psp-61-5-708

Journal of Personality and Social Psychology 1991, Vol. 61, No. 5, 708-720 Copyright 1991 by the American Psychological Association, Inc. 0022-3514/91/S3.00 Affective Influences on Partner Choice: Role of Mood in Social Decisions Joseph P. Forgas University of New South Wales Sydney, New South Wales, Australia Does mood influence our information search and decision strategies when choosing a partner? In Experiment 1 (N = 60), sad Ss preferred rewarding to competent partners and remembered information supporting that choice better. In Experiment 2 (N = 96), mood effects on information selectivity, decision speed, and processing strategy in partner choices were found. In Experiment 3 (N = 42), a computerized stimulus presentation revealed mood-induced differences in the latency, self-exposure, and eventual recall of interpersonal information. These results are interpreted as evidence for mood-induced selectivity in information search and decision strategies when making realistic partner choices. The implications of the findings for research on interpersonal relations and for contemporary affect-cognition theories are considered. Feelings often influence the quality of our social judgments, a fact well recognized by writers, artists, and philosophers throughout the ages. Descartes (1649/1961) postulated such a link between feelings and thinking almost 350 years ago, as did James (1890) in his classic Principles of Psychology. Yet psychological research on this intriguing problem is a relatively recent development and little work has been done so far on the role of moods in motivating people's interpersonal choices. Making preference choices about people is a recurring feature both of our private and of our working lives. Selecting a work partner, deciding who to sit next to at lunch today, who to approach at a cocktail party, who to invite to a dinner party, or who to ask for advice on a personal matter are examples of such person preference tasks. Affect can play a major role in such affiliative choices, even in superficial relationships (Schachter, 1959; Wenzlaff & Prohaska, 1989). The three experiments described here are explorations of the motivational role of temporary moods in interpersonal choices and the information processing strategies used by people in such decisions. Motivated processing is defined here as involving directed search strategies and selective attentional focus on information that is diagnostic of potentially rewarding outcomes and involving consequent biases in memory as well as choice outcomes. 1990). Such decisions—like most social judgments—usually require highly elaborated, constructive cognitive processing dealing with the inferred personal characteristics of others (Heider, 1958). There is now growing evidence that affect can play an important role in the way information about another person is attended to, selected, interpreted, learned, remembered, and evaluated (Bower, 1981,1991; Clore & Parrott, 1991; Fiedler, 1990,1991; Forgas, 1991a; Forgas & Bower, 1987; Forgas, Bower, & Krantz, 1984; Isen, 1984; Schwarz & Bless, 1991). Clinical research also suggests that enduring affective states, such as depression, have a marked influence on interpersonal judgments and decision making (Wenzlaff & Prohaska, 1989). Depressed people are more sensitive to risks (Pietromonaco & Rook, 1987), are more negative in their social judgments (Roth & Rehm, 1980), and tend to make more cautious, conservative choices than do nondepressives (Weary & Williams, 1990). Of course, the role of affect in interpersonal behavior has been recognized for some time. As Zajonc (1980) argued, affect is likely to be the primary medium of interpersonal behavior, and affective reactions often precede and inform subsequent cognitive elaborations (Clore & Parrott, 1991). Early work by Schachter (1959) and Sarnoff and Zimbardo (1961) showed that preference for partners in a similar predicament may be part of a motivated strategy to control an aversive emotional state. In fact, interacting with others in a matching mood may be intrinsically satisfying (Locke & Horowitz, 1990). In general, motivated interpersonal choices can serve to selectively enhance or impair mood and self-esteem (Tesser, 1986; Wenzlaff & Prohaska, 1989). Despite extensive recent interest in social cognition, the cognitive processes underlying such motivated partner choices have received little attention. Affect and Interpersonal Preferences The decision of whether to engage in an interaction with a potential partner is one of the most common yet complex and demanding tasks in everyday life. Most social relationships by definition involve recurring positive partner choices at every level from unilateral awareness to enduring intimacy (Levinger, Support for this research by the Australian Research Council, the German Research Foundation, and ZUMA, Mannheim, Germany, and comments on an earlier version of this article by Robert Wyer, Jr., Klaus Fiedler, and Norbert Schwarz are gratefully acknowledged. Correspondence concerning this article should be addressed to Joseph P. Forgas, School of Psychology, University of New South Wales, P.O. Box 1, Kensington 2033, Sydney, New South Wales, Australia. Electronic mail may be sent to forgas@hydra.maths.unsw.oz.au. Motivated Choice Strategies Models of information accessibility and memory underlie most current explanations of social judgmental processes (cf. Forgas, 1981). However, considerable evidence now also shows that affective states do have a major influence on the processing of social information (Bower, 1991; Clore & Parrott, 1991; Fiedler, 1991; Forgas. 1991 a; Forgas & Bower, 1987; Isen, 1984; 708 AFFECT AND PARTNER CHOICE 709 Schwarz & Bless, 1991). There is also a growing recognition that motivation plays an important role in many cognitive processes, particularly when it comes to realistic social decisions (Erber & Fiske, 1984; Tesser, 1986). Early evidence for motivated changes in attitudes and self-perception comes from cognitive dissonance research (Wicklund & Frey, 1981), an emphasis echoed by more recent attitude formulations (Tesser & Shaifer, 1990). Motivated thinking also plays a central role in intergroup behavior (Brewer & Kramer, 1985), stereotyping (Tajfel & Forgas, 1981), decision making (Janis & Mann, 1977), reasoning (Kunda, 1990), persuasive communication (Petty, Gleicher, & Baker, 1991), and impression formation (Brewer, 1988; Fiske & Neuberg, 1990). In memory research, Sanitioso, Kunda, and Fong (1990) recently found that motivational states enhanced the accessibility of memories, and Parrott and Sabini (1990) suggested that motivational factors underlie the selective recall of mood-incongruent memories they observed. Even quite subtle changes in motivational objectives may result in dramatic information-processing consequences, as Martin's recent set/reset theory suggests (1986). However, little or no work has been done so far on the role of mood states in motivated interpersonal choices. Motivated processing is only one of several alternative strategies available to people when making social choices. In a recent multiprocess model of mood effects on social judgments, Forgas (1991b) identified four alternative processing strategies in everyday judgments: (a) a direct access strategy, when crystallized, preexisting judgments are directly retrieved from memory; (b) a motivated processing strategy, when preexisting goals guide a selective search for and guide use of information to serve a motivational objective; (c) a heuristic processing strategy, when a simplified, effort-minimizing strategy is adopted; and (d) a substantive processing strategy, when subjects deal with the available information in a relatively unbiased manner, relying on learning, associative, and memory processes to arrive at a novel judgment. Which of these processing strategies is adopted depends on a combination of factors, such as the preexisting mood state and the personal relevance of the task. Furthermore, we may distinguish between at least two kinds of motivated processing: (a) a general motivation, for example to minimize effort, to be accurate, or to avoid distortions due to selective priming (Isen, 1984; Martin, 1986; Martin, Seta, & Crelia, 1990); and (b) specific motivation, involving the use of selective and targeted search strategies to arrive at a particular outcome (Kunda, 1990). These experiments focus on the latter process, particularly on the influence of negative mood and personal relevance in triggering targeted interpersonal preferences. Although little work has been done so far on mood effects on selective information processing, the role of affect in motivation has received some attention. In terms of a mood maintenance model, people in a positive mood may find cognitive tasks aversive, motivating them to avoid or simplify problems and use less effortful decision strategies when possible (Clark & Isen, 1982; Isen, 1984; Isen & Means, 1983; Schwarz & Bless, 1991). Furthermore, the experience of positive (and sometimes, negative) moods may also reduce cognitive capacity, mandating simplified processing strategies (Isen, 1984; Mackie & Worth, 1991). Negative mood may also function as a generic cue signal- ing potentially problematic or threatening situations and informing the individual that particularly careful, effortful, and systematic information processing strategies are required (Schwarz & Bless, 1991). In contrast, little is known about mood effects on specific, targeted processing strategies. We expected that the combination of dysphoric mood and a personally relevant task would be most likely to trigger specific motivated processing and the associated selective information search and decision strategies. The interactive role of personal relevance, affect, and motivation in social judgments has received some attention in recent years (cf. Tesser, Pilkington, & Melntosh, 1989). Because other people are perhaps the single major source of rewards—and occasionally, of punishments—for most of us (Argyle, 1986), interpersonal preferences clearly