Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

Journal of Experimental Psychology: Copyright 1992 by the American Psychological Association, Inc.

Learning, Memory, and Cognition 0278-7393/92/S3.00


1992, Vol. 18, No. 2, 379-390

Remembering Pictures: Pleasure and Arousal in Memory


Margaret M. Bradley, Mark K. Greenwald, Margaret C. Petry, and Peter J. Lang
University of Florida

Incidental memory performance for pictures that varied along the affective dimensions of
pleasantness and arousal was assessed. For both an immediate and delayed (1 year later) free-
recall task, only the arousal dimension had a stable effect on memory performance: Pictures
rated as highly arousing were remembered better than low-arousal stimuli. This effect was
corroborated in a speeded recognition test, in which high-arousal materials encoded earlier in
the experiment produced faster reaction times than their low-arousal counterparts. Pleasantness
affected reaction time decisions only for pictures not encoded earlier. These results suggest that
whereas both the dimensions of pleasantness and arousal are processed at initial encoding, long-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

term memory performance is mainly affected by arousal.


This document is copyrighted by the American Psychological Association or one of its allied publishers.

It has been demonstrated numerous times that emotional indicating better memory for pleasant materials, is at the heart
stimuli vary along two primary dimensions of affective val- of a series of experiments that corroborate this phenomenon
ence and arousal (e.g., Russell, 1980; Tellegen, 1985; Lang, with verbal stimuli (Matlin & Stang, 1978).
Bradley, & Cuthbert, 1990). Using the semantic differential, Running parallel to reports of improved memory for trau-
Osgood, Suci, and Tannenbaum (1957) showed that the bi- matic or happy stimuli are a number of experiments that have
polar factors of pleasantness and intensity accounted for most explicitly focused on the arousal dimension (see Craik &
of the variability in judgments of affective text. These dimen- Blankstein, 1975, and Eysenck, 1976, for reviews). In these
sions of emotional language map easily onto the behavioral studies, arousal has been variously operationalized as the rated
dimensions of direction (approach or avoidance) and vigor arousal of the stimulus materials, the magnitude of a physio-
(i.e., mobilization) advocated by a biphasic organization of logical index of the subject's arousal level (e.g., electrodermal
emotional responses (Hebb, 1949; Konorski, 1967; Lang et responses), or by the presence of a constant stimulus back-
al., 1990). Dimensional views of emotion are parsimonious ground (e.g., white noise). The general finding is that verbal
in that, rather than assuming independent, specific emotional items associated with higher arousal at encoding result in
states (e.g., fear, anger, and joy), two primary dimensions better memory performance on a later (especially long-term)
define the spectrum of emotional behavior. memory test. Thus, whereas affective memory has been stud-
The questions addressed in the current article exploit this ied at either extreme of the valence dimension (i.e., pleasant
organization of emotion to determine the contribution of the or unpleasant), or along an arousal continuum, a systematic
valence dimension, the arousal dimension, and their interac- exploration of the contribution of each dimension to memory
tion to memory performance. Past research has tended to performance is lacking.
focus on memory for a particular type of emotional event A necessary component of such experimentation is the
(e.g., sad, happy, traumatic) rather than using a dimensional presence of emotionally evocative stimuli that are distributed
analysis. In assessing recall of traumatic events, for example, in the two-dimensional affective space defined by pleasantness
Christianson and Loftus (1987) found a memory advantage and arousal. In the current study, the emotional materials are
for the occurrence of a traumatic situation, compared to a color photographic slides, drawn from the International Af-
neutral one. The purported clarity and stability in memory fective Picture System (IAPS; Lang, Ohman, & Vaitl, 1988).
for traumatic events has led some researchers (Brown & Kulik, This standardized collection of pictures, gathered from a
1977; Bohannon, 1988) to posit a special "flashbulb memory" variety of sources, samples contents across a wide range of
mechanism that veridically records moments of trauma. Al- emotional and semantic categories. Contents include animals,
though receiving somewhat less attention, research on mem- nature scenes, erotica, household objects, expressive human
ory for happy events also produces evidence of increased faces, weapons, mutilated bodies, and others. Affective val-
memorability, relative to neutral. The "Pollyanna" effect, ence and arousal ratings for each slide have been obtained in
previous rating studies, allowing precise placement of these
stimuli in a two-dimensional affective space, which Figure 1
illustrates (see Bradley, Greenwald, & Hamm, in press; Green-
This research was supported by National Institute of Mental Health wald, Cook, & Lang, 1989; Lang, Greenwald, Bradley, &
(NIMH) Grants MH37757, MH41950, and MH43975 to Peter J. Hamm, in press).
Lang.
In addition to assessing judgments of affective experience,
We would like to thank Alfons Hamm for his assistance in data
collection and interpretation in Experiment 2 and Sven-Ake Chris- a number of studies have convincingly demonstrated that
tianson and two anonymous reviewers for their helpful comments. emotional responses to these materials—measured by psycho-
Correspondence concerning this article should be addressed to physiological and behavioral reactions—reliably covary with
Margaret M. Bradley, Box 100165, Health Science Center, University the dimensions of valence and arousal (Bradley et al., in press;
of Florida, Gainesville, Florida 32610. Greenwald et al., 1989; Lang et al., in press). Facial electro-
380 BRADLEY, GREENWALD, PETRY, AND LANG

Pleasant

5 10 15 20 25

Pleasantness Rating
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Unpleasant

Arousal rating

Figure 1. The left panel illustrates the distribution of 240 rated slides (currently composing the
International Affective Picture System; Lang et al., 1988) in the two-dimensional space formed by mean
ratings of pleasantness and arousal; the right top panel illustrates facial corrugator electromyographic
responses (defined as the average change over a 6-s slide-viewing interval from a 1 -s baseline immediately
preceding slide onset) to slides ranked by pleasantness, whereas the right bottom panel depicts mean
skin conductance responses to slides ranked by rated arousal (Greenwald et al., 1989).

myographic responses, particularly the corrugator ("frown"; Lang, 1990; Hamm, Stark, & Vaitl, 1990; Lang et al., in
see Figure 1) and zygomatic ("smile") muscles, reliably vary press). In the experiments reported here, memory perform-
with changes in pleasantness ratings. As the pleasantness of ance for these pictorial materials is assessed using an imme-
the slide decreases, corrugator activity increases and zygo- diate free-recall task, a delayed free-recall task, or a speeded-
matic activity decreases, indexing the affective valence dimen- recognition task. Incidental memory performance is always
sion. The size of the reflexive eyeblink to a startle probe (e.g., assessed after an initial encoding phase in which the subject
a loud noise burst) presented during viewing of these slide rates each slide on the dimensions of pleasantness and arousal.
materials is also sensitive to the affective valence dimension: Predictions for memory performance are motivated by past
The magnitude of the reflexive eyeblink increases as the research and are relatively straightforward. To the extent that
unpleasantness of the slide foreground material increases the affective valence dimension differentially affects memory
(Langetal., 1990). performance, a main effect for the pleasantness of the pictorial
Whereas these somatic responses are specifically tied to stimulus should be obtained. If memory processes favor neg-
differences in the rated pleasantness of the slides, skin con- ative events, as some data indicate, free recall as well as
ductance responses reliably covary with changes in rated recognition speed should be facilitated for these materials. On
arousal. As judgments of arousal increase, the magnitude of the other hand, if the Pollyanna hypothesis is correct, better
the skin conductance response to the slide increases as well, memory performance should be obtained for pleasant stimuli
as Figure 1 convincingly demonstrates. The arousal dimen- than for unpleasant materials.
sion also appears to be related to differences in interest or Because the arousal dimension has been associated with
attention, because factor analysis (Lang et al., in press) has improved memory performance in the past, an effect of
indicated that interest ratings and the duration of time a arousal is predicted such that increases in arousal will be
subject chooses to view the slide relate to the same factor as associated with increased free-recall performance for these
the magnitude of electrodermal reactivity. materials and faster recognition speed. To the extent that
Importantly, these studies validate the dimensional relation arousal is the only salient dimension underlying memory
in affective picture processing using a range of dependent performance, this effect should occur for both pleasant and
measures. In addition, the same patterns in physiological unpleasant materials.
response have been replicated using different subsets of IAPS If, as some data seem to suggest, traumatic (i.e., unpleasant
materials, which implies freedom in sampling from this stim- and highly arousing) events are different from all others, an
ulus collection (Greenwald et al., 1989; Bradley, Cuthbert, & interaction of valence and arousal is expected in both free
PLEASURE AND AROUSAL IN MEMORY 381

