The influence of mood on the intensity of emotional responses ...

COGNITION AND EMOTION, 2001, 15 (6), 725–747

The influence of mood on the intensity of emotional responses: Disentangling feeling and knowing Roland Neumann, Beate Seibt, and Fritz Strack University of WuÈrzburg, Germany

The results of three experiments suggest that pre-existing mood increases the intensity of affectively congruent emotions while dampening the intensity of incongruent emotions independent of attributional knowledge. This result was obtained using a new method for inducing mood states unobtrusively and with minimal or no cognitive concomitants. The results of Experiment 1 revealed that for participants who were exposed to positive feedback a pre-existing positive mood led to stronger feelings of pride in comparison to negative mood. The results of Experiments 2 and 3 suggest that pre-existing mood directly influences the experience of subsequently elicited emotions independent of what one knows about the causes of this feeling. When participants were required to differentiate between the funniness of a cartoon and their subjective humour response, mood influenced only the latter judgement (Experiment 2). In Experiment 3, reminding participants of the mood induction resulted in a contrast effect in judging the funniness of a cartoon. However, the pre-existing mood continued to exert an assimilation effect on the overt mirth response. In conclusion, these results suggest that the feeling and knowledge component are partly independent bases of emotional responses.

Moods and emotions can both be conceived as affective feelings and many researchers have used these terms interchangeably. Others have pointed out that moods and emotions represent entirely different mental states (Isen, 1990). According to our position, both views are partly correct. On one hand, moods and emotions share the affective feeling component and are therefore partly similar in their underlying representation, on the other, moods and emotions are Correspondenc e should be addressed to Roland Neumann, UniversitaÈt WuÈrzburg, Lehrstuhl fuÈr Psychologi e II, RoÈntgenring 10, D-97070 WuÈrzburg, Germany; e-mail: [email protected] wuerzburg.de This research was supported by Deutsche Forschungsgemeinschaf t Grant NE-721/1-1. We would like to thank Wolfgang Danneil, Uta Deppe, Eva-Maria Gortner, Christian Harneit, Yvonne KoÈrner, Marcus Neumann, Sonja Offermanns, and Anna Sequeira for their help in data collection. Special thanks go to Hans-Peter Erb, Roland Deutsch, Jens FoÈrster, Martin Schilcher, Sabine Stepper, and Bob Wyer who gave valuable suggestions. # 2001 Psychology Press Ltd http://www.tandf.co.uk/journals/pp/02699931.html DOI:10.1080/02699930143000266

726

NEUMANN, SEIBT, STRACK

different because emotions presuppose that one knows the origin of the affective feeling whereas moods do not necessarily hinge on that knowledge. To have an emotion does not simply mean to feel pleasant or unpleasant but one also knows the cause of this feeling (Frijda, 1994), for example, we are angry at someone or afraid of something. Whereas knowledge about the origin is a defining feature of emotions, moods lack such a focus and therefore can be the basis for a variety of judgements (Frijda, 1994; Ortony & Clore, 1989). In previous publications, we proposed that all feelings share a common mental representation which differs from the mental representation of knowledge in important aspects (Strack & Gonzales, 1993; Strack & Neumann, 1996). Feelings predominantly have experiential representations that may vary in intensity and possess perceptual properties. In contrast, knowledge is coded in noetic representations that are activated in an all-or-none fashion and can be characterised as propositions. Based on these assumptions, we suggest that moods and emotions share their experiential representation but differ in their noetic representation. Because emotions hinge on different partly independent representations they were described as syndromes consisting of at least a knowledge and an experiential component (e.g., Scherer, 1984). In contrast, moods lack the knowledge component and consist exclusively of an experiential component. This difference is reflected in the determinants of moods and emotions. Whereas emotions necessarily hinge on cognitive processes moods can also emerge from changes in physiological states (Gold, MacLeod, Frier, & Deary, 1995). Weiner (1986) suggested that emotions emerge from a sequence of cognitive processes that become increasingly specific. According to his approach, in a first step general positive or negative reactions emerge from the attainment or nonattainment of a desired goal. The phenomenal quality of these outcome-dependent and attribution-independent ‘‘primitive’’ emotions is then shaped by subsequent causal ascriptions. For example, perceiving a negative outcome as controllable by others triggers the emotion of anger, whereas feelings of guilt follow from the perception that the causes of a negative outcome are controllable by oneself (Neumann, 2000; Weiner, 1986). Thus, the relation of feelings to imagined or real objects is shaped by attributional processes that occur temporally after the elicitation of primitive emotions. From our point of view, such primitive emotions are similar to moods in that they consist of an either positive or negative feeling and are not directed towards an object. Therefore, the sequence described in Weiner’s approach can be reframed such that moods are transformed into emotions by attributional processes. Whether emotions actually emerge from such a sequential process is not clear up to now. Nevertheless, if moods can be transformed into emotions by attributional processes it is conceivable that pre-existing moods fuse with feelings stemming from an emotion-eliciting event. Notably, one important feature of conscious experience is that it is impossible to experience different feelings at

INFLUENCE OF MOOD ON EMOTIONS

727

the same point in time (Marcel, 1983). Nevertheless, it is not uncommon that in the flow of daily experience feelings emerge from multiple events which may occur together or in quick succession. Because feelings last for some time and subsequently elicited feelings do not entirely replace pre-existing feelings it is reasonable to assume that subsequent feelings can fuse with pre-existing ones in such situations. This is why feelings can be conceived as an accumulated unitary outcome of several different processes. Because of the perceptual properties of feelings, pre-existing affective states can intensify or dampen feelings stemming from subsequent processes. Support for the assumed additivity of affect comes from affective priming research showing that nonconsciously processed affect adds up (Murphy, Monahan, & Zajonc, 1995). More specifically, Murphy et al. (1995) have shown in a series of studies that repeated exposure and the valence of subliminally exposed primes exert combined effects on subsequent targets. Extending this notion, we suggest that not only unconstrained affect but all affective feelings show this additivity, whether they are experienced as moods or as emotions. We claim that whenever a new affective stimulus is encountered, the resulting affective feeling is a function of the valence and intensity of the affective feeling present at the time of the encounter and the valence and intensity of the affective feeling elicited by the new stimulus. In accordance with Zillmann’s (1983) excitation transfer approach we assume that residuals of prior experience can influence subsequent experiences. However, different to that approach we assume that as well as arousal, the valences of the experiences matter in that congruent hedonic valences add up whereas incongruent valences should cancel each other out. Thus, the valence and intensity of pre-existing feelings might in part determine the intensity of subsequently elicited emotions. In line with these expectations, Branscombe (1985) found that the hedonic valence of the emotional state that had been elicited by one movie influenced the emotional response to a second movie in a congruent direction. Specifically, the second movie elicited more happiness if the first one had also elicited happiness and it elicited more sadness if the first movie evoked the same feeling. Moreover, emotions of opposite hedonic valence cancelled each other out. Thus, watching a happy movie levelled out feelings of sadness that stemmed from the first movie, whereas watching a sad movie levelled out the effects of a previously seen happy one. Although these findings support the assumption that affective feelings can influence each other depending on the congruency of their underlying hedonic valence they do not provide clear evidence for the mediating mechanisms. Different from our conceptualisation Branscombe (1985) suggested that the effect is mediated by inferential processes rather than by a transfer of the hedonic feeling. She assumed that ‘‘the cognitive label associated with the prior emotional state has an effect on the subsequent emotional state’’ (Branscombe, 1985, p. 166). Thus, participants might have simply reasoned that two positive emotions result in a more intensive positive emotion than just one

