is a fundamental
tension in psychology, generally (and in social cognition particularly) between the
analytic and synthetic approaches to theory development. Researchers working
within the analytic tradition attempt to more carefully circumscribe or delimit the
conditions under which any given effect will occur (or, to use a slightly different
language, under which any given psychological process will operate). In contrast,
those working within the synthetic tradition seek "to bring together apparently
disparate observations under a common theoretical umbrella. Rather than delim-
iting and differentiating, synthesists seek to generalize by taking a theoretical
proposition and seeing how far they can appropriately run with it" (Berkowitz 6k
Devine, 1989, p. 497).
There is no doubt that the current culture in psychology (and certainly in social
cognition) favors analytic, as opposed to synthetic, approaches. There is a very
legitimate argument, however, about whether this state of affairs is most propaedeu-
tic to a more complete understanding of human behavior and experience. I do not
believe it is. I think, at least in social cognition, we need more of a balance, and
Bargh has contributed to providing it in an important way.
Although Bargh appeals to the brilliance of Skinner, let me appeal to the
brilliance of Einstein. As I noted elsewhere (Srull, 1993), in addition to being a great
theorist, Albert Einstein was an excellent philosopher of science, and one from
whom psychologists can learn a great deal. Einstein once said that the whole
purpose of science is to discover unity in diversity—to find the underlying common-
alities in phenomena that appear on the surface to be quite different. The laws of
206 Srull
gravity and thermodynamics, for example, are elegant because of their simplicity
and their ability to explain so much with so little.
I believe it is safe to assume (although Bargh may quarrel with me about this)
that, compared to physical phenomena, psychological phenomena are much more
overdetermined, even under the most controlled laboratory conditions possible.
One implication of this is that we cannot expect psychological laws to be nearly as
clean or simple or unconstrained as physical laws. Still, it seems that our primary
objective should be to find (psychological) unity in (psychological) diversity. Bargh
has attempted this in a domain that is both difficult to study and crucial to
understand, and I don't think we can do anything other than admire his effort.
I believe that Bargh has made a number of more molecular contributions as well.
The emphasis he gives to inhibitory processes is long overdue (certainly in social
cognition), his point that goals and behavioral responses should be thought of in
terms of their corresponding mental representations will become increasingly
influential, his argument that explicit evidence should be required for claims of
conscious mediation is well taken, and his statement that "it is easy to fall into the
trap of thinking that the only effect that an experimental manipulation is having is
the one that is being measured" (chap.l, p. 48) deserves to become something of a
social cognition mantra.
In short, Bargh has made considerable and important progress along a number
of dimensions. As friends, relatives, editors, and colleagues constantly remind me,
however, there is a big difference between making progress and finishing the job.
Therefore, with a view toward what will be required to finish the job, I turn my
attention to several salient and (in my opinion) confusing issues. Ironically, I believe
that Bargh has understated his case in several respects, and overstated it in others.
THE CENTRALITY OF GENERALIZATION:

PERCEIVING SAMENESS IN WHAT IS DIFFERENT
Cognitive psychologists have devoted considerable time and effort to understanding

various aspects of generalization and categorization (see e.g., Barsalou, 1992).
Unfortunately, however, this has generally not been true within social cognition
proper. More specifically, although Bargh's model touches on a number of ways in
which generalization is important, it is never articulated fully and the precise role
it plays is left unaddressed. I believe this is the single most fruitful avenue for future
research and theoretical analysis. In other words, although I am not challenging
(for the moment) any of Bargh's specific claims, I do think that this is one aspect
of the model—and by far the most important aspect of the model—that is under-
stated.
There is an old saying to the effect of "people never walk in the same river twice."
I assume that this highlights the fact that, over time, the environment changes, as
do the individuals involved. The objective environment is continually being modi-
fied and, even if it were not, one's subjective experience of it would constantly
change (if only slightly) as a function of development and new experience.
Bringing this idea a little closer to home, the same is true of the social environ-
ment. Two social situations are never identical—they are always different. Still, we
know that people form equivalence classes (in the social domain, as in any other),
and they treat dissimilar situations as functionally equivalent. How they navigate
this balance in perceiving sameness in difference is very important to understand.
At the very beginning of this chapter, Bargh characterizes research on automatic
cognitive processes as "the search for specifiable if-then relations between situations
and psychological effects (chap. 1, p. 3)." He uses, as one classic example, the case
of helping behavior. Our behavior will be directed toward providing aid if the other
person needs our help and if the other is attractive and if we are the only person
around, but not if there are others in the vicinity who could help. But in constructing
a mental model of the situation, it is not immediately clear how we utilize our
categories of "needs help" or "is attractive" or "in the vicinity." The specifics will
always change from one situation to another, and how do we generalize our past
experiences? A n analogous thing occurs on the behavioral end. The "then provide
aid" category leaves us at a very high level of generality. Once again, the specifics
will always be unique to any particular situation.
In his initial analysis of preconscious processes, Bargh cites Lord Whitehead's
claim and, following the logic of Shiffrin and Dumais (1981), points out that
preconscious processes develop "out of one's frequent and consistent mental,
emotional, motivational, and behavioral reactions to a given set of environmental
features" (chap. 1, p. 10, italics added). However, the environmental features
associated with seeing letters on a computer screen (see Shiffrin 6k Dumais, 1981)
are going to be very constrained relative to a social setting such as an emergency
situation. Also, when Latane and Rodin (1969) found that subjects helped (or did
not help) a woman who fell off a ladder, I doubt whether their reactions represented
ones that were frequent and consistent. The same argument pertains to any
bystander intervention study.
Bargh then points out that, although reactions to environmental stimuli are
initially effortful and require conscious attention, over time, the requirement of
conscious attention diminishes "given that the same categories or evaluations or
goals are always selected in response to those features." (chap. 1, p. 10). Although
I am not quarreling with his conclusion, I do think that how such equivalence classes
are formed, modified, and utilized is a very important, yet poorly understood issue.
Indeed, I believe this is a question that is ripe to be explored in a variety of ways in
future research. As I alluded to earlier, the same could be said about many areas of
social cognition.
One can expect that the processes involved in generalization will prove to be
elusive and exceedingly complex (see e.g., Luger, 1994). It is probably best to
conceptualize generalization as being composed of (at least) two dimensions. One
can be thought of as a vertical dimension (which might be referred to as abstraction),
and one can be thought of as a horizontal dimension (which might be referred to
as inclusiveness). Consider, for example, the perception of an acquaintance as
"aggressive." He or she might physically assault you, push you backward but not hit
you, place his or her hands on your shoulders forcefully but not push you, poke his
208 Srull
or her finger in your chest to emphasize his or her disagreeableness, call you nasty
names, yell and argue loudly, or quietly mope around and refuse to speak to you.
AH of these things could be considered aggressive, but at very different levels of
intensity and abstraction.
Inclusiveness pertains to the range of behaviors that could be included within a
level of abstraction. How forceful does the placing of hands on the shoulders need
to be, how loud does the argument need to become, and how nasty does the name
need to be to be included within the category of "aggressive?" It is important to keep
in mind that, within the social domain, generalization will often—probably most
often—occur along interpretive dimensions rather than physical ones.
I should point out, in the context of fairness, that Bargh hints at these issues at
several points throughout his chapter. Most notably, he discusses the importance of
the psychobgical situation and cites the work of Koffka (1925), Lewin (1935), and
Mischel (1973). Although these conceptualizations may be a good starting point
for a detailed analysis of generalization, I do not believe they come anywhere near
to what will be required for a complete and well-articulated conceptualization of
social cognition.
Bargh also acknowledges the importance of generalization peripherally in the
context of goals (see his footnote 1, chap. 1, p. 8). I am prone to agree with him
when he argues later that "if an individual nearly always pursues the same goal within
a given situation, that goal will come eventually to be preconsciously activated within
that situation, independently of the individual's conscious purposes at that later
time" (chap. 1, p. 30, italics added). I would quickly add, however, that this will
only be true in a nontrivial sense, if both the goal and the situation are defined at
a very abstract (and therefore deeply subjective) level.
THE NEED FOR RESTRAINT: PERCEIVING DIFFERENCE

IN WHAT APPEARS TO BE THE SAME
A l l of my comments about generalization thus far pertain to aspects of the model

that are understated and in need of further development. Before leaving the general
issue, however, I should mention one aspect of Bargh's analysis that I believe is, at
best, premature and, at worst, dramatically overstated. In this case, the issue is not
one of generalization on the part of the subject, but generalization on the part of
the theorist.
Bargh addresses toward the end of his chapter the question of whether social
information processing goals can be activated preconsciously and subsequently
pursued nonconsciously. To demonstrate this, he argues, the existence and role of
conscious intent must first be eliminated. He then presents the results of two
experiments that have recently been reported by Chartrand and Bargh (1995).
The first experiment attempted to replicate a well-known study by Hamilton,
Katz, and Leirer (1980). Subjects in their study read a series of concrete behaviors
in random order that fell into four general categories: religious (e.g., set his alarm
Saturday night so that he would be sure to make it to church), intelligent (e.g., was
able to solve the puzzle quickly), athletic (e.g., played racquetball after work), and
sociable (e.g., had a party for some friends last week). Before receiving the informa-
tion, some subjects were told that they were participating in a memory experiment
and their task was to remember as much of the information as possible. Other
subjects were told that they were participating in an experiment on impression
formation, and their task was to form a coherent impression of the target person.
Hamilton et al. (1980) found that the latter group recalled more of the informa-
tion, despite of the fact that the memory test was unexpected, and showed more
category clustering by trait (Roenker, Thompson, & Brown, 1971) in their recall
protocols.
Chartrand and Bargh (1995) replicated this study but they eliminated the
conscious-processing goal. Instead of receiving "memory set" or "impression set"
instructions, subjects received a scrambled sentence test modeled after that devel-
oped by Srull and Wyer (1979, 1980). Embedded within the scrambled sentence
test were several key words. Specifically, although some subjects were exposed to
words relevant to forming an impression (e.g., opinion, personality, and evaluate),
others were exposed to words relevant to memory (e.g., absorb, retain, and remem-
ber).
As Bargh reports, "our results replicated those of Hamilton et al. (1980) exactly.
That is, participants whose impressions formation goal had been primed recalled
significantly more of the behaviors than did participants in the memorization
condition. Moreover, their recall protocols showed significantly higher clustering
according to trait category" (chap. 1, p. 33). For now I simply raise the question of
whether it is fair to say that an impression formation goal had been primed in the
Chartrand and Bargh procedure. Also, what concrete evidence is there that
exposure to words like "opinion" or "evaluate" leads to a specific objective to form
an impression of another person? The same question, obviously, can be raised with
respect to the other condition as well.
A second experiment by Chartrand and Bargh (1995) capitalized on previous
work using a general person memory paradigm. For our purposes, three particular
findings are relevant. First, previous studies found that subjects given an impression
formation objective form online impressions and make subsequent impression
judgments that are not mediated by the specific items they are able to recall (Bargh
6k Thein, 1985; Hastie 6k Park, 1986; Lichtenstein & Srull, 1987). Second, subjects
given an impression formation objective show greater differentiation in their
judgments of targets who differ on specific trait dimensions (Bargh & Thein, 1985;
Srull & Wyer, 1989). Finally, subjects given an impression formation objective show
a marked tendency to recall behaviors that are incongruent with the general
impression than behaviors that are congruent or irrelevant to the impression (Srull,
1981; Srull, Lichtenstein, & Rothbart, 1985; Srull & Wyer, 1989).
Chartrand and Bargh conducted a basic person memory experiment but they did
not include a conscious impression formation goal. Rather, some subjects were
subliminally primed with impression related stimuli, using a procedure modeled
after Bargh and Pietromonaco (1982). However, the data from these subjects
showed all three of the effects previously noted.
210 Srull
I find these studies to be absolutely fascinating. They are brilliant in their

simplicity, creative in their execution, and powerful in their results. I wonder,
however, whether Bargh is overstating his case in terms of using them to draw strong
and premature theoretical conclusions. In the original Hamilton et al. (1980) study,
for example, subjects in the impression formation condition presumably showed
more category clustering because they were mentally reorganizing the information
along trait dimensions (relating one sociable behavior to another, one religious
behavior to another, and so on). But they were doing this as a means to an end. It
makes sense to engage in such activity if one's goal is to form a coherent impression
of the target. There is a purpose to the act.
It is not obvious to me, however, that Chartrand and Bargh primed a goal in the
sense of a purpose to the act or a means to an end. This is certainly one possibility,
but I much prefer to leave it as a possibility—or open question—rather than an
established fact.
Bargh uses these studies to draw very strong theoretical conclusions. I worry
about this particularly because, in any relatively unexplored domain such as the
present one, overstating the case is, in the long run, much more problematic than
understating it. It is roughly analogous to making a Type I or Type II error.
Understating the case is an easy mistake to rectify. One can always go back and fill
in the gaps, better articulate the argument at a finer level of detail, disambiguate
what is conceptually confusing, and so on. But overstating the case, or prematurely
committing oneself to a theoretical conclusion that turns out not to be valid, is
much more difficult to overcome. In my opinion, it is a much better strategy to go
slowly, remain skeptical, and eliminate every plausible alternative interpretation
before drawing such a strong theoretical inference. Remember the mantra.
At any rate, this is one of the few components of the model in which I do not
feel comfortable drawing such a strong inference and, for the moment at least, I
would prefer to withhold judgment in terms of any specific theoretical commitment.
There is a flip side to this decision. The reward for delaying gratification will be an
absolutely fascinating area in which to continue working.
RATIONAL AND IRRATIONAL CONCERNS

OVER ECOLOGICAL VALIDITY
Conceptual issues related to ecological validity run throughout Bargh's analysis of