have a major impact on our happiness and well-being (Strack, Schwarz, & Argyle, 1991). We expected that realistic and personally relevant choices, such as decisions about interpersonal preferences in a dysphoric mood, are most likely to involve the kind of motivated information processing in which judges selectively look for and find information that supports an already established preference (Forgas, 199 Ib). This is a case when, in a sense, preferences indeed do come to guide inferences (Zajonc, 1980). Predictions What are the cognitive consequences of such motivated choice strategies? In the following three experiments, we explored how mood-induced motivated processing functions in personally relevant interpersonal choices. The interpersonal preference task we used required subjects to choose a partner from several available candidates either for themselves (personally relevant condition) or for somebody else (personally irrelevant condition) while experiencing positive, neutral, or negative mood that was induced in a prior and ostensibly unrelated study. We expected targeted, motivated processing strategies to be adopted by dysphoric subjects when they were making a personally relevant choice. Specifically, we hypothesized that motivated processing in dysphoria involves (a) selective search for information about the rewarding (interpersonal) qualities of potential partners, (b) longer self-selected exposure to and better learning of such details, (c) superior recall of information relevant to the motivational goals, (d) use of a global impression formation strategy rather than a systematic comparison-by-features strategy when selecting partners, and (e) overall faster and more efficient choices due to the directed search strategies used. In contrast, positive mood or a personally irrelevant choice were not expected to lead to such targeted information search and decision strategies. Experiment 1 The first study in this series was designed as an initial demonstration of mood effects on interpersonal preferences and the judgmental strategies used. In previous work, either no mood effects on the outcome of complex (albeit hypothetical) choices (Isen & Means, 1983) or somewhat inconsistent, context-specific results (Forgas, 1989) were found. This experiment sought (a) 710 JOSEPH P. FORGAS many questions as they can, without worrying about difficult or confusing items that may need to be revised." After the 5-min interval allowed for completing the questionnaire, subjects were given a scoring sheet that showed the "correct" answers, as well as bogus performance standards validated for "this group of subjects." For the positive mood group, performance between 7-13 correct answers was described as "average," and over 13 correct answers was described as "above average —your verbal skills are well above average on this task according to the validation standards." For the negative mood group, average performance was defined as 27-30 correct answers and above average performance was denned as more than 30 correct answers. Less than 27 correct answers was described as "below average—your verbal skills are below the average on this task according to the validation standards." As all subjects made 14-25 correct answers, those in the positive condition all received above average marks and positive feedback and those in the negative mood condition received below average marks and negative feedback. In the control condition, subjects were simply thanked for their help and told that their responses would be helpful in revising the test. This part of the procedure was concluded by the administration of a "postexperimental questionnaire" (in fact, a mood-validation measure) asking subjects to rate how they felt on three 7-point scales (happy-sad, good-bad, satisfied-dissatisfied) embedded among six other distractor items enquiring about general reactions to the experiment. We found these three mood-check scales to be reliable and consistent measures of mood in previous studies (cf. Forgas, I991b). Finally, subjects were thanked for their help and the first experimenter left the room. A different experimenter then entered and introduced the preference task as a study in problem solving. Subjects were told that for this task, each of them was to work closely with another person as a team, and they were told that it was therefore particularly important that they should choose their partners from among four people currently present in the department who volunteered to help with the study. Each of the available partners was described in a brief personal profile that was based on evaluations of that person by friends, acquaintances, and colleagues. In the personally relevant condition, subjects were instructed to select the most suitable partner for themselves; in the personally irrelevant condition, they were told, "Because of difficulties in scheduling, we decided to ask each student to select a partner for the person in the next experimental session." Next, subjects were provided with the four partner descriptions arranged in random order. They were allowed 10 min to complete this task, which was sufficient in all cases to select a partner. Subjects were next asked to answer a questionnaire (in reality, a filler task) before their free-recall memory for the target characters was assessed. They were given four blank sheets of paper and asked to "write down, as accurately as possible, the description of each of the people you have read about. Try to remember the descriptions word for word, but if you cannot remember exactly, do the best you can." This task was completed in not more than 10 min. Debriefing. The procedure was concluded with a detailed debriefing designed to eliminate the residual effects of the mood induction. Care was taken to explain the aims and rationale of the study in a friendly and informal atmosphere, and the possible perseverance of false feedback manipulations was explained. All subjects accepted and understood the rationale for the procedure, and no evidence of residual negative effects was detected. The debriefing also established that none of the subjects suspected a link between the mood manipulation and the decision task and that the mood manipulation was accepted at face value by all subjects. to explore whether short-term mood states exert a significant motivational influence on interpersonal preferences and (b) to assess selective memory effects for the superior recall of interpersonal rather than task information, consistent with the hypothesized selective attention to and motivated processing of such details. Subjects in a happy, neutral, or sad mood were provided with information about four potential partners and were asked to select one partner either for themselves (personally relevant condition) or for somebody else (irrelevant condition). The outcome of their choices and their memory for the details of the targets were assessed to establish the role of mood and motivated processing strategies in such preference choices. Method Design and subjects. The study is based on a 3 X 2 between-subjects design, with mood (happy, control, and sad) and personal relevance (high and low) as the independent variables. Subjects were 60 undergraduate students who participated in the study as part of their course requirements. Equal numbers of men and women were included in each of the three mood conditions. As a preliminary analysis of variance (ANOVA) revealed no main or interaction effects due to subjects' sex. data for men and women were combined. Stimuli. Four realistic target descriptions, each consisting of 12 short sentences, were constructed. The first and the last statements were affectively neutral (e.g.. "John lives in an apartment"). Of the remaining 10 statements, 5 contained information about the target's socioemotional, interpersonal qualities (e.g., friendliness and popularity: "John has many friends") and 5 contained information about task competence (e.g., skill and intelligence: "John usually does well on his exams"). So that the characters were realistic and plausible, each character contained either positive interpersonal characteristics and negative task characteristics (a likable but incompetent person) or the reverse (a dislikable but competent character). Two parallel character description sets were constructed, and descriptions in the matched sets were identical in all respects except the substitution of one word that changed the valence of the statement (e.g., good for bad, intelligent for unintelligent, etc.). The two sets were randomly assigned to individual subjects and occurred with the same frequency within each mood condition. Mood manipulation and procedure. Subjects were run in small groups of 2-4 people and were told that two brief but unrelated experiments would be conducted during a 1-hr session to save the subjects time. The first experiment (in reality, the mood manipulation) was introduced as a test of verbal abilities. Subjects had 5 min to complete a series of 33 sentences of the kind "Car is to road as train is to . . . ," with four multiple-choice answer alternatives provided. The sentences ranged from easy to difficult, with the last eight questions, although plausible, having no determinate answers (e.g., "Bread is to butter as river is to. . ."). An earlier pilot study (Forgas, 1989) established that on the average, 19 questions are completed in 5 min, with everyone completing at least 13 questions in that time. Mood was manipulated through the instructions and feedback provided to subjects about their above average or below average performance in a modification of a previous successfully developed and used technique (cf. Forgas & Bower, 1987; Forgas, Bower, & Moylan, 1990). All instructions and feedback information were provided in writing, allowing a high level of standardization of the mood manipulation procedure. In the negative condition, the cover sheet informed subjects that "most people find these questions quite easy and complete all items in the available time." In the positive condition, subjects were instructed