728


and therefore inferred that they were more amused. Instead, we assume that preexisting moods or emotions can influence the experiential component of emotions independent of the knowledge component. Unfortunately, however, the procedures used in Branscombe’s experiment do not allow to distinguish between an inferential approach and a transfer of hedonic valence account because she examined the influence of pre-existing emotions on subsequent emotional responses. Because emotions include both feelings and knowledge, it is unclear whether pre-existing feelings can influence the experiential component of emotions in the absence of inferences or not. Examining the effect of pre-existing moods on subsequently elicited emotions and making sure participants do not detect the mood manipulation should eliminate this interpretational ambiguity. This does not mean, however, that inferential mechanisms do not play a role in emotional responses. As was pointed out previously emotions are not solely a function of the experiential component but also depend on attribution processes which are partly independent of these feelings and form the knowledge component of an emotion. In line with Weiner (1986), we suggest that inferential mechanisms follow rather than precede affective feelings.

Feeling as information in emotion-relevant judgements It is important to note that usually individuals are not aware of the possibility that their current feeling might stem from more than one source. Instead, they attribute the entire feeling to the last eliciting experience. When judging an emotion-eliciting event or stimulus, individuals therefore use by default the intensity of their current feeling as a source of information. An assimilation of the judgement to prior mood should result. However, if reminded of the potentially biasing effect of prior mood on their judgement, individuals should try to correct for it. This prediction can be derived from the feeling-asinformation approach (Schwarz, 1990), which claims that feelings are used in evaluative judgements as long as they are seen as an immediate response to the object of the judgement. Whenever individuals see their feeling as irrelevant to the judgemental task they will no longer use it in evaluative judgements. Accordingly, we propose that the appraisal of the emotion-eliciting object or event can be regarded as a judgemental task that involves both the current feeling and knowledge about the feeling. To test our assumption that moods influence the experiential component of an emotion leaving the knowledge basis unchanged, we can therefore use the judgement of the emotion-eliciting event as a dependent measure. If pre-existing moods affect only the experiential component of subsequently elicited emotions, the appraisal of the emotion-eliciting event should be contrasted away from the mood influence when it is made salient.


729

A different prediction, however, arises from Bower’s network model (Bower, 1981) which claims that bidirectional links exist between affective states and cognitions so that moods can activate congruent cognitive contents at both the encoding and retrieval stage of information processing. From that point of view, pre-existing moods should activate congruent evaluations of an event or object so that it is perceived as funnier, riskier, sadder, etc. The appraisal of the event or object should be assimilated to the mood even if assessed along with the feeling itself or if the mood manipulation is made salient. In sum, if a preexisting mood changes the knowledge component of an emotional reaction along with or instead of the experiential component, the judgements of the emotion-eliciting event should show a different pattern than the one which we predict. Before turning to the individual studies, we present a brief overview of the three studies conducted to test our prediction.

OVERVIEW In our framework, emotional responses consist of two partly independent components, namely, feeling and knowledge. Our central aim is to show that pre-existing moods can influence the experiential component of subsequently elicited emotions independent of the knowledge component. Three experiments were conducted to examine whether congruent feelings add up whereas incongruent feelings exert a dampening impact on each other. In the first study, we wanted to find out if unobtrusively induced mood states exert an influence on the subsequently elicited emotion of pride. In the second study, the same induction procedure was used to study whether global moods influence the experiential component of humour responses while leaving the knowledge component unchanged. The third study tested the assumption that reminding participants of the mood induction procedure results in a contrast effect on humour judgements. At the same time, we expected the overt mirth response as an indicator of the experiential component to still show the postulated intensification in the case of compatible valences of mood and emotion and dampening in the case of incompatible ones.

STUDY 1 The aim of the first study was to examine whether pre-existing moods can intensify congruent emotional responses. To guarantee that participants do not know the origin of their affective experience, it is necessary to induce a mood state unobtrusively and with minimal or no cognitive concomitants. Such a procedure has been developed in a previous series of studies (Neumann & Strack, 2000). It involves the use of the human voice which has been shown to be affectively contagious (Hatfield, Cacioppo, & Rapson, 1992). In this experimental paradigm, participants are exposed to a speech about a philosophical topic that is subtly spoken with either a happy or a sad voice while the

730


content of the speech is held constant. In order to avoid that participants’ attention is directed to the origin of their mood state, the subjective feeling is not assessed after the mood induction. A series of studies was conducted to ascertain that the listeners’ mood state is indeed altered by the vocal expression of the speaker (Neumann & Strack, 2000). All of them revealed reliable mood changes as a function of exposure to the different versions of the speech. In one study, for instance, participants rated their mood on a 10-point scale after listening to the speech. Listening to the subtly happy voice resulted in a higher subjective mood rating (M = 7.0) than listening to the subtly sad voice (M = 4.4), F(2, 27) = 4.77, p < .02. No discrete emotion was elicited by the vocal expression. When participants were asked to indicate the extent to which they felt each of several discrete emotions, no effects were obtained (all Fs < 1). Moreover, when participants were additionally asked how much the voice had changed their feelings, no effect was obtained (F < 1), (Neumann & Strack, 2000, study 1). We concluded from these findings that a mood state rather than a discrete emotion is induced by the employed procedure. Therefore, this procedure seems appropriate to induce a mood state unobtrusively in the present studies. In the first study, the influence of this unobtrusively induced mood on the intensity of the subsequently elicited emotion of pride was examined. In contrast to an undifferentiated mood, pride is only felt if a positive outcome is internally attributed (Weiner, 1986). We expect that a happy mood intensifies the feeling of pride compared to sad mood. According to our conceptualisation, the current feeling should be applied only when judging the intensity of the elicited emotion but not for judging other emotions. Such a prediction can be derived from the feeling-as-information approach (Schwarz, 1990) which holds that feelings which are already attributed to a possible cause do not influence other judgements (Keltner, Ellsworth, & Edwards, 1993). Thus, knowing the origin of one’s current feeling eliminates the need to explain it in terms of another discrete emotion. Therefore, a pre-existing mood state should not exert an influence on every emotion assessed but exclusively on the emotion that is actually elicited. According to Zillmann (1978), carry-over effects from one affective state to the next can also occur due to excitation transfer, which means that the arousal components of both affective states add up. Note that this process should occur irrespective of the valence of the first affective state. We predict, however, that happy and sad moods have opposite effects on subsequently elicited emotions, irrespective of their arousal qualities. To induce the emotion of pride, participants in the present study were given an achievement task. After this task, half of the participants received an above average feedback about their individual achievement whereas the other half received a below average feedback. We expected that only those participants who received positive feedback while being in a positive mood should report the feeling of pride. For those in a negative mood receiving positive feedback, mood


