Pos Mood

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 27

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/225886952

Adult Attachment Style and Cognitive Reactions to


Positive Affect: A Test of Mental Categorization and
Creative Problem Solving

Article in Motivation and Emotion · September 2000


DOI: 10.1023/A:1005606611412

CITATIONS READS

139 1,711

2 authors, including:

Mario Mikulincer
Reichman University (IDC) Herzliya
481 PUBLICATIONS 44,734 CITATIONS

SEE PROFILE

All content following this page was uploaded by Mario Mikulincer on 23 May 2016.

The user has requested enhancement of the downloaded file.


P1: vendor
Motivation and Emotion [me] PP036-292899 December 29, 2000 16:17 Style file version Nov. 19th, 1999

Motivation and Emotion, Vol. 24, No. 3, 2000

Adult Attachment Style and Cognitive Reactions


to Positive Affect: A Test of Mental Categorization
and Creative Problem Solving
Mario Mikulincer1,2 and Elka Sheffi1

Three studies examined the moderating effect of attachment style on cognitive re-
actions to positive affect inductions. In Study 1 (N = 110), participants completed
attachment style scales, were asked to retrieve a happy or a neutral memory, and,
then, performed a categorization task. Study 2 (N = 120) used the same affect
induction, while examining creative problem solving in the Remote Associates
Test. Study 3 (N = 120) replicated Study 2, while using another affect induction
(watching a comedy film) and controlling for trait anxiety scores. Overall, securely
attached persons reacted to positive affect with broader categorization and bet-
ter performance in creative problem-solving tasks. Anxious–ambivalent persons
showed an opposite pattern of cognitive reactions to positive affect, and avoidant
persons showed no difference in their cognitive reactions to positive and neu-
tral affect inductions. The discussion emphasizes the role that attachment-related
strategies of affect regulation may play in episodes of positive affect.

In recent years, the construct of adult attachment style, as assessed by self-report


measures, has been used to explain individual differences in the process of af-
fect regulation (see Fuendeling, 1998; Mikulincer & Florian, 1998, for reviews).
However, most of the studies have exclusively focused on the experience and man-
agement of negative affect, and no systematic effort has been devoted to examining
the relevance of attachment style for episodes of positive affect. The current series
of studies attempts to fill in this empirical gap and to examine the role that adult
attachment style may play in moderating the well-documented cognitive reactions
to positive affect (Isen, 1987, 1993).
1 Departmentof Psychology, Bar-Ilan University, Ramat Gan, Israel.
2 Address
all correspondence to Mario Mikulincer, Department of Psychology, Bar-Ilan University,
Ramat Gan 52900, Israel; e-mail: mikulm@mail.biu.ac.il.

149

0146-7239/00/0900-0149$18.00/0 °
C 2000 Plenum Publishing Corporation
P1: vendor
Motivation and Emotion [me] PP036-292899 December 29, 2000 16:17 Style file version Nov. 19th, 1999

150 Mikulincer and Sheffi

ATTACHMENT STYLE AND AFFECT REGULATION

One of the basic assumptions of Bowlby’s theory (1969, 1973, 1980) is that
the quality of interactions with significant others results in internal working models
that guide the process of affect regulation. In this view, the activation of attachment
behaviors in times of need (proximity-seeking) usually goes together with anxiety
and anger, and the termination of these behaviors following distress management
may elicit feelings of relief, love, and gratitude. As a result, interactions with
significant others who are responsive to one’s attachment needs may lead to the
experience of more and longer episodes of positive affect, the development of
positive feelings toward these persons, and the reliance on support-seeking as a
mood-repair device. However, when significant others are unavailable in times of
need, one may experience chronic anxiety and anger and develop other defensive
strategies rather than support seeking. In Bowlby’s (1973) terms, these experiences
are internalized into broad regulatory strategies guide a person in the management
of emotional arousal (Collins & Read, 1994).
In examining the above ideas in adulthood, most of the studies have fo-
cused on a person’s attachment style—stable patterns of relational cognitions
and behaviors—and have adopted Hazan and Shaver’s (1987) typology of secure,
avoidant, and anxious–ambivalent style. Recently, Brennan, Clark, and Shaver
(1998) concluded that this typology reflects two underlying dimensions: avoid-
ance and anxiety. Persons scoring low on these two dimensions exhibit the secure
style and are characterized by a more positive history of interactions with signifi-
cant others and more confidence in others’ availability in times of need than persons
scoring high on either the avoidance or the anxiety dimension. Persons scoring high
on the avoidance dimension exhibit the avoidant style and are characterized by in-
security concerning others’ goodwill, compulsive self-reliance, and preference for
social and emotional distance. Persons scoring high on the anxiety dimension
exhibit the anxious–ambivalent style and are characterized by insecurity concern-
ing others’ responses, a desire for emmeshed relationships, and fear of rejection.
Bartholomew and Horowitz (1991) distinguished a subgroup of insecure persons
who score high on both anxiety and avoidance dimensions (fearful persons) and
exhibit a combination of features of the avoidant and anxious–ambivalent styles.3
Using Hazan and Shaver’s (1987) typology, Mikulincer, Orbach, and Iavnieli
(1998) delineated the affect regulation strategies that characterize each attachment
style. Secure attachment consists of constructive and flexible regulatory strategies.
Secure persons seem to rely on the basic guidelines of the attachment system:
acknowledge emotional arousal, engage in instrumental actions, ask for others’

3 Inour study, we did not focus on the fearful subgroup, because these persons do not possess a unique
strategy of affect regulation and indiscriminately move between avoidant and anxious strategies.
Moreover, our findings implied that persons scoring high on the avoidance and anxiety dimensions
(fearful) did not differ significantly from those scoring high on only the anxiety dimension (anxious–
ambivalent).
P1: vendor
Motivation and Emotion [me] PP036-292899 December 29, 2000 16:17 Style file version Nov. 19th, 1999

Attachment, Positive Affect, and Cognition 151

support, and hope for successful management of the situation. Persons scoring
high on attachment anxiety tend to regulate affect by hyperactivating attachment
needs and distress-related cues. Anxious–ambivalent persons desperately attempt
to minimize distance from others and to achieve love and security via clinging and
controlling actions. They also over-attend to distress cues, hyperactivate negative
emotions and thoughts, and fail to detach from inner pain. Persons scoring high
on attachment avoidance tend to regulate affect by deactivating attachment needs
and distress cues as well as by emphasizing self-reliance. Avoidant persons escape
from close relationships, deny attachment needs, pursue autonomy and control,
suppress negative or unpleasant thoughts, and rely on repressive mechanisms.
The hypothesized regulatory strategies have received strong empirical sup-
port in studies that focus on the experience of emotional episodes (see Fuendeling,
1998, for a review). First, secure persons have been found to react to negative affect
with weaker physiological arousal than insecure persons did (e.g., Dozier & Kobak,
1992; Feeney & Kirkpatrick, 1996; Mikulincer, 1998). Second, secure persons tend
to experience negative affect in a more functional way and to direct more attention
to positive aspects of the situation than insecure persons do (Mikulincer, 1998).
Third, whereas secure persons acknowledge negative emotions without being over-
whelmed by them and show high accessibility and differentiated processing of
them, persons scoring high in attachment anxiety experience an overwhelming
arousal of these emotions and an undifferentiated spreading of this arousal to ir-
relevant emotional themes (Mikulincer & Orbach, 1995). Persons scoring high in
attachment avoidance tend to distance themselves from emotion-laden material
and to show low accessibility of negative emotions, lack of interoceptive aware-
ness, and a restrictive control of emotional expression (e.g., Brennan & Shaver,
1995; Kotler, Buzwell, Romeo, & Bowland, 1994; Mikulincer & Orbach, 1995).
Strong support for the hypothesized regulatory strategies is also found in stud-
ies that examine ways of coping with stressful events (see Mikulincer & Florian,
1998, for a review). These studies have shown that secure people cope with these
events by relying on support-seeking strategies (e.g., Fraley & Shaver, 1998; Priel
& Shamai, 1995; Simpson, Rholes, & Nelligan, 1992). They have also documented
avoidant persons’ deactivating strategy: These persons tend to deal with distress
arousal by suppressing negative thoughts and relying on distancing coping (e.g.,
Fraley & Shaver, 1997; Mikulincer, Florian, & Weller, 1993; Radecki-Bush, Farrel,
& Bush, 1993). These studies have also corroborated anxious–ambivalent persons’
hyperactivating strategy: These persons tend to deal with distress arousal by rumi-
nating on negative thoughts and focusing on their emotional state (e.g., Birnbaum,
Orr, Mikulincer, & Florian, 1997; Mikulincer & Florian, 1995; Pistole, 1995).
Recent studies have shown that attachment-related regulatory strategies are
also manifested in information processing (Green-Hennessy & Reis, 1998;
Mikulincer, 1997). Secure persons’ constructive and flexible strategies have been
manifested in a positive attitude toward cognitive exploration as well as in cogni-
tive openness, even when new data may contradict existing knowledge. In contrast,
P1: vendor
Motivation and Emotion [me] PP036-292899 December 29, 2000 16:17 Style file version Nov. 19th, 1999