automaticity. Although he occasionally addresses them implicitly, he never does so
directly. Unfortunately, space limitations permit only a brief discussion of these
issues. Nevertheless, there are several points concerning ecological validity that are
worth developing.
Research in social cognition is often criticized because, it is argued, it has limited
ecological validity with respect to actual social interaction. Most of the criticism
revolves around the fact that the type of laboratory tasks used share no topographi-
cal similarity to anything found in real social life. Such criticisms are sophomoric
and, although Bargh addresses them only indirectly, it is clear that he recognizes
the fallacy of this argument more than most. In fact, the ecological validity of any
investigation must be judged in terms of the psychological processes that are
activated, not in terms of how superficially similar one act (or stimulus) is to another.
McGuire (1973, 1983) discussed how experiments gain their power—and their
beauty—from the fact that, in very fundamental ways, they are artificial. The entire
logic of them is to use their artificiality to gain subtlety in observation and
diagnosticity in understanding. Even though he does so only implicitly, Bargh does
an excellent job of demonstrating the relevance of his tasks to ecologically repre-
sentative social psychological processes.
There is another aspect of ecological validity, however, that Bargh does not
comment on, even indirectly, and I fear that he may leave a false impression.
Ironically, it concerns what was a criticism of Skinnerian psychology as well. To state
it as simply as possible, just because a particular (behavioral) phenomenon can be
shown in the laboratory to be produced by a particular (psychological) process, it
does not mean that, when one observes the phenomenon in a natural setting, the
genesis of the behavior lies in the same psychological process investigated.
For example, it is clear that an eyeblink can be produced through operant
conditioning. That does not mean, however, that when we observe an eyeblink in
the grocery store, it is fruitful (or fair) to start hypothesizing about the person's prior
learning history. Similarly, careful demonstrations of automatic processes in the
laboratory do not, in any way, mean that they dominate in everyday life. Because
the mental apparatus evolved for a particular set of purposes, it would be surprising
if they didn't play some role in the natural ecology, but specifying the parameters of
that role is an entirely different matter.
What Bargh has demonstrated quite elegantly, and in case after case, is that
nonconscious processes are capable of producing identifiable behavioral and emo-
tional phenomena, not that they necessarily do. I may walk slowly because a well-
articulated stereotype of the elderly has just been activated, but I may also walk
slowly because of a conscious, deliberative decision that throwing down my pen,
leaving my desk, and getting some air will help me gain a little perspective on life.
Similarly, I may like "surly J. B." because I've just been thinking about a tenacious
newspaper reporter, but I may also like him because of his New York swagger and
the unique combination of confidence and erudition he exudes.
I would be surprised if Bargh disagreed with any of this. After all, no one would
suggest that conscious processes never play a role in discrimination or other forms
of prejudicial (or altruistic, aggressive, etc.) behavior. In short, issues related to
ecological validity are more complex than they sometimes seem. I believe that Bargh
has met one challenge brilliantly, but that there is a much more difficult issue lurking
in the background.
SEPARATING PRINCIPLE, PRACTICE, AND PROMISE
There is a danger in making statements about complex issues that are short, cryptic,
and undeveloped. I'm afraid that Bargh fell victim to this danger and, because of
212 Srull
space limitations, I'm sure that I will as well. Unfortunately, what we will be left
with is two people doing little more than teasing the reader.
The gist of my argument is this: most of Bargh's introductory comments about
social psychology, experimental psychology, and psychology in general are, at best,
exaggerations and oversimplifications. Like most provocative statements, they
contain a kernel of truth, but they also provide the foundation for fundamental
misunderstanding.
Let me begin with his characterization of social psychology. It is true that social
psychology has always been very experimental in nature. Situational variables of
one type or another are manipulated and their effects on behavior (broadly
conceived) are assessed. In this sense, social psychology has always been concerned
with discovering various types of if-then relations.
However, whereas all of this is true in a descriptive sense, it is not true—and it
never been true—in a conceptual sense. When Bargh claims that "social psychol-
ogy's natural focus [is] on the situational determinants of thinking, feeling, and
doing" (chap. 1, p. 1) he is, I am convinced, leaving a misleading impression. His
argument is a noteworthy oversimplification in a descriptive or methodological
sense, but an extraordinary oversimplification in a theoretical sense. For one thing,
at least since the time of Lewin (1938), Bruner (1957), and Heider (1958), it has
been well-understood that trying to separate out the situation from the perceiver
is inherently artificial. Even more important, however, social psychologists were
always concerned with such issues as interpersonal communication, various as-
pects of social interaction, and changes in behavior (and opinions, attitudes,
beliefs, etc.) over time. In addition, they always recognized the dynamic elements
of social psychological processes that generalize over time and situations, the roles
of constructing one's own environment, bidirectional causality in social interac-
tion, feedback loops, and so on. Also, although it is true that the general
experimental paradigm treats individual differences as error variance, social psy-
chologists always showed great concern with individual-level variables (or proc-
esses) that have transsituational effects or interact with specific situational
parameters.
I think that Bargh has exaggerated these points in order to give strength to his
argument and, as I've mentioned, there is some truth to what he says. It is an
incomplete truth, however, and although his points may be representative in terms
of general method, they are also misleading with respect to what have always been
the central conceptual concerns of social psychology.
THE ROLE OF THE PAST IN THE PRESENT
Much of the same can be said with respect to Bargh's description of the historical
development of experimental psychology. Once again, he makes his points with
truths that are accurate but incomplete and, from a historical perspective, poten-
tially misleading.
Although his comments about the serial stage model are well-taken, for
example, I do not believe anyone ever took the postulation of discrete stages as
strongly as Bargh suggests. The "control processes" of Atkinson and Shiffrin
(1968), for example, were an explicit recognition of the fact that the model
could not be strictly serial. In a broader sense, when I review the stage model of
Atkinson and Shiffrin (1968), the working memory model of Baddeley and
Hitch (1974), the production model of Anderson (1976), the PDP model of
Rumelhart and McClelland (1986), and overlay them with critical papers by
people like Posner and Snyder (1975), Shiffrin and Schneider (1977), Logan
(1988), and others, I see a very logical progression of conceptual understanding.
Assuming the position of one more impatient than I, it may have been slow, but
it was certainly steady and, as Bargh has so elegantly demonstrated, this is a hard
business.
Finally, I would like to address Bargh's more general comments on conscious-
ness. He writes in a very provocative statement, for example, that "Early
cognitive models, in other words, equated cognition with conscious cognition
(see Bowers, 1981; Lazarus, 1982), and we have been cleaning up after this
misconception ever since" (chap. 1, p. 50). Although I understand his point
about the debate over perceptual defense, as a general claim, his argument is
vastly exaggerated. Twenty years ago, for example, Lachman, Lachman, and
Butterfield (1979) reviewed the precursors to the development of cognitive
psychology and stated without hesitation that most of what we do goes on
unconsciously. They indicate that it is the exception, not the rule, when
thinking is conscious. They also point out that unconscious processing is
phylogenetically prior and constitutes the product of millions of years of evolu-
tion, whereas conscious processing is in its evolutionary infancy. I believe once
again that, although there is some truth in Bargh's statement, it is an incomplete
truth, and one that can easily be misconstrued.
A DENOUEMENT IS WISHFUL THINKING
I mentioned earlier that, in my opinion, Bargh could have been a bit more
understanding of his progenitors. It should be clear from my comments about social
and experimental psychology why I believe this is true. I also mentioned that Bargh
has produced a very important and perceptive intellectual effort. He deserves our
praise and admiration. We should all remember, however, that scientific progress is
cumulative, evolutionary, and usually made in the trenches. Bargh has stood on the
intellectual shoulders of many who have gone before, and instead of thinking of the
present treatise on the automaticity of everyday life as a denouement, we should
think of it as what it is—a fair and critical analysis of past research, a cogent
statement of our current understanding, a piquant theoretical analysis, and a
stimulant for a further, and even more refined, analysis of the intricacies and
vicissitudes of human social behavior and mental life.
214 Srull
REFERENCES
Anderson, J. R. (1976). Language, memory, and thought. Hilllsdale, NJ: Lawrence Erlbaum Associates.
Atkinson, R. C , & Shiffrin, R. M . (1968). Human memory: A proposed system and its control processes.
In K. W. Spence &. J. T. Spence (Eds.), Advances in the psychology of learning and motivation (Vol. 1).
Baddeley, A . D., & Hitch, G . (1974). Working memory. In G . Bower (Ed.), Recent advances in learning
and motivation (Vol. 8). New York: Academic Press.
Bandura, A . (1986). Social foundations of thought and action: A social cognitive theory. Englewood Cliffs,
NJ: Prentice-Hall.
Bargh, J. A . , & Pietromonaco, R (1982). Automatic information processing and social perception: The
Bargh, J. A . , & Thein, R. D. (1985). Individual construct accessibility, person memory, and the
1129-1146.
Barsalou, L. W (1992). Cognitive psychology: An overview for cognitive scientists. Hillsdale, NJ: Lawrence
Erlbaum Associates.
Berkowitz, L., & Devine, R G . (1989). Research traditions, analysis, and synthesis in social psychological
theories: The case of dissonance theory. Personality and Social Psychology Bulletin, 15, 493-507.
Bowers, K. S. (1981). Knowing more than we can say leads to saying more than we can know: O n being
implicitly informed. In D. Magnusson (Ed.), Toward a psychology of situations: An interactional
perspective. Hillsdale, NJ: Lawrence Erlbaum Associates.
Bruner, J. S. (1957). O n perceptual readiness. Psychological Review, 64, 123-152.
Buss, A . H . (1980). Self-consciousness and social anxiety. San Francisco: Freeman.
Chartrand, T L., & Bargh, J. A . (1995). Automatic activation of impression and memorization goals:
Nonconscious goal priming reproduces effects of explicit task instructions. Unpublished manuscript, New
York University.
Crick, F. (1994). The astonishing hypothesis: The scientific search for the soul. New York: Simon & Schuster.
Dennett, D. C . (1994). Consciousness explained. New York: Little, Brown.
Dennett, D. C . (1995). The mystery of consciousness: A n exchange. The New York Review, December
21.
Deutsch, M . , & Collins, M . E. (1951). Interracial housing: A psychological evaluation of a social experiment.
Minneapolis: University of Minnesota Press.
Edelman, G . (1992). The remembered present: A biological theory of consciousness. New York: Basic Books.
Edelman, G . (1994). Bright air, brilliant fire: On the matter of the mind. New York: Basic Books.
stereotypes. Psychological Review, 102, 4-27.
Hamilton, D. L., Katz, L. B., & Leirer, V O . (1980). Organizational processes in impression formation.
In R. Hastie, T M . Ostrom, E. B. Ebbesen, R. S. Wyer, D. L. Hamilton, & D. E. Carlston (Eds.), Person
memory: The cognitive basis of social perception (pp. 121-154). Hillsdale, NJ: Lawrence Erlbaum
Associates.
Hastie, R., & Park, B. (1986). The relationship between memory and judgment depends on whether
the judgment task is memory-based or on-line. Psychological Review, 93, 258-268.
Heider, F. (1958). The psychology of interpersonal relations. New York: Wiley.
Hovland, C . I., Janis, I. L., & Kelley, H . H . (1953). Communication and persuasion: Psychological studies
of opinion change. New Haven, C T Yale University Press.
Koffka, K. (1925). Die Grundlagen der psychischen Entwicklung. Osterwieck: Zickfeldt.
Lachman, R., Lachman, J. L . , & Butterfield, E. C . (1979). Cognitive psychology and information processing:
An introduction. Hillsdale, NJ: Lawrence Erlbaum Associates.
Latane, B., & Rodin, J. (1969). A lady in distress: Inhibiting effects of friends and strangers on bystander
intervention. Journal of Experimental Social Psychobgy, 5, 189-202.
Lazarus, R. S. (1982). Thoughts on the relations between emotion and cognition. American Psychobgist,
37, 1019-1024.
Lewin, K. (1935). A dynamic theory of personality. New York: McGraw-Hill.
Lewin, K. (1938). The conceptual representation and the measurement of psychobgical forces. Durham, N C :
Duke University Press.
Lichtenstein, M . , & Srull, T. K. (1987). Processing objectives as a determinant of the relationship
between recall and judgment. Journal of Experimental Social Psychobgy, 23, 93-118.
Logan, G . D. (1988). Toward an instance theory of automatization. Psychobgical Review, 95, 492-527.
Luger, G . F. (1994). Cognitive science: The science of intelligent systems. San Diego: Academic Press.
McGuire, W. J. (1968). Personality and attitude change: A n information-processing theory. In A . G .
Greenwald, T. C . Brock, & T M . Ostrom (Eds), Psychobgical foundations of attitudes (pp. 171-196).
San Diego: Academic Press.
McGuire, W. J. (1969). T h e nature of attitudes and attitude change. In G . Lindzey & E. Aronson (Eds.),
Handbook of social psychobgy (2nd ed., Vol. 3, pp. 136-311). Reading, M A : Addison-Wesley.
McGuire, W. J. (1973). T h e yin and yang of progress in social psychology: Seven koan. Journal of
McGuire, W. J. (1983). A contextualist theory of knowledge: Its implications for innovation and reform
in psychological research. In L. Berkowitz (Ed.), Advances in experimental social psychobgy (Vol. 16).
Mischel, W. (1973). Toward a cognitive social learning reconceptualization of personality. Psychobgical
Review, 80, 252-283.
Penrose, R. (1995). Shadows of the mind: A search for the missing science of consciousness. New York: Oxford
University Press.
Posner, M . I., & Snyder, C . R. (1975). Attention and cognitive control. In R. L. Solso (Ed.), Information
processing and cognition: The Loyola symposium. Hillsdale, NJ: Lawrence Erlbaum Associates.
Ringelmann, M . (1913). Recherches sur les moteurs animes: Travail de l'homme. Annales de I'lnstitut
National Agronomique, 2e serie,tomXII, 1-40.
Roenker, D. L., Thompson, C . P., & Brown, S. C . (1971). Comparison of measures for the estimation of
clustering in free recall. Psychobgical Bulletin, 76, 45-48.
Rosenfield, I. (1994). The strange, familiar, and forgotten: An anatomy of consciousness. New York: Vintage
Press.
Rumelhart, D. E., & McClelland, J. L. (1986). Parellel distributed processing: Expbrations in the microsome-
ture of cognition. Cambridge, M A : M I T Press.
Searle, J. R. (1995a). T h e mystery of consciousness. The New York Review, November 2.
Searle, J. R. (1995b). The mystery of consciousness: Part II. The New York Review, November 16.
Shiffrin, R. M . , & Dumais, S. T (1981). T h e development of automatism. In J. R. Anderson (Ed.),
Cognitive skills and their acquisition. Hillsdale, NJ: Lawrence Erlbaum Associates.
Shiffrin, R. M . , & Schneider, W. (1977). Controlled and automatic human information processing: II.
Snyder, M . (1987). Public appearances/private realities: The psychobgy of self-monitoring. San Francisco:
Freeman.
Srull, T. K. (1981). Person memory: Some tests of associative storage and retrieval models. Journal of
Experimental Psychobgy: Human Learning and Memory, 7, 440-463.
Srull, T K. (1993). Identifying the causes, correlates, and consequences of concept activation: Looking
back and looking ahead. Psychobgical Inquiry, 2, 204-217.
Srull, T. K., Liechtenstein, M . , & Rothbart, M . (1985). Associative storage and retrieval processes in
person memory. Journal of Experimental Psychobgy: Learning Memory, and Cognition, 11,316-345.
Srull, T K . . & Wyer, R. S. (1979). T h e role of category accessibility in the interpretation of information
about persons: Some determinants and implications. Journal of Personality and Social Psychobgy, 37,
1660-1672.
216 Srull
Srull, T. K., & Wyer, R. S. (1980). Category accessibility and social perception: Some implications for
the study of person memory and interpersonal judgment. Journal of Personality and Social Psychology,
38, 841-856.
Srull, T. K., & Wyer, R. S. (1989). Person memory and judgment. Psychological Review, 96, 58-83.
Triplett, N . (1897-1898). T h e dynamogenic factors in pacemaking and competition. American Journal
of Psychology, 9, 507-533.
Zawadski, B. (1948). Limitations of the scapegoat theory of prejudice. Journal of Abnormal and Social
Psychobgy, 43, 127-141.
Chapter 14
Automatic but Conscious:

That Is How We Act Most of the Time
Joseph Tzelgov
Ben Gurion University of the Negev
Since 1985 psychologists deepened their understanding of the automatic c o m -

ponents of h u m a n processing. Bargh (1989, 1992) contributed significantly to
this development, and his analysis of automaticity, i n particular, affected how I
and my colleagues think about this aspect of psychological processing (Tzelgov
& Yehene, 1994; Tzelgov, Yehene, & N a v e h - B e n j a m i n , i n press). Therefore, I
find it very easy to agree with his basic argument, "that m u c h of everyday
l i f e — t h i n k i n g , feeling, and doing—is automatic i n that it is driven by current
features of the environment (i.e., people, objects, behaviors of others, settings,
roles, norms, etc.) as mediated by automatic cognitive processing of those
features . . . " (chap. 1, p. 2).
Provocative as this argument sounds, i n my view it reflects a basic feature of
h u m a n behavior. I believe automatic processes are responsible for a significant
portion of our actions i n the social and cognitive domains.
Psychologists frequently do not distinguish between automaticity and the
absence of consciousness, even to the point of using the terms automatic and
unconscious interchangeably (e.g., K i h l s t r o m , 1987). A s an another example,
the two terms are frequently referred to as identical w i t h i n the framework of the
process dissociation procedure (Jacoby, 1991). Bargh's (1992) position is similar.
In contrast, I believe we are dealing with two different, although related,
concepts. In particular, I do not agree with the coupling of automatic processing
w i t h the absence of consciousness. S u c h coupling is unnecessary for the basic
argument and inconsistent with our current knowledge about automatic proc-
essing. A framework that conceives the relation between automaticity and
consciousness as different but related concepts is presented i n the second part
of this chapter.
217
218 Tzelgov
WHAT IS "AUTOMATIC"
Feature-List Definitions
T h e early definitions of automaticity (e.g., Hasher & Z a c k s , 1979; Posner & Snyder,
1975 ) were based on a list of features that a process should have (or, i n fact, lack)
in order to be defined as automatic. Accordingly, the absence of consciousness was
one of the three features, the other two being the absence of attention and the
absence of intentionality, which were common to all early definitions of automat-
icity. T h e assumption that this definition is true enabled cognitive psychologists to
put phenomena based o n differing psychological mechanisms—from preattentive
processing (Neisser, 1967) via well-practiced cognitive or perceptual-motor skill
(e.g., Newell & Rosenbloom, 1981) to encoding i n memory (Hasher & Zacks,
1979)—under a single theoretical umbrella, which was widened even further by
social psychologists (see Bargh, 1989). A s a by-product of such definitions, a
two-process framework of psychological processing evolved, with classification of
processes as automatic versus controlled (Schneider & Shiffrin, 1977).
Over the years, however, it became clear that, contrary to this contrast, (some)
automatic processes are controlled to some degree (see Logan, 1985, and N e u m -
man, 1984, for a theoretical analysis, and Logan & Zbrodoff, 1979, and Tzelgov,
H e n i k , 6k Leiser, 1990, for empirical examples). Several investigators (e.g., Bargh,
1992; Carr, 1992; Neumann, 1984) pointed out that the three features used to
define automaticity almost never hold simultaneously, which, i n turn, led some to
challenge the usefulness of the very concept of automaticity (Y. Tsal, personal
communication, September, 1995) and caused others to renew the quest for a valid
definition of automaticity.
A Minimalist Definition of Automaticity

O n e approach was to propose a definition that classifies as automatic only a subset
of the phenomena classified as automatic by the feature list approach. Logan's
(1989, 1992) construet-oriented 1 definition of automaticity i n terms of his (1988)
instance theory and Neumann's (1984) definition apply only to skill-based automat-
icity; that is, automaticity that is due to a learning process. A n alternative approach
led to a mimimalist definition of automaticity i n terms of a single feature that would
still allow one to keep the variety of automatic phenomena under a single umbrella.
Bargh (1989) pointed out that ballisticity (Logan & Cowan, 1984)—a feature of a
process, to run to completion once started, without the need of conscious monitor-
ing—is common to all automatic processes, and he proposed (1992) its adoption
as the definition of automaticity. Jacoby, Ste-Marie, and Toth (1993) proposed
defining a process as automatic if it produces the same effects whether they are i n
concert with, or i n opposition to, one's intention.
1
See Tzelgov and Yehene (1996) for a discussion of construct oriented approaches for defining
automaticity.
14. Automatic but Conscious 219
Similarly to Bargh (1992), Tzelgov, Yehene, and Naveh-Benjamin (in press)

defined a process as automatic if it runs without conscious monitoring. Such a
definition immediately raises the question, how do we know that a given psycho-
logical process runs without being monitored? To clarify this point, let me make a
distinction between two modes of automatic processing: Automatic processing is
intentional when the process is part of the task's requirements; it is autonomous when
the process is not part of the task's requirement. Most cognitive psychologists agree
that when a sentence is read for meaning by a skilled reader, the processing of
individual words is automatic (e.g., Jacoby, Levy, &Steibach, 1992). However, the
meaning of the sentence cannot be grasped if the individual words are not processed
at some level. In this case, the automatic processing of the words is intentional. O n
the other hand, the inability to ignore the meaning of the word while reporting its
color, as indicated by the Stroop (1935) effect is caused by autonomous automatic
processing of the color word. Tzelgov, Yehene, and Naveh-Benjamin (in press)
suggested using autonomous automatic processing as a criterion for automaticity:
A process is automatic if, and only if, it can be shown to act in the autonomous
mode. When the same process appears in the intentional mode, we infer that it is
automatic by assuming its activity under such conditions is identical with its activity
in the autonomous mode.
It is this criterion for automaticity I have in mind in this chapter. Let me start by
trying to apply it to social processes. The fact that people pick up facial expressions
when not required to (Murphy & Zajonc, 1993) is a case of autonomous automatic
processing. This serves as an indication that picking up facial expressions is
automatic. We, therefore, may assume that, when people involved in social inter-
action pick up facial expressions intentionally, this process is also automatic.
ARE WE CONSCIOUS OF "PRECONSCIOUS"

AUTOMATIC PROCESSING?
The processes Bargh discusses in the target chapter are classified as preconscious;
that is, they "require only the triggering proximal stimulus event and occur prior to
or in the absence of conscious awareness of that event" (chap. 1, p. 6). The
assumption that humans are not conscious of the stimuli that trigger "preconscious"
automatic processing is based on two premises: (a) preconscious processing is like
preattentive processing, and (b) preconscious automatic processing is elicited by
subliminal stimuli. In this section, I challenge both of these assumptions.
Are Preattentive and Preconscious Processes Alike?

Preattentive processes result in parsing the perceived space on the basis of elemen-
tary visual features and the relations among them, such as similarity or proximity
(Treisman, 1985). They precede object perception, which in most cases will require
spatial indexing, that cannot be done preattentively (Ullman, 1984); therefore they
precede consciousness "by definition." Bargh argues that the notion of preattentive
220 Tzelgov
processes can be extended to social situations and that, preconscious processes in

the domains of social cognition are similar, if not identical, to preattentive processes.
But are they?
The Place of Learning in Preattentive and Preconscious Processes. Preattentive

processes are innate and apparently insensitive to learning. Treisman, Vieira, and Hayes
(1992) showed that learning does not change the (attentive) processing of conjunctive
features into preattentive processing. Similarly, Logan (1992) showed that practice that
results in automatic retrievalfrommemory does not lead to priority learning, that is, it
does not increase the ability of the stimulus to attract attention automatically (Shiffrin
& Schneider 1977). Logan (1992) also pointed out that priority learning may be seen
as a possible mechanism for changing processing to preattentive. In contrast, precon-
scious automaticity is skill-based; that is, it is acquired in a learning process: "Through
frequent and consistent activation by the environment, social constructs . . . become
capable of being activated by the relevant stimulus information . . . " (Bargh, 1989, p.
12). Thus one difference between preattentive and preconscious processes is that the
former are innate and apparently cannot be acquired by learning, whereas the latter
result from a learning process.
Is the Stimulus Enough for Preconscious Processing?