731

should dampen the experience of pride. Consistent with Weiner’s assumption that pride is only felt if a positive outcome is attributed internally, participants receiving negative feedback should not feel any pride at all.

Method Participants. A total of 40 students from the University of Trier served as participants. Approximately half of the participants were female. Participants were recruited for a study on ‘‘attentional resources and academic achievement’’. They were paid DM5 (approximately US$3.60 at the time) for their participation. Design. The experimental design was a 2 £ 2 factorial comparing Voice (sad vs. happy) and Feedback (above average vs. below average) with both factors varied between participants. Participants were run in groups of up to four and each participant was randomly assigned to experimental conditions. Procedure. Participants were recruited for a validation study of a new achievement test. After their arrival in the laboratory they were seated on separate tables with partitions in between to obstruct their view on each other. Then, they were informed that the test was being developed as a measure of the ability to study at a university. It was explained to them that their task was to listen to a difficult speech, trying to remember it as correctly and completely as possible, while doing a dexterity task at the same time. They were told that this is a task which allows to predict future achievements at the university. As dexterity task, we used the Finger Dexterity Test by O’Connor (1932). Participants were instructed to insert three metal pins into each of 100 holes in a wooden board. They were told that the number of correctly recollected sentences from the text in relation to the number of pins would be a sensitive indicator of attention in the context of academic tasks. Participants were requested to perform the dexterity test until a specific word occurred in the speech. They then listened to one of the two pre-tested audiotapes which lasted about 4 minutes and was presented via earphones. The audiotapes served to unobtrusively manipulate participants’ mood state. Participants in the happy-voice condition listened to the speech presented in a slightly happy voice while those in the sad-voice condition heard the same speech in a slightly sad voice. Based on pre-tests, listening to a happy voice was expected to raise participants’ mood while a sad voice was expected to deteriorate it. After listening to the speech, participants were instructed to count the number of holes with three pins and indicate it in a questionnaire where the alleged average achievement of other participants was noted. To prompt the experience of pride, participants in the above average feedback condition were given a low

732


standard of comparison, whereas participants in the below average feedback condition were given a high standard. Pre-tests had revealed that participants were able to insert between 35 and 55 holes during the speech. Thus, in the above average feedback condition, the number given in the questionnaire as the average achievement was 20. It was expected that this feedback sets a standard which results in a positive evaluation of participants’ own performance and therefore elicits pride. In the below average feedback condition, participants were informed that the average number of inserted pins was 60. It was expected that participants would not be proud because they were led to believe that their performance was below average. At this point, participants were informed that we were interested in affective factors which might have influenced their performance. To assess their feelings of pride together with other emotions, participants were given five emotion adjectives and asked to indicate on 10-point rating scales ranging from 0 (not at all) to 9 (very much) the degree to which they were feeling each emotional state. The emotion adjectives were (in order of their presentation) sad, proud, happy, angry, and anxious. To control for a mediating role of excitation transfer (Zillmann, 1978) participants were asked to further indicate the degree to which they felt aroused. Before participants left the laboratory, they were closely questioned for suspicion. In stepwise debriefing, participants were informed about the actual purpose of the study and about the contrived nature of the feedback. When confronted with the actual purpose, most participants uttered surprise and no participant mentioned any suspicion. Some participants mentioned that they rather gave credit to the alleged purpose of the study than to the real one.

Results Dexterity performance. The number of holes containing three pins was used as a measure of manual dexterity. On average, participants inserted 37 pins. Participants’ performance were analysed in a 2 (Feedback: below average vs. above average) £ 2 (Voice: sad vs. happy) ANOVA. The means did not differ significantly from each other (F < 1). As expected from pre-testing, participants’ relative performance was always lower than the standard in the below average feedback condition and clearly higher than the standard in the above average feedback condition. Subjective feelings. To assess various qualities of affect after the session, participants were asked to report the intensity of their feelings on adjective checklists. Separate 2 (Feedback: below average vs. above average) £ 2 (Voice: sad vs. happy) ANOVAs for these variables yielded only one significant effect: The manipulations of the present experiment only influenced reported feelings of pride and left the remaining variables including the


733

reported arousal unaffected (all Fs < 1), (for comparisons of the individual means, see Table 1). Inspection of Table 1 reveals that participants who both received above average feedback and listened to a happy voice reported the most pride. The main effect for the mood manipulation on the reported feelings of pride was marginally significant, F(1, 38) = 3.4, p > .07, and the main effect for the feedback manipulation was not significant, F(1, 38) = 1.9, p > .18, the predicted interaction between the feedback and the mood manipulation was reliable, F(1, 38) = 4.1, p < .05.

Discussion Results provide support for the hypothesis that unobtrusively induced moods influence the intensity of a subsequently elicited emotion. Participants felt prouder when receiving above average feedback in a good mood than when receiving it in a bad mood. Moreover, as predicted, mood did not exert a global effect on all assessed affective experiences. Instead, it altered specifically the intensity of the emotion which was elicited by the appraisal of the situation. Thus, consistent with previous findings, feelings that are already attributed to a possible source do not influence other judgemental dimensions (Keltner, et al., 1993). As predicted, participants in the below average feedback condition did not show any pride as they were not in a pride-eliciting situation. Furthermore, in line with Branscombe’s (1985) research we obtained evidence that incongruent hedonic valences level each other out. That is, participants who received positive feedback in a negative mood did not report any

TABLE 1 Subjective feelings as a function of feedback and voice in Study 1 (n = 10 for each group)

Emotion Group

Sad

Proud

Happy

Angry

Anxious

Aroused

Below average Happy Sad

2.3 2.1

1.5 1.6

4.2 4.9

2.1 1.1

2.4 1.8

3.1 2.5

Above average Happy Sad

2.0 2.5

3.6a 1.2

3.9 3.8

1.8 1.1

1.3 1.9

2.7 3.3

Note: The scales ranged from 0 (note at all) to 9 (very much). a The only reliable effect was the Mood 6 Feedback interaction with Proud as the dependen t measure. Single comparisons revealed that this cell differs from all the others in this column.