recall and recognition. In this case, negative, highly arousing were selected from IAPS (Lang et al., 1988). The Appendix contains
events should produce superior recall and faster recognition a complete listing of the pictorial stimuli used in this study, indexed
speed than slides in any other quadrant of the two-dimen- by the IAPS number and accompanied by a brief semantic descrip-
sional emotional space. tion.
Each slide was rated on the dimensions of valence, arousal, and
dominance using a pencil-and-paper version of the Self-Assessment
Experiment 1 Manikin (SAM) rating system (Lang, 1980; Hodes, Cook, & Lang,
1985). SAM is designed to represent the range of the subject's emo-
tional response to a stimulus, and as Figure 2 illustrates, the valence
In Experiment 1, a free-recall task was used to assess and arousal dimensions were ordinally scaled withfivefiguresrepre-
memory performance. After rating a series of 60 slides on the senting each dimension. Ratings were made by placing an X on or
dimensions of valence and arousal, an incidental free-recall between any of the figures, producing a scale that ranged from 1 to
test was conducted both immediately and 1 year after the 9. The valence dimension depicted a figure that ranged from happy
rating session. This experiment allows an assessment of the to unhappy. The corresponding SAM figures ranged from smiling
effects of valence and arousal on both short-term and long- with raised eyebrows to frowning with knitted eyebrows. The arousal
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

term memory performance. dimension ranged from excited to calm. The corresponding SAM
This document is copyrighted by the American Psychological Association or one of its allied publishers.

figures ranged from having an active body and eyes wide open to
having an inactive body and closed eyes. A third SAM panel (not
Method pictured in Figure 2) required the subjects to rate their feelings of
dominance in reaction to the slide, with the SAMfigureranging from
Subjects. Eighty-nine University of Florida undergraduates (48 small in size (controlled) to large (in control); this dimension is highly
women, 41 men) participated voluntarily to fulfill a course require- correlated with pleasantness for static pictorial stimuli and is not
ment. Each subject viewed one of four randomized slide orders. The central to the questions addressed here.
data from 2 male subjects who did not provide clear descriptions of Each of the three dimensions was displayed on each page in the
recalled slides in the immediate memory test were excluded from ratings booklet; the order of the SAM dimensions on each sheet was
analysis (final n = 87). 'randomized, whereas the booklets were identical for all subjects. Each
In the delayed free-recall portion of the experiment, 59 (30 women, slide was rated on all three dimensions. Because relatively fewer
29 men) of the 87 subjects were successfully contacted by phone 1 ratings occurred at the nongraphic indices of the rating scale (i.e.,
year later. Two subjects (1 woman and 1 man) remembered partici- between the figures), the 9-point scale was transformed postexperi-
pating in the experiment but could not recall any of the slides viewed. mentally into a 5-point scale by using 5 as the neutral rating and
The data from an additional 3 males were not used because specific collapsing ratings 1 and 2 into a single rating, and similarly, 3 and 4,
descriptions of slides could not be provided. This filtering resulted in 6 and 7, and 8 and 9.
a total of 54 subjects for use in the long-term memory analyses. Rating procedure. The study was conducted in groups that ranged
Materials. Sixty color, photographic slides depicting various con- in size from 7 to 14 subjects, roughly balanced for sex. On arrival,
tents were selected to sample a wide range in emotional space. Five subjects were seated approximately 2.2 m to 3.6 m from the projection
additional slides were selected to serve as practice slides. All slides screen; the size of the slide image projected onto the projection screen

-• n - - - - -

_J 1 —i —1

Figure 2. Paper-and-pencil version of the Self-Assessment Manikin (SAM; Lang, 1980) used to rate
affective valence (top row) and arousal (bottom row) of the slide stimuli for Experiment 1.
382 BRADLEY, GREENWALD, PETRY, AND LANG

was approximately 1.2 m x 1.8 m. After obtaining informed consent, (i.e., a black snake) and one of a white and gray snake, and (b) two
the instructions for the experiment were read. pictures of guns, one aimed at the viewer, one aimed obliquely. To
Each trial consisted of a sequence of 3 projected slides. First, the the extent that the subject recalled both members of the category,
subject viewed a preparation slide for 5 s. This slide was used to there was no problem in assigning a correct recall for both candidates.
instruct the subject to turn to the correct page in the ratings booklet If only one was remembered (e.g., snake), recall was assigned to the
and then to focus attention on the projection screen (actual slide stimulus most frequently mentioned by this sample of subjects. This
instruction: "Rate the next slide on page "). The second slide occurred rarely; in any case, both members of these categories tended
in the sequence was the to-be-rated slide and was projected for 6 s. to be rated similarly in valence and arousal level, which is the main
The subject was instructed to view the slide for the entire time that it focus of this investigation.
was on the projection screen. After the offset of this slide, a third slide
in this sequence was projected for 15 s, which instructed the subject
to "Please rate the slide on all three dimensions." During this rating
Results
period, each subject was instructed to rate his or her actual emotional Rating information. The Appendix lists mean affective
reaction during viewing of the to-be-rated slide. Projection times were
valence and arousal ratings for each slide in this study; the
electronically controlled by a prerecorded synchronization tape
nature of the resulting affective space is highly similar to that
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

played on a Singer Caramate carousal slide projector. After these


This document is copyrighted by the American Psychological Association or one of its allied publishers.