that "most people find these questions increasingly difficult and rarely complete more than 10 items in the available time." The control group members were told that they "should try to complete as Results and Discussion Mood validation. The overall ANOVA of the combined mood self-ratings, F(2, 57) = 11.35, p < .01, indicated signifi- AFFECT AND PARTNER CHOICE 711 cant mood differences between the happy (M— 2.52) and control (M = 3.56) groups, F(\, 38) = 8.47, p < .01, as well as between the control and sad (M= 4.74) groups, F(l, 38) = 9.02, p < .01, confirming that the mood manipulation was indeed effective in inducing significantly different positive and negative mood states. Interpersonal preferences. Mood effects on preference for partners with good task or good interpersonal qualities was first analyzed. Results showed a significant mood main effect, F(2, 57) = 8.92, p < .01. Happy subjects were more likely to select task-competent partners (55%) than control subjects were (48%), F(l, 38) = 6.21, p < .025. Sad subjects in turn preferred interpersonally rewarding rather than task-competent partners (42%), compared with the control group, F(2, 38) - 5.37, p < .01. The significant interaction of mood and the personal relevance of the choice further underscores this effect, F(2, 54) = 15.01, p < .01. It seems that dysphoric subjects strongly preferred potentially rewarding, socially skilled partners when making a choice for themselves (67%), but not necessarily for others (49%), F(l, 18) = 11.20, p < .01. In contrast, happy subjects preferred competent rather than socially skilled partners both for themselves (57%) and for others (53%), F(l, 18) = 1.38, ns, consistent with the implicit requirements of a collaborative task. This result offers rare evidence for mood effects on the outcome of interpersonal preference choices. Previous studies of mood effects on hypothetical decisions did not find conclusive evidence for mood effects on such outcomes (Isen & Means, 1983), probably due to the hypothetical decision context used (e.g., choice between fictitious cars) and the lack of personal involvement in the outcome. In contrast, our results suggest that in personally relevant choices, affect is likely to color people's preferences. The current results also suggest that, consistent with Schachter's (1959) work over 30 years ago, mood effects on interpersonal choices are likely to be mediated by motivational factors. Selecting a potentially rewarding instead of a task-competent partner, despite the requirements of the situation, suggests that dysphoric subjects applied a motivated processing strategy in their preferences. This may be due to the selective search for and greater attention to information about potentially rewarding personal qualities or to the selective avoidance of information about highly competent potential partners. Such a strategy should ultimately result in better memory for interpersonal details, the hypothesis we examined next. Memory for task versus interpersonal information. The processing strategy of dysphoric subjects paying relatively more attention to information about the rewarding qualities of potential partners should have as one of its consequences the better recall of such details. In contrast, happy subjects engaged in simplified processing (Isen, 1984; Schwarz & Bless, 1991) or happy subjects making personally irrelevant choices should show no such thematic preference in their recall of information. Each information unit recalled about the target characters was scored on a 0-3 scale (ranging from incorrect recall to perfect recall) by two independent raters who achieved an interrater reliability of .792. Results on the basis of the average of the two ratings showed a significant mood main effect on memory. Happy subjects correctly recalled more details (49%) than did control subjects (45%), F(l, 38) = 4.44, p < .05, who in turn remembered more than did negative mood (38%) subjects, F(l, 38) = 4.87, p < .05. The personal relevance of the choice also influenced memory, with significantly better recall when the choice was personally relevant (47%) rather than irrelevant (41%), F(l, 58) = 11.52, p < .01. We also found evidence for a mood-congruent memory bias, F(2,54) = 7.03, p < .01. Happy subjects recalled relatively more positive (54%) than negative (44%) details, f(18) = 7.68, p < .01, and sad subjects remembered more negative (41 %) than positive information (35%) about the target characters, <(18) = 5.31, p< .01. This pattern of mood-congruent memory is consistent with the selective and more detailed processing of mood-consistent details implied by various mood-priming models (cf. Bower, 1991; Isen, 1984). In previous studies, my colleagues and I found that people actually spend more time reading and encoding mood-consistent rather than inconsistent information, leading to the eventual better recall of such details (Forgas & Bower, 1987). Of particular interest here were the significant interactions among mood, personal relevance, and memory for task versus social information, F(2, 54) = 21.35, p < .01. As predicted, dysphoric subjects recalled significantly more social (48%) rather than task information (36%), but only when the choice was personally relevant, ((18) = 8.25, p < .01. There was no such memory effect for either control (46% vs. 48%) or happy (49% vs. 53%) subjects, or for any of the mood conditions when making a personally irrelevant choice (Figure 1). This pattern is consistent with the assumption that more negative moods are more likely to lead to motivated processing strategies in which people selectively look for and learn interpersonal information that is relevant to mood improvement. In the present case, searching for a partner with positive personal qualities and better remembering such details are consistent with our assumption of a motivated processing strategy. Alternatively, dysphoric subjects may have selectively avoided dealing with information about competent others as a defensive strategy of avoiding unfavorable comparisons. The use of negative feedback about performance in our mood manipulation may have contributed to this tendency. However, in several recent studies, it has been suggested that avoidance of unfavorable comparisons with others is not a reliable feature of dysphoria. For example, Wheeler and Miyake (1990) found that sad subjects were more likely to engage in unfavorable "upward" comparisons, a pattern of judgments also common in depression (Ottaviani & Beck, 1988). In one of our recent studies, we also found sad subjects to be more likely to make self-deprecatory upward comparisons and to overestimate the performance of others (Forgas et al, 1990). Experiment 1 was thus successful in showing that short-term moods do have a significant effect on people's interpersonal preferences and their memory for information about others. Selectively remembering the social qualities of potential partners in dysphoria is consistent with a directed, motivated search strategy, in which attention is focused on interpersonal rather than task-relevant information. However, the evidence for such processes at this point is rather indirect. Experiments 2 and 3 were designed to obtain more specific evidence for the 712 66 JOSEPH P. FORGAS the external validity of Experiment 1, as interpersonal choices may often involve considerably greater complexity and information overload. As target descriptions in Study 1 were not based on standard features by which partners could be directly compared, the observed memory effects may have been con45 R E C 50 A L L founded by the differential effort subjects spent in trying to interpret and decode different verbal descriptions. In Experiment 2, the number of targets was increased to 8, more detailed target descriptions were developed to simulate realistic person preference tasks, and characteristics were presented in comparative terms, making across-target comparisons possible. Instead of consistently likable or dislikable and competent or inHappy Control MOOD CONDITION competent target character qualities, the revised stimulus set involved qualities that represented a continuum to be inferred from multiple characteristics. 3. Experiment 2 also relied on a revised mood-manipulation procedure. The false-feedback mood manipulation used in Experiment I may have selectively triggered interpersonal choices that were partly due to the thoughts primed by a recent experience of failure or were confounded by the direct motivational consequences of an achievement event, rather than mood per se. To control for this possibility, we used a new mood induction strategy in Experiment 3 that involved no success or failure manipulation and was unlikely to selectively prime personally relevant information (Forgas et al., 1990). We used short, preselected films with consistent affective consequences as mood manipulators. Method Happy Control Sad MOOD CONDITION Figure 1. The effects of mood and personal relevance on the recall of social and task information. (A: personally relevant choice; B: personally irrelevant choice. Inf. = information.) operation of motivated processing strategies in interpersonal choices. Experiment 2 In the previous experiment, we found significant mood effects on motivated interpersonal choices and corresponding memory effects. Experiment 2 was designed to obtain more direct and detailed evidence about the order and selectivity of subjects' information search strategies and the precise decisionmaking processes they used. The basic design of Experiment 1, manipulating mood and the personal relevance of the choice was retained. Major changes were made in three areas: procedure, stimuli, and mood manipulation. 1. A new procedure that allowed the step-by-step monitoring and recording and the subsequent detailed analysis of the information search and decision-making strategies adopted by happy, sad, and control subjects was developed. 