734


pride. This might be due to the fact that when assessing their pride, individuals use both, the knowledge and the experiential component of their emotional reaction. Although those sad participants might know that the above average feedback in our study can elicit the emotion of pride, they do not experience the positive experience that is necessary to claim that one is actually proud. Participants in the below average feedback condition were somewhat more happy than those in the above average feedback condition. This could be due to the fact that participants often engage in conversational strategies while answering questionnaires and try to differentiate two subsequent responses (e.g., Strack Schwarz & WaÈnke, 1991). It is unlikely that this reflects different experiences. The fact that happy, but not sad participants in the above average feedback condition judged themselves to be proud is incompatible with the assumption that mood effects on the intensity of a subsequent emotional response are due to the arousal which accompanies the mood, because happy and sad participants had equal levels of arousal. The data justify the conclusion that listening to the speech had no differential impact on participants’ arousal. Thus, the data support the hypothesis that it is the valence, rather than the arousal of the pre-existing mood, which determines the direction of influence on subsequently elicited emotions. In sum, the results of Study 1 provide evidence that successively elicited affective states add up if they have the same valence, and cancel each other out if they possess opposite valences. A possible alternative explanation, though, would be that the pre-existing mood actually changed the interpretation of the situation, in this case the appreciation of the performance. Participants in a positive mood might have evaluated their own performance as more outstanding than participants in a bad mood who might have seen it as only slightly better than the average performance, despite equal feedback in both conditions. To rule out this possibility we conducted a second experiment where we assessed the cognitive evaluation of a situation and the emotional reaction to it separately.

STUDY 2 Our aim in Study 2 was to replicate the results of Study 1 with a different target emotion, namely, humour responses. At the same time, we wanted to tap more directly the mechanisms underlying mood effects on emotions. Although Study 1 provided evidence that the intensity of an emotion can be influenced by a prior mood induction it is not clear whether this is due to the knowledge or to the experiential component of emotional responses. We cannot rule out the possibility that the evaluation of the emotion-eliciting event mediated the effect of mood on emotional intensity. To overcome this interpretational ambiguity of Study 1, we assessed emotional intensity and the evaluation of the humourous stimuli separately in Study 2. This should allow us to determine if indeed a


735

pre-existing mood state influences exclusively the experiential component but not the knowledge component of an emotion. Prior research has demonstrated that individuals usually rely on their feeling when either judging their subjective emotion or the emotion-eliciting event (Olsen, 1992; Olsen & Roese, 1995). They do, however, have access to the two components of emotional responses if they are required to differentiate between their more subjective feeling and their more objective knowledge about an emotional event. If, for example, individuals are required to judge both their subjective feeling and the eliciting event, they rely on their feeling only for judging their subjective emotional experience but not for the appraisal of the eliciting event. This has been shown in studies which required participants to distinguish between the ‘‘objective’’ features of an emotional stimulus and the ‘‘subjective’’ feeling state it elicited (e.g, Gasper & Clore, 1998; Gavanski, 1986; Strack, Martin, & Stepper, 1988). When do people differentiate between their feeling and their knowledge? Such situations occur when several questions which tap partially overlapping knowledge bases are asked in the same conversational context. In these cases, the use of the same basis to judge two similar questions would contradict conversational norms, that is, that of nonredundancy (Grice, 1975). As an example, Schwarz, Strack, and Mai (1991) asked participants to report their happiness with their dating life before reporting their satisfaction with life as a whole. When these two questions were framed as belonging to different conversational contexts, the correlation between the two was considerably higher than when presented in the same context. That is, information already provided in the specific question was excluded from answering the general question. By the same vein, when an individual is specifically asked to report his/her feeling, he/she should exclude it from any more general judgement, such as judging an emotion-eliciting event, in that same conversational context, basing the general judgement on other knowledge instead. The feeling, then, is no longer a relevant basis for that judgement. This was first shown by Gavanski (1986) who instructed participants to indicate how amused they felt while watching the a cartoon and how funny this cartoon was from an objective point of view. The funniness question focuses on the knowledge-based evaluation of a cartoon whereas the amusement question records participants’ affective experience. In line with this assumption, Gavanski (1986) found that the repeated exposure to a cartoon decreased the subjective amusement but had no influence on the appraised funniness of the cartoon. Thus, even though the subjective amusement as measured by the rating scale and the overt mirth response decline with repeated exposure (due to adaptation) participants nevertheless know that the cartoon keeps its capability to evoke humour responses in others. To sum up, the method first introduced by Gavanski (1986) seems valid for separating subjective and objective reactions to emotional stimuli. The amusement measure is sensitive to manipulations of the intensity of

736


the affective feeling while the funniness measure remains unchanged by such manipulations. If our assumption about the process whereby pre-existing moods influence subsequently elicited emotions is correct, our mood manipulation from Study 1 should therefore have no impact on the funniness measure while at the same time producing an assimilation effect on the amusement measure. Specifically, we hypothesise that humourous stimuli evoke more amusement after participants have been exposed to the happy than to the sad voice. In contrast, the funniness should not be influenced by the prior mood. A further shortcoming of Study 1 is that it does not clarify whether the obtained results were due to both augmentation as well as reduction effects. Based on our assumption about the additivity of the experiential component, we predicted that good mood would boost the experience of pride, while bad mood would spoil it. However, the results of Study 1 could be due to either one of these processes as no control group with neutral mood was employed. To overcome this interpretational ambiguity, we introduced in Study 2 a group who listened to the audiotaped speech with a neutral emotional expression. This allows to examine whether hedonic compatibility between the pre-existing mood and emotion exerts an intensifying effect and incompatibility a dampening effect.