instructions, 5 practice slides were presented; then, the 60 experimen- obtained in previous studies (see Figure 1). Table 1 lists the
tal slides were presented and rated. No mention of a subsequent mean proportion of slides rates at each of the five positions
memory test was given at this time. along the valence and arousal dimensions. To adjust for
Immediate-recall procedure. Following the rating procedure, each differences in recall that are produced by the number of slides
subject was given an incidental free-recall test, in which he or she was rated at each position on the scale, the number of slides
instructed to write down, in any order, a word or phrase describing recalled at each level of valence (or arousal) was divided by
each experimental slide that could be remembered (e.g., cow, baby, the number of slides rated at that level for each subject. These
or couple kissing). The subject was told to provide a clear description data were subjected to an analysis of variance in which gender
so that the recalled slide could be identified; the free-recall period was (male or female) of the subject was a between-subjects vari-
5 min. At the conclusion of this recall period, the subject was
able, rating level was a within-subject variable, and the pro-
debriefed, paid (with course credit), and thanked.
portion of slides recalled at each of the five positions along
Delayed-recall procedure. Approximately 1 year after participat
ing in the slide rating and immediate-recall experiment, each subject the scale (i.e., valence or arousal) was the dependent measure.
who could be contacted by telephone was given an incidental free - Unless otherwise noted, a significance level of .05 was used.
recall test, in which he or she was instructed to again recall, in any Immediate memory. Memory performance was first as-
order, a word or phrase describing each slide that could be remem- sessed as a function of the rated arousal of the slide materials
bered from the slide-viewing procedure the year before. The subject (see Table 1). Level of arousal had a significant effect on
was instructed to continue recalling slides for a 3-min period. At the immediate-recall performance, F(4, 340) = 20.39, MSe =
conclusion of this recall period, the subject was debriefed and thanked .036. Newman-Keuls post hoc comparisons indicated that
again.
slides rated as highly arousing were remembered better than
Scoring. The recall sheets were scored on the basis of the subject's
slides rated in all lower categories. In addition, slides rated as
description of the slide; the same criteria applied to scoring immedi-
ate- and free-recall data, and the same person scored both sets of moderately arousing were remembered more often than those
data. Correct recall was scored if the description of a particular slide in the neutral category.
was able to be clearly linked to a slide that had been shown. In almost A second analysis showed that the pleasantness of the slide
all instances, the descriptions were completely clear. The main stimuli also had a significant effect on the slides recalled, F(4, 340) =
subject to confusion were (a) two pictures of snakes, one of a cobra 14.39, MSe = .033. A significant quadratic trend indicated

Table 1
Proportion of Slides Rated and Recalled at Each of the Five Levels of Valence
or Arousal Ratings
Arousal dimension
Highly Moderately Moderately Highly
unarousing unarousing Neutral arousing arousing
Proportion rated .18 .18 .25 .27 .12
Immediate recall .46 .45 .41 .49 .65
Delayed recall .07 .07 .07 .09 .20

Valence dimension
Highly Moderately Moderately Highly
unpleasant unpleasant Neutral pleasant pleasant
Proportion rated .15 .18 .29 .21 .17
Immediate recall .56 .41 .42 .46 .55
Delayed recail .13 .06 .06 .06 .10
Mean arousal rating 6.22 5.01 3.51 4.77 6.04
(1-9 scale)
PLEASURE AND AROUSAL IN MEMORY 383

that memory for both pleasant and unpleasant slides was .55 1
Immediate Recall
greater than that for neutral slides, F(l, 85) = 52.61, MSe =
.03 (see Table 1). Newman-Keuls comparisons corroborated
O
this, showing that slides rated as highly pleasant and highly 1 .50
unpleasant were remembered better than those in the middle
categories. However, there was no specific advantage in recall • PLEASANT
for either highly pleasant or highly unpleasant materials, as .45
• UNPLEASANT
these did not differ significantly from each other.
Table 1 lists the mean arousal ratings for slides in each of
the five valence categories. From these data it is clear that the .40
level of rated arousal differs as a function of pleasantness,
^(4, 344) = 62.69, MSC = 1.65, and may be mediating the Delayed recall
.15
effect of valence on memory. This hypothesis is supported by
a significant quadratic trend for these ratings across levels of
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

affective valence, F(l, 86) = 234, MSt = 1.58, demonstrating


This document is copyrighted by the American Psychological Association or one of its allied publishers.

10
that both highly pleasant and highly unpleasant slides pro-
duced high-arousal ratings (as well as better memory perform-
ance). 05
To determine whether there was any contribution of pleas-
antness to memory performance, while controlling for inten-
sity, the following data set was constructed. First, the 60 slides
were rank ordered by valence ratings for each subject. The Low High
ranks for slides given identical valence ratings were decided Arousal
on the basis of the group valence means. The top 30 slides Figure 3. Mean proportion of slides recalled at immediate (top
were assigned to the pleasant category; the bottom 30 slides panel) and delayed (bottom panel) recall as a function of the rated
were assigned to the unpleasant category. Within each valence pleasantness and arousal category of the slide stimuli.
category, slides were then ranked by their arousal ratings. Ties
in this ranking were decided on the basis of the group arousal
means. The top 15 slides in each of the two valence categories arousal ratings for these materials. The interaction of valence
were designated high-arousal slides; the bottom 15 were des- and arousal was clearly not significant, and no main effects
ignated low-arousal slides. This procedure produced, for each or interaction with the subject's gender approached signifi-
subject, a mean recall score for four categories of slides, cance.
defined by the combination of valence (pleasant or unpleas- Long-term memory. The proportion of slides recalled in
ant) and arousal (high or low), with 15 slides in each category. the long-term memory test are presented in Table 1. The
The mean arousal ratings for slides assigned to these four pattern of recall 1 year later is remarkably similar to the effects
categories were first analyzed to determine if this procedure obtained on the immediate memory test. In the dimensional
was successful in controlling arousal across valence categories. analyses, the main effect of arousal was again significant, F(4,
As expected, although there was a large difference in mean 208) = 11.69, MSC = 016. Newman-Keuls comparisons
arousal ratings for slides in the low- and high-arousal cate- indicated that slides rated as highly arousing were still remem-
gories, F(l, 86) = 539.59, MSe = .72, pleasant and unpleasant bered better than slides given lower arousal ratings 1 year
slides were not generally different in mean level of arousal (F later. Similarly, the effect of affective valence was again sig-
< 1). A significant interaction between slide pleasantness and nificant, F(4, 208) = 5.52, MSe = .012, as was the quadratic
arousal level, F(\, 86) = 22.20, MSC = .38, however, indicated trend for this dimension, F(l,52) = 14.55, MSC = .015.
that although pleasant and unpleasant slides in the high- Once again, to assess any influence of pleasantness without
arousal category were not significantly different in rated the contributing effect of arousal, a within-subject recall score
arousal (M = 5.83 and 6.06, respectively), slides composing for the 2 x 2 factorial combination was derived as described
the low-arousal category showed a small but significant effect above. These data can be seen in Figure 3 (bottom panel).
of valence, F{\, 86) = 9.19, MS* = .71. Pleasant slides, The significant main effect of arousal indicates that highly
however, were rated as only slightly more arousing (M = 4.01) arousing slides were recalled better than low-arousal slides,
than unpleasant slides (M = 3.6). even after a retention interval of 1 year, F( 1,52) = 7.44, MSt
The mean proportions of slides recalled as a function of = .005. The main effect of valence and the interaction of
these valence and arousal groupings are depicted in Figure 3 valence and arousal were not close to significance (both Fs <
(top panel). The arousal level of the slide significantly affected 1), indicating that the slight advantage in memory for pleasant
recall, with memory for high-arousal slides better than mem- material obtained on an immediate test was lost 1 year later.
ory for low-arousal slides, F(l, 85) = 31.97, MSe = .014. Memory for specific contents. Slides were grouped, post
There was a marginally significant effect of valence, indicating hoc, into nine semantic categories, and immediate free recall
that pleasant slides were remembered slightly better than as a function of semantic category was investigated. The
unpleasant slides, F(l, 85) = 3.87, p = .053, MSt = .019, categories and the number of slides occurring in each category
which is somewhat consistent with the slight difference in were as follows: attractive females (« = 3), attractive males (n
384 BRADLEY, GREENWALD, PETRY, AND LANG