Preattentive processing, being a bottom-up phenomenon, requires only an external
stimulus, but is that true for preconscious automaticity? Full control of behavior by
an external stimulus seems to be an exception rather than a rule in human behavior.
The only additional example of full control of behavior by an external stimulus I
can think of (except for preattentive processing) again comes from the perceptual
domain and deals with an innate process: exogenous orienting of attention (see
Rafal & Henik, 1994). By contrast, skill-based automaticity depends also on the
state of the organism rather than only on the external stimulus. Thus, Neumann
(1984) defined performance as automatic "whenever parameters for action are
specified by skill and the processing system" (p. 282). I wish to emphasize that even
the elicitation of conditioned reflexes depends, at least to some extent, on additional
factors, and not just on the C S 3 (see Locurto, 1981, for a review). Therefore, I find
it hard to accept Bargh's (1989) view that "preconscious processes require only the
triggering proximal stimulus" (p. 11). Recent findings of D'Agostino and Beegle
(1996) are consistent with my claim that the state of the organism is an important
factor in preconscious processing—so are the findings of Gilbert and Hixon (1991),
which show that cognitive busyness decreases the likelihood a particular stereotype
will be activated.
2
From this point on, I use the term preattentive to denote the set of the operations resulting in object
perception, and reserve the term preconscious for the processes discussed in the target chapter.
This point is important, because, as I suggest in the second part of this chapter, classical conditioning
may be one of the learning mechanisms responsible for the preconscious automaticity in the social
context.
Is Preconscious Automatic Processing

Based on Subliminal Perception of Stimuli?
M y frank answer to this question is simply, "I don't know." I am not familiar enough
with the research i n social cognition concerning subliminal perception research.
However, according to my knowledge of the cognitive literature, there is no
up-to-date, consistent body of knowledge supporting the notion of subliminal
perception, mainly because much of the research cited as supporting this notion is
open to alternative interpretations (Holender, 1986). Marcel's (1983) results show-
ing a Stroop effect (the clearest example of automaticity i n the domain of reading)
without word detection are extremely hard to replicate and many, including my
colleagues and me (Tzelgov, Porat, & Henik, 1994), failed to do so. In particular,
we showed that under conditions of short exposure durations, the Stroop effect is
constrained to trials i n which subjects correctly identified the word, and to subjects
who were able to identify the words above chance level. T h e argument is still going
on; Merikle and Reingold (1990), working within the S D T framework, found that
both words and nonwords were recognized for detected stimuli (i.e., i n case of hits),
but only words were recognized for undetected stimuli (i.e., i n case of misses) and
they suggested this supports the idea of an unconscious perception that is qualita-
tively different from a conscious one. Their interpretation was based o n the
assumption that stimulus detection provides an adequate index of consciousness.
This assumption, however, was challenged by Theios and Haase (1994), who
showed that identification without detection can be elegantly accounted for by
Macmillan's and Creelman's (1990) independent observations model without
assuming that detection corresponds to conscious awareness. In contrast, analyses
by Greenwald, Klinger, and Schuh (1995) point toward the possibility that sublimi-
nal perception may be a reliable, although weak phenomenon. But if we are dealing
with a relatively weak phenomenon, it is hard to see how it can be responsible for
a significant portion of our behavioral repertoire, as Bargh correctly proposes i n the
target chapter.
Are Stimuli in the Social Context

Processed Differently?
M y argument in the previous section was based on the assumption that stimuli in
social situations are analyzed similarly to stimuli in nonsocial contexts. T h a t is not
necessarily the case; i n social situations, emotional aspects of the situation may be
more important and dominate processing. It was suggested that the processing of
affective information requires minimal perceptual analysis (Zajonc, 1980; see also
Murphy, & Zajonc, 1993) and that processing of such information is faster ( O h m a n ,
Dimberg, & Esteves, 1989). Murphy and Zajonc (1993) reported affective priming
by facial expression of stimuli presented for 4 ms, and they reviewed neuroanatomi-
cal evidence i n favor of the independence of processing affective versus cognitive
222 Tzelgov
information, at least in the case of facial stimuli (see Murphy & Zajonc, 1993). They
provide convincing data that their subjects did not recognize the presented faces,
which may be interpreted as the faces not being consciously perceived. This may
be true. But it may be, given that emotional information is processed faster and
independently of cognitive information, that the very short exposure durations
could be enough for perceiving the emotional information without perceiving the
face. Thus, under such condition subjects are aware of the relevant stimulus—that
is, the emotional information. I develop this idea in a moment, but at this point, I
suggest that similar things may be happening in many social situations classified as
preconscious. As I discuss in what follows, the information relevant for automatic
processing is not perceived without awareness. Rather, its perception reflects a
different kind of awareness.
Does "Preconscious" Automatic Processing

Differ From Postconscious?
Taken together, the existing data do not support Bargh's assumption that precon-
scious automaticity is not based on the conscious perception of the stimulus. Such
a conclusion immediately raises the question of whether we should distinguish
between preconscious and postconscious automatic processing (Bargh, 1989,
1992). My answer to this question is positive: As emphasized by Bargh, postcon-
scious automatic processing is due to residual effects. A given postconscious
automatic process is primed by some other process that precedes it, in addition to
all the factors (internal and external) required for preconscious automaticity. Thus,
although the pre- as opposed to postconscious terminology may be misleading, the
distinction is important and useful. A n example of this distinction from the domain
of reading would be the difference between automatic reading of the prime and the
(automatic component of) the relatedness effect in the semantic priming paradigm
(see Neely, 1991, for a review). Reading words may be automatic and it apparently
happens in case of the prime in the semantic priming paradigm. My colleagues and
I (Friedrich, Henik, & Tzelgov, 1991) showed that this effect is independent of the
spread of activation within the semantic network. Although the processing of the
prime is preconscious according to Bargh's (1989, 1992) typology, the spread of
activation, as indicated by the shorter RTs when the prime and the target are related,
is postconscious, according to his typology.
ON THE RELATIONS BETWEEN AUTOMATICITY

AND CONSCIOUSNESS
In this section, I outline a framework for evaluating the relation between automat-
icity and consciousness. My approach to consciousness as presented here is based
on the mentalist view of consciousness developed by Dulany (1991,1996). Further-

more, although I do not dare to provide a formal defintion of consciousness, my use
of the term is similar to Bisiach's (1988) notion of "the access of parts or processes
of a system to its parts or processes" (p. 103), on the one hand, and to the notion
of phenomenogical awareness, as discussed by Airport (1988), on the other.
Are We Conscious of Psychological Processing?
Psychological processes can be either automatic or nonautomatic, but can they be

similarly classified as conscious or unconscious? My answer to this question is
negative (see also Dulany, 1991; Mandler, 1991; Velmans, 1991). I believe that we
are not more conscious of psychological processes than of any biological processes
responsible for our bodily functions. Psychological, in contrast to biological, proc-
esses, are characterized by the "stuff' they act on: information. We are not aware
of the processes that enable us to decode spoken language (Chomsky, 1980) or to
perceive the objects around us (Rock, 1983), and yet we are conscious of the world
around us. Taking as a starting point that psychological processing may be described
as applying procedures stored in terms of production rules such as those in
Anderson's (1983) A C T * theory, on factual information stored in declarative
memory in terms of propositions, Kihlstrom (1987) pointed out that, although "we
may be aware of the goals of and the conditions of procedures and the products of
their executions we are not aware of the operations themselves" (p. 1446). Thus,
automatic and nonautomatic processes are alike; we are not conscious of either of
them.
One could challenge this argument by asking, if this assumption is correct, how
can one describe the process behind selecting a specific move—for example, in a
chess game? My response is that one cannot. I believe that what people do under
such conditions is try to reconstruct a process they believe happened, but they have
no access to the process that really happened; that is in responding to such a
question, one tries to simulate the process.4
There is another question one could ask: Because a significant part of human
activity reflects a learning process, and human learning, at least in the cognitive
domain, is mediated by declarative knowledge of which we are fully aware (Ander-
son, 1983), how can it happen that all this knowledge "disappears" from conscious-
ness? Let me emphasize that it does not disappear. In fact, this knowledge is used
when one is trying to reconstruct a psychological process, as described previously.
Thus, human description of psychological processes are based on the declarative
representation of these processes, not of the procedural representation that it used
when this process is performed (Anderson, 1983).
4
T h i s idea is similar to Dennett's (1991) notion of consciousness being a virtual serial computer
running on a parallel computer—the brain. I do not assume, however, that the results of reconstruction
are always veridical.
224 Tzelgov
On Representations in Automatic
and Nonautomatic Processing
I believe that, i n contrast to other biological processes, psychological processes
return symbolic representations 5 as an integral part of their output, and that we are
conscious of that output (see also N a v o n , 1989). Nevertheless, the mode of
consciousness characterizing automatic and nonautomatic processing differs.
Automatic processing is performed either because it is a component of a more
complex task (and under such conditions, we refer to it as intentional), or because
it was triggered by the specific combination of the state of the organism and its
environment (Neumann, 1984) at a given moment (it is then autonomous). In both
cases, the specific process was evoked rather than deliberately initiated by the
performer, i n contrast to what happens i n the case of nonautomatic processing.
Dulany (1996) proposed referring to the two modes of consciousness that charac-
terize automatic versus nonautomatic processes as evocative and deliberative modes
of consciousness, respectively. Dulany (1991, 1996) also suggested that the delib-
erative mode of consciousness is based on propositional representation, whereas
representation i n the evocative mode is much less specific—to use his words, "it is
the sense of " (see Dulany, 1996). Thus we have a "sense o f the words that
the sentence we are trying to understand is made of, just as we have a "sense o f a
word the color of which we are reporting i n the Stroop task. Similarly, I believe we
have a "sense o f the various stimuli that trigger the preconscious phenomena Bargh
discusses; thus, I believe that humans have a "sense o f the stimuli that trigger their
stereotypes (e.g., Devine, 1989), they have a "sense o f the stimuli that activate
their attitudes (Fazio, Sabonmatsu, Powell, ckKardes, 1986), and they have a "sense
o f the stimuli that trigger their plans. 6 In the case of automatic processing, we are
evocatively conscious of the stimuli that trigger it.
Where Does Preconscious Automaticity Come From?

Taking as a starting point that automaticity i n the social context reflects a learning
process, one may ask what are the mechanisms of automaticity i n the social context?
Tzelgov, H e n i k , Sneg, and Baruch (1996) showed that skill-based automaticity—at
least i n the case of reading—may reflect either memory retrieval or algorithmic
processing. M o r e generally, assuming that our memory is organized i n terms of
different systems (see Squire, Knowlton, 6k Musen, 1993), my colleagues and I
(Tzelgov & Yehene, 1994; Tzelgov, Yehene, & Naveh-Benjamin, i n press) suggested
that learning cognitive tasks leads to a dual representation; the relevant algorithm
5
T h a t is not to say that all representations involved in psychological processing are symbolic; obviously
a significant part of psychological processes act at the sybsymbolic level (e.g., see the model of word
recognition, suggested by Van Orden's, Pennington, & Stone, 1990).
I am not implying here that they know the casual chain between the specific stimulus. What is meant
is simply that humans are aware of the stimulus causing specific behavior—not that they are aware the
specific stimulus causes the behavior.
is represented in the habit subsystem of nondeclarative memory, whereas the results

of learning are stored in declarative memory. That would mean that skill-based
automaticity represents either retrieval from declarative memory (see Logan 6k
Etherton, 1994) or from the habit subsystem of nondeclarative memory. In some
cases, automatic processing in the domain of social cognition might reflect these
subsystems (e.g., Smith 6k Lerner, 1986). This may be one source for the develop-
ment of automatic links between two semantically unrelated concepts that later on
result in automatic activation of the goals of behavior. Using an episodic priming
paradigm, Yehene and Tzelgov (1996) has showed that 40 presentations of concepts
from two different categories result in an associative link between them that causes
automatic processing of the target concept, given the prime.
In other situations, in particular when automaticity is based on emotional
information (e.g., Murphy & Zajonc, 1993), it may reflect the activity of the
conditioning subsystem of nondeclarative memory. Classical conditioning may be
responsible for the automatization of various affective responses to specific stimuli
(see Martin & Levey, 1978) and operant conditioning may be involved in activating
processing goals, as suggested in the last part of the target chapter. In both cases, I
believe that humans are evocatively conscious of the stimulus that elicits the
behavior. The evocative mode of consciousness characterizes associative learning
in general and, as such, is common to humans and other animals, although it may
be that humans recode in parallel the stimulus into a set of propositions to enable
a deliberative mode of consciousness of the stimulus.7 Gluck and Myers (1993)
proposed a computational model that provides a mechanism for such recoding. If
this model is correct, it leads to an interesting speculation: Similarly to the dual
representation of learning in the cognitive context, it may be that the results of
learning (that leads to automatization) in the social context also have a dual
representation; one in the conditioning subsystem of nondeclarative memory, and
the other in the declarative system.
On the Difference Between Monitoring and Control

In this chapter, I challenge the identification of automaticity with unconscious
processing; however, I adopt Bargh's (1989, 1992) suggestion to define automaticity
in terms of the absence of conscious monitoring. That does not mean that automatic
processing is not controlled. We do know that automatic (as well as nonautomatic)
processes are controlled by expectations (e.g., Logan, 1980; Tzelgov, Henik, 6k
Berger, 1992). I suggest distinguishing between control and monitoring. Control refers
to the sensitivity of a system to changes in inputs, which may reflect a feedback
loop, and, as such, it does not necessarily require consciousness. I would like to
preserve the term monitoring for intentional setting of the goal of behavior and to
7
D . E. Dulany (personal communication, December, 10, 1995) pointed out that a deliberative mode
of consciousness may also exist in rudimentary form in other mammals, as indicated by the works of
K ö h l e r (1927) and Tolman (1959).
226 Tzelgov
intentional testing of the output of the process. The SAS mechanism proposed
Norman and Shallice (1986) may be responsible for such activity. It is my belief that
monitoring requires a propositional representation of both the goal and the outputs.
Monitoring, as previously defined, is absent in autonomous automatic process-
ing. In the case of intentional automatic processing, monitoring applies to the unit
of behavior used to define action (Vallacher & Wegner, 1987). Defining action at
a given level implies that processing below that level is automatic (characterized by
evocative consciousness and not monitored). When the monitoring system detects
that the goals of action are not achieved, the action is redefined in terms of
subordinate units. Turning to the reading example once again, if the goal of behavior
is to read a sentence for meaning, monitoring is at the level of the sentence;
therefore, the reader is deliberatively conscious of the sentence meaning, but
evocatively conscious of the words of which the sentence is made. If, however, the
reader fails to understand the sentence, the goals for action will be redefined to the
word level and the monitoring will switch to that level; the processing of the words
will then cease to be automatic, and the reader will be deliberatively conscious of
them.
CONCLUDING REMARKS
Let me now summarize the main points I make in this chapter. I believe Bargh does
excellent work by taking automaticity out of the laboratory and into the real
situations of social interaction among people. In particular, the idea that goals of
behavior are automatically primed is both exciting and simple. My own research is
in complete agreement with the main argument of the target chapter: Most of our
behavior in everyday life is automatic. But this basic truth does not mean that
consciousness is riding into the sunset, and not only for the reasons Bargh discusses
in the last part of the target chapter.
It is true that automatic processing does not require conscious choice; in fact, it
does not require choice at all. It results from a single step retrieval of the required
output from declarative memory, or from the retrieval of the program responsible
for the behavioral unit, of which a given automatic act is made from nondeclarative
memory. In particular, in the autonomous automatic mode, we have no choice but
to behave. It is also true that we are not conscious of the very act of processing, but
this seems to be the case for all biological processes in our bodies.
As I show, there is (in my view at least) no convincing evidence in favor of the
perception of subliminal stimuli, which, in turn, leads to the conclusion that
automatic processing is based on perception of stimuli of which we are consciously
aware. The finding of Murphy and Zajonc (1993) showing affective priming under
very short exposure durations, supports not the notion of unconscious perception,
but the affective primacy hypothesis.
I also would argue that we are conscious of the output of automatic processes,
although it may be a different mode of consciousness (evocative rather than
14. A u t o m a t i c but C o n s c i o u s 227
deliberative). T h e argument that we are not conscious of the outputs of automatic