Method Participants. A total of 30 (15 female and 15 male) students at the University of Trier served as participants. They were recruited for a ‘‘test of attentional resources in academic tasks’’ and were paid DM5 (approximately US$3.60 at the time) for their participation. Design. The experimental design was a one factorial between-subjects design, using Voice (sad vs. happy vs. control) as an independent variable. Participants were run in groups of ten and each group was randomly assigned to experimental conditions. Procedure. After their arrival in the laboratory, participants were seated at different tables with a questionnaire in front of them. They were told not to open the questionnaire yet. Then they were told that they were participating in a study investigating learning abilities with different material. They were further told that we would first test their text comprehension. Then, the same mood induction procedure was used as in Study 1. Participants were required to listen to a philosophocal speech which was spoken in a sad, happy or neutral voice by a male actor. Immediately after listening to the philosophical speech, participants were told that we would now test them for their comprehension of drawings. Therefore, they were instructed to open the questionnaire, inspect the drawing presented on the next page and whenever they felt ready thereafter turn to page 3 to answer


737

questions about that drawing. Page 2 of the questionnaire showed a cartoon from the collection The Far Side by Gary Larson. On page 3, two questions were posed. First, participants were asked to indicate on a 10-point rating scale ‘‘How funny do you think the cartoon is if you try to apply an objective standard?’’ The scale ranged from 0 (not at all funny) to 9 (very funny). Second, the affective response was assessed by asking the question: ‘‘What feeling was elicited in you when you looked at the cartoon?’’ Participants were asked to respond on a 10point rating scale ranging from 0 (I felt not at all amused) to 9 (I felt very much amused). At the end of the study, participants were closely questioned for suspicion. No participant uttered any doubts concerning the cover story or mentioned any connection between the voice quality and their subjective feeling state. Then participants were paid and debriefed.

Results Participants’ responses were subjected to a single-factor between-subjects ANOVA with Vocal Expression as the independent variable (happy, neutral, sad). Ratings of funniness. Inspection of Table 2 reveals that the funniness ratings were almost identical across conditions. Not surprisingly, the appropriate one-way ANOVA was not significant (F < 1). Thus, the knowledge component of the humour response was not influenced by the mood manipulation. Ratings of amusement. In contrast, participants’ amusement ratings were clearly influenced by the experimental manipulation. The mean ratings (see Table 2) suggest that participants were more amused by the cartoon if they had previously listened to the happy voice than to the sad voice. Control participants who listened to the neutral voice gave amusement ratings that were in between the two treatment groups. The one-way ANOVA for the amusement variable

TABLE 2 Humour reaction as a function of the voice in study 2 (n = 10 for each group)

Voice Humour componen t

Sad

Neutral

Happy

Amusement Funniness

3.7a 4.1a

4.4a 4.0a

5.2b 4.1a

Note: Within rows, means not sharing a subscript differ at p = .05.

738


was significant, F(2, 27) = 3.5, p < .05. Individual comparisons between cell means yielded significant differences between the sad and happy voice condition, t(27) = 2.6, p < .02, and a marginally significant difference between the happy voice and the control condition, t(27) = 1.8, p < .07. The difference between the control and the sad condition does not reach the level of significance (t < 1).

Discussion The findings of the second study support our assumption that pre-existing moods exert an assimilating effect on subsequently elicited emotions by changing their experiential component. We observed that our mood manipulation affected the experiential component of the humour response but not the knowledge component. Specifically, participants who had been put into a good mood by being exposed to a happy voice reported feeling more amused by a cartoon than participants who had been put into a bad mood by listening to a sad voice. In line with our expectation, no such mood effect was obtained for the funniness ratings. That is, as in previous studies (Gavanski 1986; Strack et al., 1988, Study 2), participants differentiated between their more subjective and their more objective response to the cartoon, such that only their subjectively felt amusement was influenced by the mood manipulation but not the objective judgement about the funniness of the cartoon. Thus, our findings are in accordance with previous results (Branscombe, 1985; White, 1979), although our interpretation is somewhat different. Because the mood induction occurred in a maximally unobtrusive fashion it is unlikely that the influence of mood on the amusement ratings was the result of an inferential process on the basis of knowledge about the potential influence of the pre-existing mood as was suggested by Branscombe (1985). Unlike her, we assume that for an affective state to influence a subsequently elicited emotion it is not necessary for subjects to label that affective state. Rather, we propose that an emotional appraisal involves a judgemental task in which the current feeling is attributed to a salient plausible cause of it as a default option. In the next experiment we will provide support for our argument that the experiential component of a humour reaction will no longer be used as a basis for judging cartoons if the experienced feeling is attributed to the prior mood induction.

STUDY 3 In the previous study participants were required to distinguish between the knowledge and experiential component of their humour response. Although the validity of this measure was supported in several studies (Gavanski, 1986; Strack et al., 1988), we wished to further validate it and provide more evidence supporting our central claim using a different procedure. In order to show that moods exert an assimilating effect on emotions because affective feelings add


739

up, it is desirable to separately manipulate the knowledge and the experiential component of the emotional response. One technique to disentangle the feeling and the knowledge component of emotional responses is provided by the feeling-as-information approach (Schwarz, 1990). According to this account, individuals use their feeling as a basis of their judgement as long as they consider it a response to the object of the judgement. For example, Schwarz and Clore (1983) have shown that feelings stemming from weather conditions are used to judge one’s own life satisfaction. However, the impact of sunny and rainy days was eliminated when the diagnostic value of participants’ current feeling was called into question by asking them first about the weather. The same logic can be applied to our research: Being aware that the current feeling does not stem from the emotional stimulus but from a previous mood induction should lead to a correction of its judgement. As various studies have shown, such a correction process often leads to overcompensation, that is, individuals tend to contrast the judgement away from the assumed biasing influence, in this case the mood induction (Schwarz et al., 1991, Study 3; Strack, Schwarz, Bless, KuÈbler, & WaÈnke, 1993). Despite this compensation which is based on one’s knowledge about the contamination of judgements, the experiential component should still be influenced by a preexisting mood state. This is because moods directly fuse with the emotional feelings that emerge from emotion-eliciting cognitive processes. Thus, knowing that one’s current feeling does not stem from the assumed cause might change the quality of the reported emotion but not the intensity of the current feeling. To test the prediction that the experiential component of emotions varies independent of the knowledge component, a third study was conducted. In this study, which again investigated the influence of moods on humour responses, we directed participants’ attention towards the prior mood induction before asking them to judge the funniness of a cartoon. In line with the findings of the feeling-as-information approach (Schwarz, 1990), we expected that participants would try to correct for that influence. Accordingly, if participants’ attention was directed towards the mood induction procedure those who were put into a happy mood should judge the cartoon to be less funny, whereas those in a sad mood should judge the cartoon to be more funny. Without being aware that a prior mood induction might have influenced one’s affective response to the cartoon, participants in a happy mood should judge the cartoon to be more funny whereas sad participants should judge the cartoon to be less funny.1 Most importantly, however, to disentangle the independent contributions of the 1 Note, that without the requirement to differentiate between the objective and the subjective componen t of the humor response (see Study 1), participants are expected to use their subjective feeling to judge the funniness of a cartoon by default (see also Strack et al., 1988 for the same argument). Therefore, we expect that the funniness ratings reflect the influence of the subjective feeling.