= 3), mutilated bodies (n = 4), weapons (n = 5), household and delayed recall. Superior memory performance was con-
objects (n = 5), food (n = 4), animals (n = 7), nature scenes sistently obtained for slides rated as highly arousing in com-
(« = 9), and sports activities (n = 5). parison with slides that received lower arousal ratings. These
Table 2 lists the mean proportion of slides recalled in each data suggest that high arousal facilitates memory perform-
category for each gender. Not surprisingly, the mean propor- ance, at least when memory for the event's occurrence is
tion of slides recalled differed as a function of semantic probed. In Experiment 2, a speeded-recognition task was used
category. For both males and females, attractive members of to assess memory for these affective materials. To the extent
the opposite sex (i.e., female nudes and male nudes) and that the arousal effect obtained in Experiment 1 reflects the
mutilated bodies composed the top three categories recalled. operation of a general memory mechanism, and not one
More interestingly, it appeared that some categories recalled specific to the free-recall process, similar effects should be
were related to the subject's gender rather than strictly related obtained in Experiment 2. Based on the facilitatory effects of
to the arousal level of the category. arousal on free recall performance in Experiment 1, the
For males, mean recall averaged across the categories of prediction is that high arousal should aid recognition perform-
weapons, mutilated bodies, and attractive women was higher ance: As rated arousal of the material increases, recognition
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

than recall of these materials by women, F(l, 85) = 7.98, MSC speed should decrease, indicating faster recognition decisions.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

= .057, even though mutilations were rated significantly In Experiment 1, the affective valence of the slide had a
higher in arousal by women (M = 7.08) than men (M =6.13), small, marginally significant effect at immediate recall, but
F{\, 85) = 6.19, MS, = 3.13. On the other hand, the mean no effect on delayed recall. To the extent that the 15-min
arousal ratings for weapons were similar for both males (M = retention interval used in Experiment 2 is considered a long-
5.83) and females (M = 5.96). Whereas these differences in term memory test, no effects of affective valence should be
memory cannot be attributed to rated arousal, better recall of seen in the recognition reaction time for slides encoded earlier.
attractive females is consistent with the significantly higher The magnitude of the skin conductance response at encod-
arousal ratings assigned to these materials by men (M =6.91) ing indexes affective responses, especially arousal, during
in relation to women (M = 3.60), F(l, 85) = 80.21, MSe = processing of these pictorial materials. Measurement of these
2.92. sympathetically mediated responses at encoding allows an
For women, mean recall averaged across the categories of additional assessment of emotional engagement during slide
household objects, food items, and attractive men was higher processing (i.e., in addition to verbal ratings), and in Experi-
than for men, F(\, 85) = 4.61, MSe = .092. The memory bias ment 2, these responses were measured during the initial
for household objects and foods was not due to differences in encoding phase.
rated arousal, as both men and women produced practically
identical arousal ratings for these categories. Again, however, Method
the mean arousal rating for attractive males was higher for
women (M = 5.62) than for men (M = 2.99), which satisfac- Subjects. Sixty-six (33 women and 33 men) University of Florida
introductory psychology students participated as part of a class re-
torily accounts for the pattern of recall in this category, F{ 1, quirement. Subjects were randomly assigned to view one of the two
85) = 60.57, MS, = 2.45. slide series (described below). Gender was balanced across slide series.
Across the remaining categories of sports, nature scenes, One female lacking complete data was omitted from thefinalanalyses
and animals, males and females showed no differences in (final n = 65).
either proportion recalled or in mean ratings of arousal. Stimulus materials. Two sets of 21 colored photographic slides
were chosen from the IAPS (Lang et al., 1988) on the basis of affective
(valence and arousal) ratings obtained from independent subject
Experiment 2 samples. The two stimulus sets were matched for valence and arousal
ratings and (as closely as possible) semantic content. Contents of both
In Experiment 1, the rated arousal level of a slide was sets were broadly distributed in the two-dimensional emotional space.
clearly related to memory performance at both immediate One set of 21 slides served as the encoding stimuli for each subject
and therefore were the "repeated" slides (correct response = yes) in
the recognition phase. The remaining set of 21 slides then served as
Table 2 "new" foils (correct response = no) in the recognition task. The
Mean Proportion of Slides Recalled in Each of Nine particular set of slides serving as encoding stimuli was counterbal-
Semantic Categories for Men, Women, anced across subjects.
and All Subjects Apparatus and response measurement. Presentation of stimuli
Slide category Men Women All subjects and collection of data were controlled by a Digital Equipment Cor-
poration model 11/23 and an Apple He microcomputer. Slides were
Mutilated bodies .76 .68 .71 projected through a sound-attenuating port into the subject room
Weapons .51 .38 .44
.72 with a Kodak Ektagraphic HI projector. Slide duration (6 s) was
Female nudes .75 .69
controlled using an electronic shutter (rise/fall time < 5 ms). Valence
Household objects .35 .43 .40 and arousal ratings were obtained using a computerized version of
Food .29 .41 .36 SAM (Lang, 1980). The SAM display was presented on a video screen
Male nudes .58 .63 .61 located 1.6 m in front of the subject's eyes. Subjects adjusted each of
three sequential displays (representing the valence, arousal, and dom-
Animals .52 .53 .52 inance dimensions, respectively) along a 0-29 scale using a poten-
Sports events .35 .35 .35 tiometer. The order in which the ratings were displayed was randomly
Nature scenes .34 .38 .36 determined on each trial for each subject.
PLEASURE AND AROUSAL IN MEMORY 385

Skin conductance was sampled during the encoding phase of the the level of arousal across the two valence categories. Because
experiment. Skin conductance electrodes were placed adjacently on slides were assigned to the low- and high-arousal categories
the hypothenar eminence of the left palmar surface, using Beckman on the basis of these ratings, the significant difference in mean
Ag-AgCl standard electrodes filled with K-Y lubricating jelly. The arousal ratings for slides in these categories is expected, F(l,
signal was calibrated prior to each session to detect activity in the 62) = 315.6, MSt = 15.68. More important, there were no
range from 0 to 40 pS. Electrodermal activity was sampled at 10 Hz
significant effects involving affective valence in the analysis
for 2 s prior to slide onset, throughout the 6-s slide exposure, and for
2 s following slide offset. A skin conductance response (SCR) was of arousal ratings, indicating that neither pleasant and un-
defined as the largest response occurring between 0.9 and 4 s after pleasant slides in the low-arousal (M = 11.6 and 10.8, respec-
slide onset; a log transformation of logCX^ + 1) was performed to tively) nor the pleasant and unpleasant high-arousal (M =
normalize the data. Detailed description of additional indices of 20.2 and 19.9, respectively) categories were judged differen-
physiological response collected in this experiment are reported in tially arousing.
Lang et al. (in press). Reaction time. Only correct trials were used in the analysis
Encoding procedure. The subject sat in a recliner in a small, of reaction time data. In an overall analysis of variance of
sound-attenuated, dimly lit room. The slide image was projected these data, gender (male or female) was a between-subjects
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

about 1.9 m from the subject's eyes and subtended 20° of visual angle. variable, and slide status (repeated or new), slide affective
This document is copyrighted by the American Psychological Association or one of its allied publishers.