processing is based on the fact that frequently people are not able to report it
verbally, but i n order to report a given output, people have to remember it. This
implies a memory criterion of consciousness. Allport (1988) pointed out that such
a criterion is reliable only as a positive one; when we remember something, it is clear
we are conscious of it; but when we do not remember an event, it does not mean
we were not conscious of it. Therefore, such a criterion may be biased against the
evocative mode of consciousness. The representations i n this mode, because of
being just "a sense of," are less available to verbal reports due to the need to recode
information to representations more suitable to verbal reports (i.e., propositional).
In other words, it may require more time to provide a verbal description of this
output, and, therefore, there are more chances that it will be forgotten.
I believe we are conscious, or phenomenally aware of, the outputs of automatic
processing, although the representations of these outputs differ from those resulting
from nonautomatic processing. This implies that consciousness is not an unitary
phenomenon. Allport (1988) discussed various behavioral indicators of conscious-
ness and points out that, only i n the paradigmatic case, all of them hold simultane-
ously. T h u s , the feature-list approach for defining automaticity, w h i c h was
abandoned with the active help of the author of the target chapter, still dominates
thinking about consciousness. Furthermore, it does not appear that our conceptions
of consciousness will improve dramatically in the near future. T h a t is another reason
why the interchangeable use of the terms automatic and unconscious do not
contribute to our understanding of the automaticity of human action.
ACKNOWLEDGMENTS
I wish to thank G o r d o n Logan and Orit Tykocinski for very helpful discussions, and
D o n Dulany, Nachschon Meiran, Vered Yehene and the participants of my 1995
seminar o n "Unconscious cognitive processing" at B G U , for their feedback o n my
half-baked ideas.
REFERENCES
Allport, A . (1988). What concept of consciousness? In A . J. Marcel & E. Bisiach (Eds.), Consciousness
in contemporary science (pp. 159-182). New York: Oxford University Press.
Anderson, J. R. (1983). The architecture of cognition. Cambridge, M A : Harvard University Press.
cognition. Inj. S. Uleman & J. A . Bargh (Eds.), Unintended thought (pp. 3-51). New York: Guilford.
Bargh, J. A . (1992). The ecology of automaticity: Towards establishing the conditions needed to produce
automatic processing effect. American Journal of Psychology, 105, 181-199.
Bisiach, E. (1988) The (haunted) brain and consciousness. In A . J. Marcel & E. Bisiach (Eds),
Consciousness in contemporary science (pp. 101-120). New York: Oxford University Press.
Carr, T H . (1992). Automaticity and cognitive anatomy: Is word recognition automatic? American
Journal of Psychology, 105,201-238.
228 Tzelgov
Chomsky, N . (1980). Rules and representations. New York: Columbia University Press.
D'Agostino, P. R., & Beegle, W. (1996). A re-evaluation of the evidence for spontaneous trait inferences.
journal of Experimental Social Psychology, 32, 153-165.
Dennett, D. C . (1991). Consciousness explained. Boston: Little, Brown.
Dulany, D. E. (1991). Conscious representation and thought systems. In R. S. Wyer & T. K. Srull (Eds.),
Advances in social cognition (pp. 91-120). Hillsdale, NJ: Lawrence Erlbaum Associates.
Dulany, D. E. (1996). Consciousness in the explicit (deliberated) and the implicit (evocative). In J.
C o h e n & J. Schooler (Eds.), Scientific approaches to consciousness (pp. 179-212). Mahwah, NJ:
Fazio, R. H . , Sabonmatsu, D. M . , Powell, M . C , 6k Kardes, F. R. (1986). O n the automatic activation
of attitudes. Journal of Personality and Social Psychology, 50, 229-238.
Friedrich, F. J., Henik, A . , & Tzelgov, J. (1991). Automatic processing in lexical access and spread of
activation. Journal of Experimental Psychology: Human Perception and Performance, 17, 792-806.
Gilbert, D. T , & Hixon, J. G . (1991). T h e trouble of thinking: Activation and application of stereotypic
beliefs. Journal of Personality and Social Psychology, 60, 509-517.
Gluck, M . , & Myers, C . (1993). Hippocampal mediation of stimulus representation: A computational
theory. Hippocampus, 3, 491-516.
Greenwald, A . G . , Klinger, M . R., & Schuh, E. S. (1995). Activation of marginally perceptible
("subliminal") stimuli: Dissociation of unconscious from conscious cognition. Journal of Experimental
Hasher, L., & Zacks, R. T. (1979). Automatic and effortful processes in memory. Journal of Experimental
Holender, D. (1986). Semantic activation without conscious identification in dichotic listening, para-
foveal vision, and visual masking: A survey and appraisal. The Behavioral and Brain Sciences., 9, 1-66.
Jacoby, L. L. (1991). A process discrimination framework: Separating automatic from intentional uses
of memory. Journal of Memory and Language, 30, 531-541.
Jacoby, L. L., Levy, B., & Steibach, K. (1992). Episodic transfer and automaticity: Integration of
data-driven and conceptually driven processing in rereading. Journal of Experimental Psychology:
Learning Memory and Cognition, 18, 15-24.
Jacoby, L. L., Ste-Marie, D., & Toth, J. R (1993). Redefining automaticity: Unconscious influences,
awareness, and control In A . D. Baddeley, 6k Weiskrantz (Eds), Attention, selection, awareness and
control: A tribute to Donald Broadbent (pp. 261-282). Oxford, England: Oxford University Press.
Kihlstrom, J. F. (1987). T h e cognitive unconscious. Science, 137, 1445-1451.
Kohler, W. (1927). The mentality of apes. London: Routledge & Kegan Paul.
Locurto, C . M . (1981). Contributions of autoshaping to the partitioning of conditioned behavior. In C .
M . Locurto, H . C . Terrace, 6k J. Gibbon (Eds.), Autoshaping and conditioning theory, (pp. 14-16). New
York: Academic Press.
Logan, G . D. (1980). Attention and automaticity in Stroop and priming tasks: Theory and data. Cognitive
Logan, G . D. (1985). Skill and automaticity: Relations implications and future directions. Canadian
Journal of Psychology, 39, 367-386.
Logan, G . D. (1989). Automaticity and cognitive control. In J. S. Neuman 6k J. A . Bargh (Eds.),
Unintended thought (pp. 52-74). New York: Guilford.
Logan, G . D. (1992). Attention and preattention in theories of automaticity. American Journal of
Psychology, 105,317-339.
Logan, G . D., & Cowan, W. B. (1984). O n the ability to inhibit thought and action: A theory of an act
of control. Psychological Review, 91, 295-327.
Logan, G . D., & Etherton, J. L. (1994). What is learned during automatization: T h e role of attention in
constructing an instance. Journal of Experimental Psychology: Learning Memory and Cognition, 20,
1022-1050.
Logan, G . D., & Zbrodoff, N . F. (1979). W h e n it helps to be misled: Facilitative effects of increasing the
frequency of conflicting stimuli in a Stroop like task. Memory & Cognition, 7, 166-174.
Macmillan, N . A . , &. Creelman, C . D. (1990). Response bias: Characteristics of detection theory,
threshold theory and non parametric indexes. Psychological Bulletin, 107(3), 401-413.
Mandler, G . (1991). The processing of information is not conscious, but the products often are. Behavioral
and Brain Sciences, 14, 688-689.
Marcel, A . J. (1983). Conscious and unconscious perception: Experiments in visual masking and work
recognition. Cognitive Psychology, 15, 197-237.
Martin, I., & Levey, A . B. (1978). Evaluative conditioning. Advances in Behavior Research and Therapy,
1,57-102.
Merikle, R M . , & Reingold, E. M . (1990). Recognition and lexical detection without detection:
Unconscious perception? Journal for Experimental Psychology: Human Perception and Performance, 16,
574-583.
Murphy, S. T , & Zajonc, R. B. (1993). Affect cognition and awareness: Affective priming with optimal
and suboptimal stimulus exposures. Journal of Personality and Social Psychology, 64, 723-739.
Navon, D. (1989). O n the importance of being visible: O n the role of attention in a mind viewed as an
archaic intelligence system: I Basic tenets. European Journal of Cognitve Psychology,1,191-213.
Neely, J. H . (1991). Semantic priming effects in visual word recognition: A selective review of current
findings and theories. In D. Besner & G . W. Humphreys (Eds.), Basic processes in reading: Visual word
recognition (pp. 264-335). Hillsdale, NJ: Lawrence Erlbaum Associates.
Neisser, U . (1967). Cognitive psychology. New York: Appelton Century Croft.
Neumann, O . (1984). Automatic processing: A review of recent findings and a plea for an old theory.
In W. Prinz 6k A . F. Sanders (Eds.), Cognition and automatic processing (pp. 255-283). Berlin:
Springer-Verlag.
Newell, A . , & Rosenbloom, P. S. (1981). Mechanisms of skill acquisition and the law of practice. In J.
R. Anderson (Ed), Cognitive skills and their acquisition (pp. 1-55). Hillsdale, NJ: Lawrence Erlbaum
Associates.
Norman, D. A . , & Shallice, T. (1986). Attention to willed action: Willed and automatic control of
behavior. In R. J. Davidson, G . E. Schwartz, 6k R. L. Shapiro (Eds.) Conscious and self regulation;
Advances in research) (Vol. 4, pp. 1-18). New York: Plenum.
Ohman, A . , Dimberg, U., & Esteves, F. (1989). Preattentive activation of aversive emotions. In T. Archer
& L. G . Nillson (Eds.). Aversion avoidance and anxiety: Perspectives on aversively motivated behavior.
Posner, M . I., & Snyder, C . R. R. (1975). Attention and cognitive control. In R. L. Solso (Ed.), The
Loyola Symposium (pp. 55-85). Hillsdale: Lawrence Erlbaum Associates.
Rafal, R., & Henik, A . (1994). T h e neurology of inhibition: integrating controlled and automatic
processes. In D. Dagenbach 6k T. H . Carr (Eds.), Inhibitory processes in attention, memory and language.
(1-51). New York: Academic Press.
Rock, I. (1983). The logic of perception. Cambridge, M A : M I T Press.
Schneider, W , & Shifrrin, R. M . (1977). Controlled and automatic human information processing: I.
Detection search and attention. Psychological Review, 84, 127-190.
Shifrrin, R. M . , & Schneider, W (1977). Controlled and automatic human information processing: II.
Perceptual learning, automatic attending, and a general theory. Psychological Review, 84, 127-190.
Smith, E. R., & Lerner, M . (1986). Development of automatism of social judgment. Journal of Personality
and Social Psychology, 50, 246-259.
Squire, R. L., Knowlton, B., & Musen, G . (1993). T h e structure and organization of memory. Annual
Review of Psychology, 44, 453-495.
Stroop, J. R. (1935). Studies of interference in serial verbal reactions. Journal of Experimental Psychology,
18, 643-662.
230 Tzelgov
Theios, J., & Haase, S. J. (1994, November). A signal detection theory interpretation of one type of unconscious
perception. Paper presented at the 35 th annual Meeting of the Psychonomic Society, St. Louis, M O .
Tolman, E. C . (1959). Principles of purposive behavior. In S. Koch (Ed.), Psychology: A study of a science
(Vol. 1, pp.92-158). New York: McGraw-Hill.
Treisman, A . (1985). Preattentive processing in vision. Computer Vision Graphics and Image processing,
31, 156-177.
Treisman, A . , Vieira, A . , & Hayes, A . (1992). Automaticity and preattentive processing. American
Journal of Psychology, 105, (341-362).
Tzelgov, J., Henik, A . , & Berger, J. (1992). Controlling Stroop effect by manipulating expectation for
color related stimuli. Memory & Cognition, 20, 727-735.
Tzelgov, J., Henik, A . , & Leiser, D. (1990). Controlling Stroop interference: Evidence from a bilingual
task. Journal of Experimental Psychology: Learning, Memory and Cognition, 16, 760-771.
Tzelgov, J., Henik, A . , Sneg., R., & Baruch, O . (1996). Unintentional reading via the phonological route:
The Stroop effect with cross-script homophones. Journal of Experimental Psychobgy: Learning, Memory
and Cognition. 11, 339-369.
Tzelgov, J., Porat, Z., & Henik, A . (1994, November). Is perceiving the word necessary for the Stroop effect?
Paper presented at the 35th annual meeting of the Psychonomic Society, Saint Louis, M O . Tzelgov,
J., & Yehene, V. (1994, September). Memory, skill-based automaticity and transfer. Paper presented at
the 7th meeting of the European Society of Cognitive Psychology, Lisbon, Portugal.
Tzelgov, J., & Yehene, V. (1996). Skill-based automaticity: From definitions toward a theoretical framework.
Manuscript in preparation.
Tzelgov, J., Yehene, V , & Naveh-Benjamin, M . (in press). From automaticity to memory and vice versa:
O n the relations between automaticity and memory. In J. Brzezinski, B. Krause & Maryszewski (Eds.),
Idealization in psychobgy: Poznan studies of the sciences and the humanities. Amsterdam-Atlanta: Rodopi.
Ullman, S. (1984). Visual routines. Cognition, 18, 97-159.
Vallacher, R. R., & Wegner, D. M . (1987). What do people think they are doing? Action identification
and human behavior. Psychobgical Review, 94, 3-15.
Van Orden, G . C , Pennington, B. F. & Stone, G . O . (1990). Word identification in reading and the
promise of subsymbolic psycholinguistic. Psychobgical Review, 97, 488-522.
Velmans, M . (1991). Is human information processing conscious? Behavioral and Brain Sciences, 14,
651-725.
Yehene, V , & Tzelgov, J. (1996). The two factor model of automaticity: A test using transfer paradigm.
Psychobgia, 5, 121-138.
Zajonc, R. B. (1980). Feeling and thinking. American Psychobgist, 35, 151-175.
Chapter 7 5
Reply to the Commentaries
John A . Bargh
New York University
I admit that right after I read Mischel's commentary (chap. 11, this volume), I had
my office swept for the presence of surveillance equipment. His description of some
of my underlying motives in writing the target chapter is uncannily, if not eerily,
accurate. Yes, I wanted to push a particular point of view as hard as I could, that
automatic processes play a far larger role in everyday thinking, feeling, and doing
than our social psychological models suggest. My aim was to raise the possibility
that many of the processes believed to be the product of conscious intention and
oversight could in fact be nonconsciously produced; to wit, even the activation and
operation of intentions (goals) themselves could occur without an act of will.
Mischel was not alone in voicing his suspicions that the forcefulness of my
arguments was more tactical than heartfelt (see also Srull, chap. 13 and Gardner
& Cacioppo, chap. 8), and so I was not too surprised when the office search turned
up nothing in the way of eavesdropping apparatus. There was a perception among
several commentators that my stated claims were more radical than my real beliefs,
and on some points they perhaps were. So it would be best if I now own up to this
and make clear my own opinions about both the extent of automatic influence in
daily life, and the purpose and functions of consciousness.
Baumeister and Sommer (chap. 3), although expressing agreement that conscious
involvement is relatively infrequent, take exception to my insinuation that it is entirely
absent. As did Srull, (chap. 13), Gardner and Cacioppo (chap. 8), and Mischel (chap.
11), they call me on the carpet for pushing a "single-cause" model of social psychology:
situations to the exclusion of person variables, and automatic forces to the exclusion of
any conscious or controlled processes. It is on this latter point that I am vulnerable to
the charge of being more tactical than sincere in my arguments. But the former point
reduces to a misunderstanding caused by my lack of precision as to what I meant.1
1
It was by no means the only imprecision in terminology in the target chapter. After writing several chapters
devoted nearly exclusively to the various definitional qualities of automaticity (Bargh, 1984, 1989, 1994), I
thought, for once, that I could play fast and loose. But Smith's commentary (chap. 12) caught me out.
231
232 Bargh
DIRECT EFFECTS OF ENVIRONMENTS

WITHOUT CONSCIOUS CHOICE
In hindsight, my use of the terms situation and environment, combined with my

invocation of precognitive (Skinner) and preinteractionist (e.g., Milgram) ghosts,
was unintentionally (consciously, at least) misleading. By situation, I meant the
psychological situation, and this certainly includes the immediate internal reactions
of the individual. To me, this is what the interactionist position means, that the
objective external situation is not a cause as much as the taken meaning and internal
experience of that situation-and this can certainly vary from individual to indi-
vidual. Environmental events can directly trigger these internal reactions without
the intervention or need of conscious choice (as Smith notes in his chap. 12
commentary, this is what I meant in the target article by automatic), and as these
internal reactions can vary from individual to individual, these automatic effects,
can and do, vary as well. In fact, when these individual differences in the meanings
of external situations are taken into account, substantial cross-situational consis-
tency is shown in emotional and behavioral reactions, as Mischel's own research
shows (Mischel & Shoda, 1995).
To me, such consistency points to automatic links between the representations
of those situations and the representations of specific motivational and behavioral
responses to them. It could be the case, of course, that a person engages in a
conscious decision making process every time before acting (although this seems
unlikely, especially for the impulsive, aggressive responses Berkowitz discusses, chap.
4). But this would be unnecessary: If the environment is consistently producing the
same emotion or behavior then, as Skinner argued, we have no need for explanatory
devices such as reasoning and decision-making and choice. And to say that
conscious choice is involved each time would run counter to Baumeister and
Sommer's (chap. 3) well-taken argument that one important function of conscious-
ness is to enable overriding of habitual, usual responses.
To put this as clearly as I can, I argue that consistent environment-behavior
effects-whether produced in most people by powerful situational forces, as studied
by Milgram or Darley and Latam!, or only in some people because of their idiosyn-
cratic internal reactions to that situation, as studied by Mischel and Shoda-may
be due to automatic processes, and may not involve any conscious choice at that
moment. I did not mean to argue that such consistency was somehow proof of
automaticity and ruled out the possibility of conscious involvement. My purpose
instead was to persuade the reader to consider both the automatic and the willful
accounts on an equal playing field, without implicit prejudice towards one as the
default causal mechanism.
This view of the environment as causal agent is of course markedly different from
Skinner's, as it includes internal states and representations activated directly by
environmental events (i.e., without the intervention of conscious choice or delib-
eration). Obviously, I was fuzzy on this point in the target chapter, as Mischel (chap.
11), Srull (chap. 13), Tzelgov (chap. 14) and others understood me to be excluding
15. Reply to the Commentaries 233
the person variable from the picture entirely. In our automaticity work we often
take individual differences into account, such as in chronic accessibility effects on
person perception, and in the effects of contextual power on sexual harassment.
The direct effects of the environment are those that continue on to influence and
produce psychological and behavioral responses with no role played by conscious
choice. Such automatic effects of situations include and incorporate individual
differences in the way that situations are understood and the responses made to
them chronically in the past.
AUTOMATIC PROCESSES CAN INTERACT

WITH THE ENVIRONMENT
This brings me to a second point I did not make as lucidly as I could have in the
target chapter. The internal responses that are said to be directly activated and put
into motion by social environments are not restricted to immediate, momentary
responses. The auto-motive model holds that goals can be activated automatically
(without an act of will) and then may operate on environmental information
without one's awareness or need to monitor the goal's progress. That is, the ongoing
automatic process interacts with the environment over time; it is a much more
sophisticated view of an automatic process than the one to which several of the
commentators reacted.
Automatic processes have been viewed as simple and crude in their effects. Neisser
(1967) argued that the extent of preattentive analysis of the environment could not
be any further than the crude segmentation of the sensory field into basic objects. So
ingrained was this assumption that when Spelke, Hirst, and Neisser (1976) demon-
strated fairly amazing multitasking performance on the part of their subjects—with
practice, they could simultaneously take dictation and read a separate tract for
meaning—the authors invoked the very complexity of the tasks performed as
evidence against the explanation that the tasks had become automated.
Several of the commentaries seem to share this assumption of relatively simple-
minded automaticity. Baumeister and Sommer (chap. 3) cite my 1982 dissertation
study in which automatic activation of the self-concept was demonstrated, in which
automatic processes were described as inflexible. However, that study was just the
first step of a research program on the extent of initial, immediate environmentally
driven mental processes, and concerned the activation of a representation by
relevant input. The kinds of automatic processes described in the target chapter
show (I'd like to think) a greater degree of sophistication and flexibility in dealing
with an ongoing environment. Goals that operate automatically, just like goals that
are put into motion by an act of conscious will, interact with environmental
information over time in pursuit of that goal; driving a car for many miles while
daydreaming and later having no memory for any of it being the most commonly
experienced example of this process.
Similarly, Srull (chap. 13) and Banaji et al. (chap. 2) are resistent to the idea of
the automatic activation and operation of goals, and again I suspect that this is
234 Bargh
because automatic processes are understood and assumed to be immediate reactions