740


knowledge and the experiential component to the experience and judgement of emotions in this study, the EMG activity in the region over the zygomaticus muscle was assessed as an indicator for the feeling response. This EMG activity can be seen as a direct indicator of the affective feeling as long as no intention to control one’s facial expression comes into play. Spontaneous contraction of the zygomaticus muscle (smiling) indicates amusement, and the more amusement one feels, the stronger the contraction. We predict a main effect for the mood induction: Participants in a happy mood should show a higher activation of the zygomaticus muscle than those in a sad mood.

Method Participants and design. A total of 49 students from the University of WuÈrzburg served as participants and were paid DM5 (approximately US$3.60 at the time) for their participation. They were randomly assigned to the conditions of a 2 £ 2 design, comparing Reminding (reminding vs. no reminding) and Music (happy vs. sad). Both factors were varied between participants. Procedure. Participants were recruited for a study on ‘‘subvocalisation and mental activity’’. On their arrival in our laboratory they were informed that we were interested in the effects of distraction on activity called ‘‘silent talk’’. Therefore, they were allegedly distracted by listening to music while counting upwards silently. To extend the findings of our previous studies, music served as a mood induction procedure in this study. To induce a happy mood half of the participants listened to Garden Party by Mezzoforte. The other half of the participants listened to Adagietto for Strings by Barber, which should induce a sad mood. Prior studies in our own lab and of other researchers (Niedenthal, Halberstadt, & Setterlund, 1998) had shown that these pieces of music reliably induce the expected moods. To ostensibly measure their silent talk while counting upwards, EMG Beckmann Electrodes were fixed over the region of the zygomaticus major muscle. Then they received a Walkman and were instructed to simultaneously start the tape and start counting upwards. After exposure to the music, half of the participants in each music condition were instructed to indicate how happy the music was on a scale from 0 (sad) to 9 (happy). This was done to direct participants’ attention to the source of their mood state. Then participants were informed that we wanted to assess their subvocalisation while they were exposed to pictures. Similar to the procedure employed in the second study they were told that we selected a cartoon as pictorial stimulus that was presented on the computer. The cartoon was presented for 12 seconds followed by a separate screen with the question ‘‘How funny is the cartoon?’’ and a 10-point rating scale ranging from 0 (not at all funny) to 9 (very funny). While inspecting the cartoon, the EMG activity over the


741

zygomaticus region was continuously assessed. Following their judgement of the cartoon, participants were asked about their interpretation of the experiment. Afterwards, they were debriefed, thanked and paid.

Results No participant uttered doubts that the study was concerned with the investigation of subvocalisation. Therefore, the data of all participants were submitted to analysis. EMG scores. The EMG signal was transmitted to a personal computer and digitised at a resolution of 0.3 mV/unit at a rate of 1000 samples per second. Artifacts were removed from the raw data on the basis of video-recordings of the participants. Then, the resulting EMG samplings were averaged across the time participants watched the cartoon and an equal length of time before the cartoon was presented. Finally, for each participant an index was computed by subtracting the EMG mean before the cartoon was presented as a baseline from the mean obtained while exposed to the cartoon. This index was submitted to a 2 (happy vs. sad) £ 2 (reminding vs. no reminding of the mood induction) ANOVA. As Table 3 reveals, the averaged EMG activity was higher for those participants who had listened to the happy (M = 43,6 mV) as compared to the sad music (M = 7.8 mV). Analysis revealed a marginally significant effect for mood, F(1, 45) = 3.7, p < .06. However, reminding participants of the mood induction did not reliably affect the averaged EMG scores (F < 1). Moreover, the interaction between mood and reminding was also not significant (F < 1).

TABLE 3 Funnines s ratings and EMG-measure of amusement reaction as a function of exposure to music and reminding in Study 3

Music Reminding

Happy

Funniness measured by rating scale With 4.3a Without 6.0b

Sad

5.6b 4.0a

Amusement measured by EMG (mV) With 54.8 a 8.6b Without 30.3 a 6.8b Note: Within each dependent measure, means not sharing a subscript differ at p < .05.

742


Ratings of funniness. As depicted in Table 3, without being reminded of the mood induction, participants rated the cartoon as funnier after listening to the happy than after listening to the sad music. However, this pattern was reversed if participants’ attention was directed to the music. As predicted, participants judged the cartoons to be less funny after listening to the happy than after the sad music if they were reminded of that source of influence. The interaction was reliable, F(1, 47) = 4.5, p < .04. No other effect reached the level of significance (all Fs < 1).

Discussion The present study replicates and extends the findings of the previous two experiments. As long as the attention was not directed towards the mood induction procedure, happy participants judged the cartoon to be more funny than sad participants. Directing participants’ attention to the mood induction procedure, however, prompted a reversal on the funniness ratings. That is, if participants’ attention was directed to the mood induction procedure, those in a happy mood judged the cartoon to be less funny than those in a sad mood. Thus, reminding participants of the exposure to the music resulted in a contrast effect on the verbal reports of emotion. Unlike the funniness ratings, the experiential component was influenced only by the mood induction but not by the knowledge about the origin of the affective response. This is evident from the finding that participants who had listened to happy music showed an increased activation of the zygomaticus muscle as a response to the cartoon as compared to those who had listened to the sad music, whether they had been reminded of the origin of their mood or not. In sum, these findings suggest that the information provided by the experiential component is not always used to judge the emotion-eliciting event. If participants are aware that their current feeling does not stem from the judged emotional event but from another source, they correct for the assumed biasing influence by contrasting their judgement away from it. In sum, these findings support our assumption that appraisals of the object of an emotion can be conceived of as judgements which are based on both the experiential component and the knowledge component of emotions. The pattern of obtained results is consistent with the view that only the experiential component but not the knowledge component of an emotion is influenced by a pre-existing mood.

GENERAL DISCUSSION We suggested that emotions consist of a partly independent experiential and a knowledge component whereby the experiential component provides information about the valence and the intensity of the emotional response. In contrast, the knowledge component provides information referring to the causes and the object of the emotion. In support of these assumptions we found in the first study