2. A new set of realistic stimulus character descriptions containing more detailed descriptive information about potential partners was created. The use of only 4 targets may constrain Overview. After watching happy, control, or sad films in an allegedly unrelated experiment, subjects were asked to chose 1 partner from 8 potential candidates. The choice was either personally relevant or irrelevant. Candidates were each described in terms of 10 features: sex, age, 4 characteristics pertaining to social competence and likability, and 4 characteristics relevant to task competence. Interpersonal preferences, decision strategies, and latencies were analyzed as a function of subjects' mood. The study incorporates an overall 3 x 2 design, with mood (happy, sad, or neutral) and personal relevance (personal-impersonal) as the independent variables. Subjects. Ninety-six undergraduate students (48 men and 48 women) participated in the study as part of their course requirements, with 32 subjects assigned to each of the three mood conditions. As a preliminary ANOVA again failed to detect any main or interaction effects due to subjects' sex, data for men and women were combined for all subsequent analyses. Stimuli. A new set of 8 stimuluscharacters were developed. The aim was to more accurately simulate the usual complexity of interpersonal preference choices and to provide more tangible and realistic information about both the likability and the competence of the targets. The 8 targets were each described in terms of 10 characteristics, each printed on separate cards in "personnel files." In addition to sex and age, 4 characteristics had to do with task competence (intelligence, exam performance, performance on a project, and ratings of competence by a faculty member), and 4 characteristics described interpersonal qualities and likability (scores on a test of social skills and ratings of friendliness, likability, and popularity by peer group members). Sex and age information were randomly varied so that 4 men and 4 women between the ages of 18 and 22 were included in each file. For each descriptive feature, a range of eight comparative values, ranging from the most desirable to the least desirable, were denned. Features were assigned to targets in a Latin square design, so that the average rank value of fea- AFFECT AND PARTNER CHOICE tures for each target was the same. Assuming equal weighting for every feature, there was no rational basis for selecting any person over any other. Mood manipulation and procedure. The procedure was similar to that in Experiment 1. Subjects signed up for two "unrelated" studies. The first study, in fact the mood manipulation, was introduced as involving the selection of appropriate films for a later experiment. The films were shown on videotape, and subjects were asked to watch the tapes as they would at home. After the film, a brief mood-validation questionnaire asked subjects to rate how they felt on three 7-point bipolar scales embedded among several other distractor items: happy-sad, good-bad, and satisfied-dissatisfied. In the positive condition, the film contained edited scenes from a successful comedy series. In the neutral condition, the film was a documentary about architectural design. The negative-mood induction film contained edited scenes from a feature film dealing with a mother dying from cancer. These films were previously validated as effective in inducing the appropriate mood state (cf. Forgasetal.,1990). After the mood manipulation, a different experimenter introduced the interpersonal preference task the same way as in Experiment 1. In the personally relevant condition, subjects were asked to select the most suitable partner for themselves from 8 candidates, in the personally irrelevant condition, they were told to select a partner for the person in the next experimental session. The 8 personnel files were then presented to each subject, with the order of the files, as well as the order of the information cards contained within them randomized. Subjects were asked to sequentially number each of the information cards as they consulted them and also to rate the usefulness of that information to them at that time on 5-point not at all useful (0) to very useful (4) scales. They were told that they were permitted to look at any item of information in the folders in any order and as often as necessary to reach a decision. The beginning and the completion of the decisionmaking process for each subject was timed. This procedure to record decision-making strategies was previously pilot tested and found to be a feasible and reliable method for analyzing decision processes (cf. Forgas,1989). Debriefing. As in Experiment I, an extensive debriefing designed to eliminate any residual effects of the mood induction concluded the procedure. We found no evidence of any awareness of the design and objectives of the study. 713 Results and Discussion Effectiveness of the mood manipulation. An analysis of the combined mood self-ratings showed significant overall differences, F(2,93) = 18.13, p < .01. Differences between the happy (M= 2.74) and control (M= 3.69), F(\, 62) = 12.83, p < .01, as well as between the control and sad (M = 4.49) groups were significant in the predicted direction, F(l, 62) = 11.34, p < .01. Consistent with previous studies (cf. Forgas, 199lb), this mood manipulation was thus effective in inducing significantly different affective states in subjects. Partner preferences. Preference for partners who were competent or likable was assessed by calculating the difference between a chosen partner's average rank on the four task characteristics and average rank on the four interpersonal characteristics. This is an index of relative task versus interpersonal skill, in which positive values indicate greater task skills and negative values represent greater interpersonal skills in the preferred partner. We found a significant mood main effect for preferences, F(2, 93) = 8.92, p < .01. Dysphoric subjects overall preferred partners with desirable interpersonal qualities, F(2,62) = 6.55, p < .01, whereas happy subjects were marginally more likely than controls to choose partners with better task skills, F(2, 62) = 4.32, p < .05 (Figure 2). The significant interaction of mood and personal relevance further qualifies this effect, however, F(2,90) = 15.01, p < .01. It seems that dysphoric subjects strongly preferred potentially rewarding, skilled partners for themselves but not necessarily for others, f(30) = 8.21, p < .01 (Figure 2). This finding replicates and confirms the results of Experiment 1, with a more complex and realistic choice context and with a different mood manipulation procedure. The combination of negative mood and a personally relevant choice apparently led to a motivated decision to prefer rewarding partners for the self. No such differences in preference were found in either the control or the positive-mood groups (Figure 2). Mood effects on processing strategies. The sequential numbering and rating of each information unit allowed a precise reconstruction of how each subject reached a decision. From these records and for each subject, the following indexes were constructed: (a) overall time taken to reach a decision; (b) number of steps necessary to reach a decision; (c) proportion of information units eliminated (not looked at); (d) rate of repetition (average number of times each unit was considered); (e) proportion of interpersonal minus proportion of task information looked at; (f) average usefulness rating of interpersonal information minus average usefulness rating of task information; (g) average number of consecutive steps within the same target character (high values indicate an impression formation strategy dealing with all features of a single character before progressing to the next one); and (h) average number of consecutive steps within one information category (high values indicate a strategy of comparing targets in terms of a single feature). Obviously these measures may not be mutually independent from each other. To construct a smaller set of nonredundant measures of decision strategy, an overall factor analysis of these measures across all subjects was carried out, specifying the oblique rotation of all factors with eigenvalues greater than 1.0. A three-factor solution accounting for 64.7% of the variance represented the best combination of these measures, with inter- p A 0.3 N E R P R E F E R E N 0.2- I PERSONAL CD IMPERSONAL 0.1 I -0.1 Happy Control Sad MOOD CONDITION Figure 2. Preference for rewarding versus competent partners as a function of mood and personal relevance (higher values indicate preference for competent rather than rewarding partners). 714 0.5 JOSEPH P. FORGAS D E C I IMPERSONAL 0.4 s O 0.3 S P E E D 0.2 0.1 Happy Control Sad MOOD CONDITION Figure3. The influenceof mood and personal relevance on the speed of interpersonal preference choices (higher values indicate faster and more efficient decisions). factor correlations no greater than .20. The first factor was labeled decision speed and seemed to measure the speed and ease of the decision-making process. This factor was marked by the following measures: number of steps required to reach a decision (.61), average repetition rate (—.54), time taken to reach a decision (.52). and proportion of items not considered (.49). The second factor was labeled information selectivity, indicating a subject's tendency to preferentially look for task or interpersonal information. This factor was marked by two measures: average usefulness rating for interpersonal over task information (.63) and average proportion of interpersonal over task information considered (.57). The third factor was labeled choice strategy, indicating a judge's tendency to make acrossperson or across-feature comparisons in making a choice. This factor was marked by two indexes: average number of consecutive steps within the same person (—.58) and average number of consecutive steps within the same information category (.55). After the factor analysis, we constructed three new nonredundant dependent variables from the linear combination of the standardized original measures by using the factor score coefficients as weighting factors. These