152 Mikulincer and Sheffi

persons scoring high on the avoidance or anxiety dimensions hold a negative ap-
proach toward exploration and cognitive openness. Among avoidant persons, their
distancing from distress-related cues and their attempts to suppress personal weak-
nesses seem to result in the rejection of new stimuli, because the exploration of this
data may create a temporary state of ambiguity and distress and may lead to the
recognition that one’s beliefs are faulty. Among anxious–ambivalent persons, their
hyperactivation of the attachment system may cause them to become preoccupied
with attachment-related concerns, and they may then have insufficient resources
for engaging in cognitive exploration and elaboration of new data.
Overall, there is extensive evidence supporting the link between attachment
style and affect regulation. However, most of the reviewed studies have focused
on the management of negative affect. Although attachment-related regulatory
strategies might originally have been developed to buffer distress (Bowlby, 1988),
they may also be relevant for understanding reactions to positive affect. According
to Bowlby (1988), the quality of interactions with significant others is closely
linked with the experience of positive emotions (e.g., relief, joy, love) as well as
with a person’s cognitive state upon mood repair. The more positive the interactions
with significant others in times of need, the more secure a person will be in others’
availability, and the more relaxed he/she will be when no threat is present. In these
cases, secure persons are hypothesized to engage in playful exploration and to enjoy
cognitive activities (e.g., Ainsworth, Bell, & Stayton, 1973; Mikulincer, 1997).
Furthermore, anxious–ambivalent persons’ chronic focus on negative emotions as
well as avoidant persons’ distancing from emotion-laden material (Fuendeling,
1998) may also affect the reactions to positive affect. We followed this reasoning
and examined the extent to which attachment style moderates the well-documented
cognitive reactions to positive affect (Isen, 1987).

POSITIVE AFFECT AND COGNITIVE PROCESSES

A wealth of studies in social psychology have found that the induction of


positive affect (e.g., receiving a small gift, seeing a brief comedy film, visualizing
a pleasant event) exerts strong effects on social behavior and cognitive processes
(Isen, 1987, 1993). In the realm of cognitive functioning, positive affect has been
found to lead people to perceive more interconnections between stimuli, to make
more unusual associations between cognitions, and to use broader mental cate-
gories (Isen, 1987). Specifically, Isen and Daubman (1984) found that persons in
a positive affective state rated poor exemplars of a semantic category as better ex-
emplars of this category than did persons in a control condition, and sorted items
into larger and more inclusive categories. That is, positive affect facilitated the
perception of relatedness, even for items that are generally perceived as having
poor connections with the other items in a category. This effect was replicated in
the categorization of personality traits (Isen, Niedenthal, & Cantor, 1992).
P1: vendor
Motivation and Emotion [me] PP036-292899 December 29, 2000 16:17 Style file version Nov. 19th, 1999

Attachment, Positive Affect, and Cognition 153

Positive affect has also been found to enhance the complexity and flexibility
of mental categorization. For example, Isen (1987) cited a study in which per-
sons reacted to positive affect with a tendency to sort stimuli into more different
categories over multiple trials when asked how many ways this material could
be sorted. Accordingly, Showers and Cantor (1985) found that positive affect en-
hances the ability to make multiple alternative interpretations of social situations.
Moreover, Murray, Sujan, Hirt, and Sujan (1990) found that persons in a positive
affect condition perceived more similarities and differences between items and
generated more distinct and novel types of similarities and differences than per-
sons in a control condition did. The findings imply that positive affect not only
increases the perception of relatedness, but also the complexity of this perception.
There is also evidence that positive affect facilitates creativity and problem
solving. For example, positive affect inductions, as compared to control conditions,
have been found to lead people to give more unusual and creative associations in
a word association task (Isen, Johnson, Mertz, & Robinson, 1985) and to improve
performance in creative problem-solving tasks (Estrada, Isen, & Young, 1994;
Isen, Daubman, & Nowicki, 1987). Moreover, Estrada, Isen, and Young (1997) and
Isen, Rosenzweig, and Young (1991) found that physicians exposed to a positive
affect induction displayed more flexibility and fewer distortions in clinical problem
solving than those in a control condition did. Overall, these findings further support
the idea that positive affect leads to a more integrative and complex organization
of cognitive material.
Several mechanisms have been proposed to account for the documented ef-
fects of positive affect. Isen (1987) argued that positive affect cues positive material
in memory and that this material, which is quite extensive and varied (e.g., Boucher
& Osgood, 1969), may provide a complex context for cognitive organization and
bring divergent material into mind. As a result, this cognitive context may facilitate
an integrative and creative organization of incoming information. Alternatively,
Schwarz and Bohner (1996) contended that positive affect signals that all is going
well and that one can give up control over effortful strategies aimed at dealing with
potential dangers. Consequently, this signal may evoke a more relaxed approach
to a task, which, in turn, may result in playful exploration and a “loosening” of
cognitive strategies. This cognitive change may thus facilitate creativity and the
perception of relatedness between stimuli.
In a related vein, Murray et al. (1990) proposed that positive affect increases
the appraisal of a task as interesting and fun and may promote intrinsic interest
(e.g., Pretty & Seligman, 1984), which in turn may result in playful exploration
and creativity. According to the mood management perspective (Wegener & Petty,
1994), persons in positive affect conditions may deliberately generate creative
or unusual responses as a means of making the task more fun and interesting,
thereby maintaining their positive mood. That is, the highly integrated and creative
performance of these persons may reflect their efforts to maintain their positive
affective state.
P1: vendor
Motivation and Emotion [me] PP036-292899 December 29, 2000 16:17 Style file version Nov. 19th, 1999

154 Mikulincer and Sheffi

ATTACHMENT STYLE, POSITIVE AFFECT,


AND COGNITIVE PROCESSES

The above explanations imply that the effects of positive affect may be moder-
ated by an array of contextual and individual difference factors. These effects may
depend on the extent to which a person focuses on positive material, engages in
playful exploration, enjoys cognitive tasks, or attempts to maintain a positive mood.
Unfortunately, there is no systematic research on the moderating role of individ-
ual difference factors, because studies that have examined the above explanations
have exclusively focused on the moderating role of contextual factors (e.g., Hirt,
Levine, McDonald, & Melton, 1997; Hirt, Melton, McDonald, & Harackiewicz,
1996; Martin, Ward, Achee, & Wyer, 1993). In the current study, we focus on
attachment style while examining the role this basic individual difference factor
may play in moderating the effects of positive affect on categorization and creative
problem solving. Attachment style has been related to the cognitive accessibility
of hedonic material, the engagement in exploration, and the process of affect reg-
ulation (e.g., Fraley & Shaver, 1997; Fuendeling, 1998; Mikulincer, 1997).
Our main hypothesis was that attachment style would moderate the effects
of positive affect on cognitive processes and that these effects would mainly be
found among secure persons. These persons hold highly available and accessible
positive cognitions about the self and the world (Collins & Read, 1994). More-
over, they enjoy exploration and open their schemata to new data (e.g., Mikulincer,
1997). Whereas secure persons’ availability of positive cognitions would facili-
tate the creation of a complex cognitive context in response to a positive affect
induction, their positive approach to exploration would allow them to “loosen”
cognitive strategies under this condition and to engage in a carefree processing of
information. As a result, secure persons would react to positive affect with broader
categorization and more creative problem solving.
We also hypothesized that anxious–ambivalent persons would not show the
above reviewed effects of positive affect. The regulatory strategies of these persons
facilitate the availability and accessibility of negative material in memory (Collins
& Read, 1994), hyperactivate negative affect (Fuendeling, 1998), and prevent
the cognitive exploration of new data (Mikulincer, 1997). In our terms, anxious–
ambivalent persons’ focus on negative material and the low availability of positive
cognitions may prevent the formation of a complex cognitive context in response
to a positive affect induction. Moreover, their hyperactivation of distress-related
cues and their cognitive closure would prevent anxious–ambivalent persons from
engaging in relaxed exploration following a positive affect induction. As a result,
these persons would not react to positive affect with broader categorization and
better performance in creative problem-solving tasks.
With regard to avoidant persons, we also hypothesized that they would not
show the above-reviewed effects of positive affect. Although avoidant persons have
high accessibility of positive cognitions and engage in defensive attempts to reduce
P1: vendor
Motivation and Emotion [me] PP036-292899 December 29, 2000 16:17 Style file version Nov. 19th, 1999

Attachment, Positive Affect, and Cognition 155

negative affect (Mikulincer & Florian, 1998; Shaver & Hazan, 1993), their ten-
dency to distance from emotion-laden material may lead them to regard the arousal
of positive affect as an irrelevant input for cognitive processing. Moreover, these
persons reveal a negative approach to cognitive exploration (Mikulincer, 1997)
and are unable or unwilling to give up control and “loosen” cognitive strategies
(Shaver & Hazan, 1993). In fact, they show a clear preference for control and order
and tend to avoid ambiguous and novel stimuli (Mikulincer, 1997). Therefore, in-
ductions of positive affect would fail to produce broader categorization and better
creative problem solving performance among avoidant persons.