743

that happy mood but not sad mood intensified the feeling of subsequently elicited pride. Moreover, participants judged exclusively the activated emotion on the basis of their subjective feeling. Thus, when the feeling is already attributed to a possible eliciting event, there is no further reason to find other causes for that experience (Keltner et al., 1993). Furthermore, Study 1 reveals that it is not sufficient to just know that a specific situation is potentially capable of eliciting a specific emotion in order to feel this emotion. Instead, individuals only experience an emotion if the valence and intensity of the underlying feeling is compatible with the knowledge about the eliciting circumstances. The second study provided evidence that pre-existing mood influences the experiential component of subsequently elicited emotions independent of the knowledge component. In this study, participants were required to distinguish between their subjective response and their objective evaluation of a cartoon. We expected that the prior mood induction would exclusively influence the experiential component but not the knowledge component of the humour reaction. Consistent with this prediction we found that the subjective experience but not the objective assessment of the cartoon was influenced by the prior mood induction suggesting that the experiential component was affected independent of the knowledge component. Study 3 replicated this finding using an independent manipulation of the experiential and the knowledge component. In this study, again, the experiential component was manipulated by an unobtrusive mood induction and half of the participants were reminded of this influence before judging the funniness of a cartoon. In line with other research showing mood effects on judgements, happy mood increased the perceived funniness of a cartoon compared to sad mood. However, happy mood decreased the funniness of a cartoon compared to sad mood if participants were reminded of the prior mood induction. Most important, however, reminding participants of the prior mood induction did not affect the experiential component itself. We observed that the spontaneous expressive response was influenced only by the prior mood induction but not by the awareness of the cause of the underlying feeling. Insofar as this measure can be interpreted as an indicator of the intensity of the feeling, this finding allows the conclusion that the emotional experience continues to be affected by a preexisting mood even if individuals are aware of this biasing influence at the time of exposure to the emotional stimulus. We interpret these findings as evidence for our assumption that the experiential component and the knowledge component of emotions are guided by different processes. Therefore, they dissociate under certain circumstances. Although Bower’s (1981) network model can handle the association of both components it is not easily reconciled with this dissociation. The dissociation of both components is consistent with the feeling-as-information approach (Schwarz, 1990) insofar as the appraisal of the emotion-eliciting object can be

744


regarded as a judgmental task that involves both the current feeling and the knowledge about the feeling. Thus, taken together the present research was an endeavour to identify variables which have different impact on both components and to realise conditions under which the predicted dissociation can be tested. Drawing on Weiner’s (1986) approach, it is likely that this dissociation reflects different stages in the process that generates emotions. Weiner (1986) proposed that appraisal processes first lead to undifferentiated primitive emotions that are independent of attributional processes. We suggested that these ‘‘raw feelings’’ resemble moods that are differentiated by subsequent causal ascriptions. Hence, directing the attention to a possible irrelevant cause of one’s current feeling has no influence on the intensity of the experienced feeling because this kind of information: (a) shapes the quality rather than the intensity of emotions, and (b) comes into play at those stages of the attributional process that follow rather than precede the automatic processes that trigger primitive emotions. Note, however, that the main aim of the current studies was not to investigate the sequence of the processes that shape emotions. An alternative explanation for our findings is a process suggested by Ekman (1994), whereby the threshold for arousing an emotion is lowered by moods. This would mean that events and objects more easily elicit emotions when the individual is in a congruent mood state as opposed to a neutral or incongruent mood state. But once the threshold is overcome and an emotion is aroused, it should not be experienced more intensily in a good mood than in a bad mood. This account is able to explain our results with respect to congruency effects (i.e., the augmentation of pride and humour when preceded by positive mood). However, it cannot easily explain the incongruency effect obtained in Study 2 (i.e., the reduction in emotional intensity as a function of incongruent mood). To test the hypothesis of lowered threshold through congruent mood against our account, namely, the additivity-of-affective-feelings hypothesis, it would be necessary to measure the existence of an emotion separately from its intensity. This cannot be accomplished with the kind of rating scales usually employed for measuring emotions. If the threshold hypothesis is true, the measure indicating whether or not an emotion has been aroused should show a mood effect and the measure indicating emotional intensity should not show this effect. The effect we assume, however, should show up primarily on the intensity scale. Another conceptual difference among theories concerning mood and emotion is the assumed specificity of moods. Our conception allows for only two kinds of moods: good and bad moods (i.e., those with a positive and those with a negative valence). Other conceptions (i.e., Ekman, 1994) assume more specific moods with more specific effects, such as irritable moods which are assumed to lower the threshold for anger responses. In the reported studies we present first evidence in favour of the unspecified-moods hypothesis. We showed that the same mood can influence different emotional reactions, namely, pride and amusement. Further research is needed to corroborate these results.


745

Based on our findings it seems unlikely that people infer the intensity of their current emotion from their knowledge about their current and pre-existing feelings as was suggested by Branscombe (1985). Rather, we assume that feelings can fuse with subsequently elicited feelings and therefore possess properties that are entirely different from those of knowledge (see Murphy et al., 1995 for a similar conception). Consistent with our conceptualisation, the current findings suggest that the affective feelings present in moods and emotions are represented in the same way. Therefore, compatible feelings that are elicited in quick succession intensify each other, whereas incompatible feelings level out each other. In conclusion, emotions hinge on two interacting responses: a knowledge component (Ortony, Clore, & Collins, 1988) that provides information about possible causes and objects of emotions and an experiential component that can be characterised by the experienced hedonic valence and its intensity. In line with this distinction we proposed that cognitive operations hinge on noetic representations and affective responses hinge on experiential representations (Strack & Neumann, 1996, see also Johnson & Multhaup, 1992). Noetic representations are assumed to stem from a conceptual mode of processing, have a propositional structure, and follow the laws of semantic memory. In contrast, experiential representations are perceptual in nature and follow the laws of psychophysics. Furthermore, experiences may possess a hedonic value in that they may be pleasant or unpleasant. In this perspective, experiential representations may be elicited either by conceptual processing, like appraising a situation that is emotionally relevant, or by perceptual mechanisms, like listening to a pleasant voice. Both moods and emotions have an experiential component; emotions, however, are unique in their noetic aspect. Therefore, emotions may not only be influenced through appraising the situation but also by a noetically undifferentiated affective experience that is present at the time. Of course, the present studies only considered the influence of mood states on positive emotions. Further research is needed to demonstrate the influence on negative emotions. The emotions studied so far were intensified by positive and attenuated by negative mood. It remains to be shown if the intensity of a negative emotion indeed is heightened by negative mood and reduced by positive mood as we would predict. Manuscript received 2 January 2000 Revised manuscript received 10 February 2001

REFERENCES Baron, R.A. (1993). Reducing aggression and conflict: The incompatibl e response approach, or, why people who feel good usually won’t be bad. In Gary G. Brannigan & Matthew R. Merrens (Eds.), The undaunted psychologist: Adventures in research (pp. 203–217). Philadelphia, PA: Temple University Press.