that cannot unfold over time. Srull takes exception to Chartrand and Bargh's
(1996) conclusion that impression formation and memorization goals were acti-
vated automatically i n their experiments, and he expresses the wish that we had
considered other explanations. H e does not, however, suggest what these alterna-
tives might be, and frankly, we are at a loss as to what they would be. T h e
auto-motive model predicts that goals will operate on relevant environmental
information regardless of how they are activated, by an act of will or nonconsciously,
and i n two studies employing nonconscious priming (in one study, subliminal
priming) we were able to replicate the findings of previous experiments i n which
the same goals were explicitly given to subjects through instructions. Moreover, our
subjects showed no hint of having differentially consciously construed the experi-
mental tasks as a function of the priming manipulation. I suspect (given the lack of
alternative explanations supplied) that resistance to these and similar nonconscious
goal effect findings has more to do with implicit assumptions about the limits of
automatic processes than with the conclusions we drew from our results.
Perhaps a change of metaphor would be helpful. Automatic processes are seen
as mechanical, sort of like a computer's reactions to input from the environment,
and computers are not considered to have purposes with regard to that external
environment but to react only to external input. Thus, the metaphor of automatic
processing as pure, computer-like information processing does not help us intui-
tively see automatic processes as being interactive with the environment. T h e
metaphor change I have i n mind is from that of the computer to that of the
(nonhuman) animal—perhaps better put, from automaticity as artificial intelli-
gence to automaticity as ethology. Ethology is the study of animal behavior i n
complex interaction with environments, without (usually) the invocation of con-
scious will as an explanatory principle.
A s an example, consider the mating behavior of crickets. If a female cricket hears
the songs of two or more male crickets at the same time, she will often move i n the
direction of one but not the other. The way this has been traditionally understood
is that the female prefers one song over the other, and has chosen that male. But
researchers tested the hypothesis that this apparent active selection process is
instead due to an auditory mechanism that signals which song is the closer of the
two. They did this by building a "cricketbot" that incorporated a mechanical
signaling device, put it on wheels, and found that its behavior i n an environment
of male cricket chirps very closely approximated that of actual cricket females ("Not
clever enough?," 1996).
T h e moral of the story for my purpose is that what looks very much like the
product of conscious choice may in reality not (and need not) be. A s Tzelgov i n his
commentary (chap. 14) also argues, automatic (nonwillful) processes are just as
capable of interacting with environmental information and directing behavior as
are willful processes. The target chapter (as well as others; see Bargh & Barndollar,
1996; Bargh & Gollwitzer, 1994) has provided accounts of several studies i n which
goals such as achievement and affiliation and impression formation were toggled by
priming procedures and shown to produce the same effects as when chosen by an
act of will. As Srull (chap. 13) and Banaji et al. (chap. 2) note, to say that such goal
effects (and any other automatic effect described in the target article) can occur
automatically does not mean that they do happen automatically all of the time (to
echo Clore and Ketelaar's, chap. 6, pithy phrase, perhaps all we were doing in our
studies was "hot-wiring the social ignition system," causing it to operate by means
other than its usual method). But as the cricketbot example suggests, the fact that
both an automatic and a willful cause are possible for goal-directed phenomena
should at least give one pause before making a default ascription to an act of will.
FUNCTIONS OF CONSCIOUSNESS
The contention that consciousness is not necessary (i.e., that everything is auto-
matic) often produces a response of the form, "Okay, smartypants, then why do we
have consciousness? Huh?" Many of the commentators discuss what the role of
consciousness might be for the hypo the tically automatic world described. Baumeis-
ter and Sommer (chap. 3) argue that consciousness allows us to override automatic
processes, to not always do the usual thing. I find their notion of consciousness as
adding an element of indeterminacy to behavior and other reactions extremely
compelling and exciting. Certainly, to become too consistent in one's behavior in
reaction to the environment makes one very predictable, and other animals
(including humans) could well use this predictability to their (and against one's
own) advantage. If knowledge is power, then perfect knowledge of our responses
gives those who would eat, mislead, or otherwise use us for their purposes, great
power over us. By introducing some variability and perhaps randomness into the
equation, consciousness reduces this power and at the same time, in a manner
analogous to genetic mutations, introduces new and potentially better responses to
an environmental event. (Note that the response might equally lead to a worse
outcome than usual, which would lead one to slap oneself on the forehead later and
say "why did I do that?")
Gardner and Cacioppo (chap. 8) describe the role motivation plays in effortful
processing of persuasive messages, such that highly motivated subjects are affected
by message quality whereas nonmotivated subjects are not (Chaiken, 1980; Petty
& Cacioppo, 1984). They point to such findings as evidence of conscious and
nonautomatic causation (because the amount of effortful processing moderated the
amount of attitude change). Because an effect varies with effortful processing does
not mean the effect was produced by an act of conscious will, however. If the motive
or goal that caused the more effortful processing was automatically triggered, the
subsequent processing would be under the control of an unintended, and, in that
sense, nonconscious process. I think Gardner and Cacioppo and I are in basic
agreement on this point, for a few paragraphs later they describe high need for
cognition individuals as probably possessing such automatic motivations, which
result in greater effortful processing. Their point is that using the term automatic
236 Bargh
both for the initial activation of the goal and then for its operation (which uses
attention, much like driving a car automatically involves attentional processing of
massive amounts of information) is problematic. I certainly agree that the use of
automatic as a blanket term for both the unintentional and the efficiency aspects
of processing leads to such muddied waters (see Bargh, 1994), and that more precise
terms are needed. A t the same time, the auto-motive model holds that both the
triggering of the goal and its continued operation (demanding of effort and attention
as it does) occur in the absence of an act of conscious will.
Gardner and Cacioppo next underscore the important role of conscious proc-
essing in the development and formation of automatic processes; this was implicit
in the target chapter because automated social perception and motivational effects
were said to be put in place by frequent and consistent conscious use in the past.
My only gripe here is that this fact by itself does not justify their conclusion that
social psychological phenomena cannot be reduced to the study of automatic
processes, as long as we are talking about people of a certain age—30? 20? 10?—who
have had sufficient experience in frequently and consistently doing most things.
But getting back to the function of conscious processing to put chronic, auto-
matic effects in place (over time), I am not certain that automatic evaluation
requires frequent and consistent prior conscious evaluations. Thus, I am in agree-
ment with Smith (chap. 12) that not all automatic processes were originally
conscious and then automatized. A dissertation recently completed by Garcia
(1996) found in several experiments that the automatic attitude activation effect
occurs for entirely novel stimuli (the sounds and abstract pictures that served as
priming stimuli were assumed to be good or bad for the experimental subjects based
on normative ratings by separate groups of subjects). Obviously, the subjects never
consciously evaluated any of these objects, and the effect held even when the
subject's conscious task during the experiment had nothing to do with evaluation
at all. The adaptive advantages of classifying everything immediately in terms of
whether it is positive or negative are obvious, and so this evaluation process would
seem to be a good candidate for an automaticity "hardwired by evolution," in
Smith's phrase.
Gardner and Cacioppo's discussion of the recent neuropsychological evidence
concerning evaluative processes—especially their own work on distinct positive and
negative neural processing areas—suggests that what Garcia's studies are tapping
is the initial, immediate classification of experience as positive or negative, which
must occur if separate neural structures are responsible for their processing. Such
possibilities are intriguing and an example of how, as Gardner and Cacioppo argue,
we can gain a greater understanding of a phenomenon by examining it at neuro-
psychological as well as cognitive and social psychological levels of analysis. I found
their discussion helpful in my own understanding of what produces such immediate
evaluation of incoming, novel stimuli.
Useful as well on this score was the commentary by Clore and Ketelaar (chap.
6). They note that I was less than specific about the automatic evaluation effect
(see also Smith's comments) and perhaps in need of some help in understanding it.
I found their analysis of different levels of evaluation, and of the automatic
evaluation effect, insightful, helpful, and compelling. Their suggestion of two

evaluative stages, one involving mere classification as good or bad and the other
appraising the degree of affect in terms of current and chronic goals for the
individual, fits nicely with our data as well as with the neurological evidence coming
out of Cacioppo's laboratory.
It may be, in terms of the separate module idea, that the evaluative and the
motivational system each "evaluate" events, but on different bases: the evaluative
module in terms of "preferenda" or aesthetic-taste features (this apparently being
an innate function), and the motivational module separately appraises the event in
terms of its relevance for current and long-term goals. Or it could be that the
evaluative system passes along its primitive, good-bad categorization to the moti-
vational system for further analysis in terms of those goals. In any case, the Clore
and Ketelaar analysis provided me with much food for thought about the nature
and purpose of automatic evaluation.
ON OLD, ABANDONED ESKIMOS
Logan's commentary (chap. 10) provides a valuable educational service to the field
of social cognition. His detailed critique of resource theory should give anyone pause
who places all of their predictive eggs in that particular basket. Although none of
the effects that I reported in the target chapter were based on resource theory, Logan
is being too generous (to me) to exclude me from the set of those who have invoked
it in the past. The notion of resource limitations, and of the greater efficiency in
processing by chronically accessible constructs, served as the basis of the predictions
in the Bargh and Thein (1985) person memory study, as well as the Bargh and Tota
(1988) study of the depressive self-schema. And so, if Gilbert, Wegner, and Macrae
deserve their lumps then so do I.
In response to Logan's commentary, I should first note that the model underlying
my predictions for a long time now is more accurately termed process theory than
resource theory, and here, I think, I am on safer ground. As was discussed in the
"Preconscious and Skill Acquisition" section of the target chapter, the automatic
mental process is said to become compiled and streamlined with practice, assembled
into larger all or none structures that come to operate as a whole.
In my view, process theory has an advantage over Logan's (1988) preferred
instance theory in the area of social judgments because, according to Logan (1988),
the latter cannot account for generalized as opposed to specific effects of practice
on proceduralization (but process theory can). Smith (1994) showed general and
specific transfer effects in his research on social judgment automatization. A person
becomes faster (more efficient) at judging the honesty or aggressiveness of a
behavior the more he or she has made those kind of judgments in the past, even for
new behavioral descriptions not judged before.
Indeed, I found myself in sympathy with the main thrust of Logan's critique,
because the effects reported in the target chapter were, in fact, predicted based on
238 Bargh
exactly the kinds of priming or "residual activation" ideas he invokes in his

commentary. Evaluations, motivations, perception, and behavior were argued to
be represented in memory and to become automatically activated on the presence
of relevant environmental features. Our predictions, such as in the rude-polite
study, or the stereotype-behavior studies, or the achievement priming studies, rely
on the idea of residual activation just as does Logan's analysis. He notes that some
readers may find that idea implausible, but I believe that it will instead find a
receptive audience among those who are familiar with the substantial body of
construct priming research in social cognition since 1975.
I am nonetheless left with the feeling that the future for resource theorists in
social psychology may be less frightening than the picture Logan paints here. It
may instead be more of a problem at one level of analysis than another. That is,
the idea of capacity and resource limitations has proven to be of great heuristic
value within social cognition. It generated intuitively plausible and empirically
supported ideas about the greater role played by the self-concept, stereotypes,
and attributional processes in noisy and busy real-world environments than in
informationally impoverished laboratory settings. Discoveries about the perva-
sive use of stereotyping and dispositional attribution processing by Gilbert and
Macrae and others, and more recently by Baumeister and Sommer (personal
communication) in the area of one's ability to self-regulate, have been based on
the limited capacity idea.
As Logan notes, resource theory generated the predictions that were confirmed
in those studies. If resource theory turns out to not be true at a more molecular
level, as the review by Logan certainly suggests is already the case, then one is left
with the problem of accounting for all of these results. Something is changing the
outcome of social information processing when people are otherwise busy or
distracted. Although resource or capacity differences may turn out to not be that
"something", this does not change the fact that the phenomenon of interest was
shown to be different under these very real and commonplace overload conditions.
Logan's (1988) instance theory model of automaticity was a major breakthrough
in models of automaticity, and one I read and thought about quite carefully when
it appeared, for the obvious reason that it was so directly related to my own research.
As Logan noted then, the predictions of instance theory and capacity theory are
the same in many, if not most, cases, although Logan generated and found support
for novel predictions from instance theory that resource theory could not make. In
applying instance theory to my own information overload experiments, however, I
found that it made the identical prediction as did resource theory. The same is true,
I suspect, for the studies by Gilbert and others.
There is a deeper theme and message in Logan's commentary that I wholeheart-
edly endorse, and which deserves repeating. This is that resource theory alone as a
basis for predictions about automatic processes is very limited and not very precise.
In social cognition research, resource theory has been used to generate predictions
concerning the efficiency of a process. But there are other aspects of an automatic
process, such as its unintentionality and its ability to guide information processing,
evaluation, or behavior as an unseen hand, to which resource theory does not speak
at all. O n the other hand, process theory, with its notion of proceduralization or
compilation, allows one to generate predictions in all of these aspects of automat-
icity (see Vera & Simon, 1993). I think this is the heart of Logan's message, and it
deserves to be taken to heart by the rest of us.
HOW SPECIFIC (OR GENERAL) ARE ENVIRONMENTS?
Although those experimental outcomes summarized in the target chapter are, to

me, strongly suggestive of the role automatic processes play in everyday life, they
are only the first wave of such evidence. They answer the first question of whether
all aspects and qualities of social life can be produced nonconsciously. But much
more is left to be known. One crucial question is raised most completely by Srull
(chap. 13) and by Hardin and Rothman (chap. 9), but also by Banaji et al., (chap.
2), and Berkowitz (chap. 4): O f what level of specificity are the "sets of environ-
mental features" argued to trigger these automatic processes? When does a push
become a shove, or a smile a show of affection? As Hardin and Rothman phrase the
question, "Which elements of information will actually guide judgment and behav-
ior?" (p. 143).
Again, as described in the target chapter, the external features are not as
important as the internal meanings or representations they activate; it is these
internal meanings that become chronically linked to emotional and behavioral
responses. Banaji et al. (chap. 2) refer to these as rnicroenvironrnents.
However, there do exist theoretical positions concerning the bandwidth of these
internal categories. Bruner (1957), for instance, argued that the more accessible
the internal representation, the greater the range of behaviors that are perceived
to fall within its domain. Moreover, the research by Mischel, Shoda, Wright, and
colleagues suggests a basic level of behavior categorization that is linked to fairly
high consistency: not "hostility," for example, but "being insulted" or "being
reprimanded." As this finer level of behavior categorization leads to higher observed
consistency in the subject's own behavior, this is probably the level of specificity of
feature sets to which automatic responses are linked.
CULTURE AND CONTROL
Speaking of forces that create consistency in behavior, Cohen (chap. 7) argues (see
also Banaji et al., chap. 2) that I may have underestimated the degree of automatic
control by not including mention of cultural transmission of behavioral rules,
norms, and especially understandings of events. As Cohen's recent book (Nisbett
& Cohen, 1996), as well as his commentary makes clear, the meaning of certain
classes of acts is taught and shared within a culture, and the appropriate responses
to it also very closely defined (see also Haidt, Koller, ck Dias, 1993). Much earlier,
240 Bargh
Barker and Wright (1954) showed just how much variance in behavioral variation
could be accounted for just by knowing the setting in which the behavior occurred;
in one Midwest U.S. town, there was great variation between the barber shop,
church, school, and street, and it swamped individual variation within these
settings. Social historians such as Foucault (1979) and Giddens (1984) described
in detail the implicit background controls operating on behavior within any culture
or society, such that its participants create it each morning anew without thinking
about it.
The automaticity of social norms and settings was included as a component of
automatic motivations (Bargh, 1990) but as I had no data of my own to present, it
was not discussed in the target chapter. Cohen's commentary fills that gap by
highlighting this important implicit source of automatic influence, and points the
way for a potentially fruitful collaboration between automaticity and cultural
psychology researchers in the future.
APPLICABILITY AND THE INTERACTION

OF AUTOMATIC FORCES
We do need to further research the precise environmental conditions that produce

automatic effects on evaluation, motivation, perception, and behavior. But now
that such effects were shown to exist for each psychological system in isolation, a
critical "second wave" research question becomes, how do these various precon-
scious effects interact with each other? With evaluations, goals, perceptions, and
behavioral tendencies all being activated in parallel in immediate response to the
external environmental situation, what determines which of these will control
responses, impact most heavily on judgments, and so on? What happens when the
implications of one preconscious process runs counter to another, such as if a person
has a chronic motive to be egalitarian but nonetheless possesses a cultural stereo-
type of the group to which a target individual belongs (Moskowitz, Wasel, Goll-
witzer, & Schaal, 1996)?
Such questions are the focus of Hardin and Rothman's (chap. 9) excellent
commentary. In my opinion, their approach to an answer in terms of the applicability
to the situation of the various forms of preconscious activation will prove to be a
seminal contribution.
Take, for instance, the "interruption" study (Bargh, Chen, & Burrows, 1996)
mentioned in their commentary. The concept of rudeness or of politeness was
primed in a preliminary experiment, and this manipulation was found to predict the
probability of the subject interrupting the experimenter later on down the hallway.
But note that we set up the interruption situation to be one that was applicable to
the primed concepts of rudeness and politeness, just as we set up the situation in
the Bargh and Gollwitzer (1994) experiments to be applicable both to achievement
or to affiliation tendencies. Residual effects of context (priming) on subsequent
15. Reply to the C o m m e n t a r i e s 241
responses will only occur to the extent the subsequent situations are applicable to
their expression. A s Hardin and Rothman emphasize, this principle puts important
constraints on the expression of automatically triggered tendencies.
One important moderator of such constraints, as Hardin and Rothman's analysis
indicates, is whether the automatic effect is being driven on-line by the current
environment, or is a residual effect of a previous environmental context. Here is
where, to answer Berkowitz's (chap. 4) important query, the distinction between
preconscious and postconscious really matters. A s Hardin and Rothman note, when
the prime (or "trigger") and the target of the automatic process are the same, then
accessibility and applicability reduce to the same thing: The relevance of the
situation to the (preconscious) automatic process is not i n question, as it is the very
situation to which the automatic process was repeatedly linked i n the past. Post-
conscious automaticity, on the other hand, is a temporary state created by recent
activation of a process. Here is where the subsequent situation needs to be
applicable to the process or else it will not have an influence. In Bargh et al. (1996)
Experiment 3, for example, i n which subjects were primed subliminally with
African-American faces, they subsequently showed greater hostility to the experi-
menter i n reaction to a (mild) provocation, but not just prior to that provocation
(though the stereotype was already activated).
AWARENESS OF THE STIMULUS

VERSUS AWARENESS OF ITS EFFECT
O n the topic of pre - versus postconscious automaticity, there is a common confusion

between preconscious and subliminal that is the basis for much of Tzelgov's (chap.
14) commentary. In a 1989 chapter, I defined preconscious processes as those that
"require only the triggering proximal stimulus event and occur prior to or i n the
absence of conscious awareness of that event" (p. 11). This does not mean a person
is never aware of stimuli that trigger preconscious processes; i n fact I argued i n
other places (1992, 1994) that the subliminality or supraliminality of the activating
event makes no difference to the effect (as shown i n the two Chartrand & Bargh
[1996] studies discussed i n the target chapter). W h a t does matter is whether the
person is aware of the potential influence of the automatic process, not awareness
of the stimulus event. T h e key and often overlooked phrase i n the 1989 definition
is "prior to." Preconscious processes are those that occur prior to one's conscious
awareness of the stimulus event, they occur during the formation of the percept,
and are part of the "givens" with which we start any control or conscious processing
about the event, and they are experienced phenomenally as "out there"—fluently,
in other words, as qualities of the stimulus not of any internal reactions or
processing of the event. Happily, once "preconscious" is not read as "subliminal,"
most of Tzelgov's objections to the target article disappear. I should stress that he
is not alone in his interpretation of what I meant i n 1989; as a result I have had
since then to be much more complete and careful to distinguish the two concepts.
242 Bargh
DISSOCIATED SYSTEMS VERSUS A GENERAL THEORY
I found Smith's proposal and description of a connectionist preconscious analysis

system (chap. 12) very appealing, and think he and I have fewer areas of disagree-
ment than may appear. The model in his Fig. 12.1 is highly similar to that in the
target chapter; both approaches speak of modularity and interrelations between the
modules. I certainly never intended to speak of mysterious or ineffable processing
systems or rules, just experimental effects that could not be readily accounted for
in terms of the models with which I was familiar. Smith (chap. 12, see also Kunda
&Thagard, 1996) shows that there may be a general cognitive model, the principles
of which can generate the various findings described in the target chapter, and also
incorporate the idea of separate modules. I found his commentary extremely helpful
as a framework with which to more systematically understand the evidence we
generated, using a uniform set of operating principles.
RELATIONS BETWEEN CONSCIOUS