STUDY 1

In Study 1, we focused on Isen and Daubman’s (1984) finding that positive


affect leads to acceptance of less typical exemplars as members of a semantic
category, while examining whether attachment style moderates this facilitation
effect. Participants filled out measures of attachment style, were exposed to a
positive or neutral affect induction, and performed Isen and Daubman’s (1984)
categorization task. In this task, the dependent variable was the number of poor
exemplars of semantic categories, according to Rosch’s (1975) norms, that were
accepted as members of the categories. Our predictions were

1. Replicating Isen and Daubman’s (1984) findings, participants in the pos-


itive affect condition would accept poorer exemplars as members of a
semantic category than participants in the neutral affect condition.
2. The above effect would be moderated by attachment style. Whereas
secure persons would show the predicted difference between positive
and neutral affect conditions, anxious–ambivalent and avoidant persons
would show no significant effect of positive affect on categorization of
poor exemplars.

Method

Participants

One hundred and ten undergraduate social sciences students from Bar-Ilan
University (78 women and 32 men ranging in age from 19 to 32, median = 23)
participated in the study without receiving a monetary reward.

Materials and Procedure

Participants were approached during lecture time and invited to participate in


a social psychology study. They completed two scales tapping the endorsement of
a specific attachment style and continuous scores along the avoidance and anxiety
P1: vendor
Motivation and Emotion [me] PP036-292899 December 29, 2000 16:17 Style file version Nov. 19th, 1999

156 Mikulincer and Sheffi

Table I. Means and Standard Deviations of Mood and Categorization Measures According to
Induced Affect and Attachment Style (Study 1)
Neutral affect condition Positive affect condition
Secure Avoidant Anxious Secure Avoidant Anxious
Measures (n = 29) (n = 17) (n = 9) (n = 33) (n = 14) (n = 8)

Reported mood
M 4.66a 4.64a 4.78a 5.67b 5.64b 4.13a
SD 1.37 1.17 1.48 0.92 1.34 1.25
Number of poor exemplars accepted as category members
M 4.79a 4.88a 5.00a 6.43b 5.35a 2.50c
SD 2.42 2.18 2.01 2.12 2.46 1.93
Representativeness ratings of poor exemplars
M 5.48a 5.93a 6.27a 7.03b 6.29a 4.44c
SD 1.72 1.51 1.58 1.62 1.32 1.64

Note. Means within a row that have different letters were significantly different ( p < .05).

dimensions (Hazan & Shaver, 1987; Mikulincer, Florian, & Tolmacz, 1990). Both
scales assessed global attachment style in adulthood rather than attachment orien-
tation in a specific relationship. These two scales were given in random order.
In one scale, participants read Mikulincer et al.’s (1990) Hebrew version of
Hazan and Shaver’s (1987) three descriptions of attachment styles and endorsed the
description that best described their feelings in close relationships. This scale has
been found to be a reliable and valid classificatory measure of adult attachment style
(e.g., Mikulincer et al., 1990). The frequencies of self-classified attachment styles
in the current sample were as follows: 57% of the sample classified themselves
as secure (n = 62), 28% as avoidant (n = 31), and 15% as anxious–ambivalent
(n = 17). No significant difference was found in this distribution across conditions
(see ns in Table I).4
The second scale consisted of 10 items, tapping the attachment dimensions
of anxiety and avoidance (Brennan et al., 1998). The decision to add this scale was
based on Fraley and Waller’s (1998) findings that categorical measures do not pro-
vide a complete picture of variability in attachment style. Furthermore, Brennan et
al. (1998) showed that a two-dimensional model underlies most measures of adult
attachment style. On this basis, we employed the avoidant and anxiety subscales of
Mikulincer et al.’s (1990) scale, which were constructed by decomposing Hazan
and Shaver’s (1987) prototypical descriptions into items corresponding to Brennan
et al.’s (1998) definitions and items. Five items tapped attachment anxiety (e.g.,
“I often worry that my partner does not love me,” “I find that other persons are

4 No gender difference was found in the distribution of attachment style as well as in the anxiety and
avoidance scores. The statistical control of gender in the analyses of the main dependent variable
did not affect the significant effects of induced affect and attachment style. In addition, interactions
between induced affect, gender, or attachment scores or a combination of these were not significant,
implying that gender did not have any moderating effect. These findings were replicated in Studies 2
and 3.
P1: vendor
Motivation and Emotion [me] PP036-292899 December 29, 2000 16:17 Style file version Nov. 19th, 1999

Attachment, Positive Affect, and Cognition 157

reluctant to get as close as I would like”) and the other five items tapped attach-
ment avoidance (e.g., I’m somewhat uncomfortable being close to other persons,”
“I find it difficult to trust other persons in close relationships”). Participants rated
the extent to which an item described their feelings in close relationships using
a 7-point scale, ranging from 1 (not at all) to 7 (very much). This scale has been
found to be highly reliable and valid (e.g., Mikulincer et al., 1993).
In our sample, a factor analysis with Varimax rotation explained 58% of
the scale variance. Whereas the first factor (40%) included the five avoidance
items (loading >.40), the second factor (18%) included the five anxiety items
(loading >.40). Cronbach’s Alphas for each factor implied acceptable internal
consistency (.76, .77). Hence, two scores were computed by averaging items that
belonged to a factor. The Pearson correlation between these scores showed a
significant but moderate-to-weak association, r (108) = .27, p < .01.
Upon completing the above scales, participants were randomly divided into
two conditions according to affect induction. In the positive affect condition (n =
55), they were asked to remember a happy event, to bring it to their mind, to
visualize it, and to provide a written description of the event. Participants provided
a wide variety of events, such as the receipt of a satisfactory grade in an exam,
romantic dating, and a birthday party. In the neutral affect condition (n = 55),
participants were asked to remember the last time they were at a drug store, to
bring it to mind, to visualize it, and to provide a written description of the event. All
the participants in the two conditions successfully completed the task and reported
having no difficulties in remembering the targeted event. The recalled events were
content analyzed independently by the authors and one-way analyses of variance
(ANOVAs) revealed no significant difference between the three attachment-style
groups in the length of the description, the recall of interpersonal events (yes/no),
the number of emotional phrases used, and the number of positive affective states
and positive adjectives/traits mentioned. This lack of significant attachment-style
differences was replicated in both the positive affect and neutral affect conditions.
Following the recall task, all the participants completed measures check-
ing on affect induction. They rated their current mood on a 7-point scale, rang-
ing from 1 (very negative) to 7 (very positive). They also rated the pleasantness
of five unfamiliar words on 7-point scales, ranging from 1 (very unpleasant) to
7 (very pleasant). These pleasantness ratings have been used before as an indirect
assessment of mood (e.g., Isen, Daubman, & Nowicki, 1987). Cronbach’s α for
the five unfamiliar words was high (.85), allowing us to compute a total score by
averaging the five ratings.
Next, participants were informed that they would perform a cognitive task,
with no information provided concerning the ego-relevance of the task or the
assessed cognitive skills. Then they performed Isen and Daubman’s (1984) cat-
egorization task. They were instructed that they would rate items on a 10-point
scale, indicating the degree to which they felt an item belonged or did not belong
to a category. They were told that a rating of 1 meant that an item “definitively does
P1: vendor
Motivation and Emotion [me] PP036-292899 December 29, 2000 16:17 Style file version Nov. 19th, 1999

158 Mikulincer and Sheffi

not belong to the category” and that a rating of 10 meant that an item “definitively
does belong to the category.” They also received a break-point between 5 and 6:
Whereas a rating of 5 could be used when an item did not belong to a category but
was very similar to members of the category, a rating of 6 could be used when an
item did belong to a category but was not a very good example of it.
After reading the instructions, participants received a sheet containing the
name of a category (e.g., clothing), nine exemplars, and a 10-point scale for each
exemplar. The 9 items included 3 excellent exemplars of the category, 3 mod-
erately good exemplars, and 3 poor exemplars (e.g., shirt, tie, and ring for the
clothing category, respectively). These items were selected according to prototyp-
icality norms identified by Rosch (1975), as well as Israeli norms concerning these
rankings (Mikulincer, Kedem, & Paz, 1990).5 The first item was always an excel-
lent exemplar of the category; the remaining items were presented in a random
order. This procedure was repeated for three categories—clothing, furniture, and
vehicles. The order of the categories was randomized across participants.

Results and Discussion

The data were analyzed in two ways. First, we examined the contribution of
the attachment typology to mood and categorization ratings via two-way ANOVAs
with induced affect and self-classified attachment style as the factors. Second, we
examined the contribution of the two continuous attachment scores to mood and
categorization ratings by hierarchical regressions. In these regressions, the pre-
dictors were induced affect (a dummy variable contrasting the positive affect and
the neutral affect conditions), attachment anxiety, and attachment avoidance. The
main effects of the predictors were introduced in Step 1, the two-way interaction
in Step 2, and the three-way interaction in Step 3.