746


Baron, R.A., & Ball, R.L. (1974). The aggression-inhibitin g influence of nonhostil e humor. Journal of Experimantal Social Psychology, 10, 23–33. Bower, G.H. (1981). Mood and memory. American Psychologist, 36, 129–148. Branscombe, N. (1985). Effects of hedoni c valence and physiological arousal on emotion: A comparison of two theoretical perspectives. Motivation and Emotion, 2, 153–169. Clore, G., Schwarz, N., & Conway, M. (1994). Affective causes and consequence s of social information processing. In R.S. Wyer & T.K. Srull (Eds.), Handboo k of social cognition (pp. 323– 417). Hillsdale, NJ: Erlbaum. Ekman, P. (1994). Moods, emotions, traits. In P. Ekman & R.J. Davidson (Eds.), The nature of emotion (pp. 56–58). New York: Oxford University Press. Frijda, N.H. (1994). Varieties of affect: Emotions and episodes, moods, and sentiments. In P. Ekman & R.J. Davidson (Eds.), The nature of emotion (pp. 59–67). New York: Oxford University Press. Gasper, K., & Clore, G.L. (1998). The persistent use of negative affect by anxious individuals to estimate risk. Journal of Personality and Social Psychology 74, 1350–1363. Gavanski, I. (1986). Differential sensitivity of humor ratings and mirth responses to cognitive and affective component s of the humor response. Journal of Personality and Social Psychology, 51, 209–214. Gold, A.E., MacLeod, K.M., Frier, B.M., & Deary, I.J. (1995). Changes in mood during acute hypoglycemi a in healthy participants. Journal of Personality and Social Psychology, 68, 498– 504. Grice, H.P. (1975). Logic and conversation . In P. Cole & J.L. Morgan (Eds.), Syntax and semantics: Speech acts (pp. 41–58). New York: Academic Press. Hatfield, E., Cacioppo, J.T., & Rapson, R.L. (1992). Primitive emotional contagion. In M.S. Clark (Ed.), Emotion and social behavior. Review of personality and social psychology (Vol. 14, pp. 151–177). Newbury Park, CA: Sage. Isen, A. (1990). The influence of positive and negative affect on cognitive organization: Some implications for development . In N.L. Stein, B. Leventhal, et-al. (Eds.), Psychological and biological approaches to emotion (pp. 75–94). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc. Johnson, M.K., & Multhaup, K.S. (1992). Emotion and MEM. In S.A. Christianson (Ed.), The handboo k of emotion and memory: Research and theory (pp. 33–66). Hillsdale, NJ: Erlbaum. Keltner, D., Ellsworth, P.C., & Edwards, K. (1993). Beyond simple pessimism: Effects of sadness and anger on social perception. Journal of Personality and Social Psychology, 64, 740–752. Marcel, A. (1983). Conscious and unconsciou s perception: An approach to the relations between phenomena l experience and perceptual processes. Cognitive Psychology, 15, 238–300. Murphy, S.T., Monahan, J.L., & Zajonc, R.B. (1995). Additivity of nonconsciou s affect: Combined effects of priming and exposure. Journal of Personality and Social Psychology, 69, 589–602. Neumann, R. (2000). The causal influences of attributions on emotions: A procedural priming approach. Psychologica l Science, 11, 179–182. Neumann, R., & Strack, F. (2000). ‘‘Mood contagion’’: The automatic transfer of mood between persons. Journal of Personality and Social Psychology, 79, 211–223. Niedenthal, P.M., Halberstadt, J.B., & Setterlund, M.B. (1998). Being happy and seeing ‘‘happy’’: Emotional state mediates visual word recognition. Cognition and Emotion, 11, 403–432. O’Connor, J. (1932). Finger dexterity test. Chicago, IL: Stoelting. Olsen, J.M. (1992). Self-perception of humor: Evidence for discounting and augmentation effects. Journal of Personality and Social Psychology, 62, 369–377. Olsen, J.M., & Roese, N.J. (1995). The perceived funniness of humorous stimuli. Personality and Social Psychology Bulletin, 21, 908–913. Ortony, A., & Clore, G.L. (1989). Emotions, moods, and conscious awareness. Cognition and Emotion, 3, 125–137.


747

Ortony, A., Clore, G.L., & Collins, A. (1988). The cognitive structure of emotions. Cambridge, UK: Cambridge University Press. Ortony, A., & Turner, T.J. (1990). What’s basic about basic emotions? Psychologica l Review, 97, 315–331. Scherer, K.R. (1984). On the nature and function of emotion: A componen t approach. In K.R. Scherer & P. Ekman (Eds.), Approache s to emotion. Hillsdale, NJ: Erlbaum. Schwarz, N. (1990). Feelings as information: Informational and motivational functions of affective states. In E.T. Higgins & R.M., Sorrentino (Eds.), Handbook of motivation and cognition: Foundations of social behavior (Vol. 2, pp. 527–561). New York: Guilford Press. Schwarz, N., Bless, H., Strack, F., Klumpp, G., Rittenauer-Schatka, H., & Simons, A. (1991). Ease of retrieval as information: Another look at the availability heuristic. Journal of Personality and Social Psychology, 61, 195–202. Schwarz, N., Strack, F., & Mai, H.P. (1991). Assimilation and contrast effects in part-whole question sequences: A conversational logic analysis. Public Opinion Quarterly, 55, 3–23. Schwarz, N., & Clore, G. (1983). Mood, misattribution, and judgments of well-being. Informative and directive functions of affective states. Journal of Personality and Social Psychology, 45, 513–523. Strack, F., & Gonzales, M.H. (1993). Wissen und FuÈhlen: noetische und experientielle Grundlagen heuristischer Urteilsbildung. In W. Hell, K. Fiedler, & G. Gigerenzer (Eds.), Kognitive TaÈuschungen (pp. 291–315). Heidelberg: Spektrum Akademische r Verlag. Strack, F., Martin, L., & Stepper, S. (1988). Inhibiting and facilitating conditions of human smile: A nonobtrusive test of the facial feedback hypothesis. Journal of Personality and Social Psychology, 54, 768–777. Strack, F., & Neumann, R. (1996). ‘‘The spirit is willing but the flesh is weak’’: Beyond mind-body interactions in human decision-making. Organizational Behavior and Decision Making, 65, 300– 304. Strack, F., Schwarz, N., Bless, H., KuÈbler, A., & WaÈnke, M. (1993). Awareness of the influence as a determinant of assimilation versus contrast. European Journal of Social Psychology, 23, 53–62. Strack, F., Schwarz, N., & WaÈnke, M. (1991). Semantic and pragmatic aspects of context effects in social and psychologica l research. Social Cognition, 9, 111–125. Weiner, B. (1986). An attributional theory of motivation and emotion. New York: Springer. White, L. (1979). Erotica and aggression: The influence of sexual arousal, positive affect, and negative affect on aggressive behavior. Journal of Personality and Social Psychology, 37, 591– 601. Zillmann, D. (1978). Attribution and misattribution of excitatory reactions. In J.H. Harvey, J. Ickes, & R.F. Kidd (Eds.), New directions in attribution research (Vol. 2, pp. 335–368). Hillsdale, NJ: Erlbaum.