The subject was told he or she would view a series of slides differing
valence (pleasant or unpleasant), and slide arousal (high or
in emotional content. The subject was asked to attend to the slide for
the entire exposure time (6 s), to perform SAM ratings after each low) were repeated measures. There were no significant effects
trial, and to relax during varying interslide intervals (i.e., 25-35 s). involving gender in any of these analyses. Figure 4 illustrates
Each subject viewed the slide series in one of three varied orders. the mean reaction time (RT) data for new and repeated slides
After instructing the subject to rest quietly, the experimenter exited as a function of the affective valence and arousal level of the
the subject room, and the slide sequence was presented. No mention slide category.
of a memory test was given at this time. For the RT data, there was a marginally significant main
Recognition phase. After a 15-min retention interval (during effect of slide status, F(l, 58) = 3.83, p = .06, MS, = 596,132,
which the subject completed several unrelated questionnaires), in- indicating the mean RT for repeated slides (yes responses)
structions for the incidental recognition phase were given. Two game was slightly faster than for new slides (no responses). This
paddles were affixed (using Velcro) on the armrests of the subject's
effect could arise because these responses were executed with
chair, and the subject was instructed that another series of slides
would be viewed and memory for each slide assessed. the subject's dominant hand or, more probably, due to savings
For each recognition trial, a slide was presented on the screen, and in encoding time (i.e., these slides were repeated).
the subject was told to decide as quickly as possible whether the slide More important, significant interaction effects indicated
had been seen earlier in the experiment. If the slide had been seen that the valence and arousal level of the slide had different
before (repeated slide), the subject pressed the button associated with effects on RT, depending on whether the slide was repeated
the dominant hand. If the subject had not seen the slide previously or new. A significant interaction between slide status and
(new slide), the button associated with the nondominant hand was arousal level, F(l, 58) = 31.03, MSe = 64,958, indicated that
pressed. Both speed and accuracy were emphasized in the instructions. slide arousal had opposing effects on recognition speed: For
After making the recognition decision, the subject was allowed to slides that were seen before, RT was facilitated by high arousal,
view the slide as long as desired. A simultaneous press on both game
F(l, 58) = 9.72, MSC = 58,215, whereas for new slides, RT
paddles terminated the slide presentation. The subject then made an
interest rating and an emotion rating (a forced choice selection of one was delayed by high arousal, F(\, 58) = 22.01, MSe = 71,625.
of seven emotions) for each slide (for discussion of these ratings, see Similarly, a significant main effect for slide valence, F(\,
Lang et al., in press). 58) = 6.01, MSt = 95,550, was modified by a significant
Data reduction. Four slides were used in which two identical faces
occurred with different facial expressions (e.g., angry face/neutral
face). Subsequently, it was determined that these slides were difficult Repeated Slides New Slides
to categorize as new or repeated and resulted in extremely long "yes' response no' responses
reaction time responses. These items were therefore excluded from
further analyses.
Individual subject ratings for affective valence and arousal were
obtained for slides seen in the encoding phase only (i.e., repeated • Unpleasant
slides). To compare new and repeated slides, as well as to compare • Pleasant
these data directly to the recall data from Experiment 1, the recog-
nition data were assigned to a 2 x 2 category breakdown formed by
factorially combining valence (pleasant or unpleasant) and arousal
level (high or low) on the basis of group means. The slides were first
rank ordered by the group valence means; the top 9 and bottom 10
slides were designated pleasant and unpleasant slides, respectively.
Within each valence category, the slides were ranked by the group
arousal means and then divided into high- and low-arousal categories.

AROUSAL AROUSAL
Results
Figure 4. Mean reaction times for recognizing repeated slides (left
Rating information. The mean arousal ratings for slides panel) or new slides (right panel) as a function of slide pleasantness
assigned to the four affective categories were analyzed to assess and arousal in a speeded-recognition task.
386 BRADLEY, GREENWALD, PETRY, AND LANG

interaction wiih slide status, F(\, 58) = 4.00, MSe = 116,158. Skin conductance at encoding
Unpleasant slides resulted in longer RT than pleasant slides,
but only when the slide was new, F(\, 58) = 6.90, MSe =
150,122. If the slide had been previously presented in the
experiment (during the encoding phase), the valence of the
slide had no effect on retrieval speed, F(l, 58) < 1.
Accuracy. The overall error rate in this experiment was
low (.04). However, in an analysis of the mean errors made
for each of the slide categories, a main effect for gender, F( 1,
58) = 4.08, MSC = .01, indicated that women were slightly
more error-prone than men (mean error rates of .05 and .03, o.i
respectively). A significant main effect of arousal, F{\, 58) =
5.81, MS; = .009, indicated that slightly, but significantly,
more errors were made on slides categorized as low-arousal Figure 5. Mean skin conductance responses during encoding for
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

(M = .05) in relation to high-arousal slides (.03). The arousal slides categorized by pleasantness and arousal level.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

variable did not interact with slide status, F(l, 58) < 1, which
indicates that high arousal was associated with fewer recog- lus arousal on memory performance (Craik & Blankstein,
nition errors, regardless of slide status. Because high arousal 1975). For instance, Maltzman, Kantor, and Langdon (1966)
had differential effects on reaction time as a function of found greater recall for high-arousal words, in comparison
whether a slide was new or repeated, the lack of an interaction with low-arousal words, on both an immediate memory test
in error rate suggests that a trade-off of speed with accuracy and a delayed test administered 30 min later. Importantly,
is unlikely for the effects of arousal obtained here. the retention interval in their study was a between-subjects
Encoding and skin conductance responses. Skin conduct- variable, indicating better long-term memory performance for
ance responses were measured during the initial presentation high-arousal items, in the absence of an earlier memory test.
of each slide in the encoding phase of the experiment. Figure Of the studies reviewed by Craik and Blankstein (1975),
5 illustrates the mean skin conductance responses at encoding however, none covary pleasantness and arousal as in the
as a function of the breakdown of the slides into valence and present set of experiments, and none assess memory for
arousal categories. As found in the past, high-arousal slides affective pictures. The data obtained here suggest their con-
led to significantly greater magnitude skin conductance re- clusions extend to these materials and, furthermore, that
sponses than did low-arousal slides, F( 1,64) = 80.37, AfSe = affective valence is not a controlling factor.
.02. In addition, unpleasant slides prompted slightly greater A number of studies in the past have only found a positive
skin conductance responses at encoding than did pleasant effect of high arousal on long-term memory tests; on imme-
slides, F(l, 64) = 3.5, MS, = .02, p = .06. As Figure 5 diate tests, poorer memory performance was obtained (Craik
illustrates, there was no interaction between these variables. & Blankstein, 1975). In those experiments, arousal level has
Therefore, skin conductance responses at encoding tended to often been defined by the magnitude of the subject's electro-
show the same pattern as was obtained for reaction times of dermal response at encoding, rather than by stimulus arousal.
new, encoded slides (in the context of the recognition task): Walker's (1958) action decrement theory accounted for this
Both the affective valence and arousal of the slide were pattern by suggesting that highly arousing events lead to a
apparent in responses during encoding. longer consolidation process in memory, which, in turn,
promotes better long-term memory. However, during consol-
General Discussion idation, retrieval is inhibited, which produces poor short-term
memory results. In an additional analysis here, we assessed
If emotional events are considered to be organized by the the reaction time data in Experiment 2 as a function of the
dimensions of affective valence (pleasantness) and arousal, magnitude of electrodermal reactivity during slide encoding.
the data obtained here demonstrate that memory retrieval is The same conclusion was supported: High levels of arousal,
most sensitive to differences in the arousal level of previously as measured by skin conductance responses, speeded later
encoded emotional stimuli. In both immediate and delayed recognition decisions. These data support Walker's account,
free recall, high-arousal materials led to better memory per- if one characterizes the 15-min retention interval used here
formance than materials rated lower in arousal (Experiment as long-term. On the other hand, there is no evidence in any
1). In a recognition task, high arousal facilitated recognition of our data that high arousal impairs memory performance,
decisions for slides encoded earlier in the experiment (Exper- even on memory tests that occur relatively soon after the
iment 2). Thus, although many factors affect memory per- encoding episode.
formance, the dimension of arousal (as indexed by verbal More recent studies have also concluded that high arousal
ratings and electrodermal responses) accounts here for sub- can impair, rather than facilitate, memory. The differences
stantial variability in remembering emotional stimuli. are primarily paradigmatic, and a close look suggests past
research is not inconsistent with the present data. For instance,
Arousal and Memory Christianson and his associates (Christianson & Nilsson, 1984;
Christianson, 1986; Christianson, Nilsson, Mjorndal, Penis,
Paradigms using verbal items as the to-be-remembered & Tjellden, 1986) utilize a methodology in which slides
material have also demonstrated facilitator/ effects of stimu- depicting traumatic or erotic events are accompanied by
PLEASURE AND AROUSAL IN MEMORY 387