Manipulation Check

Whereas the ANOVA for pleasantness ratings revealed no significant effects,


the ANOVA conducted on global mood rating yielded the expected significant main
effect for induced affect, F(1, 104) = 10.87, p < .01, with persons in the posi-
tive affect condition reporting more positive mood (M = 5.43) than those in the
neutral affect condition (M = 4.67). However, this effect was qualified by a signif-
icant interaction between induced affect and attachment style, F(2, 104) = 3.34,
p < .05. Tests for Simple Main Effects revealed that the effect of positive affect on
mood rating was significant only among secure, F(1, 104) = 16.68, p < .01, and
5A pre-test conducted on 20 undergraduate students confirmed the prototypicality ranking of the se-
lected exemplars. Moreover, the ratings made in the current study also corroborated Rosch’s (1975)
ranking. Excellent exemplars were rated as more representative of a category (M = 9.77) than moder-
ately good exemplars (M = 8.08), which in turn were rated as more representative than poor exemplars
(M = 6.11). These differences were statistically significant, F(2, 218) = 346.43, p < .01.
P1: vendor
Motivation and Emotion [me] PP036-292899 December 29, 2000 16:17 Style file version Nov. 19th, 1999

Attachment, Positive Affect, and Cognition 159

avoidant persons, F(1, 104) = 4.90, p < .05, but not among anxious–ambivalent
persons (see Table I). Among anxious–ambivalent persons, no significant differ-
ence was found between the two experimental conditions.
The hierarchical regression for pleasantness ratings revealed no significant
main effects or interactions. However, the regression for the global mood ratings
revealed a significant main effect for induced affect, Beta = .29, p < .01, and a
significant interaction for induced affect × anxiety, Beta = −.86, p < .01. No
other effects were significant. Within-condition partial correlations (controlling
for the other attachment rating) revealed that attachment anxiety was significantly
related to the mood rating in the positive affect condition, r (52) = −.34, p < .05,
but not in the neutral condition, r (52) = .02, ns. That is, the higher a person’s
attachment anxiety, the worse his/her mood after a positive affect induction. This
pattern of associations paralleled the ANOVA findings.

Categorization

Although participants provided ratings for good, moderate, and poor exem-
plars, our main dependent variable was the categorization of poor exemplars. Like
Isen and Daubman (1984), we included good and moderate items as filler items.
In fact, ANOVAs and regressions performed on mean ratings of moderate and
excellent exemplars yielded no significant effects for affect induction and attach-
ment scores. We therefore focused only on poor exemplars and computed the two
scores proposed by Isen and Daubman (1984). First, we counted the number of
poor exemplars a participant rated 6 or above. This score reflected the extent to
which poor exemplars were accepted as category members. Second, we averaged
the ratings of the 9 poor exemplars, reflecting the extent to which poor exemplars
were perceived as similar to category members.
The ANOVAs indicated that the main effect for induced affect approached
significance for the number of accepted poor exemplars,F(1, 104) = 2.09, p =
.09, and was significant for the rating of these exemplars, F(1, 104) = 5.19, p <
.05. As expected, the induction of positive affect led participants to rate these
exemplars as more similar to category members (M = 6.46) than those in the
neutral affect condition (M = 5.75). In addition, there was a significant interaction
for induced affect and attachment style as predictors of the number of accepted
poor exemplars, F(2, 104) = 5.68, p < .01, and the rating of these exemplars,
F(2, 104) = 7.24, p < .01.
Tests for Simple Main Effects revealed an interesting pattern of differences.
For secure persons, a positive affect induction (compared with a neutral affect
induction) led to acceptance of more poor exemplars in a category, F(1, 104) =
7.98, p < .01, and to higher ratings of similarity between these exemplars and
category members, F(1, 104) = 12.92, p < .01 (see Table I). In contrast, induced
positive affect led anxious–ambivalent persons to accept fewer poor exemplars,
F(1, 104) = 5.27, p < .05, and to rate these exemplars as less similar to category
P1: vendor
Motivation and Emotion [me] PP036-292899 December 29, 2000 16:17 Style file version Nov. 19th, 1999

160 Mikulincer and Sheffi

members, F(1, 104) = 5.43, p < .05 (see Table 1). For avoidant persons, no sig-
nificant difference in categorization measures was found between affect conditions.
The hierarchical regressions also revealed a marginally significant main effect
for induced affect for the number of accepted poor exemplars, Beta = .19, p < .10,
and a significant main effect for the ratings of these exemplars, Beta = .21, p <
.05. The interaction between induced affect and anxiety was also significant for
the number of accepted poor exemplars, Beta = −.89, p < .01, as well as for
the ratings of these exemplars Beta = −1.16, p < .01. No other effects were sig-
nificant. Partial correlations revealed that attachment anxiety was inversely and
significantly related to the categorization scores in the positive affect condition
(r (52) = −.31, p < .01 for acceptance of poor exemplars, r (52) = −.38, p < .01
for ratings of these exemplars), but not in the neutral condition (rs of −.01 and
−.03, ns). This pattern paralleled the ANOVA findings.

Conclusions

The findings partially supported our predictions. On the one hand, as ex-
pected, secure persons reacted to positive affect with broader categorization, and
avoidant persons failed to show any effect of positive affect on categorization. On
the another hand, unexpectedly, anxious–ambivalent persons showed narrower
categorization (less acceptance of poor exemplars in a category) upon this induc-
tion. In addition, anxious–ambivalent persons’ reports of mood were not affected
by induced positive affect. One could argue then that the events these persons
recalled were less happy than those recalled by secure and avoidant persons. How-
ever, this argument does not fit the content analysis of the written accounts that
revealed no significant attachment-style differences in the emotional tone of the
recalled events. Alternatively, one could argue that anxious–ambivalent persons
were just less sensitive to the positive affect induction. However, this argument
does not fit the finding that anxious–ambivalent persons’ categorization was still
affected by positive affect. It seems that although mood reports of these persons
did not show any effect of induced positive affect, the induction had a significant
effect on their cognitive operations.
Before drawing any conclusion, we chose to replicate the findings using other
cognitive tasks. For this purpose, Study 2 focused on the effects of positive affect
on creative problem solving and examined the moderating role of attachment style.

STUDY 2

In Study 2, we focused on Isen et al.’s (1987) finding that positive affect fa-
cilitates performance in creative problem-solving tasks, while examining whether
attachment style moderates this facilitation effect. Participants filled out measures
of attachment style, were exposed to either a positive or a neutral affect induc-
tion, as in Study 1, and then performed the Remote Associates Test (Mednick,
P1: vendor
Motivation and Emotion [me] PP036-292899 December 29, 2000 16:17 Style file version Nov. 19th, 1999

Attachment, Positive Affect, and Cognition 161

Mednick, & Mednick, 1964) used in Isen et al.’s (1987) study. In this task, par-
ticipants received 21 items, varying in the level of difficulty, and were asked to
provide the correct associate to each item. As in Isen et al.’s (1987) study, the main
dependent variable was the number of correct responses participants provided for
moderately difficult items. Easy and highly difficult items were filler items and
no prediction was made for them because of a possible restricted range of perfor-
mance in these conditions (i.e., a ceiling effect with easy items and a floor effect
with highly difficult items). Our predictions were as follows:

1. Replicating Isen et al.’s (1987) findings, participants in the positive affect


condition would provide more correct responses to moderately difficult
items in the Remote Associates Test than participants in the neutral affect
condition would.
2. This effect would be moderated by attachment style. Whereas secure per-
sons would show the predicted effect of induced positive affect, avoidant
persons would show no significant effect of positive affect. For anxious–
ambivalent persons, we made no ad hoc predictions because findings of
Study 1 were at odds with our initial predictions.

Method

Participants

One hundred and twenty undergraduate social sciences students from Bar-
Ilan University (91 women and 29 men ranging in age from 19 to 32, median =
23) participated in the study without receiving a monetary reward.

Materials and Procedure

Participants were approached during lecture time. The general instructions


and the two attachment style scales were identical to those given in Study 1.
In the self-classification procedure, 56% of the sample classified themselves as
secure (n = 67), 27% as avoidant (n = 32), and 17% as anxious–ambivalent (n =
21). No significant difference was found in this distribution between conditions
(see ns in Table II). In the 10-item attachment scale, Cronbach’s Alphas were
.73 for the anxiety items and .71 for the avoidance items, implying acceptable
internal consistency. Thus, two scores were computed by separately averaging
the anxiety and avoidance items. The association between the attachment scores
was not significant (r = .04). After completing these scales, participants were
randomly divided into two conditions (positive affect, neutral affect) according
to the procedure described in Study 1, with 60 participants in each condition.
Participants then rated their mood and the pleasantness of five unfamiliar words
(see Study 1). Cronbach’s α for the five words was high (.81), so, we averaged the
five ratings into a total score.
P1: vendor
Motivation and Emotion [me] PP036-292899 December 29, 2000 16:17 Style file version Nov. 19th, 1999

162 Mikulincer and Sheffi

Table II. Means and Standard Deviations of Mood and Remote Associates Test Scores According
to Induced Affect and Attachment Style (Study 2)
Neutral affect condition Positive affect condition
Secure Avoidant Anxious Secure Avoidant Anxious
Measures (n = 34) (n = 15) (n = 11) (n = 33) (n = 17) (n = 10)

Reported mood
M 4.85a 4.47a 4.64a 5.54b 5.59b 4.01a
SD 1.05 1.36 1.12 1.03 1.01 1.15
Number of correct responses for moderately difficult items
M 1.94a 1.67a 2.09a 3.09b 2.64ab 0.50c
SD 2.04 1.68 1.57 2.24 1.93 0.71

Note. Means within a row that have different letters were significantly different ( p < .05).