three new dependent measures, decision speed, information selectivity, and choice strategy were then subjected to a series of ANOVAs. Decision speed. Mood had a significant main effect on decision speed, F(2, 93) = 5.24, p < .01. Happy subjects reached a decision significantly faster, in fewer steps, with fewer repetitions, and with fewer information units considered than did sad subjects, F(l, 62)= 5.76,/><.025, or control subjects, F(l, 62) = 7.87, p < .01, although differences between these two groups were not significant (see Figure 3). A significant interaction between mood and the personal relevance of the task, F(2, 90)= 6.68, p < .01, qualifies this result. In a happy mood, subjects were relatively more efficient in reaching an impersonal rather than a personal decision, ((30) = 6.97, p < .01. This pattern was reversed in negative moods, when subjects took less time to find a partner for themselves than they did for others, /(30) = 4.36, p < .01. This nonobvious effect is consistent with our expectation that the generally slower information processing often found in dysphoria may be counterbalanced by the kind of directed, selective information search strategies charac- teristic of the motivated processing of personally relevant choices. Consistent with our predictions, the benefits of this directed information search process applied only to self-relevant but not to personally unimportant decisions (Figure 3). Information selectivity. Did sad subjects selectively look for interpersonal qualities in making their choices? In terms the proportion of interpersonal versus task information consulted and their relative usefulness rating, the answer seems to be yes, F(2, 93) = 6.98, p < .01 (Figure 4). Results showed the greater preference for task information in positive mood subjects, F(l, 62) = 5.98, p < .025, and a greater preference for interpersonal information in negative mood subjects when they were compared with controls, F(\, 62) = 6.31, p < .025. The significant interaction of mood and personal relevance reveals, however, that this difference was largely due to dysphoric subjects paying disproportionate attention to interpersonal qualities when choosing a partner for themselves rather than others, F(2,90) = 7.23, p < .01. It seems that dysphoric subjects selectively looked at and rated as more useful interpersonal rather than task information about potential personal partners, f(30) = 8.05, p < .01. This information selectivity effect, not evident in either the control or the positive mood groups, is consistent with sad subjects engaging in motivated information search strategies, looking for and rating as more useful information about the potentially rewarding interpersonal qualities of others. Such a preference for social information is also consistent with their eventual choice of socially rewarding over task-competent partners. Choice strategies. Was there a link between subjects' mood and the particular decision strategy they adopted in making their choice? In selecting a partner, subjects could either consecutively deal with information about one person to construct a "global" impression or choose to compare several potential partners in terms of one selected characteristic ("comparison by features"). Overall, subjects experiencing a negative rather than a positive mood, F(2, 93) = 7.89, p < .01, and those making a personally relevant rather than irrelevant decision, F(l, 94) = 9.45, p< .01, were significantly more likely to engage in a global, impression-formation strategy rather than a comparison-by-features strategy. We also found a significant Mood X Personal Relevance interaction, F(2,90) = 7.67, p< .01 (Figure I N F O R M A T I O N B I A S M PERSONAL 0.2 CD IMPERSONAL 0.15 0.1 — Jj 0.05 0 n f\fi Happy Control n 1 Sad MOOD CONDITION Figure 4. Biased information choice in interpersonal decisions: the effects of mood and personal relevance (positive values indicate preference for task vs. social information). AFFECT AND PARTNER CHOICE 715 5), showing that the personal relevance of the task had a greater effect on decision strategies in negative moods than in control or positive moods. This pattern provides direct evidence that interpersonal choice strategies are significantly influenced by a person's mood and the personal relevance of the outcome. Dysphoric subjects making a personal choice were more likely than those making an impersonal choice to rely on a within-target, impression formation strategy, F(l, 30) = 9.31, p < .01. We expected this group to be most influenced by their self-serving motives to find a rewarding partner, and the search for a global impression rather than an analytic, comparison-by-features strategy served this purpose well. In contrast, choices by happy subjects and by those making an impersonal choice were more likely to be based on a comparison-by-features process. These results are not necessarily inconsistent with other evidence suggesting more systematic, detailed information-processing styles in negative moods (Forgas & Bower, 1987; Schwarz & Bless, 1991), as both the impression formation and the comparison-by-features strategies are compatible with detailed, systematic, or simplified heuristic processing styles. In conjunction with the previous results, it seems that dysphoric subjects focused on identifying interpersonally rewarding rather than task-competent partners for themselves, selectively looked for interpersonal information, and sought to form a global impression of potential partners in terms of interpersonal features. These findings go a considerable way toward establishing the characteristics of motivated choice strategies and linking such processes to transient mood states. However, motivated processing should also be reflected in the actual time spent by subjects consulting various kinds of information categories in making their choices. This was examined in the next experiment, in which the results are replicated and further extended by using a computer-controlled stimulus presentation procedure that also allows the measurement of reaction times and decision latencies. Experiment 3 The previous two experiments were successful in showing that temporary mood has a major influence on interpersonal preferences. We found that in selecting a partner, people use different information search and decision-making strategies, depending both on their mood and the personal relevance of the choice. We designed Experiment 3 to provide additional evidence about the nature of these processes by using a more sensitive methodology that allowed the controlled presentation of stimulus information and the exact measurement of learning and decision latencies. The first objective of Experiment 3 was to replicate the motivated memory effects observed in Experiment 1, this time in conjunction with the careful monitoring of subjects' learning and attention strategies. Second, we substantially improved the procedure developed in Experiment 2 to measure the information selection strategies used. The need for subjects to number and rate each of the information cards they encountered was somewhat intrusive and may have interfered with the selection process itself. In Experiment 3, we developed a sensitive computer-administered procedure for presenting target information to allow the precise measurement of subjects' preferred decision strategies without interfering with the process itself. In addition, this procedure allowed the collection of reaction time data to assess the informational search latencies associated with different moods and task relevance. In Experiment 3, we also incorporated revised mood-manipulation stimuli. With any particular mood-manipulation technique, there is always the possibility that factors other than mood—such as personal memories, cognitions, or motivations —are also influenced. One way of controlling for this possibility is to use a multimethod approach to ensure that findings are not artifactual but can be reliably obtained with slightly different induction materials. A new set of mood-induction films were developed and used in Experiment 3 in an effort to increase the generalizability of the findings. Method Overview. Subjects were shown a new set of mood manipulation films designed to elicit positive, negative, or neutral mood in an ostensibly separate experiment. Next, they were again asked to make interpersonal choices for themselves or for others. The overall design and partner descriptions were the same as in Experiment 2. However, through the computerized presentation of the target information, it was possible to measure both the time taken to select and consult each information unit and the precise sequence of the information search process. Subjects. Forty-two female subjects participated in the experiment as part of their course requirements, with 14 subjects in each mood condition. As no significant sex differences were found in the previous experiments, the use of female subjects only was unlikely to limit the validity of the findings. Mood manipulation and procedure. The overall experimental design was similar to that in Experiment 2. Subjects signed up to participate in two "unrelated" studies. The aims of the first study, in reality the mood manipulation, were described as the selection of films for later experiments. Subjects were instructed to watch the videotapes as they normally would at home. This time, the happy film showed scenes from a different comedy episode, the control film showed a lecture on art, and the sad film showed edited scenes from a film depicting a young woman suffering a debilitating sporting accident. A brief postexperimental questionnaire asking for mood self-ratings inserted D E C I - IMPERSONAL s 0 N S T R A T E Q Y Happy Control MOOD CONDITION Figure 5. Decision-making strategy as a function of mood and personal relevance (positive values indicate preference for impression formation vs. comparison-by-features strategies). 