AND AUTOMATIC PROCESSES
Smith also describes various ways in which conscious and automatic processes can
interact; other commentators note how consciousness can "override" an automatic
process if there is sufficient attention and also motivation to do so. Tzelgov's (chap.
14) commentary also contains some thoughtful analysis of the relations between
automaticity, control, autonomy, and awareness. In the target chapter, I focused
primarily on establishing the pervasiveness of preconscious determinants of various
aspects of mental and social life, and did not give full attention to the topic of how
automatic and conscious or control processes interact. Recently, however, Wegner
and I attempted to do that (Wegner & Bargh, 1997).
We analyzed the possible forms of interaction between control (conscious) and
automatic processes in terms of four essential relations: multitasking, overriding,
launching, and transformation. Multitasking refers to the operation of automatic and
control processes in parallel. The remaining three relations, importantly, can go
both ways, in that an automatic process can override a control one (as in intrusions)
just as a control process can override an automatic one (as when a person regulates
an automatically activated stereotype). A n automatic process can launch a control
process (as in orienting attention to salient information), and a control process can
launch an automatic process (as in well practiced judgments or motor skills; we
called this delegation). Finally, a control process can become transformed into an
automatic one with sufficient practice (automatization), and an automatic process
can be transformed back into a control process (disruption) when environments
change and no longer support them. The important point about all this for present
purposes is that each of these relations is two-way; relations between automaticity
and conscious processes are not as simple as conscious processes always dominating
automatic ones.
FREE WILL
I disagree with Smith (chap. 12) that the findings described in the target chapter
should have no bearing on the question of free will. If we define automaticity in
terms of direct environmental effects for which no conscious involvement is
required, clearly these effects are caused by some force other than an act of will. It
might be countered that, even then, preconscious effects can be controlled, and
therefore an individual has free will concerning their expression (Fiske, 1989), but
in my opinion this position is overstated. Such acts of control require an awareness
of the possibility of being influenced in ways other than those of which one is aware,
and such occasions are few and far between. We may be well-informed of the
possibility because we are the ones who study the implicit and automatic influences
such as stereotyping, but anyone who has ever attempted to explain to a lay friend
or relative that mental events can occur and affect their judgments and behavior
without their knowing about it can attest that it is a difficult task indeed. What
even the best-intentioned and open-minded individual will do when confronted
with such an idea is to examine their autobiographical memory, find no cases in
which they were influenced without knowing it (of course!), and reply, "Uh-uh, not
me, Jack."
Nor do I equate free will with the moral or legal concept of responsibility. The
question of free will is an existential one and fair game for philosophers and
scientists; we need not shy away from it as scientists (nor from the implications of
research that suggest we may have less of it than we'd like) because it might have
moral or legal implications we might dislike. At least, I would hope we would not.
(For an excellent treatment of the two senses of "free will" and the confusion of the
societal with the scientific meanings, see Prinz, 1996).
Finally, I strongly object to considering error variance as somehow evidence in
support of the existence of free will. Error variance clearly can come from many
sources, including, it should be said, nonconscious influences that we are not
measuring or have not yet discovered. To credit free will with all or even most of
our unexplained variance it is to commit precisely the kind of conceptual error that
Skinner exposed in the opening quote of the target chapter. A l l of the causes of an
effect that we do not know about we tend to locate within the individual's conscious
choice. In our subjects we call this tendency by such terms as fundamental attribution
error and correspondence bias. I do not see it becoming any less of an error when we
do it ourselves.
99 AND 44
/ 100 % AUTOMATIC
I pushed, and was pushed back in turn (in both directions!) The commentaries
educated and persuaded me in many ways, and I hope the target chapter and this
reply also caused some synapses to grow. Given the quality and expertise of the set
244 Bargh
of contributors the editor was able to recruit for this task, and the extensive, varied,
and thoughtful feedback they provided, this experience was extremely valuable to
me. I am much in the debt of the series editor and the commentators for their
scrutiny and consideration of my ideas and conjectures.
Although the process invites contention—that's why they call it a target chap-
ter—now I hope some rapprochements have been made. Bloodied but unbowed, I
gamely concede that the commentators did push me back from a position of 100%
automaticity—but only to an Ivory ©soap bar degree of purity in my beliefs about
the degree of automaticity in our psychological reactions from moment to moment.
Why still such a high percentage? Because control processes themselves can be
triggered "automatically" in that they do not need to be put in place by an act of
will (Barsalou, 1992; Wegner 6k Bargh, 1997); because the extent to which cultural
and societal norms implicitly direct our perceptions, evaluations, and behavior is
very great, and is in addition to the routinized forms of automatic control argued
for in the target chapter; because we adhere to environments in the moment to
moment present via automatic processes. Although automaticity keeps us tied to
the present, consciousness is floating ahead in time, setting up strategic automatic
contingencies for the future where they might be needed (novel or problematic
situations), to keep us responding fluently and appropriately in that present when
it comes (Bargh & Gollwitzer, 1994).
I believe if one is scrupulously honest about the number of times per day that
one actually takes more than a half-second to make a decision (one signature of a
control or nonautomatic process), the number could be counted on one's fingers.
(Preparing replies to sets of commentaries is excepted from this general rule.) This
is a very small percentage (0.56%?) of all the perceptions, behaviors, judgments,
evaluations and intentions one constantly makes each day. When it does hap-
pen—when we do override the automatic process — these occasions are memorable
and salient precisely because they are effortful and unusual. As a consequence, we
are misled by the greater availability of these occasions in memory into hugely
overestimating how often we really do engage in acts of deliberate control.
ACKNOWLEDGMENTS
Preparation of this reply was supported in part by grant SBR-9409448 from the
National Science Foundation. Correspondence concerning this article can be
addressed to the author at Department of Psychology, New York University, 6
Washington Place, Seventh Floor, New York, N Y 10003.
REFERENCES
Anderson, J. R. (1982). Acquisition of cognitive skill. Psychological Review, 89, 369-406.

Bargh, J. A . (1984). Automatic and conscious processing of social information. In R. S. Wyer & T. K.
Srull (Eds.), Handbook of Social Cognition, (Vol. 3, pp. 1-43). Hillsdale. NJ: Lawrence Erlbaum
Associates.
Bargh,]. A (1989). Conditional automaticity: Varieties of automatic influence in social perception and
cognition. [n]. S. Uleman &]. A Bargh (Eds.), Unintended thought (pp. 3-51). New York: Guilford.
Bargh,]. A (1990). Auto-motives: Preconscious determinants of social interaction. [n E. T. Higgins &
R. M. Sorrentino (Eds.), Handbook of motivation and cognition (Vol. 2, pp. 93-130). New York: Guilford
Press.
Bargh, J. A (1992). Does subliminality per se matter to social psychology? Being unaware of the stimulus
versus unaware of its interpretation. In R. Bomstein & T. Pittman (Eds.) , Perception without awareness
(pp. 236-255). New York: Guilford.
Bargh, J. A. (1994). The four horsemen of automaticity: Awareness, intention, efficiency, and control
in social cognition. [n R. S. Wyer & T. K. Srull (Eds.), Handbook of social cognition (2nd ed., Vol. 1,
pp. 1--40). Hillsdale, N]: Lawrence Erlbaum Associates.
Bargh, J. A, & Barndollar, K. (1996). Automaticity in action: The unconscious as repository of chronic
goals and motives. [n P. M. Gollwitzer &]. A Bargh (Eds.), The psychology of action (pp. 457--481).
New York: Guilford.
Bargh, J. A, Chen, M., & Burrows, L. (1996). Automaticity of social behavior: Direct effects of trait
construct and stereotype activation on action. Journal of Personality and Social Psychology, 71, 230--244.
Bargh,]. A, & Gollwitzer, P. M. (1994). Environmental control of goal-directed action: Automatic and
strategic contingencies between situations and behavior. Nebraska Symposium on Motivation, 41,
71-124.
Bargh, J. A, & Pietromonaco, P. (1982). Automatic information processing and social perception: The
Journal of Personality and Social Psychology, 43, 437--449.
Bargh, J. A, & Thein, R. D. (1985). Individual construct accessibility, person memory, and the
1129-1146.
Bargh,]. A, & Tota, M. E. (1988). Context-dependent automatic processing in depression: Accessibility
54,925-939.
Barker, R. G., & Wright, H. F. (1954). Midwest and its children: The psychological ecology of an American
town. Evanston, IL: Row, Peterson.
Barsalou, L. W (1992). Cognitive psychology: An overview for cognitive scientists. Hillsdale, NJ: Lawrence
Erlbaum Associates.
Bruner, J. S. (1957). On perceptual readiness. Psychological Review, 64, 123-152.
Chaiken, S. (1980). Heuristic versus systematic information processing and the use of source versus
message cues in persuasion. Journal of Personality and Social Psychology, 39, 752-766.
Chartrand, T. L., & Bargh,]. A. (1996). Automatic activation of impression formation and memorization
goals: Nonconscious goal priming reproduces effects of explicit task instructions. Journal of Personality
and Social Psychology. 7I, 464--478.
Fiske, S. T. (1989). Examining the role of intent: Toward understanding its role in stereotyping and
prejudice. In]. S. Uleman &J. A Bargh (Eds.), Unintended thought (pp. 253-283). New York: Guilford.
Foucault, M. (1979). Discipline and punish: The birth of the prison (A Sheridan, Trans.). New York:
Vintage.
Garcia, M. (1996). Automatic evaluation of novel stimuli. Unpublished doctoral dissertation, New York
University.
Giddens, A (1984). The constitution of society: Outline of the theory of stTUCturation. Berkeley: University
of California Press.
Haidt, ]., Koller, S. H., & Dias, M. G. (1993). Affect, culture, and morality, or is it wrong to eat your
dog? Journal of Personality and Social Psychology, 65, 613-{i28.
Higgins, E. T., Rholes, W S., &]ones, C. R. (1977). Category accessibility and impression formation.
Journal of Experimental Social Psychology, 13, 141-154.
Kunda, Z., & Thagard, P. (1996). Forming impressions from stereotypes, traits, and behaviors: A
parallel-constraint-satisfaction theory. Psychological Review, /03,284-308.
246 Bargh
Mischel, W., & Shoda, Y. (1995). A cognitive-affective system theory of personality: Reconceptualizing
the situations, dispositions, dynamics, and invariance in personality structure. Psychological Review,
102, 246-268.
Moskowitz, G . B., Wasel, W., Gollwitzer, R M . , & Schaal, B. (1996). The interaction of chronic egalitarian
motives with automatic stereotype activation. Unpublished manuscript, Princeton University.
Neisser, U . (1967). Cognitive psychology. New York: Appleton-Century-Crofts.
"Not clever enough?" (1996, May 18). The Economist.
Nisbett, R. E., & Cohen, D (1996). Culture of honor: The psychology of violence in the south. Boulder, C O :
Westview Press.
Petty, R. E., & Cacioppo, J. T (1984). The effects of involvement on responses to argument quantity
and quality: Central and peripheral routes to persuasion. Journal of Personality and Social Psychobgy,
46, 69-81.
Prinz, W. (1996). Explaining voluntary action: T h e role of mental content. In P.Machamer & M . Carrier
(Eds.), Phibsophy and the Sciences of the Mind. New York: Cambridge University Press.
& T . K. Srull (Eds.), Handbook of social cognition (2nd ed., Vol. 1, pp. 99-152). Hillsdale, NJ: Lawrence
Erlbaum Associates.
Spelke, E., Hirst, W., & Neisser, U . (1976). Skills of divided attention. Cognition, 4, 215-230.
Srull, T K., & Wyer, R. S., Jr. (1979). T h e role of category accessibility in the interpretation of information
about persons: Some determinants and implications. Journal of Personality and Social Psychobgy, 37,
1660-1672.
7-48.
Wegner, D. M . , & Bargh, J. A . (1997). Control and automaticity in social life. In D. T Gilbert, S. T
Fiske, & G . Lindzey (Eds.), Handbook of social psychobgy (4th ed.). Boston, M A : McGraw-Hill.
Author Index
A 227, 234, 236, 237, 240, 241, 242, 244,

244, 245,246
Abelson, R. E, 100, 103 Barker, R. G . , 240, 245
Abramson, L. Y., 50, 53 Barnden, J. A . , 197, 200
A c h , N . , 41, 53 Barndollar, K., 36, 40, 41, 42, 49, 54, 234, 245
Akimoto, A . A . , 152, 155 Baron, L., 125, 130
Allport, D. A . , 165, 176, 223, 227, 227 Barsalou, L. W., 2,52,55,193,200,206,214,244,
Alwin, D . E , 37, 60 245
Andersen, S. M . , 152, 153 Bartlett, F. C . , 122, 128, 130
Anderson, J. R., 3, 11,29,53, 171, 176,213,214, Baruch, O . , 224, 230
223,227,244 Bator, R. J., 32, 49, 55
Anderson, N . H . , 51, 53, 108, 118 Baumeister, R. F., 81, 231, 233, 235, 238
Arendt, H . , 129, 130 Beegle, W., 220, 228
Armour, V , 147, 154 Begg, I., 147, 154
Arnold, M . B., 110, 118 Bern, D. J., 4, 55, 85, 86, 93, 126, 130
Asanuma, C , 189, 202 Berger, J., 225, 230
Asch, S., 149, 153 Berkowitz, L., 7, 16,55, 8 3 , 8 4 , 8 5 , 8 8 , 8 9 , 9 3 , 9 3 ,
Ashby, F. C . , 160, 161, 162, 174, 178 94, 205,214, 232,239, 241
Ashmore, R. D., 147, 153 Berntson, G . G . , 26, 27, 55, 133, 138, 141, 190,
Atkinson, J. W., 5, 34, 39, 40 201
Atkinson, R. C . , 50, 54,213,214 Berry, J., 122, 131
Berscheid, E., 20, 60
Biernat, M . R., 152, 155
B Bigler, R., 152, 154
Birch, D., 5, 40, 54, 399
Baddeley, A . D., 213, 214 Bisiach, E., 223, 227
Baldwin, M . E., 21, 54 Bjorklund, D. E , 157, 176
Balota, D. A . , 22, 23, 46, 54 Blair, I.V., 71,72, 74
Banaji, M . S., 1, 5, 12, 57, 67, 69, 70, 71, 72, 73, Bless, H . , 94, 145, 149, 155, 193, 201
74, 147, 148, 151, 154, 204, 233, 235, Bobrow, D . G . , 161, 178
239 Bodenhausen, G . V , 15, 18,58, 150,156, 193,201
Bandura, A . , 16, 54, 204,214 Boies, S.J., 159, 163, 178
Bargh, J. A . , 1,2,5,6, 7, 8, 11, 12, 13, 14, 15, 17, Bond, R. R , 14, 22, 35, 46, 49, 54, 150, 154
18, 19, 20, 21, 22, 23, 24, 25, 27, 28, 31, Bonilla, S., 150, 156
33,34,35,36,38,39,40,41,42,44,45, Borgida, E., 151, 155
46,47,49, 50, 51,54,56,57,60,61,67, Bormann, C . , 7, 60
68,69, 70, 71, 72, 74, 75, 76, 77, 78, 79, Bowdle, B. E, 126, 130
80,81,83,84,85,86,87,88,89,90,93,
Bower, G . H . , 94, 115,118,119
95,96,98,100,101,102,105,106,107,
Bowers, K. S., 50, 55,213,214
108, 109, 110, 111, 112, 113, 116, 117,
Boysen, S. T., 138, 141
118, 122, 124, 126, 127, 133, 134, 135,
Bradley, M . M . , 26, 27, 28, 58, 138, 141
136, 137, 138, 139, 140, 143, 144, 145,
Branscombe, N . R., 7, 60, 196, 202
146, 147, 148, 149, 150, 151, 154, 155,
Brehm, J. W., 32, 61
156, 157, 158, 168, 175, 176, 181, 182,
Brendl, C . M . , 184, 186
183, 184, 185, 186, 187, 188, 190, 191,
Brewer, M . B., 12, 14, 18, 45, 55, 114, 118
196, 197, 198, 199, 200, 201, 204, 205,
Broadbend, D. E., 158, 159, 176
206, 207, 208, 209, 210, 211, 212, 213,
Brown, S. C . , 209,215
214, 217, 218, 219, 220, 222, 225, 226,
Bruner, J. S., 13, 14, 55, 212, 214, 239, 245
247
248 Author Index
Bryan, W. L., 167, 176 Creelman, C.D., 221, 229