verbal information related to the pictured event. Memory In the experimental literature, paradigms have tended to
performance for this peripheral verbal information is typically assess memory either for traumatic events (e.g., Christianson
poorer when the visual context is highly arousing than when & Loftus, 1987; Christianson & Nilsson, 1984; Brown &
it is neutral. Similarly, Loftus and Burns (1982) presented, Kulik, 1977; Bohannon, 1988) or for pleasant events (e.g.,
via a series of slides, a situation designed to be traumatic (i.e., Christianson, 1986)—often compared to neutral—but seldom
a small boy gets shot) or nontraumatic (i.e., the boy crosses a for both types of materials in the same experimental context.
parking lot unharmed). Recognition memory for a peripheral This is unfortunate, because differences in memory perform-
visual detail (a number on the boy's jersey) produced poorer ance are difficult to interpret when the neutral comparison
performance in the traumatic version. Similar data were stimulus differs on both the valence and arousal dimensions.
obtained by Clifford and Hollin (1981) in a test of detailed In the present study, the level of arousal was controlled
identification of the perpetrator of a violent crime. Detterman across different levels of affective valence. In this case, the
(1975) and his colleagues (Detterman & Ellis, 1972) found only effect of pleasantness on memory retrieval was at im-
poor memory performance for unrelated words that followed mediate free recall, and this effect was only marginally signif-
or preceded presentation of a single arousing stimulus (e.g., a icant. Pleasantness of the stimuli had no impact either on
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

loud shout or a photograph of a nude).' Across all of these delayed free recall or in recognizing items encoded earlier in
This document is copyrighted by the American Psychological Association or one of its allied publishers.

studies, memory for specific and mainly peripheral detail was the experiment. A similar null effect of pleasantness on mem-
assessed, and it is in this situation that high arousal produced ory performance was recently obtained by Thompson (1985),
poor memory performance. who found no effect of pleasantness when assessing memory
When probing memory for the occurrence of the primary for naturally occurring personal events. In addition, Bradley
event itself, however, Christianson and Loftus (1987) dem- and Baddeley (1990) found no difference between pleasant
onstrated that traumatic events are recalled better than neutral and unpleasant words, equated for concreteness and fre-
ones. In one experiment, subjects were shown a film of a bank quency, at immediate or delayed recall. Finally, there was no
robbery in which the ending was violent (i.e., two men were indication in our data that negative, highly arousing events
shot) or nonviolent. Six months later, subjects shown the were remembered differently (i.e., better) than other types of
traumatic version were more likely to remember the thematic events, as flashbulb theories might predict.
content of the film. These data are consistent with those For slides never seen before, affective valence did have a
presented here, although the effects of an event opposite in significant impact on reaction time: Unpleasant slides resulted
affective valence (e.g., pleasant) were not explored in that in significantly longer decision times than pleasant slides.
study. The current finding that pleasantness of the event does Similar effects of valence and arousal were obtained for the
not strongly affect memory performance leads us to suggest skin conductance data: Both high-arousal slides and unpleas-
that the memory advantage found for traumatic events may ant slides prompted greater reactions at encoding. However,
be primarily due to the high-arousal level, rather than the affective judgments of arousal did not differ for pleasant and
unpleasantness, of the traumatic episode. unpleasant materials.
More important, one implication from our data is that the There are two interesting ramifications of these data. First,
better memory found for information central to an arousing they indicate that the dimensions of valence and arousal are
event may be related to the physiological reaction engendered both salient at encoding. This is consistent with earlier data
during encoding of these types of stimuli. Components of the indicating dimensional discrimination in reactivity to these
event that are associated with an emotional reaction (as slide materials (Greenwald et al, 1989). Second, they suggest
indexed by skin conductance responses or other measures of that the judged level of arousal is probably not solely deter-
responsivity) may be better remembered than features lacking mined by electrodermal reactivity at initial encoding. As
this reactive property. This hypothesis can be tested by meas- mentioned earlier, both the amount of free viewing time and
uring physiological and behavioral responses to component ratings of interest also relate to the arousal factor, attesting to
elements of the slide stimuli used here. Whatever the outcome other influences underlying this affective dimension. It is
of such an investigation, the focus is clearly directed toward likely that judgments of arousal act as a summary statistic,
intense affective stimuli, regardless of their affective valence. incorporating elements in addition to that engendered by the
discrete sympathetic response (measured as skin conductance
change) at encoding. The arousal-related effects on memory
Pleasantness and Memory

Interestingly, folklore encourages the view that there is a


strong relation between affective valence and memory per- ' This raises a question concerning the effect of one slide on what
formance. For instance, the adage that one "views the past followed or preceded it in the current study. The design of this
through rose-colored glasses" attributes a central role to pleas- experiment was not intended to specifically assess this issue; in fact,
antness by proposing that memory favors pleasant events. different slide orders were used to counterbalance effects due to slide
sequence. To the extent this issue could be assessed in post hoc
Similarly (although the emphasis was reversed), Sigmund analyses, however, no additional mnemonic effects due to sequencing
Freud's views regarding repression were singularly tied to were obtained. Concerning physiological responses, other studies (e.g.,
degraded memory for events that were negative in affect. Greenwald et al., 1989; Lang et al., in press) have demonstrated full
Conversely, the lore underlying flashbulb memory pinpoints recovery within the 10- to 20-s interslide interval used here; in the
traumatic events as those that are remembered very precisely current study, such recovery is even more likely, given the presence
and for a long period of time. of the postslide SAM rating period.
388 BRADLEY, GREENWALD, PETRY, AND LANG

performance in our data appear to reflect this broader concept pothesis that increasing an animal's arousal level by electrical
of arousal. or chemical intervention after training facilitates perform-
ance. Enhancement in retention has been obtained both for
Arousal, Pleasantness, and Memory: Some Theoretical aversive and appetitive training tasks (Steinberg, Isaacs, Gold,
Comments & McGaugh, 1985), supporting the idea that memory is
facilitated similarly by arousal, regardless of affective valence.
As Lang et al. (1990) emphasized, the verbal dimensions of On the basis of such data, Gold and McGaugh (1975) pro-
affective valence and arousal can be directly related to the posed a memory system in which motivationally relevant
behavioral dimensions of direction and vigor. In this scheme, events (i.e., those that are highly arousing) produce endoge-
pleasant events are defined as those that primarily engage nous physiological consequences that can facilitate later mem-
approach or appetitive behaviors, whereas unpleasant events ory performance.
are those prompting withdrawal, avoidance, or defensive ac- From a cognitive viewpoint, elaboration may be a sufficient
tions. The arousal dimension is related to the vigor of the explanatory device. Slides rated as highly arousing (i.e., in-
behavioral disposition that is currently active and can range tensely emotional) produce large skin conductance responses,
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

from a level of extreme emergency to that of calm expediency. appropriate facial electromyographic responses, increased car-
This document is copyrighted by the American Psychological Association or one of its allied publishers.