716 A R 0.3 JOSEPH P. FORGAS T N E R ^'*- • PERSONAL EZHiMPERSONAL - p R F E R E N C E 0.1 0 -01 fl Happy Control I Sad MOOD CONDITION Figure 6 Interpersonal preference as a function of mood and personal relevance (higher values indicate preference for rewarding rather than competent partners). among distractor questions (inquiring about general reactions to the study) was administered at the conclusion of the mood manipulation. The second stage of the study was introduced the same way as in Experiment 2. However, the information about the potential partners was provided on a computer screen rather than in personnel files. Information about the targets was presented in a matrix form, with individual targets in the columns and descriptive features in the rows. Subjects were told that they may view any item of information about any target and in any order by selecting the cell corresponding to the intersection of that target and the feature they wanted to find out about. For example, to read about Characteristic 2 of Target 3, they would have to select Cell 23 on the keyboard. After viewing each information unit, they were asked to press the space bar to get back to the overall matrix before selecting the next information unit they wished to look at. Subjects were given a number of practice trials to familiarize themselves with the procedure. The computer was programmed to record not only the precise sequence in which information units were looked at, but also the time subjects spent studying each item of information, the time it took to make each choice, and the total time spent until arriving at a partner selection choice. Once the choice task was completed, subjects were asked to complete a brief questionnaire—in fact an interference task—before a free-recall test was administered. At this time, subjects were asked to recall and write down as much as they could about each of the target characters they had read about on each of eight sheets of paper. The procedure concluded with a debriefing session along the lines described in the previous experiments. No evidence of subject awareness of the hypotheses or manipulations was detected. to competent partners was measured the same way as in Experiment 2. Consistent with the previous results, sad subjects once again preferred partners with relatively better interpersonal than task qualities than did happy subjects, F(\, 26) = 5.93, p < .025, or control subjects, F(\, 26) = 4.83, p < .05, and did so more for themselves than for others, F(2, 36) = 7.72, p < .01 (Figure 6). This pattern confirms the findings of the previous experiments using a different mood manipulation stimulus and a different selection procedure, thus attesting to the robustness of mood effects on interpersonal preference choices. Decision speed. Mood also had a significant effect on decision speed (operationalized the same way as in Experiment 2). Happy subjects reached a decision faster, in fewer steps, and with fewer repetitions than did sad subjects, F(l, 26) = 6.92, p < .025, or control subjects, F(l, 26) = 5.83, p < .025. We also found a significant Mood X Personal Relevance interaction, F(2, 36) = 10.06, p < .01. This showed that in a happy mood, subjects were faster in making impersonal rather than personal choices, F(l, 12) = 7.32, p < .025, whereas the reverse was the case in a negative mood, when subjects made personal choices in less time than they made impersonal choices, F(l, 12)= 5.40, p < .05 (Figure 7). The fact that sad subjects were faster in making decisions with personal consequences may at first appear counterintuitive. However, this outcome is consistent with processing benefits inherent in motivated processing, as it involves a highly directed and selective search for specific kinds of information. Mood effects on information selectivity. Subjects'relative preference for interpersonal versus task-related information was calculated by comparing the proportion of interpersonal with the proportion of task-relevant information units looked at by each subject. There were no main effects, and we also failed to find a Mood X Personal Relevance interaction effect. However, a clear trend toward dysphoric subjects looking at more interpersonal information than task information when choosing a partner for themselves was evident in the data. There may be at least two plausible reasons why this interaction was significant in Experiment 2 but did not reach significance here. First, the measure of information selectivity used here did not include ratings of the perceived usefulness of each information unit, as Results and Discussion Validation of the mood manipulation. Combined mood selfratings were subjected to an overall ANOVA, F(2, 39) = 12.22, p < .01. Results showed the expected significant mood differences between the happy (M = 3.04) and control groups (M = 3.71), F(l, 26) = 9.43, p < .01, and the sad (M = 4.29) and control groups, F(l, 26) = 6.67, p < .025. Consistent with previous evidence (Forgas et al, 1990), the audiovisual mood manipulation was again found to be effective in inducing significantly different good or bad moods in subjects. Interpersonal preferences. Preference for likable as opposed o S D E C I S I 0.5 -s- PERSONAL IMPERSONAL 0.4 N 0.3 p E E D 0.2 0.1 Happy Control Sad MOOD CONDITION Figure 7. The effects of mood and personal relevance on the speed and efficiency of interpersonal choices (higher values indicate faster and more efficient choices). AFFECT AND PARTNER CHOICE in Experiment 2. Second, the computerized procedure presented subjects with the layout of the complete information matrix before and after each choice, thus making the predicted thematic bias in selections perhaps less likely to occur. With this procedure, information selectivity should be more readily revealed by the amount of time subjects chose to spend looking at interpersonal versus task information and the time they took to select each category, rather than the number of such units examined. We looked at these results next. Latency of exposure to social versus task information. Reading times were significantly influenced by subjects' mood, F(2, 39) = 23.26, p < .01. Compared with the control group, dysphoric subjects were overall slower (6.7 s vs. 7.9 s), F(\, 26) = 11.72, p < .01, and positive mood subjects were faster (5.8 s), F(l, 26) = 18.46, p < .01, in dealing with one item of information. Similar patterns of slower, less efficient information processing in dysphoria and faster processing in happy moods have been found in other studies both with depressed and with normal subjects (Bower, 1991; Forgas & Bower, 1987; Isen, 1984; Isen & Means, 1983; Schwarz & Bless, 1991). Of greater interest here was the significant interaction found between mood and the personal relevance of the choice, F(2, 36) = 5.99, p < .01. Not only were happy subjects generally faster in dealing with an information unit, but also this tendency was strongest when the choice was irrelevant (5.2s) rather than relevant (6.4 s) to them personally, F(1,12) = 6.81, p < .025. This tends to suggest that reports of more efficient processing in positive moods (cf. Isen & Means, 1983) may be confounded with the lack of personal relevance typical of the decision tasks studied in the past. The apparent benefits of positive mood effects for decision efficiency may be reduced or even eliminated as the personal relevance of the task increases. Consistent with our expectations, we also found a significant interaction effect between mood, the personal relevance of choice, and selective attention to task or interpersonal information by subjects, F(2, 30) = 7.11, p< .01. When the choice had no personal consequences, the time spent looking at social or task information in all mood conditions was approximately the same (Figure 8). With a personally relevant choice, however, sad subjects spent significantly longer examining interpersonal rather than task information (8.2 s vs. 7.2 s), t(l2) = 10.30, p < .01, and there was a small but nonsignificant trend in the opposite direction by happy subjects (6.2 s vs. 6.6 s). This pattern suggests that there is a significant tendency to focus on interpersonal information that occurs only in dysphoria when choosing a personal partner. In conjunction with previous results, this confirms the existence of motivated choice strategies at the micro level of selective exposure to potentially beneficial information (Figure 8). Latency of selecting task versus social information. In addition to measuring how long subjects take to look at different information units, the computer also recorded the time elapsed before making each selection (the time between returning to the matrix and making a new choice). We expected motivated processing to involve reduced choice latencies for preferred information. Consistent with evidence for the generally faster processing of information in positive moods, there was a significant mood main effect on choice latencies, with sad subjects 717 ! O I 8.5 7-5 Hippy Control MOOD CONDITION - SOCIAL INF. -*- TASK INF. | '•' 7.5 5.5 Happy Control MOOD CONDITION Figure S. Self-exposure to social versus task information about potential partners: the effects of mood and personal relevance (in seconds). (Inf. = information.) taking longer to make a selection (5.31 s) than control (4.88 s) and happy (4.43 s) subjects, F(2, 39) = 14.17, p < .01. Once again, we found significant interactions among mood, personal relevance, and latency of choice for interpersonal or task information, F(2, 30) = 5.51, p < .01. As expected, sad subjects made up their minds faster to look at interpersonal rather than task information when the choice was personally relevant (4.79 s vs. 5.58 s), F(l, 12) = 6.99, p < .025, but not when it was irrelevant to them (5.55 s vs. 5.31 s), F(l, 12) = 1.31, ns. Happy subjects had a slight tendency to choose task information faster than social information both in personally relevant (4.22 s vs. 4.53 s) and in irrelevant (4.24 s vs. 4.75 s) choice situations, with similar but smaller differences for control subjects (4.54 s vs. 4.91 s and 4.92 s vs. 5.19 s). These results are consistent with the notion that sad subjects, when making a personal choice, were most likely