Bryant, J., 62 Crelia, R. A., 198,201
Bundesen, C , 166, 176 Crick, F.H., 189,202,204,214
Burrows, L, 17, 18, 19, 28, 38, 49, 54, 240, 241, Crites, S. L., 133, 139,141
245 Cuniff, M., 88, 94
Buss, A. H., 203, 214 Cuthbert, B. N., 26, 27, 28, 58, 138, 141
Butterfield, E. C , 213, 214
Buttram, R. T, 145, 156, 196, 202 D
D'Agostino,P.R., 49, 50, 56, 59, 145, 154, 220,
C 228
Damasio, A. R., 114, 119
Cacioppo, J. T., 26,27,55,133,134,135,138,139, Darley, J. M., 2, 4, 56, 64, 74, 135, 141
141,190,201, 231, 235, 236, 237,246 Dasen,P.R., 122, 131
Campbell, D. T., 188, 201 Davies, A., 164, 178
Cantor, N., 20, 59, 183, 185, 186 DeCoster, J., 200, 202
CarlsmithJ. M., 2, 56 Del Boca, F. K., 147, 153
Carlston, D. E, 10, 34, 55 Dennett, D. C , 204,214, 223,228
Carpenter,P.A., 157, 177 Deutsch, D., 159, 176
Carr,T.H.,218,227 Deutsch, J. A., 159, 176
Cartwright, D., 44, 55 Deutsch, M., 203,214
Carver, C. S., 16, 17, 48, 55, 88, 95, 96, 98, 100, Devine,P.G., 5, 7, 14, 15, 18, 19, 56, 69, 74, 78,
J02 81,93,94,123,130,136,141,147,154,
Cave, K. R., 166, 179 105,214, 224,228
Chachere, J. G, 84, 94 Dias, M. G., 239, 245
Chai, A., 40,41,49,54 Dimberg,U.,221,229
Chaiken, S., 5, 7,20,21,22,23,24,25,27,28,31, Donnerstein, E., 85, 94
35,37,46,49,51,54,55,56, 71, 74,86, Dorman, C , 191, 194, 201
94, 108, 109, 118, 134, 141, 146, 148,Dovido, J.E, 70,74,149, 153,154
154,156, 235,245 Draguns, J., 8, 57
Chambers, W., 16, 17, 48, 55 Duan, C , 151, 152, 156
Chappell, M., 200, 201 Dulany, D. E., 223, 224, 225, 228
Chartrand, T. L., 5,33,49,56,208,209,214,234, Dull, V, 18, 55
241,245 Dumais, S. T, 10, 60, 207,215
Chen, J., 49, 56 Dunbar, K., 171, 173, 176
Chen, M., 17, 18, 19, 20, 27, 28, 38, 49, 54, 56, Duncan, J., 161, 176
144,148,154, 240, 241,245 Dunn, J. C , 10, 40, 56
Chen, S., 20, 31, 35, 37, 49, 56 Dunton, B. C , 72, 74, 148, 150,154
Cherry, E. C , 159,176 Durkheim, E., 127,130
Chichetti, R, 23, 139,141
Chomsky, N., 223, 228 E
Churchland,P.S., 189, 191, 194, 201 Eagly,A. H., 5,56, 134, 141
Cialdini, R. B., 32, 35, 37, 49, 55, 56 Edelman, G., 204, 214
Clark, A., 193, 195, 196, 197, 201 Edwards, K.,21,56, 88, 94
Clark, M. S., 115, 119 Ellsworth,P.C , 88, 94
Clore, G. L, 86, 94, 105, 106, 109, 110, 112, 113, Elman, J. L, 189, 197, 201,202
115,118, 119, 146,155, 235, 236, 237 Erber, R., 2, 56
Cohen, D., 125, 126, 129, 130, 131, 239, 246 Erdelyi, M. H.,51,56
Cohen, J. D., 171, 173, 176 Erdley.C. A., 49, 56, 145, 154
Cohen, Y., 166, 178 Esses, V.M., 149,154
Collins, A., 105, 109, 112, 119 Esteves, E, 221,229
Collins, A. M., 23, 44, 56 Etherton, J. L., 225, 229
Collins, M. E., 203, 214 Evans, N., 70, 74
Conway, M., 110, 119
Cook, T. D, 188, 201
F
Cooper, J., 146, 156
Cosmides.L., 10,61, 111, 119
Cowan, W.B.,218,228 Fagot, B., 152,154
Craik.EI., 157, 176 Faludi, S., 129, 130
Author Index 249
Fazio, R. R , 5, 7, 8,20, 21,22, 26,49,56, 71, 72, Greenwald, A . G . , 1, 6, 12, 57, 67, 69, 71, 72, 73,
74, 148, 150,154,193,202, 224,228 74, 148, 151,154, 204,214, 221,228
Fein, S., 153, 155 Griffitt, W., 86, 94
Feinstein, J., 135, 141 Gurtman, M . B., 18, 59
Feldman, S., 183, 185, 186
Felleman, D. J., 192, 201 H
Festinger, L , 2, 56
Haase, S. J., 221, 230
Fischer, G . W., 155
Haddock, G . , 149, 154
Fiske, S. T , 2,5,15,49,51,56,61,187,201,243,
Haidt, J., 239,245
245
Hamilton, D. L , 33, 57, 147, 154, 209, 210, 214
Fitz, D., 92, 94
Hansen, R. D., 5, 57
Fitzgerald, L F, 45, 57
Hardin, C . , 69, 70, 71, 72, 74, 145, 147,148, 149,
Flavell, J. R , 8, 57
151,154,155, 239, 240, 241
Fodor, J. A . , 9, 57
Harford, R. A . , 76, 77, 81
Ford, T. E., 152, 154
Harnisfenger, K. K., 157, 176
Forgas.J. R, 115, 119
Harter, N . , 167, 176
Foucault, M . , 240, 245
Hasher, L , 154, 168, 177,218, 228
Franklin, S., 188, 195, 201
Hashtroudi, S., 146, 155
Franzel, S. L., 166, 179
Hastie, R., 15,34,57, 152, 155, 209
Freud, S., 68, 74, 116, 119
Hayes, A . , 167, 178, 220, 230
Friedrich, F. J., 222, 228
Hayes-Roth, B., 30, 57
Frijda, N . R , 85,94, 106, 119 Heatherton, T. E, 77, 81
Froming,W.J., 16, 17,48,55 Hebb, D. O . , 7, 30, 48,57
Fyock, J., 152, 154 Heckhausen, R , 31, 32, 37, 39, 57
Hefferline, R. E , 76, 77, 81
G Heider, E , 106,119,212,214
Henik, A . , 168,177,214,118,220,221,222,225,
Gaertner, S. L., 70, 74, 149, 153, 154 228,229,230
Galanter, E., 8, 10, 59, 101, 102 Hernstein, R. J., 85, 94
Ganellen, R.J., 16, 17,48, 55 Herr, R M . , 17,57
Garcia, M . , 236, 245 Herscovitch, R, 139, 141
Gardner, W. L., 133, 139, 141, 231, 235, 236 Hicks, R., 166, 177
Gates, R L , 123, 129, 130 Higgins, E. T , 5, 7, 12, 13, 14, 15, 17, 35, 36, 37,
Gaudiano, R, 191, 194, 201 38, 39, 40, 41, 49, 54, 57, 60, 61, 144,
Gazzaniga, M . , 30, 50, 57, 139, 141 145, 146, 150, 154, 155, 184, 186, 195,
Geen, R. G . , 89, 94 196, 197, 198, 200,201,245
Gelade, G . , 163, 165, 166, 178 Hilgard, E. R., 16, 30, 57
George, M . S., 139, 141 Hill, T , 67, 74, 152,155
Gerard, H . B., 65, 74 Hilton, J. L., 152, 153, 155
Gibson, B. D., 152, 155 Hinton, G . E., 189, 202
Giddens, A . , 240,245 Hintzman, D. L , 50, 57
Gilbert, D. T., 5,13,15,51,57,114,119,136,141, Hirst, W., 166,177, 233,246
146, 154, 157, 176, 198,201,220,228 Hitch, G . , 213, 214
Giner-Sorolla, R., 20, 31, 35, 37, 49, 56 Hixon, J. G . , 220, 228
Glass, B., 151, 152, 156 Hodges, S., 20, 58
Gluck, M . , 225, 228 Hodges, S. D., 193, 202
Goldstein, D., 147, 154 Holender, D., 221, 228
Gollwitzer, R M . , 8, 31, 32, 36, 39, 40, 41, 49, 55, Horner, M . S., 37, 58
57,61,234, 240, 244, 245,246 Horwitz, B., 139, 141
Gopher, D., 161, 164, 165, 177 Hovland, C . I., 203, 214
Gorman, T. E , 150, 156 Humphreys, M . S., 196, 200, 201, 202
Gotay, C . C . , 88, 94 Hymes, C , 22, 23, 27, 46, 49, 51, 54, 108, 109,
Govender, R., 20, 21, 22, 25, 27, 28, 54, 71, 74, 118
148,154
Gray, J. A . , 159, 176
I
Green, M . , 42, 54 Isen, A . M . , 115, 119
Greene, R R , 101, 102 Iwata,J.,23,26, 58, 139, 141
250 Author Index
Iyengar, S., 128, 131 Kohler, W., 225, 228

Kolb, B., 139, 141
Koller, S. R , 239, 245
J
Knill, D . S . , 13,57, 157, 176
Kühl, J., 23,58
Jackson, D . R , 37, 58
Kumar, R, 15, 34, 57
Jackson, J. R., 72, 74, 148, 150, 154
Kunda, Z., 242, 245
Jacoby, L. L., 9,42,58, 70, 74,157,177,188,201,
217,218,219, 228
James, W., 16, 58, 167, 169, 177 L
Janis, I. L , 203, 214
Jarvis, B., 135, 141 LaBerge, D., 167, 168, 177
Jaynes.J., 76, 77,81 Lachman, J. L , 213, 214
Johnson, M . K., 9, 23, 58, 155 Lachman, R., 213, 214
Johnson-Laird, P. N . , 106, 119 Laird, J. D.,88, 94
Jones, C . R., 7, 14, 49,57, 144, 145, 155,245 Lang, P. J., 26, 27, 28,58, 138, 141
Jones, E. E., 9, 45, 58, 65, 68, 74 Lashley, K. S., 16, 53, 58
Jordon, M . I., 189, 202 L a t a n é , B., 2, 4, 56, 64, 74, 135,141, 207, 215
Jung, C . G . , 30, 58 Lazarus, R. S., 8, 50, 58, 109, 119, 213, 215
Just, M . A . , 157, 177 LeDoux, J. E., 7, 23, 24, 26, 27,58, 114,119, 139,
141
• Leinbach, M . , 152, 154

Leirer, V. O . , 33, 57, 208, 209, 210, 214
Leiser, D., 218, 230
Kahneman, D., 157, 158, 159, 163, 168, 177
LePage, A . , 16, 55, 88, 94
Kail, R., 157, 177
Lerner, M . , 7, 60, 225, 229
Kalmar, D., 166, 177
Levey, A . B., 225, 229
Kantowitz, B. H., 162, 177
Levine, J. M . , 197,201
Kaplan, B., 8, 61
Levy, B., 219, 228
Kardes, F. R., 7, 8, 20, 21, 22, 25, 56, 71, 74, 224,
Lewicki, R, 67,74, 145, 152, 155
228
Lewin, K., 7, 9, 26, 27, 39, 58, 208, 212, 215
Karniol, R., 5, 50, 58
ü b e n , I., 152, 154
Karp, L , 115, 119
Liberman, A . , 5, 56
Katz, L B., 33, 57, 208, 209, 210,214
Lichtenstein, M . , 34, 58, 60, 209, 215
Kawamoto, A . H . , 189, 202
Lindsay, D. S., 146, 155
Keenan, B., 76, 77, 81
Linville, R, 149, 155
Keil, F., 198, 201
Litt, J., 24, 54
Kelley, C . M . , 70, 74
Locurto, C . M . , 220, 228
Kelley, R H . , 5, 44, 51, 58, 61, 203, 214
Loftus, E. E , 23,44,56
Keltner, D., 88, 94
LoftusJ., 10,58
Kerr,T., 147, 154
Logan, G . D., 157, 158, 167, 169, 171, 172, 177,
Ketelaar, T., 235, 236, 237
179, 198,201, 213, 215, 218, 220, 225,
Ketter, T. A . , 139, 141
228,229, 237, 238, 239,246
Kihlstrom, J. E , 217, 223,228
Lombardi, W. J., 14, 22, 35,36,38,39, 40,46, 49,
Kim, J. K . , 2 3 , 58
54,57, 150,154, 196,200,201
Kinder, D. L , 128, 131
King, G . A . , 13, 57 Lopez, D . F . , 2 1 , 5 4
Kinsbourne, M . , 166, 177 Lorch, R. F., Jr., 22, 23, 44, 54, 58
Kipnis, D., 45, 47, 58 Luevano, M . , 150, 156
Kirsch, J., 119 Luger, G . E , 207, 215
Kirsner, K., 10, 40, 56 Lui, L , 18, 55
Kitayama, S., 130, 131 Lynch, L , 151, 152, 156
Klatsky,R. L , 152, 153
Klein, S. B., 10, 58 M
Klinger,M. R., 221, 228
Klumpp, G . , 149, 155 McClelland, D. C , 37, 50, 59
Klüver, R , 53, 58 McClelland, J. L., 171, 176, 189, 192, 193, 194,
Knowlton, B., 224, 229 198, 200,201,202,1X1,215
Koffka, K., 7, 9, 16, 58, 208,214 McDonel, E. C , 49, 56
Author Index 251
M c D o w d . J . M . , 157, 176 Newell, A . , 3,6,11,28,50,59,171,177,218,229

McGuire, W. J., 203, 211,215 Newman, L. S., 13,51,59,61
MacKay, D. G . , 171,177 Newsom, J. T., 35, 56
M c K o o n , G . , 71,74, 148,155 Newton, L, 65, 74
McLaughlin, J. R, 70,74 Niedenthal, P. M . , 7, 20, 21,51, 59
MacLeod, C . M . , 172, 177 Nisbett, R. E , 1, 9, 45, 50, 58, 59, 121, 122, 126,
Macmillan, N . A . , 229 129, 130, 131, 143, 145, 155, 181, 186,
McNamara, T. P., 148, 155 239,246
McNaughton, B. L , 198, 200, 201 Norman, D. A . , 95, 97, 99, 102, 159, 161, 177,
Macrae, C . N . , 15, 18,58 178, 189,202,229
Malamuth, N . M . , 46, 58, 151, 155
Mandler.G., 108, 119, 223,229
Manis, M . , 152, 155
O
Marcel, A . J., 24, 59, 221,229
Oatley,K., 106, 119
Markus, H . , 193, 201
Ogden, W C . , 168, 178
Markus, H . R., 130, 131
O h m a n , A . , 117, 119, 221, 229
MarshalLGoodell, B. S., 139, 141
O'Reilly, R. C . , 198,201,201
Martin, L, 225, 229
Ornstein, R., 190,202
Martin, L. L., 5, 59, 146, 156, 198, 201
Ortony, A . , 105, 109, 112, 119
Martin, R. C . , 148, 156
Orwell, G . , 129, 131
Martindale, C . , 5, 59
Osgood, C . E , 24, 26, 27, 28,59, 107,119
Masson, M . E , 196, 201
Ovsiankina, M . , 39, 59
Mavin, G . H . , 13, 57
M e l c h n e r . M . J . , 162, 178
Merikle,P.M.,221,229 P
Meyer, D. E , 44, 59
Milgram, S., 2, 59, 64, 74 Paap, K. R., 168, 178
Miller, G . A . , 8, 10, 59, 100, 101, 102, 166, 177 Parekh, R L, 139, 141
Miller, J., 165, 177 Park, B., 34, 57, 152, 155, 209,214
Miller, N . , 91, 94 Pashler, H., 166, 170, 178
Mills, C . J . , 14, 59 Pavelchak, M . , 49, 56
Milne, A . B., 15, 18,58 Peake, R, 184, 186
Mischel, W., 7,9,59,183,184,185,186,208,215, Pelham, B . W . , 13,57, 157, 176
231,232,239, 246 Peng, K., 123, 126, 131
Moray, N . , 159, 163, 177 Pennington, B. F., 224, 230
Morris, M . W., 123, 131 Penrose, R., 204, 215
Moskowitz,G. B.,5,13,51,61,146,156,240,246 Perdue, C . W , 18, 59
Movellan, J. R., 192, 202 Pervin, L. A . , 5, 59
Munro, P. W., 189, 202 Petty, R. E , 134, 135, 139, 141,246
Murphy, S. T , 7, 20, 21,23,25,59, 114,119, 219, Piaget, J., 16, 59
221,222,225,226, 229 Pietromonaco, P., 7, 14, 15,34,55, 157, 176, 209,
Murray, H., 37, 59 214, 245
Murray, J., 152, 153 Pinker, S., 197, 202
Murre, J. M . , 194, 202 Pittman, T. S., 59
Musen, G . , 224, 229 Poortinga, Y., 122, 131
Myers, C . , 225, 228 Porat, Z., 221, 230
Posner, M . I., 21, 53, 60, 135, 141, 158, 163, 166,
N 167, 168,178,213,215,218, 229
Post, R. M . , 141
Powell, M . C , 7, 8, 20, 21, 22, 25,56, 71, 74, 224,
Naveh-Benjamin, M . , 217, 219, 224, 230
228
Navon, D., 161, 164, 165, 166, 177, 224, 229
Powers, W. T , 98, 100, 102
Neely,J. H . , 2 1 , 4 4 , 59,71,74, 222
Pratto.F., 13,14,18, 20,21,22,24,25,27,28,49,
Negroponte, R , 10, 11, 59
54, 60,71,74, 148,154
Neil, W . T , 166, 177
Presser, S., 128, 131
Neisser, U . , 8, 52, 59, 218, 229, 233, 246
Pribram, K. R , 8, 10, 59, 101, 102
Nelson, T. E., 152, 155
Price, R. H.,95, 102
Neuberg,S. L , 17,59
Priester, J. R., 26, 55
Neumann, O . , 218, 220, 224, 229
252 Author Index
Prince, A . , 197, 202 Sanbonmatsu, D. M . , 7, 8, 20, 21, 22, 25, 56, 71,
Prinz, W., 16, 60, 243, 246 74,152,155, 224,228
Pryor, J. B., 4 5 , 4 6 , 4 7 , 5 5 , 6 0 , 1 5 1 , 1 5 4 , 1 5 5 Sanders, I. M . , 128, 131
Sartre, J. R, 80,81
Sasaki, I., 152, 155
Schachter, S., 2, 60
R
Schank, R. C , 100, 103
Scheier, M . E , 95, 96, 98, 100, 102
Rabin, D . E., 189, 202 Scherer,K. R., 112, 119
Rafal, R., 220, 229 Schiller, R, 158,178
Ratcliff.R., 71,74, 148,155 Schmidt, H . , 166,178
Raymond, P, 7, 22, 23, 24, 27, 28, 45, 46, 47, 49,
Schneider, W , 30, 60, 135, 141, 167, 168, 171,
50,51,54,55, 108, 109,118, 151,154
178,213,215,218, 220,229
Razran, G , 86, 94
Schuh, E. S., 221,228
Reaves, A . L , 126,131
Schuman, H., 128, 131
Reed, J. S., 124,131
Schvaneveldt, R. W , 44, 59
Reeves, A . , 163, 178
Schwartz, E , 158, 178
Regan, J. E., 168, 178
Schwartz, R , 110, 117, 118, 120
Reingold, E. M . , 221,229
Schwarz, N., 86,94,105, 110,113, 114, 115,119,
Reiss, D.J., 23,26, 58, 139, 141
126, 130, 145, 146, 149, 155, 156, 193,
Resnick, I. B., 197, 201
200
Reuman, D. A . , 37, 60
Searle, J. R., 204, 215
Rholes, W. S., 7, 14,49,57,144, 145,155,245
Sedikedis, C , 145, 155
Rilling, M . , 2, 60
Segali, M . R , 122, 131
Ringelmann, M . , 203, 215
Seidenberg, M . S., 192, 200, 202
Riskind.J. H . , 8 8 , 9 4
Sejonowski,T.J., 189, 191, 194, 201
Risse, G , 23, 58
Seligman, M . E., 50, 53
Robbins, M . , 115, 119
Seta, J. J., 198, 201
Rock, I., 223,229
Shalker,T.E., 115, 119
Rodin, J., 207, 215
Shallice, T., 53, 60, 99,102,103,229
Rodriguez, M . L , 184, 186
Shastri, L , 197,202
Roenker, D. L., 209,215
Sheerer, D., 31, 35, 56
Roman, R.J., 146, 156
Sheehan, K., 88, 94
Rosch, E.,95, 102
Shelton,J., 148,156
Roseman, I. L , 85, 94
Sherman, S.J., 49, 56, 147, 154
Rosenbloom, P S., 3, 11, 28, 59, 171, 177, 218,
Shiffrin, R. M , 10, 30, 50, 54, 60, 135, 141, 167,
229
168, 171, 178, 207, 213, 214, 215, 218,
Rosenfield, I., 204,215
220, 229
Roskos-Ewoldsen, D. R., 193, 202
Shoda, Y., 183, 184, 185,186, 232, 239, 246
Ross, L D . , 65, 74,121, 122,131, 143,155, 181,
Shulman, D., 88, 94
186
Simon, H. A . , 29,41,50, 52,59,61,96,103, 106,
Ross, M., 1, 5, 50, 58
119,147,156, 239,246
Rothbart, M . , 34, 60, 209,215
Singer, J. A . , 2, 19, 60
Rothman, A . J., 69, 71, 74, 145, 147, 149, 151,
Skinner, B. E , 1, 2, 52, 60, 63, 70, 73, 74
154,155, 239, 240,241
Sloman, S. A . , 197, 199, 202
Ruble, D., 152, 155
Smith, E. R., 3, 5, 6, 7, 11, 12, 28, 60, 145, 152,
Rudman, L. A . , 151, 155
156, 189, 190, 194, 196, 197, 200, 202,
Rueckl, J. G , 192, 193, 202
225, 229, 231, 232, 236, 237, 242, 243,
Rumelhart, D . E., 50,60, 189,192,201,202, 213,
246
215
Smolensky, R, 197, 202, 231
Russell, B., 44, 60
Sneg, R., 224, 230
Snyder, C R., 21,53, 60,135,141, 167, 168, 178,
S 213,215,218,229
Snyder, M . , 20, 60, 203,215
Salovey, R, 19, 60, 149, 155 Solarz, A . , 27, 28, 60
Salthouse, T. A., 157,177,178 Solomons, L. M . , 167, 178
Samuels, S.J., 167, 168, 177 Sommer, K. L , 231, 233, 235, 238
Sorrentino, R. M . , 37, 60
Author Index 253
Spelkc, E , 233, 246 Trope, Y , 114, 120