According to this analysis, high intensity signals a motivation- diovascular activity, and ratings of heightened interest and
ally relevant event that can involve either an appetitive or attention, in relation to low-arousal slides (Greenwald et al.,
aversive environmental transaction. 1989; Bradley et al., 1990). If one attributes a similar infor-
The data obtained in this experiment (and others as well) mational role to physiological response information as is
suggest that arousal is a complex concept. There have been typically attributed to other types of information (Lang, 1979,
several attempts to detail, for instance, distinctions between 1984), the episodic network encoding of an arousing picture
cerebral, behavioral, and sympathetically mediated compo- can be considered more elaborate than one representing a
nents (e.g., Lacey, 1967). One can also distinguish between nonarousing slide. Assuming this type of network elaboration
stimulus arousal (as manipulated here) and nonstimulus- increases the probability of accessing a stored representation,
related subject arousal (e.g., by using an injection of adrenalin; both the higher free recall and faster recognition performance
Christianson et al., 1986). As discussed above, variables relat- obtained here could result.
ing to attention, interest, or effort may also be involved. If elaboration (involving active responses) is responsible for
Regardless of its underlying multiplicity, factor analyses con- the advantage found in memory for highly arousing materials,
sistently confirm the existence of such an organizing dimen- one prediction is that neutral materials exposed to such proc-
sion in emotion. Furthermore, this dimension seems to be essing (e.g., by inducing physiological or behavioral respond-
readily apprehended by subjects, who can easily assess the ing) could be equated, in terms of memory strength, to
intensity of their affective experience, independent of its emotional stimuli. On the other hand, high emotional arousal
valence. The data obtained here clarify its mnemonic effects: engendered by sympathetic activation (aversive or appetitive
Memory for the occurrence of an emotional stimulus associ- in orientation) may be singularly different from response
ated with high arousal is better than for a stimulus rated low elaboration of nonaffective materials. If this were the case,
in arousal, regardless of its pleasantness. neutral stimuli should tend to be consistently inferior in terms
A memory system sensitive to the arousal level of an event of memory performance. Clearly, these alternative hypotheses
is a broadly functional survival tool. Behaviors that demand can be evaluated.
high mobilization of resources, regardless of whether these are To the extent that the elaborated information represents
directed toward approaching a desired object (e.g., a food bodily responses to arousing content, this proposed processing
source when hungry) or fleeing from a feared one (e.g., a mechanism is also consistent with evidence that concurrently
predator in one's environment), are good candidates for mem- presented, peripheral stimulus information does not appear
ory storage. Such intense actions are probably strongly related to be well-represented in episodic memories of arousing
to survival, in terms of both preservative behaviors (such as events. Activation of emotional responses to arousing stimuli
eating, drinking, procreating, etc.) and protective ones, and may interfere with deployment of other behaviors necessary
may prove useful in future interactions with the environment. for encoding peripheral detail (e.g., in this study, scanning the
On the other hand, events low in arousal may often be visual display). If this were the case, high reactivity to an
unimportant or redundant with respect to successful environ- arousing stimulus would produce an elaborated representa-
mental maneuvers, resulting in low priority for memory stor- tion (and good memory performance) for elements associated
age. with the arousal response and poor memory for information
The underlying mechanisms relating arousal to memory relying on other types of behavioral operations at encoding.
are potentially multiple. Neurophysiological factors undoubt- Future investigations that probe memory for either details or
edly play a role in facilitating memory performance. For central components of the arousing stimuli used here can
instance, Kety (1972) has proposed that processes associated easily be implemented. The primary conclusion from the
with encoding emotional stimuli affect the circulation of current study is an empirical one: When emotion is defined
neurochemical substances that increase the efficiency of mne- by the dimensions of pleasantness and arousal, the major
monic processes. Gold and McGaugh (1975) and their asso- predictor of future memory performance is not the valence
ciates (McGaugh et al.. 1979) have explicitly tested the hy- of the specific emotion but its intensity.
PLEASURE AND AROUSAL IN MEMORY 389

References Kety, S. (1972). Brain catecholamines, affective states, and memory.


In J. L. McGaugh (Ed.), The chemistry of mood, motivation, and
Bohannon, J. N., III. (1988). Flashbulb memories for the space shuttle memory (pp. 65-80). New York: Plenum Press.
disaster: A tale of two theories. Cognition, 29, 179-196. Konorski, J. (1967). Integrative activity of the brain: An interdiscipli-
Bradley, B. P., & Baddeley, A. D. (1990). Emotional factors in nary approach. Chicago: University of Chicago Press.
forgetting. Psychological Medicine, 20, 351-355. Lacey, J. I. (1967). Somatic response patterning and stress: Some
Bradley, M. M., Cuthbert, B. N., & Lang, P. J. (1990) Startle reflex revisions for activation theory. In M. M. Appley & R. Trumbull
modification: Emotion or attention? Psychophysiology 27, 513— (Eds.), Psychological stress (pp. 14-36). New York: Appleton-
523. Century-Crofts.
Bradley, M. M., Greenwald, M. K., & Hamm, A. O. (in press). Lang, P. J. (1979). Presidential address, 1978: A bio-informational
Affective picture processing. In N. Birbaumer & A. Ohman (Eds.), theory of emotional imagery. Psychophysiology, 16, 495-512.
The structure of emotion: Psychophysiological, cognitive, and clin- Lang, P. J. (1980). Behavioral treatment and bio-behavioral assess-
ical aspects. Toronto: Hogrefe & Huber. ment: Computer applications. In J. B. Sidowski, J. H. Johnson, &
Brown, R., & Kulik, J. (1977). Flashbulb memories. Cognition, 5, T. A. Williams (Eds.), Technology in mental health care delivery
73-99. systems (pp. 119-137). Norwood, NJ: Ablex.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Christianson, S.-A. (1986). Effects of positive emotional events on Lang, P. J. (1984). Cognition in emotion: Concept and action. In C.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