Next, participants received instructions similar to those received in Study 1


and performed a Hebrew version of Mednick et al.’s (1964) Remote Associates
Test. This test was used in Isen et al.’s (1987) study and was translated into Hebrew
by Nevo and Lewin (1978). The Hebrew version was found to be a reliable and valid
measure of creative problem solving (Nevo & Lewin, 1978). Participants received
21 items of varying difficulty. Each item consisted of three words followed by a
blank space. Participants were asked to provide, in the blank space, a word that
related to each of the three words given in the item. Participants were also told
that there was only one correct answer for an item and that they should not spend
more than 1 or 2 min on an item.
Based on Israeli norms for the test (Nevo & Lewin, 1978) and a pretest, our
version of the Remote Associates Test included 7 easy, 7 moderately difficult,
and 7 very difficult items.6 The items were presented in a random order. The
dependent variable was the number of correct responses for the seven moderately
difficult items. As expected, statistical analyses performed on the number of correct
responses for easy and difficult items yielded no significant effects for induced
affect or attachment scores.

Results and Discussion

Manipulation Check

The two-way ANOVA for induced affect and attachment style on the mood
rating revealed the expected significant main effect for induced affect, F(1, 114) =
8.50, p < .01, with participants in the positive affect condition reporting more
positive mood (M = 5.30) than did participants in the neutral affect condition (M =
4.70). This effect was qualified by a significant interaction, F(2, 114) = 4.29,
6A pre-test conducted on 20 undergraduate students confirmed the difficulty level of the 21 items.
Moreover, the difficulty level of the 21 items was corroborated in the performance of the current
sample. Specifically, easy items were more likely to be correctly answered (M = 5.07) than moderately
difficult items (M = 2.21), which in turn were more likely to be correctly answered than very difficult
items (M = 0.49). These differences were statistically significant, F(2, 238) = 295.45, p < .01.
P1: vendor
Motivation and Emotion [me] PP036-292899 December 29, 2000 16:17 Style file version Nov. 19th, 1999

Attachment, Positive Affect, and Cognition 163

p < .05. Tests for Simple Main Effects replicated the findings of Study 1: The
effect of positive affect on mood rating was found only among secure, F(1, 114) =
7.41, p < .01, and avoidant persons, F(1, 114) = 7.19, p < .01, but not among
anxious–ambivalent persons (see Ms in Table II).
The ANOVA conducted on pleasantness ratings also yielded a significant
main effect for affect induction, F(1, 114) = 6.09, p < .01, with the positive affect
condition leading to more pleasant ratings (M = 3.56) than does the neutral affect
condition (M = 3.08). The main effect for attachment style was also significant,
F(2, 114) = 11.62, p < .01. Scheffe tests (α = .05) indicated that secure persons
provided more pleasant ratings (M = 3.66) than avoidant persons (M = 3.17),
who, in turn, provided more pleasant ratings than anxious–ambivalent persons
(M = 2.42). The interaction was not significant, suggesting that positive affect
was induced in all three attachment groups.
The hierarchical regression for the mood rating revealed significant main
effects for induced affect, Beta = .23, p < .05, and attachment anxiety, Beta =
−22, p < .05, as well as a significant interaction between these two factors,
Beta = −.53, p < .05. No other effects were significant. As in Study 1, partial
correlations revealed that attachment anxiety was significantly and inversely re-
lated to mood in the positive affect condition, r (57) = −.37, p < .01, but not in the
neutral condition (−.08). The hierarchical regression for pleasantness ratings re-
vealed significant main effects for induced affect, Beta = .20, p < .05, attachment
anxiety, Beta = −.21, p < .05, and attachment avoidance, Beta = −.20, p < .05.
None of the interactions were significant. These effects paralleled the ANOVA
findings.

Remote Associates Test

The ANOVA for induced affect and attachment style indicated that the main
effect for induced affect approached statistical significance, F(1, 114) = 3.21, p =
.07, with participants in the positive affect condition performing better on moder-
ately difficult items (M = 2.53) than did participants in the neutral affect condition
(M = 1.90). In addition, there was a significant interaction between induced affect
and attachment style, F(2, 114) = 4.18, p < .05. Tests for Simple Main Effects
revealed a pattern of effects similar to those obtained in Study 1. Among secure per-
sons, the induction of positive affect, as compared to the neutral affect condition,
led to better performance on moderately difficult items, F(1, 114) = 4.82, p < .05
(see Table II). In contrast, among anxious–ambivalent persons, the induction of
positive affect led to worse performance on these items, F(1, 114) = 8.56, p < .01
(see Table 2). Avoidant persons showed no significant difference between the two
affect conditions in creative problem solving.
The hierarchical regression conducted on the number of correct responses for
moderately difficult items revealed a significant main effect for induced affect,
Beta = .22, p < .05, and a significant interaction between induced affect and
P1: vendor
Motivation and Emotion [me] PP036-292899 December 29, 2000 16:17 Style file version Nov. 19th, 1999

164 Mikulincer and Sheffi

attachment anxiety, Beta = −.71, p < .05. Partial correlations revealed that at-
tachment anxiety was significantly and inversely related to creative problem-
solving performance in the positive affect condition, r (57) = −.35, p < .01, but
not in the neutral condition (−.04). No other effects were significant, implying that
only variability along the anxiety dimension was significantly related to creative
problem solving, regardless of a person’s score on the avoidance dimension. This
pattern of associations replicated the ANOVA findings.

Conclusions

The findings replicated those of Study 1. Secure persons reacted to positive


affect with improved performance in a creative problem-solving task, anxious–
ambivalent persons with impaired performance in this task, and avoidant persons
failed to show any effect of positive affect on creative problem solving. However,
despite the fact that the effects of attachment style were replicated across different
dependent variables, there were two issues that needed to be examined before
a firm conclusion could be reached. First, Studies 1 and 2 used the same affect
induction and it was important to replicate these effects using a different affect
induction. Second, it was important to examine whether these effects are unique
to attachment style or may reflect the action of other personality factors, such as
trait anxiety. In fact, trait anxiety may be related to the dimension of attachment
anxiety (Brennan & Shaver, 1995) and may exert an effect on cognitive processes
(Sarason, 1975). Study 3 was designed to deal with these issues.

STUDY 3

In Study 3, we attempted to replicate the findings of Study 2, while controlling


for trait anxiety and employing a different induction of positive affect—watching
a comedy video. This manipulation has been used in previous studies and has been
found to induce positive affect (Isen, 1987). Participants filled out measures of trait
anxiety and attachment style, watched a comedy or a documentary video, and then
performed the Remote Associates Test. Our predictions were identical to those of
Study 2.

Method

Participants

Another independent sample of 120 undergraduate social sciences students


from Bar-Ilan University (84 women and 36 men ranging in age from 19 to 30,
median = 23) participated in the study without receiving a monetary reward.
P1: vendor
Motivation and Emotion [me] PP036-292899 December 29, 2000 16:17 Style file version Nov. 19th, 1999

Attachment, Positive Affect, and Cognition 165

Table III. Means and Standard Deviations of Remote Associates Test


Scores According to Induced Affect and Attachment Style (Study 3)
Attachment styles
Secure Avoidant Anxious

Number of correct responses for moderately difficult items


Neutral affect condition
n 36 15 9
M 1.86a 2.07a 2.01a
SD 1.92 1.71 1.70
Positive affect condition
n 32 18 10
M 3.47b 2.44ab 0.70c
SD 1.98 1.38 1.16

Note. Means with different letters within rows and columns were signif-
icantly different ( p < .05).

Materials and Procedure

The study consisted of two sessions. In the first session, conducted during lec-
ture time, participants completed the two Attachment Style measures described in
Study 1. In the self-classification measure, 57% of the sample classified themselves
as secure (n = 68), 28% as avoidant (n = 33), and 15% as anxious–ambivalent
(n = 19). No significant difference was found in this distribution between condi-
tions (see ns in Table III). Regarding the 10-item attachment scale, Cronbach’s αs
were .75 for anxiety items and .76 for avoidance items, allowing us to compute
two total scores by averaging the relevant items. The association between the two
attachment scores was not significant (r = .12, ns). In this session, participants
also completed a Hebrew version of the trait form of the State-Trait Anxiety Ques-
tionnaire (STAI, Spielberger, Gorsuch, & Lushene, 1970). This scale consisted of
20 statements tapping the cognitive, affective, and behavioral manifestations of
anxiety. Participants rated the extent to which they agreed with each statement on
a 4-point scale, ranging from 1 (totally disagree) to 4 (totally agree). Cronbach’s
α for the 20 items in our sample was adequate (.79), allowing us to compute a trait
anxiety score by averaging the 20 items.
One week later, participants were escorted individually into a testing room by
a new experimenter who was blind to their attachment style, and were randomly
divided into two conditions. In the positive affect condition (n = 60), participants
watched a 10-min humorous video skit by a popular comedy group. In the neutral
affect condition (n = 60), participants watched a 10-min documentary video on
animal life. After the film, participants rated their mood and the pleasantness of
five unfamiliar words (see Study 1). Cronbach’s α for the five pleasantness ratings
was high (.84). Next, participants performed the Hebrew version of the Remote
P1: vendor
Motivation and Emotion [me] PP036-292899 December 29, 2000 16:17 Style file version Nov. 19th, 1999