to engage in motivated processing, selectively looking for interpersonal information and taking less time to choose such details. Recall data. At the end of the decision task, the experimenters gave subjects a memory test asking them to recall as 718 JOSEPH P. FORGAS much information as possible about each target. Items recalled were scored from 0 to 3, ranging from incorrect recall to near perfect recall by two independent raters who achieved an interrater reliability of .826. Results showed that mood had only a weak overall main effect on recall performance, with the proportion of information correctly recalled being somewhat higher in positive (43.5%) than in neutral (38.5%) or sad (40%) moods, F(2,39) = 4.92, p < .025. However, consistent with the results of Experiment 1, memory was overall much better for subjects who made a personally relevant (47%) rather than an irrelevant (35%) judgment, F(l, 40) = 5.93, p< .025, confirming that personal relevance is a major factor in facilitating recall. Overall, subjects remembered task (41.25%) and interpersonal (40.5%) details almost equally well, F(l, 41) = .81, ns. However, consistent with our expectations, we did find significant interaction effects among mood, the personal relevance of the choice, and memory for task rather than interpersonal details, F(2, 30) = 8.29, p < .01. This interaction showed that when the task was not personally relevant, recall of task and interpersonal details was not significantly different in happy (39% vs. 34%), control (36% vs. 32%), or sad (37% vs. 32%) moods. When the outcome was personally relevant, however, subjects in a positive mood recalled relatively more task information (54% vs. 46%), t(\2) = 6.52, p < .01, and subjects in a negative mood remembered relatively more interpersonal than task details (53% vs. 39%), t(\ 2) = 11.07, p < .01. These memory effects are largely consistent with, and are likely to be the consequence of, the previous findings showing information selectivity and longer self-exposure to interpersonal details by dysphoric subjects. Affective influences on decision strategies. Using the index described in Experiment 2, we found that subjects selecting a partner for themselves were again significantly more likely to rely on a within-target impression formation strategy in their choices, F(l, 39) = 7.52, p < .01, and this tendency was most marked for subjects in a negative rather than a positive mood state, F(2, 36) = 16.31, p < .01. The interaction between mood and personal relevance was only marginally significant, however, indicating a slight tendency for sad subjects to be more likely to adopt an impression formation strategy when making choices for themselves, but not for others, F(2, 36) = 3.76, p < .05 (Figure 9). This finding is consistent with the results of Experiment 2. The somewhat weaker effects obtained here are probably due to the different choice procedures made possible by the computerized presentation of information, making it relatively easier for subjects to make nonconsecutive selections from the information matrix. I 0.21S 0.16 T R A T E G Y o.i 0.05 o Happy Control MOOD CONDITION Figure 9. Decision-making strategy as a function of mood and personal relevance (positive values indicate preference for impression formation vs. comparison-by-features strategies). General Discussion The three experiments reported here offer convergent evidence for the importance of temporary moods in motivating selective interpersonal preference choices. It was found that the combination of sad moods and personally relevant choices clearly led to motivated processing strategies involving a directed search for information relevant to rewarding outcomes, self-serving interpersonal choices, and superior memory for such thematically salient information. Of course, different judgmental contexts may lead to somewhat different motiva- tional objectives (Wenzlaff& Prohaska, 1989). These effects are markedly different from the kind of automatic, dispassionate, and rational processing strategies explicitly or implicitly assumed by some early models of social cognition (cf. Forgas, 1981). For example, traditional affect-priming models predict relatively open, unbiased information search and processing strategies and can account fairly well for mood effects on many less personally involving choices (Bower, 1981; Forgas & Bower, 1987; Isen, 1984). However, it seems that very different and more motivated choice strategies may be used when the decision becomes personally relevant (Tesser et al, 1989). Many everyday social judgments are performed in the service of specific goals and involve some degree of motivated processing (Kunda, 1990; Martin, 1986; Tesser, 1986). Like evidence from other fields, our results suggest that specific motivational objectives may play a major role in determining information search and processing strategies in many real-life judgments and decisions (Brewer, 1988; Fiske & Neuberg, 1990; Forgas et al., 1990; Mackie & Worth, 1991; Petty et al, 1991; Wenzlaif & Prohaska, 1989). Interest in motivated thinking has a venerable history in social psychology. Explanations of cognitive processes involved in decision making (Janis & Mann, 1977), dissonance reduction (Wicklund & Frey, 1981), intergroup behavior (Brewer & Kramer, 1985), persuasion (Petty et al., 1991), and stereotyping (Tajfel & Forgas, 1981) have traditionally included a motivational component. The role of motivational factors in social judgments has also been explicitly recognized (Brewer, 1988; Fiske & Neuberg, 1990; Tesser, 1986). More recently, the role of motivation in memory processes has received growing attention (Parrott & Sabini, 1990; Sanitioso et al., 1990). Contemporary theoretical formulations by Brewer (1988), Fiske and Neuberg (1990), Kunda (1990), Martin (1986), and Tesser (1986) have contributed much to incorporate motivated processes in models of social cognition. Surprisingly, despite considerable interest in affect and cognition in recent years (Bower, 1991; Clore & Parrott, 1991; Forgas, 1991a; Isen, 1984; Schwarz & Bless, 1991), very little work has been done so far on the role of moods in motivated interpersonal choices. AFFECT AND PARTNER CHOICE 719 One likely reason why the kind of mood-induced motivated processing demonstrated here was not documented earlier is that few, if any, researchers looked at mood effects on choices with meaningful personal consequences. Yet there is evidence from other fields that personal relevance has a major impact on various judgmental strategies (Tesser et al., 1989). Consistent with our expectations, it was the combination of high personal relevance and negative mood that was most likely to result in motivated choices. Some processing differences due to positive or negative mood are predicted both by mood control (Clark & Isen, 1982) and by mood-as-information theories (Clore & Parrott, 1991; Schwarz & Bless, 1991). According to the latter, affect may serve as information predisposing people to effortful or effortless or to motivated or dispassionate information processing strategies (Clore & Parrott, 1991; Schwarz & Bless, 1991). However, neither of these models deals with the issue of personal relevance or specifically predicts the kind of selective, thematically biased information search and processing we found here. It seems that an adequate explanation of mood effects on social cognition and judgments may have to take into account the variety of alternative processing strategies available to people when they are dealing with realistic problems. Such a multiprocess model, building on earlier formulations by Brewer (1988), Fiedler (1990), Petty et al. (1991), and others has been recently proposed by Forgas (1991 b). This model identifies four distinct processing strategies (direct access, motivated, heuristic, and substantive) and makes explicit allowance for the joint effects of mood and personal relevance in triggering motivated processing strategies. The results of these three experiments provide clear support for the existence of such targeted choice strategies. Indeed, it may be useful to make a further distinction between processing strategies that are (a) genuinely unmotivated, (b) passively influenced by a general motivational state, and (c) actively goal-oriented in service of a specific motivational objective, as was found here. Some basic cognitive processes such as spreading activation or affect priming seem automatic and generally unmotivated and clearly belong in the first category (Bower, 1991). An interesting example of the second kind of process is Martin's (1986) ingenious set/reset formulation. This model proposed that mood-incongruent biases or "contrast effects" may be due to the "suppressed use of primed concepts" (Martin et al., 1990, p. 28) as people are motivated to form an impression of a target unbiased by their prior exposure to contextual stimuli. Thus, a nonspecific motivation for accuracy may lead to the reversal of the usual priming effects. Finally, the third category, information processing in the service of a specific motivational objective, has received surprisingly little attention to date. Our results documenting the selective information search and decision strategies used by subjects involving the active recruitment of evidence to satisfy a specific motivational goal offer a useful insight into the dynamics of motivated thinking. The demonstration of significant processing differences that are due to mood when making interpersonal choices is also of direct relevance to research on interpersonal behavior and relationships. At every stage of development (Levinger, 1990), relationships between people are maintained by the numerous small, recurring choices and decisions partners make. The interdependence of affect and cognition in relationship maintenance and development is now receiving increasing attention (Fincham & Fletcher, 1991). The kind of microanarysis of individual interpersonal choices presented here offers a promising experimental paradigm for studying real-life influences on partner choices