memory. Scandinavian Journal of Psychology, 27, 287-299. E. Izard, J. Kagan, & R. B. Zajonc (Eds.), Emotions, cognitions,
Christianson, S.-A., & Loftus, E. F. (1987). Memory for traumatic and behavior (pp. 192-228). Cambridge, England: Cambridge Uni-
events. Applied Cognitive Psychology, 1, 225-239. versity Press.
Christianson, S.-A., & Nilsson, L.-G. (1984). Functional amnesia as Lang, P. J., Bradley, M. M., & Cuthbert, B. N. (1990). Emotion,
induced by a psychological trauma. Memory & Cognition, 12, 142— attention, and the startle reflex. Psychological Review, 97, 377—
155. 398.
Christianson, S.-A., Nilsson, L.-G., Mjorndal, T., Perris, C, & Lang, P. J., Greenwald, M. K., Bradley, M. M., & Hamm, A. O. (in
Tjellden, G. (1986). Psychological versus physiological determi- press). Looking at pictures: Evaluative, facial, visceral, and behav-
nants of emotional arousal and its relationship to laboratory in- ioral responses. Psychophysiology.
duced amnesia. Scandinavian Journal of Psychology, 27, 300-310. Lang, P. J., Ohman, A., & Vaitl, D. (1988). The international affective
Clifford, B. R., & Hollin, C. R. (1981). Effects of the type of incident picture system [Photographic slides]. Gainesville: University of
and number of perpetrators on eyewitness memory. Journal of Florida, The Center for Research in Psychophysiology.
Applied Psychology, 66, 364-370. Loftus, E. F., & Burns, T. E. (1982). Mental shock can produce
Craik, F. I. M., & Blankstein, K. R. (1975). Psychophysiology and retrograde amnesia. Memory & Cognition, 10, 318-323.
human memory. In P. H. Venables & M. J. Christie (Eds.), Re- Maltzman, I., Kantor, W., & Langdon, B. (1966). Immediate and
search in psychophysiology (pp. 388-417). New York: Wiley. delayed retention, arousal and defensive reflexes. Psychonomic
Detterman, D. K. (1975). The Von Restorff effect and induced Science, 6, 445-446.
amnesia: Production by manipulation of sound intensity. Journal Matlin, M., & Stang, D. (1978). The Pollyanna principle: Selectivity
of Experimental Psychology: Human Learning and Memory, 1, in language, memory and thought. Cambridge, MA: Schenkman.
614-628. McGaugh, J. L., Gold, P. E., Handwerker, M. J., Jensen, R. A.,
Detterman, D. K., & Ellis, N. R. (1972). Determinants of induced Martinez, J. L., Meligeni, J. A., & Vasquez, B. J. (1979). Altering
amnesia in short-term memory. Journal of Experimental Psychol- memory by electrical and chemical stimulation of the brain. In M.
ogy, 95, 308-316. A. B. Brazier (Ed.), Brain mechanisms in memory and learning:
Eysenck, M. W. (1976). Arousal, learning, and memory. Psychologi- From the single neuron to man (pp. 151-164). New York: Raven
cal Bulletin, 83, 389-404. Press.
Gold, P. E., & McGaugh, J. L. (1975). A single-trace, dual-process Osgood, C, Suci, G., & Tannenbaum, P. (1957). The measurement
view of memory storage processes. In D. Deutsch & J. A. Deutsch of meaning. Urbana: University of Illinois.
(Eds.), Short-term memory (pp. 355-378). San Diego, CA: Aca- Russell, J. (1980). A circumplex model of affect. Journal of Person-
demic Press. ality and Social Psychology, 39, 1161-1178.
Greenwald, M. K., Cook, E. W., & Lang, P. J. (1989). Affective Sternberg, D. B., Isaacs, K. R., Gold, P. E., & McGaugh, J. L. (1985).
judgment and psychophysiological response: Dimensional covar- Epinephrine facilitation of appetitive learning: Attenuation with
iation in the evaluation of pictorial stimuli. Journal of Psychophys- adrenergic receptor antagonists. Behavioral and Neural Biology,
iology, 3, 51-64. 44, 447-453.
Hamm, A. O., Stark, R., & Vaitl, D. (1990). Classical fear condition- Tellegen, A. (1985). Structures of mood and personality and their
ing and the startle probe reflex [Abstract]. Psychophysiology, 27, relevance to assessing anxiety, with an emphasis on self-report. In
537. A. H. Tuma & J. D. Maser (Eds.), Anxiety and the anxiety disorders
Hebb, D. O. (1949). The organization of behavior: A neuropsycholog- (pp. 681-706). Hillsdale, NJ: Erlbaum.
ical theory. New York: Wiley. Thompson, C. P. (1985). Memory for unique personal events: Effects
Hodes, R. L., Cook, E. W., & Lang, P. J. (1985). Individual differences of pleasantness. Motivation & Emotion, 2, 277-289.
in autonomic response: Conditioned association or conditioned Walker, E. L. (1958). Action decrement and its relation to learning.
fear? Psychophysiology, 22, 545-560. Psychological Review, 65, 129-142.

(Appendix follows on next page)


Appendix
List of Slides, Affective Ratings, and Free Recall for All Stimuli Used in Experiment 1
IAPS Mean pleasantness Mean arousal Proportion of Ss
number Description rating rating recalling slide
109 Attacking snake 3.70 5.90 .49
111 Black cobra 3.84 5.96 .52
123 Spider 4.61 4.03 .56
159 Horse 7.18 4.74 .60
161 Rabbit 7.69 3.98 .48
164 Coyote 6.16 5.18 .38
167 Cow 5.82 3.33 .63
213 Angry woman 4.08 5.02 .20
215 Man with baby 7.92 5.00 .37
221 Neutral face 4.70 3.08 .30
250 Old man 6.16 3.61 .64
251 Old woman 6.91 4.00 .70
301 Mutilated face 1.71 2.88 .74
310 Burn victim 1.60 6.49 .75
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

314 Mutilated body 1.83 3.20 .61


This document is copyrighted by the American Psychological Association or one of its allied publishers.

315 Bloody finger 2.26 6.55 .75


321 Surgery 4.49 5.39 .17
461 Romantic couple 7.29 5.54 .44
465 Couple in bed 6.96 5.56 .62
560 Mountains 7.57 5.19 .60
582 River 7.33 4.61 .39
583 Beach at sunset 8.00 4.92 .77
620 Man with gun 3.20 5.82 .56
623 Aimed pistol 2.37 7.39 .59
690 Bomber aircraft 4.76 5.64 .43
693 Missile range 4.39 4.88 .31
700 Rolling pin 5.00 2.42 .32
705 Hair dryer 4.93 2.75 .45
706 Trash can 4.43 2.55 .45
715 Umbrella 4.72 2.61 .51
717 Light bulb 5.14 3.21 .25
723 Turkey dinner 7.38 5.52 .41
726 Strawberry pie 7.21 6.03 .35
727 Chocolate soda 7.53 5.88 .37
728 Wine 7.20 4.46 .31
750 Urban building 5.33 5.17 .32
751 City landscape 6.05 4.52 .39
803 Ski jump 7.33 7.35 .48
808 Sailing 7.73 6.65 .53
809 Gymnast 7.02 5.71 .30
812 Tennis player 7.09 4.85 .16
813 Pole vaulter 6.58 5.49 .29
900 Cemetery 2.55 4.06 .05
90! Barbed wire 3.94 4.14 .31
904 Starving child 1.67 3.10 .51
905 Plane crash 2.43 6.36 .35
907 Solemn boy 5.01 3.63 .53
908 Electric wires 4.07 4.05 .31
909 Auto exhaust 3.56 3.97 .28
911 Dirty puddle 3.76 4.88 .14
914 Animal carcass 2.19 5.38 .37
916 Soldier 3.23 5.87 .30
918 Dead seal 2.99 5.02 .38
919 African woman 3.90 3.91 .49
Mean pleasantness Mean arousal Recall
rating rating
Erotic Stimuli Males Females Males Females Males I-emales
422 Woman in swimsuit 8.02 5.29 7.17 3.63 .80 .60
429 Nude woman 7.61 3.67 7.20 4.10 .95 .93
203 Woman on beach 7.51 6.02 6.24 3.08 .51 .54
449 Nude man 4.29 6.27 2.85 6.06 .90 .98
452 Man in swimsuit 4.76 7.04 2.68 5.48 .33 .40
453 Man in bed 4.46 6.19 3.15 5.31 .51 .52
Note. IAPS = International Affective Picture System, Lang et al., 1988.

Received December 3, 1990


Revision received June 28, 1991
Accepted August 23, 1991

You might also like