166 Mikulincer and Sheffi

Associates Test (see Study 2). The dependent variable was the number of correct
responses for moderately difficult items.7

Results and Discussion

Manipulation Check

The ANOVAs for induced affect and attachment style revealed the expected
significant main effect for induced affect on both global mood, F(1, 114) =
9.82, p < .01, and pleasantness rating, F(1, 114) = 14.88, p < .01, with the pos-
itive affect induction producing more positive mood (M = 5.04) and more pleasant
ratings (M = 4.07) than a neutral condition (M = 4.06, M = 3.23, respectively).
The main effect for attachment style was also significant on both global mood,
F(2, 114) = 3.27, p < .05 , and pleasantness ratings, F(2, 114) = 4.91, p < .05.
Scheffé tests indicated that secure and avoidant persons reported more positive
mood (M = 4.76, M = 4.98) and more pleasant ratings (M = 3.94, M = 3.86)
than anxious–ambivalent persons (M = 3.91, M = 3.17). The interaction was not
significant for the two mood measures (F < 1), implying that the comedy video
produced more positive affect than the documentary video in all of the attachment-
style groups.
The hierarchical regressions revealed significant main effects of induced af-
fect, Beta = .26, p < .01 for global mood, and Beta = .32, p < .01 for pleasant-
ness rating. The main effect of attachment anxiety was also significant, Beta =
−.21, p < .05 for global mood, and Beta = −.24, p < .05 for pleasantness rat-
ing. Neither the main effect of attachment avoidance nor the interactions were
significant. These associations paralleled the ANOVA findings.

Remote Associates Test

The ANOVA performed on the number of correct responses for moderately


difficult items replicated the findings of Study 2. The main effect for induced af-
fect was significant, F(1, 114) = 5.59, p < .01, with the positive affect condition
leading to better performance (M = 2.70) than does the neutral affect condition
(M = 1.94). Again, the interaction effect was significant, F(2, 114) = 5.28, p <
.01. Among secure persons, the positive affect induction, as compared to the
neutral condition, led to more correct responses on moderately difficult items,
F(1, 114) = 11.55, p < .01 (see Table III). In contrast, the positive affect induc-
tion led anxious–ambivalent persons to perform worse on moderately difficult
7 Asin Study 2, ANOVAs performed on correct responses for easy and very difficult items yielded no
significant effect for induced affect or attachment style. In addition, no significant association was
found between these measures and attachment scores.
P1: vendor
Motivation and Emotion [me] PP036-292899 December 29, 2000 16:17 Style file version Nov. 19th, 1999

Attachment, Positive Affect, and Cognition 167

items, F(1, 114) = 3.77, p < .05 (see Table III). There was not a significant dif-
ference between the two affect conditions for avoidant persons.
The hierarchical regression conducted on the number of correct responses
for moderately difficult items revealed a significant main effect for induced af-
fect, Beta = .20, p < .05, and a significant interaction between induced affect
and attachment anxiety, Beta = −.64, p < .05. As in Study 2, partial correlations
revealed that attachment anxiety was significantly and inversely related to creative
problem-solving performance in the positive affect condition, r (57) = −.34, p <
.01, but not in the neutral condition (−.11). No other effects were significant, im-
plying that only variability along the anxiety dimension was significantly related
to creative problem solving.
An examination of the possible confounding between trait anxiety and at-
tachment anxiety indeed yielded a significant Pearson correlation between these
two measures, r (118) = .35, p < .01. However, the findings indicated that trait
anxiety could not explain the observed interactive effects of induced affect and
attachment style on creative problem solving. First, a two-way ANCOVA revealed
that the interaction of affect induction and attachment style on the number of
correct responses for moderately difficult items was significant even after con-
trolling both trait anxiety and the interaction between induced affect and trait
anxiety as covariates, F(2, 112) = 4.73, p < .01. Second, a hierarchical regres-
sion revealed that the interaction effect of induced affect and attachment anxiety
on creative problem solving was still significant even after introducing trait anxiety
and the interaction between induced affect and trait anxiety as additional predic-
tors, Beta = −.72, p < .01. Third, additional hierarchical regressions indicated
that the main effect for trait anxiety and the interaction effect for induced affect
and trait anxiety on creative problem solving were not significant either before or
after introducing attachment scores as predictors.

Conclusions

The findings replicated the findings of Studies 1 and 2 and extended them to
a different induction of positive affect. In addition, the findings indicated that trait
anxiety could not explain the observed attachment-style differences.

GENERAL DISCUSSION

During the last decade, attachment research in adulthood has mainly focused
on attachment-style differences in the regulation of negative affect. Our study
extends attachment research and indicates that attachment style is an important
individual difference factor in explaining cognitive reactions to positive affect.
However, the findings only partially supported our hypotheses. On the one hand,
in line with our predictions, secure persons reacted to positive affect inductions
P1: vendor
Motivation and Emotion [me] PP036-292899 December 29, 2000 16:17 Style file version Nov. 19th, 1999

168 Mikulincer and Sheffi

with more liberal and inclusive criteria in the categorization of semantic stimuli
and better creative problem-solving performance. As expected, avoidant persons
showed no significant difference in categorization and creative problem solving
between positive affect and neutral affect conditions. On the other hand, anxious–
ambivalent persons showed two unexpected responses to positive affect. First, in
Studies 1 and 2, these persons failed to report enhanced positive mood as a result
of a positive affect induction. Second, in all three studies, anxious–ambivalent
persons showed less inclusive categorization and worse creative problem-solving
performance in positive than in neutral affect conditions. It is important that these
effects of positive affect on cognitive processing were obtained with different mood
manipulations.
In interpreting the observed findings, we want to propose a theoretical integra-
tion based on the “mood as an input” approach (Martin et al., 1993). In this view,
the interpretation of positive affect seems to determine the effects of positive affect
on cognitive processing. Our findings suggest the following attachment-style dif-
ferences in the interpretation of positive affect. First, secure persons may interpret
positive affect as a sign that something good is happening, which in turn may lead
them to “loosen” their cognitive strategies and then to perform cognitive tasks in
a relaxed manner and to explore unusual stimuli. Second, avoidant persons may
interpret positive affect as an irrelevant input for cognitive processing and there-
fore may not be affected by it. Third, anxious–ambivalent persons may interpret
positive affect as a danger cue, because in the past it may have caused them to let
down their guard, which resulted in painful or dangerous consequences (Shaver
& Hazan, 1993). As a result, these persons may be wary about being happy and
may react defensively against positive affect by either denying it or tightening up
their cognitive strategies and avoiding relaxed exploration of unusual stimuli. In
the following paragraphs, we elaborate on this integrative approach.
With regard to secure persons, the findings fit the overall effects of positive
mood inductions that were originally found in Isen’s studies (Isen & Daubman,
1984; Isen et al., 1987). In line with the “mood as an input” approach, we want
to suggest a three-step account of the cognitive processing of positive affect that
may explain secure persons’ reactions. First, they may interpret positive affect as
a signal that “all is going well,” that they have made progress toward their goals,
or that they are enjoying what they are doing, or both. This positive interpretation
may reflect the activation of the rich and varied positive material secure persons
hold in their semantic memory (Collins & Read, 1994). Second, secure persons
may consider the above signals as a highly relevant input for cognitive processing.
This high relevance may reflect secure persons’ habitual attention to affective cues
in interpersonal and problem-solving settings (Fuendeling, 1998). Third, secure
persons may react positively to the above signals by “loosening” their cognitive
strategies and engaging in playful and creative exploration. This reaction may re-
flect secure persons’ confidence that letting down their guard will not be dangerous.
P1: vendor
Motivation and Emotion [me] PP036-292899 December 29, 2000 16:17 Style file version Nov. 19th, 1999