and preferences. Results from experimental studies such as these are also relevant to our understanding of chronic mood disorders, such as depression. As depression usually involves a complex constellation of symptoms, causal links between mood and cognition are difficult to establish. In contrast, in experimental studies with normal subjects, we were able to generate "depressogenic" cognitions in a variety of memory, judgment, and attribution tasks in the past (Bower, 1991; Forgas & Bower, 1987; Forgas et al., 1984, 1990). The present results add a new dimension to such findings. Instead of demonstrating the usual mood-congruent judgmental bias, dysphoric subjects engaged in directed, motivated information search and choice strategies that were consistent with a desire to control or eliminate their dysphoria. There is similar evidence for motivated strategies in depressive patients directed at avoiding stressful, demanding, or potentially threatening tasks and situations (Ottaviani & Beck, 1988). However, it often appears that motivated strategies in depressive cognition involve retreat and avoidance of situations that could also lead to positive, rewarding experiences (Weary & Williams, 1990). Such motivated strategies may contribute to the kind of circular pattern of negative thought and affect that typically characterizes depression. More research exploring in detail the cognitive microstrategies used by depressives when dealing with similar interpersonal choices would be of obvious practical and clinical interest. Interpersonal choices are among the most complex and demanding cognitive tasks we face in everyday life. These studies showed that transient mood and the personal relevance of the outcome have significant effects on the way people look for, use, and remember information when making such choices. It was also found that motivated processing involving a selective search for rewarding outcomes was most likely to be adopted as a strategy by dysphoric subjects making a personal choice. Further research on this important topic is likely to be of considerable theoretical, as well as practical, interest to our understanding of interpersonal relationships and the effects of mood on cognition. References Argyle, M. (1986). The psychology of happiness. London: Heinemann. Bower, G. H. (1981). Mood and memory. American Psychologist, 36, 129-148. Bower, G. H. (1991). Mood congruity of social judgments. In J. P. Forgas (Ed.), Emotion and social judgments. Elmsford, NY: Pergamon Press. Brewer, M. (1988). A dual-process model of impression formation. In T. K. Srull & R. S. Wyer (Eds.), Advances in social cognition. Hillsdale, NJ: Erlbaum. Brewer, M. B., & Kramer, R. M. (1985). The psychology of intergroup attitudes and behaviour. Annual Review of Psychology, 36, 219-243. Clark, M. S., & Isen, A. M. (1982). Towards understanding the relationship between feeling states and social behavior. In A. H. Hastorf & 720 JOSEPH P. FORGAS sion formation. Journal of Personality and Social Psychology, 51, 493-504. Martin, L. L., Seta, J. J, & Crelia, R. A. (1990). Assimilation and contrast as a function of people's willingness and ability to expend effort in forming an impression. Journal of Personality and Social Psychology, 59, 27-37. Ottaviani, R, & Beck, A. T. (1988). Cognitive theory of depression. In K. Fiedler & J. Forgas (Eds.), Affect, cognition and social behavior. Gottingen, Federal Republic of Germany: Hogrefe. Parrott, W G., & Sabini, J. (1990). Mood and memory under natural conditions: Evidence for mood incongruent recall. Journal of'Personality and Social Psychology, 59, 321-336. Petty, R. E, Gleicher, F, & Baker, S. (1991). Multiple roles for affect in persuasion. In J. P. Forgas (Ed.), Emotion and social judgments. Elmsford, NY: Pergamon Press. Pietromonaco, P. R., & Rook, K.. S. (1987). Decision style in depression: The contribution of perceived risks and benefits. Journal of Personality and Social Psychology, 52, 399-408. Roth, D., & Rehm, L. R (1980). Relationships among self-monitoring processes, memory, and depression. Cognitive Therapy and Research, 4, 149-157. Sanitioso, R., Kunda, Z., & Fong, G. T. (1990). Motivated recruitment of autobiographical memories. Journal of Personality and Social Psychology, 59, 229-24]. Sarnoff, I., & Zimbardo, P. G. (1961). Anxiety, fear and social affiliation. Journal of Abnormal and Social Psychology, 62, 356-363. Schachter, S. (1959). The psychology of affiliation. Stanford, CA: Stanford University Press. Schwarz, N., & Bless, H. (1991). Happy and mindless, but sad and smart? The impact of affective states on analytic reasoning. In J. P. Forgas (Ed.), Emotion and social judgments. Elmsford, NY: Pergamon Press. Strack, F, Schwarz, N., & Argyle, M. (Eds.). (1991). Subjective well-being. Elmsford, NY: Pergamon Press. Tajfel, H., & Forgas, J. P. (1981). Social categorization: Cognitions, values and groups. In J. P. Forgas (Ed.), Social cognition: Perspectives on everyday understanding. San Diego, CA: Academic Press. Tesser, A. (1986). Some effects of self-evaluation maintenance on cognition and action. In R. M. Sorrentino & E. T. Higgins (Eds.), The handbook of motivation and cognition: Foundations of social behaviour (pp. 435-464). New York: Guilford Press. Tesser, A., Pilkington, C. J., & Mclnlosh, W D. (1989). Self-evaluation maintenance and the mediational role of emotion: The perception of friends and strangers. Journal of Personality and Social Psychology, 57, 442-456. Tesser, A., & Shaffer, D. R. (1990). Attitudes and attitude change. Annual Review of Psychology, 41, 479-523. Weary, G, & Williams, J. P. (1990). Depressive self-presentation: Beyond self-handicapping. Journal of Personality and Social Psychology, 58, 892-898. Wenzlaff, R. M., & Prohaska, M. L. (1989). When misery prefers company: Depression attributions and responses to others' moods. Journal of Experimental Social Psychology, 25, 220-233. Wheeler, L., & Miyake, K. (1990). The effects of depression on social comparison strategies. Unpublished manuscript, University of Rochester, NY. Wicklund, R., & Frey, D. (1981). Motivational processes in cognitive dissonance. In J. P. Forgas (Ed.), Social cognition: Perspectives on everyday understanding. San Diego, CA: Academic Press. Zajonc, R. B. (1980). Feeling and thinking: Preferences need no inferences. American Psychologist, 35,151-175. Received November 29,1990 Revision received May 28,1991 Accepted June 3,1991 • A. M. Isen (Eds.), Cognitive social psychology (pp. 72-108). New York: Elsevier-North Holland. Clore, G. L, & Parrott, G. (1991). Moods and their vicissitudes: Thoughts and feelings as information. In J. P. Forgas (Ed.), Emotion and social judgments. Elmsford, NY: Pergamon Press. Descartes, R. (1961). Passions of the soul: Essential works. (L. Blair, Trans.). New York: Bantam Books. (Original work published 1649) Erber, R., & Fiske, S. T. (1984). Outcome dependency and attention to inconsistent information. Journal of Personality and Social Psychology, 47, 709-726. Fiedler, K. (1990). Mood-dependent selectivity in social cognition. In W Stroebe & M. Hewstone (Eds.), European review of social psychology (Vol. I). New York: Wiley. Fiedler, K. (1991). On the task, the measures and the mood in research on affect and social cognition. In J. E Forgas (Ed.), Emotion and social judgments. Elmsford, NY: Pergamon Press. Fincham, F D., & Fletcher, G. J. O. (Eds.). (1991). Cognition in dose relationships. Hillsdale, NJ: Erlbaum. Fiske, S. T., & Neuberg, S. L. (1990). A continuum of impression formation, from category-based to individuating processes: Influences of information and motivation on attention and interpretation. In M. Zanna (Ed.), Advances in experimental social psychology (Vol. 23). San Diego, CA, Academic Press. Forgas, J. P. (Ed.). (1981). Social cognition: Perspectives on everyday understanding. San Diego, CA: Academic Press. Forgas, J. P. (1989). Mood effects on decision-making strategies. Australian Journal of Psychology, 41,197-214. Forgas, J. P. (Ed.). (1991a). Emotion and social judgments. Elmsford, NY: Pergamon Press. Forgas, J. P. (1991b). Affect and social perception: Research evidence and an integrative theory. In W Stroebe & M. Hewstone (Eds.) European review of social psychology. New York: Wiley. Forgas, J. P., & Bower, G. H. (1987). Mood effects on person perception judgments. Journal of Personality and Social Psychology, 53, 53-60. Forgas, J. P, Bower, G. H., & Krantz, S. (1984). The influence of mood on perceptions of social interactions. Journal of Experimental Social Psychology, 20, 497-513. Forgas, J. P., Bower, G. H., & Moylan, S. J. (1990). Praise or blame? Mood effects on attributions for success or failure. Journal of Personality and Social Psychology, 59, 809-819. Heider, F (1958). The psychology of interpersonal relations. New York: Wiley. Isen, A. M. (1984). Towards understanding the role of affect in cognition. In R. S. Wyer & T. K. Srull (Eds.), Handbook of social cognition (Vol. 3, pp. 179-235). Hillsdale, NJ: Erlbaum. Isen, A. M., & Means, B. (1983). The influence of positive affect on decision making strategy. Social Cognition, 2, 18-31. James, W (1890). The principles of psychology. New York: Holt, Rinehart & Winston. Janis, I. L_ & Mann, L. (1977). Decision making: A psychological analysis of conflict, choice and commitment. New York: Free Press. Kunda, Z. (1990). The case for motivated reasoning. Psychological Bulletin, 108, 480-498. Levinger, G. (1990, July). Figure versus ground: Macro and micro perspectives on personal relationships. Paper presented at the International Conference on Personal Relationships, Oxford, England. Locke, K. D., & Horowitz, L. M. (1990). Satisfaction in interpersonal interactions as a function of similarity in level of dysphoria. Journal of Personality and Social Psychology, 58, 823-831. Mackie, D., & Worth, L. (1991). Feeling good but not thinking straight: The impact of positive mood on persuasion. In J. P. Forgas (Ed.), Emotion and social judgments. Elmsford, NY: Pergamon Press. Martin, L. L. (1986). Set/reset: Use and disuse of concepts in impres-