Sperber, D., 127, 131 Trost, M . R., 35, 56
Sperling, G , 162, 163, 178 Turnbull, G M . , 122, 131
Spielman, L A . , 24, 49, 54, 60 Tyler, R. B., 70, 74
Squire, L , 138, 141 Tyrrell, D. J., 14, 59
Squire, R. L , 224, 229 Tzelgov, J., 217,218,219,221,222,224,225,228,
Srull, T. K., 5, 7,12,13,14,16,18,33,34,39,40, 230, 232,234, 241,242
44,48,52,58,60,61,114,120,144,156,
190, 195, 202, 205, 209, 215, 216, 231,
233, 234, 235, 239,246
Stangor, C , 151, 152, 154, 155, 156
U
Ste-Marie, D.,218, 228
Steele, C . M . , 50, 61 Uleman, J. S., 5, 13, 51, 59, 61, 143, 156
Steibach, K.,219, 228 Ullman, S., 166, 178,219, 230
Stein, G . , 167, 178 Usher, M . , 193, 194, 202
Steller, B., 31, 32, 57

Stewart, T. L , 145, 156, 196, 202
Stone, G . O . , 224, 230 V
Strack, E , 45, 55, 146, 149, 151,154, 155,156
Strauman, T. J., 5, 15,49,61 Vallacher, R. R., 226, 230
Straus, M . A . , 125, 130 Van Essen, D. G , 192, 201
Stroop, J. R., 172, 178,219,229 Van Orden, G G , 224, 230
Strum, G , 88, 94 Velmans, M . , 223, 230
Suci, G . S., 24,59, 107,119 Vera, A . H . , 29, 41, 52, 61, 147, 156, 239, 246
Sunstein, C . R., 128, 131 Veroff, J., 37, 60
Swam, T. S., 85, 94 Vieira, A . , 167, 178, 220, 230
T W
Tanke, E. D., 20, 60 Wanke, M , 193, 201

Tannenbaum, P. H . , 24, 59, 107, 119 Wedderburn, A . A . , 159, 176
Tassinary, L. G . , 139, 141 Wegner, D. M . , 11, 20,58,61, 157,178, 226,230,
Taylor, S. E , 5, 50,51,61 242, 244,246
Teasdale, J., 50, 53 Weldon, M . , 162, 177
Tesser, A . , 5, 59 Welford, A . T , 158, 170
Thagard, R, 242, 245 Werner, H . , 8, 61
Thein, R. D., 13,15,34,46,55,209,214,237,245 West, L.J., 165, 170, 179
Theios, J., 221,230 Whishaw, I. Q., 139, 141
Thibaut, J. W., 44, 61 Whitehead, A . N . , 10, 61
Thompson, C . R, 209,215 Wickens, G D., 164, 170
Thompson, E R, 146,156 Wicklund, R. A . , 32, 39, 61
Thompson, R., 139, 141 Wiest, G , 85, 94
Tice, D. M . , 77, 81 Wilensky,R.,8,61
Tipper, S. R, 166, 178 Wiles, J., 196, 202
Tolman, E. C , 225, 230 Wilkin, N . , 115, 119
Tomkins, S. S., 85, 94 Williams, G J., 72, 74, 148, 150, 154
Tooby,J.,9,61 Williams, R. J., 189, 202
Toppino, T , 147, 154 Wilson, J. Q . , 8 5 , 9 4
Tota, M . E , 3, 5, 15, 22, 35, 46, 49, 54, 55, 150, Wilson, T D . , 50,59,121,131,145,155,193,202
154, 189, 200, 237,245 Winkielman, R R, 114, 117, 118, 120
Toth,J. R, 189,201,218,228 Winter, L , 5, 13, 61
Townsend, J. T , 160, 161, 162, 174, 178 Wolfe, J. M . , 166, 179
Treisman, A . , 159, 163, 164, 165, 166, 167, 177, Wong, F. Y , 84, 94
178,219, 220,230 Wong, N . Y , 126, 131
Triandis, H . G , 123, 131 Wood, W , 84, 94
Triplett, N . , 203, 216 Wright, H . F., 239, 240, 245
Troccoli, B. T , 88, 94
254 Author Index
Wright, J. C , 184, 186 Z

Wurf, E., 193, 201
Wyer, R. S., 5, 7,12,13,14, 16,18,33,39,40,44, Zacks, R . T . , 168, 177,218, 228
48,52,61,114,120,144,150,156,190,
Zajonc, R. B., 4, 7, 20, 21, 23, 24, 25, 51, 59, 61,
195,202,209,215,216, 246
105, 114, 117, 118, 119, 120, 141, 191,
202, 219, 221, 222, 225, 226,229, 230
Zanna, M . R, 146, 149, 154, 156
Y Zarate, M . A . , 145, 150, 152, 156
Zawadski, B., 203, 216
Zbrodoff,N.F.,218, 229
Yehene, V , 217, 219, 224, 225, 230
Yonelinas, A . R, 189, 201 Zbrodoff,N.J., 172, 179
Zillman, D., 61
Zipser, D., 189, 202
Subject Index
A Automatic processing (if-then relations), 128,

188, see also Cognitive processes; C o n -
Accessibility, see Applicability; Trait categoriza- scious mediation; Social behavior
tions cultural behavior theory and, 121-124
Achievement and affiliation priming effects, environment-behavior effects and, 2-6
39-40, 240 freewill and, 181-182
chronic goal interaction and, 36-38, 47 psychological effects and, 2-3, 134
Achievement motivation, 35-36, 38, 191 Awareness, 15, 20, 34, 98, 107, 241, see also Rep-
Affective processes, 95, see also Cognitive proc- resentation
esses; Environment-behavior effects
implicit attributions and, 117-118 B
perception-behavior effects and, 192-193
unconscious processes and, 7, 106, 115—116
Behavior, 76, 95, 107, 108, 238, see oho Behavior-
Agent programming, 10-11
ism; Environment-behavior effects;
Aggression, 15, see also Power
Free will; Perception-tehavior effects;
ideomotor action principle and, 16, 87 Social psychology
perception-behavior effects and, 16-17, 19, default response, 99-100
20, 83-85,88-92, 107 emotions and, 111-113
Applicability, situational influence and, see also evaluation interface, 25-28
Conscious mediation influence of judgments on, 51, 64-65
accessibility of information and, 144, situational influence and, 3, 5, 64, 140,
145-146, 149 144-146
in person judgment and stereotyping, structure of, 100-101
147-153 Behaviorism, 1, 63, 64, 182, 186, 232
judgment/behavior and, 144-146 stimulus control and, 183, 185
learning and, 151-153
Approach/avoidance reactions, see Evaluation
Associationism, 86, 88, 93
c
Attention, 3, 10, 12,13, 99, 109, see also Resource
Categorization, 110, see also Social cognition; Trait
theory
Categorization
Attitudes, 6-9, 27, 110
Cognitive processes, 9, 189, 197, see also Percep-
Attributions, 6, 13, 45
tion-behavior effects; Stimulus control
conscious vs. unconscious, 65, 106-107, 112
issue
implicit, 117-118
affective processes and, 23-25, 50-51,
Attribution theory, 51, see also Conscious media-
191-192
tion
automatic processing and, 2-4, 123, 137,
Automaticity, 81,140,188,243-244, see also Con-
181-185,207
scious mediation; Resource theory
evaluation and, 23-25, 108, 112
as memory, 171-173
mediating, preconscious processes and, 2-8,
development of, conscious processes and,
75, 95, 182-183, 190, 224-225
52-53,98-101,222-223,
Cognitive structure, unconscious, 5, 157-158
226-227
Cognitive theory, 32, 63, 85, 190, 213, see also
learned, see Stereotyping effects
Connectionist models; Emotions
Automatic processes, see Environment-behavior
associationism and, 88, 93
effects; Unconscious automatic proc-
cultural socialization and, 121, 130
esses
social psychology and, 2, 64, 120, 181, 182,
186, 187, 189, 203-204
255
256 Subject Index
Connectionist models, 189-200, 242 Evaluation, 50, 64, 193, see also Cognitive proc-
connections modified by learning in, esses; Stimulus control issue
194-195 affective processes and, 23-24, 110, 113-114
conscious mediation and, 196-199 approach/avoidance reactions and, 12,
Conscious mediation, 217, 231, see also C o n - 26-28,48-49, 138
sciousness; Environment-behavior ef- attitude activation and, 5, 21-23
fects; Evaluation; G o a l and motive behavior interface, via motivation, 25-28
activation; Perception-behavior effects; conscious mediation and, 6, 9-12, 13, 28,
Social cognition; Social psychology 51, 107, 236-238,240
applicability and, 146-147, 150 levels of, 107-108, 112,236-237
attribution theory and, 5, 84-85 priming effects, 7, 21, 24-25, 48, 49, 107,
automatic processing and, 2-5, 121, 110-111, 115, 118
134-136, 135
connectionist models and, 196-199
F
control vs. monitoring and, 225-226
interaction of, 196-199, 242-243
Feedback, 96-97, 225
psychological situation and, 1-3, 6-10, 50,
Feeling, 1, 2, 6, 84, 97, 143-144, 182
75, 76, 223
Free will, 52
social behavior and, 6, 76-77, 84-86,
automatic processing and, 181-182
134-140, 143, 235-236
behavior and, 79, 80-31, 84-85, 187-188
testing of, 188-189
illusion of, 2, 243
Consciousness, 81, 213, see also Conscious proc-
role of, 1, 5-6
esses; Serial stage vs. parallel model
self-determination and, 97-98
automaticity definition and, 218-219
evocative vs. deliberative, 226-227
linearity of, 100-101 G
Conscious processes, see Awareness; Intention;
Free will; Conscious mediation Gender, see Stereotyping
Cultural socialization, 121-130, 239-240 Goal and motive activation, see also Impression
cognitive theory and, 121, 130 formation; Information processing goals;
cross-cultural research and, 124-127 Power
perception-behavior effects and, 122-123 chronic goal interaction and, 36-38, 47
conscious mediation and, 6-12, 29, 52, 75,
E 97-98, 233-234
goal processing and, 42-44
mental representation and, 7-8, 41-42
Emotions, 225, see also Cognitive theory
priming effects, 35, 37-38, 87, 226
cognitive theory and, 106, 108, 109
skill acquisition and, 6, 11-12, 28-29, 99
levels of, behavior and, 111-113
social cognition and, 48-49, 116-117, 138
positive vs. negative, 23-24, 108-110
Goal selection research, 5-6
preconscious processes and, 2, 3, 5-7, 9, 10,
15,20-21,85, 105-106
research, 87-88 I
unconscious processes and, 20-21, 106
Environment-behavior effects, 95, see also Auto- Ideomotor action principle, 16, 87
maticity; Behavior; Psychological situ- Impression formation, 12-13, 33-35, 49, 234
ation; Social behavior Information processing, see Representation; Re-
affective processes and, 7, 48, 84, 86, 128 source theory
automatic processes and, 233-235, 241 Information processing goals
conscious mediation and, 50-53, 64-65, individual differences in, 184-185
75-81,235,242 person perception and, 33, 49, 208-209
environmental representation and, 28-30, persuasive communication motives and,
144, 232-233 31-32, 134-135,235-236
preconscious processes and, 7, 9-11, 28, 83, primed goal states and, 3, 32-34, 39-41, 44,
207, 233-235, 240-241 83
self-regulatory behavior and, 96-98 social, activation/operation of, 33-35

unconscious processes and, 31-36, 208-209
Intention, 10-12, 16-17, 33, 112, 188
Subject Index 257
J as a situational feature, 44-45

Preattentive processes, preconscious processes
Judgments, 48, 64, 67, 76, 86, see also Applicabil- and, 8, 219-220
ity; Behavior resource theory and, 167-168
category use in, 151-152 Preconscious processes, 127, 128, 137, see also
social, 6, 114-115 Cognitive processes; Conscious media-
thought/feeling and, 143-144 tion; Emotions; Preattentive processes;
Stereotyping effects; Unconscious re-
lated topics
L consciousness of output of, 226—227
definition, 187-189
Learning, 2, 76, see also Connectionist models learning and, 224-225
applicability and, 151-153 perceptual processes and, 167-168
connectionist theory and, 194-195 resource theory and, 158-159
preconscious processes and, 224-225
social/nonsocial information and, 8-9, 221
vs. postconscious automaticity, 3, 14, 86-87,
M 222,241
Priming effects, 196, 238, see also Achievement
Memory, 7, 16, 23, 49, 76, 95, 98, 238 and affiliation priming effects; Evalu-
automaticity as, 171-173 ation; Goal and motive activation
declarative/nondeclarative, 224-225 perception-behavior effects and, 13-16, 48
goals, 33-34 trait concept activation, 17-20, 49
Mental representation, 44, 184, 193, see also Goal Psychological situation, see also Conscious media-
and motive activation; Representation tion
Mood, 9, 18, 88, 111, 114, 115, 118 environment-behavior effects and, 6-8,
Motivation, see also Achievement motivation; 232-233
Goal and motive activation; Percep-
tion-behavior effects R
evaluation-behavior interface and, 25-28
states of, 39-41
Representation, 2, 16, 23, 227, see also Mental
representation; Situational repre-
N sentations
environmental, see Environment-behavior
Neuropsychological perspectives, 23, 26-27, effect
138-140, 236 information processing and, 189-190, 192
stimulus, awareness and, 8, 66-67
Resource (attention) theory, 157-176, 197, see
P
also Attention
automaticity and, 169-171, 173-175
Perception-behavior effects, 36, 219, 221, 238,
information processing capacity and,
240, see also Aggression
160-165, 173-175
affective processes and, 192-193
preattentive processing and, 167-168
approach/avoidance motivation and, 48-49
social cognition research and, 237-239
cognitive processes and, 4, 75-76, 117, 182
conscious mediation and, 2, 5-12, 76, 107,
167-168, 182 E
cultural behavior theory and, 122-123
environment-behavior effects and, 12-13, Self-concept, 15-16, 193, 233
15-17, 50-51 Self-determination, 97-98
ideomotor action principle and, 16, 87 Self-regulation, 183-185, 192
perceptual states and, 38-40 Serial stage vs. parallel (modular) model, 6, 9-10,
priming effects and, 13-16 50-53, 72-73, 87, 100-103, 111-112,
stereotype activation and, 18-20 138,139, 162-163, 190-192, 196, 206
Postconscious automaticity, see Preconscious pro- Sexual aggression, see Power
cessing Situational effects, see B e h a v i o r ; E n v i r o n -
ment-behavior interface; Automatic
Power, 63
processing
sex association, automatic, 45-47, 50, 78
258 Subject Index
Situational representations, 29-30, 48 gender, 69, 147, 150-151, 152

Skill acquisition, see Goal and motive activation implicit-automatic, in social microenviron-
Social behavior, 20, see also Conscious mediation; ments, 68-73
Social cognition learned automaticity and, 5-7, 14, 78-79
automatic processing and, 2, 121, 134, 183 preconscious activation, 14-15, 18-20, 220
environment-behavior effects and, 15-16, Stimulus control issue, preconscious processes
64-66, 73, 183 and, 7, 9, 14, 15, 152
Social cognition, 10, 133, 144, 205, 221, see also cognitive processes and, 2, 23-24, 183-185
Goal and motive activation; Resource conscious processes and, 3, 6, 51, 219-222,
theory 226, 241
conscious mediation and, 50-51, 138-139, stimulus evaluation and, 20-21, 27-28, 108,
203-204 115
ecological validity and, 210-211 Stimulus representation, see Awareness
environment-behavior effects and, 8,
48-49,184, 237-239
T
generalization/categorization and, 70-71,
206-208
Thought, 1,2,6, 12,84, 97, 197
models, 5-6, 9
judgment and, 143-144
neuropsychological views, 139-140
processes, 76, 182
Social psychological phenomena, 1,3, 137
Trait categorizations, 6, 7, 33, see also Priming
Social psychology, 1, 83, 187, 231, see also Cogni-
effects
tive theory; Connectionist models
accessibility and, 13-14, 147-148, 195-196
automaticity research and, 3, 6, 63, 134, 140
automatic mental processes and, 63-64
conscious mediation and, 5-6, 185-186 U
experimental, 84-85, 212-213
microenvironments/microbehaviors and, Unconscious attributions, see Attributions
66-68,239 Unconscious automatic processes, 152, 172, 225,
perceiving action at a distance, 64-66 see also Affective processes; Information
redefining, 185-186 processing goals
situational influence and, see Automatic emotions and, 20-21, 106
processing Unconscious cognitive structure, 157-158
Stereotyping effects, 38, 107, 193, 238, see also
Applicability; Trait categorization

The Automaticity of Everyday Life - Advances in Social Cognition (1997)

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Automaticity of Everyday Life - Advances in Social Cognition (1997)

Uploaded by

Copyright:

Available Formats

Advances in Social Cognition,

Advances in Social Cognition,

ROBERT S. WYER, Jr.

\}J ~~I~~~~~i?G~XP Press

Published 2014 by Psychology Press

and by Psychology Press

Psychology Press is an imprint of the Taylor & Francis Group,

Copyright © 1997 by Lawrence Erlbaum Associates, Inc.

All rights reserved. No part of this book may be reprinted or

Trademark notice: Product or corporate names may be

ICover design by Kathryn Houghtaling I

The Automaticity of Everyday Life

ISBN 13: 978-0-805-81699-0 (hbk)

1 The Automaticity of Everyday Life

2 Environments and Unconscious Processes

3 Consciousness, Free Choice, and Automaticity

4 Some Thoughts Extending Bargh's Argument

6 Minding Our Emotions:

7 Ifs and Thens in Cultural Psychology

8 Automaticity and Social Behavior:

9 Rendering Accessible Information Relevant:

10 The Automaticity of Academic Life:

11 Was the Cognitive Revolution Just a Detour

13 The Vicissitudes of Social Behavior and Mental Life

14 Automatic but Conscious:

15 Reply to the Commentaries

Author Index 247

Subject Index 255

T h e potential importance of Bargh's chapter is matched by that of the compan-

— Robert S. Wyer, Jr.

The Automaticity of Everyday Life

(Skinner, 1953, pp. 6-7, 283)

In what follows, I argue that much of everyday life—thinking, feeling, and

The Essential Automaticity of Social Psychological

Bern continued o n to give several examples of studies i n which both behavioral

The Inevitability of Continued Findings

So even for social psychological models of the presumed cognitive mediating

THE PRECONSCIOUS CREATION

FIG. 1.1 Parallel forms of preconscious analysis.

A n individual's cognitive, affective, and motivational reactions to an environ-

The Interface of World and Mind

Aspects of the Psychological Situation

THE DEVELOPMENT OF THE PRECONSCIOUS

Civilization advances by extending the number of important operations which we can

A l t h o u g h Lord Whitehead's claim was made at the grand level of civilization, it

Preconscious Processes as Mental Servants

regularities demonstrated by their human user. A t the Massachusetts Institute for

The best metaphor I can conceive of for a human-computer interface is that of a

T h e present conceptualization of preconscious processes is no different: They

The Preconscious and Skill Acquisition

THE THREE FORMS OF PRECONSCIOUS ANALYSIS

We now turn to a description of the three forms or systems of preconscious analysis

Preconscious effects i n social perception—those effects that are not mediated by

categorization and understanding of a person's behavior, and consequently the

Trait Categorization of Behavioral Information

W h e n the behavioral information is clearly diagnostic of a given trait—when the

THE PERCEPTION-BEHAVIOR INTERFACE

The Principle of Ideomotor Action

The Automatic Effect of Perception on Action

instructions) to aggress against the confederate, i n that they were instructed to

Behavioral Consequences of Trait Concept Activation. We primed par-

Behavioral Consequences of Stereotype Activation. A logical derivation

Experiment 2 of Bargh, C h e n , et al. (1996) investigated the behavioral conse-

\}J I~~~i?G~XP Press