Attachment, Positive Affect, and Cognition 169

In fact, attachment security has been found to strengthen a person’s confidence


in dealing with uncertainty and to promote exploration and cognitive openness
(Mikulincer, 1997). It should be noted, however, that our study was only a first
step in documenting secure persons’ reactions to positive affect, and no data were
collected about the three proposed cognitive steps.
Following the above reasoning, secure persons’ responses to positive affect
may fit Bowlby’s (1969) idea that “felt security” during positive interactions with
warm and sensitive significant others is associated with the arousal of positive
emotions, such as relief, calmness, and gratitude. Recurrent experiences of these
interactions may form a strong link between “felt security” and positive affect
within the semantic memory network. As a result, the induction of positive affect
may reactivate a person’s sense of “felt security,” which is a basic inner resource
(Mikulincer & Florian, 1998) and may allow secure persons to loosen their cog-
nitive strategies and engage in playful exploration. This line of interpretation fits
resource-based approaches, which contend that positive affect may increase the
appraisal of the strength or adequacy of one’s resources to deal with the environ-
ment (e.g., Aspinwall, 1998; Schwarz & Bohner, 1996; Trope & Neter, 1994).
This interpretation is admittedly speculative, however, and further studies should
examine the association between positive affect and the sense of “felt security.”
Our findings are consistent with the claim that attachment security is linked
to adjustment and well-being (Shaver & Hazan, 1993). It seems that secure per-
sons not only deal actively and constructively with negative affect (Mikulincer &
Florian, 1998), but also take advantage of the enhanced creativity derived from the
arousal of positive affect. This enhanced creativity may allow secure persons to
find new and unusual ways to deal with environmental transactions, to enjoy task
performance, and to maintain positive mood. It may also allow secure persons to
adjust to environmental demands, to set realistic goals, and to avoid any cognitive
or motivational entrapment derived from the inability to revise rigid beliefs. In this
way, the enhanced creativity derived from positive affect may be an additional fac-
tor underlying secure persons’ ability to maintain mental health and psychological
well-being.
With regard to avoidant persons, the findings indicated that positive affect
had no significant effect on mental categorization and creative problem solving.
From a “mood as an input” perspective, these findings suggest that avoidant per-
sons do not regard positive affect as a relevant input for cognitive processing. This
dismissing reaction may reflect avoidant persons’ tendency to distance themselves
mentally from affective material (Dozier & Kobak, 1992), which may lead them
to consider affective cues as an irrelevant input for cognitive processing. Alter-
natively, the dismissal of positive affect may result from a discrepancy between
the information implied by this affective cue (i.e., one can let down one’s guard)
and avoidant persons’ reluctance to “loosen” their cognitive strategies. Accord-
ing to Mikulincer (1997), the habitual regulatory strategies of avoidant persons
P1: vendor
Motivation and Emotion [me] PP036-292899 December 29, 2000 16:17 Style file version Nov. 19th, 1999

170 Mikulincer and Sheffi

block cognitive exploration and the incorporation of new evidence and unusual
thoughts, because these thoughts may challenge prior knowledge and lead to con-
fusion. Moreover, avoidant persons’ overemphasis on self-reliance (Bowlby, 1988)
may lead them to reject any input that demands a revision of their beliefs. As a re-
sult, these persons may dismiss the information implied by positive affect, because
it goes against their habitual cognitive closure and rigidity. Future studies should
explore in greater depth the mechanisms underlying avoidant persons’ dismissal
of positive affect.
With regard to anxious–ambivalent persons, the findings were unexpected.
We did not expect that their reports of mood would be unaffected by positive
affect inductions. Moreover, we did not expect that they would show more narrow
categorization and worse creative problem solving in positive than in neutral affect
conditions. However, the “mood as an input” approach can provide a plausible post
hoc explanation of the findings.
According to Bowlby (1988), one of the antecedents of the anxious–
ambivalent style is the failure of significant others to provide a “secure base”
when exploration becomes dangerous and to alleviate distress in these circum-
stances. As a result, anxious–ambivalent persons may exaggerate the appraisal of
the potential threats involved in playful exploration and may have serious doubts
about their ability to deal with these threats (Mikulincer, 1997). These painful
experiences may also lead anxious–ambivalent persons to adopt a rigid, hyper-
vigilant attitude towards novelty and uncertainty, and to perceive themselves as
vulnerable and helpless whenever they let down their guard (Mikulincer, 1997).
On this basis, we want to suggest that positive affect, which signals that “all is
going well” and favors the loosening of cognitive strategies, may be perceived as a
danger cue and may reactivate anxious–ambivalent persons’ distressing memories
and worries concerning such cognitive “loosening.” This negative interpretation
of positive affect may in turn activate defensive reactions aimed at either denying
this affective experience (“I do not feel good, so there is no signal that all is going
well”) or tightening up cognitive strategies. By either means, anxious–ambivalent
people may prevent distress resulting from worries about cognitive relaxation and
openness.
The finding that anxious–ambivalent persons failed to report enhanced mood
following a positive affect induction may reflect their defensive attempts to deny
this affective experience and its cognitive implications. The findings that positive
affect inductions narrowed anxious–ambivalent persons’ mental categorization and
impaired their creative problem-solving performance may reflect their defensive
tightening up of cognitive strategies. It is possible that anxious–ambivalent persons
are wary of being happy and that they defend themselves from the recognition
that “all is going well,” as well as from the cognitive relaxation that follows this
recognition. Of course, this is only a post hoc explanation that demands further
research on anxious–ambivalent persons’ wariness of being happy as well as on
the defensive functions of their reactions.
P1: vendor
Motivation and Emotion [me] PP036-292899 December 29, 2000 16:17 Style file version Nov. 19th, 1999

Attachment, Positive Affect, and Cognition 171

It is noteworthy that anxious–ambivalent persons’ failure to report enhanced


mood following a positive affect induction was found when this induction consisted
of the recall of a happy life event (Studies 1 and 2), but not when it consisted of
watching a comedy video (Study 3). One can speculate that it may be easier to deny
the experience of a positive affect upon the recall of a happy personal event than
in response to a comedy video. Whereas the recall of a positive event is a private
experience about which a person may feel ambivalent, watching a comedy video
is a public activity that may have little personal significance and may create social
demands to report positive mood. This latter tendency may be exacerbated among
anxious–ambivalent people who tend to be highly sensitive to social pressure
(Shaver & Hazan, 1993).
Other alternative explanations can be offered. First, one might argue that
anxious–ambivalent persons might have had difficutly coming up with a positive
autobiographical story. As a result, they might have discounted any affect they
experienced from the story. Second, one might argue that anxious–ambivalent per-
sons really entered a positive affect state, but that they were reluctant to report that
they were feeling good. This self-presentational bias may fit anxious–ambivalent
persons’ tendency to present themselves as helpless and weak (Shaver & Hazan,
1993). In any case, the findings suggest that these persons may be wary about
being happy and that this wariness may have important implications for the self-
regulation process in other life domains. For example, this wariness may underlie
the way anxious–ambivalent persons set goals, organize cognitions and behaviors
during goal-directed activities, and react to goal attainment.
It might be questioned whether the above pattern of findings was uniquely
attributable to the anxious–ambivalent style or whether it is an effect of more gen-
eral trait anxiety. This is a reasonable concern, because trait anxiety seems to be
related to worse performance in cognitive tasks (Sarason, 1975). Moreover, our
findings revealed a significant association between trait anxiety and attachment
anxiety. However, the association between attachment anxiety and creative prob-
lem solving performance remained the same even after statistically controlling
for trait anxiety and its interaction with induced affect (Study 3). Moreover, trait
anxiety had no significant effect on creative problem solving. This finding seems
to rule out the possibility that trait anxiety underlies the observed attachment-style
differences and provides unique support to the attachment perspective.
It is worth mentioning that our laboratory inductions of positive affect might
have been shorter, less intense, and less personally relevant than the arousal of
positive affect in naturalistic settings. Although previous studies have used sim-
ilar laboratory inductions of positive affect, these possible differences may be
relevant for understanding the observed attachment-style differences. It is possi-
ble that insecure persons would show the typical reactions to positive affect only
upon the experience of intense, relevant, and long episodes of positive affect, be-
cause only those episodes can counteract their habitual distress and defensiveness.
Other researchers should attempt to replicate our findings using other types of
P1: vendor
Motivation and Emotion [me] PP036-292899 December 29, 2000 16:17 Style file version Nov. 19th, 1999

172 Mikulincer and Sheffi

positive affect inductions. They might also examine attachment-style differences


in the effects of positive affect on decision-making, helping, persuasion, and social
judgments. Overall, the current findings provide valuable information about the
involvement of adult attachment style during episodes of positive affect.

REFERENCES

Ainsworth, M. D. S., Bell, S. M. V., & Stayton, D. J. (1973). Individual differences in the strange-
situation behavior of one-year olds. In L. S. Stone, H. T. Smith, & L. B. Murphy (Eds.), The
competent infant. New York: Basic Book.
Aspinwall, L. G. (1998). Rethinking the role of positive affect in self-regulation. Motivation and
Emotion, 22, 1–32.
Bartholomew, K., & Horowitz, L. M. (1991). Attachment styles among young adults: A test of a
four-category model. Journal of Personality and Social Psychology, 61, 226–244.
Birnbaum, G. E., Orr, I., Mikulincer, M., & Florian, V. (1997). When marriage breaks up—Does attach-
ment style contribute to coping and mental health? Journal of Social and Personal Relationships,
14, 643–654.
Boucher, J., & Osgood, C. E. (1969). The Pollyanna hypothesis. Journal of Verbal Learning and Verbal
Behavior, 8, 1–8.
Bowlby, J. (1969). Attachment and loss: Attachment. New York: Basic Books.
Bowlby, J. (1973). Attachment and loss: Separation, anxiety and anger. New York: Basic Books.
Bowlby, J. (1980). Attachment and loss: Sadness and depression. New York: Basic Books.
Bowlby, J. (1988). A secure base: Clinical applications of attachment theory. London: Routledge.
Brennan, K. A., Clark, C. L., & Shaver, P. R. (1998). Self report measurement of adult attachment:
An integrative overview. In J. A. Simpson & W. S. Rholes (Eds.), Attachment theory and close
relationships (pp. 46–76). NY: Guilford.
Brennan, K., & Shaver, P. R. (1995). Dimensions of adult attachment, affect regulation, and romantic
relationship functioning. Personality and Social Psychology Bulletin, 21, 567–583.
Collins, N. L., & Read, S. J. (1994). Cognitive representations of attachment: The structure and function
of working models. In K. Bartholomew & D. Perlman (Eds.), Attachment processes in adulthood
(pp. 53–92). London: Jessica Kingsley.
Dozier, M., & Kobak, R. R. (1992). Psychophysiology in attachment interviews: Converging evidence
for deactivating strategies. Child Development, 63, 1473–1480.
Estrada, C., Isen, A. M., & Young, M. J. (1994). Positive affect improves creative problem solving
and influences reported source of practice satisfaction in physicians. Motivation and Emotion, 18,
285–299.
Estrada, C., Isen, A. M., & Young, M. J. (1997). Positive affect facilitates integration of information
and decreases anchoring in reasoning among physicians. Organizational Behavior and Human
Decision Processes, 72, 117–135.
Feeney, B. C., & Kirkpatrick, L. A. (1996). Effects of adult attachment and presence of romantic
partners on physiological responses to stress. Journal of Personality and Social Psychology, 70,
255–270.
Fraley, R. C., & Shaver, P. R. (1997). Adult attachment and the suppression of unwanted thoughts.
Journal of Personality and Social Psychology, 73, 1080–1091.
Fraley, R. C., & Shaver, P. R. (1998). Airport separations: A naturalistic study of adult attach-
ment dynamics in separating couples. Journal of Personality and Social Psychology, 75, 1198–
1212.
Fraley, R. C., & Waller, N. G. (1998). Adult attachment patterns: A test of typological model. In
J. A. Simpson & W. S. Rholes (Eds.), Attachment theory and close relationships (pp. 77–114).
NY: Guilford.
Fuendeling, J. M. (1998). Affect regulation as a stylistic process within adult attachment. Journal of
Social and Personal Relationships, 15, 291–322.
P1: vendor
Motivation and Emotion [me] PP036-292899 December 29, 2000 16:17 Style file version Nov. 19th, 1999

Attachment, Positive Affect, and Cognition 173

Green-Hennessy, S., & Reis, H. T. (1998). Openness in processing social information among attachment
types. Personal Relationships, 5, 449–466.
Hazan, C., & Shaver, P. R. (1987). Romantic love conceptualized as an attachment process. Journal of
Personality and Social Psychology, 52, 511–524.
Hirt, E. R., Levine, G. M., McDonald, H. E., & Melton, R. J. (1997). The role of mood in quantitative
and qualitative aspects of performance: Single or multiple mechanisms? Journal of Experimental
Social Psychology, 33, 602–629.
Hirt, E. R., Melton, R. J., McDonald, H. E., & Harackiewicz, J. M. (1996). Processing goals, task
interest, and the mood-performance relationship: A mediational analysis. Journal of Personality
and Social Psychology, 71, 245–261.
Isen, A. M. (1987). Positive affect, cognitive processes, and social behavior. Advances of Experimental
Social Psychology, 20, 203–253.
Isen, A. M. (1993). Positive affect and decision making. In M. Lewis & J. M. Haviland (Eds.), Handbook
of emotions (pp. 261–277). New York: Guilford Press.
Isen, A. M., & Daubman, K. A. (1984). The influence of affect on categorization. Journal of Personality
and Social Psychology, 47, 1206–1217.
Isen, A. M., Daubman, K. A., & Nowicki, G. P. (1987). Positive affect facilitates creative problem
solving. Journal of Personality and Social Psychology, 52, 1122–1131.
Isen, A. M., Johnson, M. M. S., Mertz, E., & Robinson, G. F. (1985). The influence of positive affect
on the unusualness of word associations. Journal of Personality and Social Psychology, 48, 1–
14.
Isen, A. M., Niedenthal, P. M., & Cantor, N. (1992). An influence of positive affect on social catego-
rization. Motivation and Emotion, 16, 65–78.
Isen, A. M., Rosenzweig, A. S., & Young, M. J. (1991). The influence of positive affect on clinical
problem solving. Medical Decision Making, 11, 221–227.
Kotler, T., Buzwell, S., Romeo, Y., & Bowland, J. (1994). Avoidant attachment as a risk factor for
health. British Journal of Medical Psychology, 67, 237–245.
Martin, L. L., Ward, D. W., Achee, J. W., & Wyer, R. S. Jr. (1993). Mood as input: People have to
interpret the motivational implications of their moods. Journal of Personality and Social Psychol-
ogy, 64, 317–326.
Mednick, M. T., Mednick, S. A., & Mednick, E. V. (1964). Incubation of creative performance and
specific associative priming. Journal of Abnormal and Social Psychology, 69, 84–88.
Mikulincer, M. (1997). Adult attachment style and information processing: Individual differences in
curiosity and cognitive closure. Journal of Personality and Social Psychology, 72, 1217–1230.
Mikulincer, M. (1998). Adult attachment style and individual differences in functional versus dysfunc-
tional experiences of anger. Journal of Personality and Social Psychology, 74, 513–524.
Mikulincer, M., & Florian, V. (1995). Appraisal and coping with a real-life stressful situation: The
contribution of attachment styles. Personality and Social Psychology Bulletin, 21, 408–416.
Mikulincer, M., & Florian, V. (1997). Are emotional and instrumental supportive interactions beneficial
in times of stress? The impact of attachment style. Anxiety, Stress, and Coping, 10, 109–127.
Mikulincer, M., & Florian, V. (1998). The relationship between adult attachment styles and emotional
and cognitive reactions to stressful events. In J. A. Simpson & W. Rholes (Eds.), Attachment
theory and close relationships (pp. 143–165). NY: Guilford.
Mikulincer, M., Florian, V., & Tolmacz, R. (1990). Attachment styles and fear of personal death: A
case study of affect regulation. Journal of Personality and Social Psychology, 58, 273–280.
Mikulincer, M., Florian, V., & Weller, A. (1993). Attachment strategies, and posttraumatic psycholog-
ical distress: The impact of the Gulf war in Israel. Journal of Personality and Social Psychology,
64, 817–826.
Mikulincer, M., & Orbach, I. (1995). Attachment styles and repressive defensiveness: The accessibility
and architecture of affective memories. Journal of Personality and Social Psychology, 68, 917–
925.
Mikulincer, M., Orbach, I., & Iavnieli, D. (1998). Adult attachment style and affect regulation: Strategic
variations in subjective self-other similarity. Journal of Personality and Social Psychology, 75,
436–448.
Murray, N., Sujan, H., Hirt, E. R., & Sujan, M. (1990). The influence of mood on categorization: A
cognitive flexibility interpretation. Journal of Personality and Social Psychology, 59, 411–425.
P1: vendor
Motivation and Emotion [me] PP036-292899 December 29, 2000 16:17 Style file version Nov. 19th, 1999

174 Mikulincer and Sheffi

Nevo, B., & Lewin, A. (1978). The Remote Associates Test—A measurement tool for creative thinking.
Megamot, 24, 87–98.
Pistole, M. C. (1995). College students’ ended love relationships: Attachment style and emotion.
Journal of College Student Development, 36, 53–60.
Pretty, G. H., & Seligman, C. (1984). Affect and the overjustification effect. Journal of Personality and
Social Psychology, 46, 1241–1253.
Priel, B., & Shamai, D. (1995). Attachment style and perceived social support. Personality and Indi-
vidual Differences, 19, 235–241.
Radecki-Bush, C., Farrel, A. D., & Bush, J. P. (1993). Predicting jealous responses: The influence of
adult attachment and depression on threat appraisal. Journal of Social and Personal Relationships,
10, 569–588.
Rosch, E. (1975). Cognitive representations of semantic categories. Journal of Experimental Psychol-
ogy: General, 104, 192–233.
Sarason, I. G. (1975). Anxiety and self-preoccupation. In I. G. Sarason & C. D. Spielberger (Eds.),
Stress and anxiety (Vol. 2, pp. 27–44). New York: Wiley.
Schwarz, N., & Bohner, G. (1996). Feelings and their motivational implications: Moods and the action
sequence. In P. M. Gollwitzer & J. A. Bargh (Eds.), The psychology of action: Linking cognitions
and motivation to behavior (pp. 119–145). New York: Guilford Press.
Shaver, P. R., & Hazan, C. (1993). Adult romantic attachment: Theory and evidence. In D. Perlman &
W. Jones (Eds.), Advances in personal relationships (pp. 29–70). London: Jessica Kingsley.
Showers, C., & Cantor, N. (1985). Social cognition: A look at motivated strategies. Annual Review of
Psychology, 36, 275–305.
Simpson, J. A., Rholes, W. S., & Nelligan, J. S. (1992). Support seeking and support giving within
couples in an anxiety-provoking situation: The role of attachment styles. Journal of Personality
and Social Psychology, 62, 434–446.
Spielberger, C. D., Gorsuch, R. L., & Lushene, R. E. (1970). Manual for the State-Trait Anxiety
Inventory. Palo Alto, CA: Consulting Psychological Press.
Trope, Y., & Neter, E. (1994). Reconciling competing motives in self-evaluation: The role of self-control
in feedback seeking. Journal of Personality and Social Psychology, 66, 646–657.
Wegener, D. T., & Petty, R. E. (1994). Mood management across affective states: The hedonic contin-
gency hypothesis. Journal of Personality and Social Psychology, 66, 1034–1048.

View publication stats

You might also like