«Feelings-as-Information Theory Norbert Schwarz University of Michigan January 2010 To appear in P. Van Lange, A. Kruglanski, & E. T. Higgins (eds.), ...»
Feelings as information -- 1
University of Michigan
To appear in
P. Van Lange, A. Kruglanski, & E. T. Higgins (eds.),
Handbook of theories of social psychology.
Completion of this chapter was supported by a fellowship at the Center for Advanced Study in
the Behavioral Sciences. Address correspondence to Norbert Schwarz, ISR, University of
Michigan, 426 Thompson St, Ann Arbor, MI 48106-1248; firstname.lastname@example.org Feelings as information -- 2 Abstract. Feelings-as-information theory conceptualizes the role of subjective experiences – including moods, emotions, metacognitive experiences, and bodily sensations – in judgment. It assumes that people attend to their feelings as a source of information, with different feelings providing different types of information. Whereas feelings elicited by the target of judgment provide valid information, feelings that are due to an unrelated influence can lead us astray. The use of feelings as a source of information follows the same principles as the use of any other information. Most important, people do not rely on their feelings when they (correctly or incorrectly) attribute them to another source, thus undermining their informational value for the task at hand. What people conclude from a given feeling depends on the epistemic question on which they bring it to bear; hence, inferences from feelings are context sensitive and malleable.
In addition to serving as a basis of judgment, feelings inform us about the nature of our current situation and our thought processes are tuned to meet situational requirements. The chapter reviews the development of the theory, its core propositions and representative findings.
INTRODUCTIONHuman thinking is accompanied by a variety of subjective experiences, including moods and emotions, metacognitive feelings (like ease of recall or fluency of perception), and bodily sensations. Feelings-as-information theory provides a general framework for conceptualizing the role of these experiences in human judgment. It was initially developed to account for the influence of happy and sad moods on evaluative judgment. However, the theoretical principles of the initial mood-as-information work (Schwarz & Clore, 1983) could be fruitfully applied to other types of feelings and developed into a more comprehensive conceptualization of the interplay of feeling and thinking. This chapter summarizes what hasbeen learned.
The first section provides a short personal account of the theory’s development and places its assumptions in their historical context (for more detailed discussions of other theoretical approaches see Clore, Schwarz, & Conway, 1994; Schwarz & Clore, 2007). The second section presents the theory’s postulates and the third section reviews representative findings.
Feelings as information -- 3
A LOOK BACK:
THE DEVELOPMENT OF FEELINGS-AS-INFORMATION THEORYAs many readers know from personal experience, our lives look better on some days than on others, even though nothing of any obvious importance has changed. In my case, the upbeat or gloomy mood induced by a sunny or rainy day is sufficient to do the trick. Trying to understand this experience as a graduate student at the University of Mannheim, Germany, in the late 1970’s, I turned to what social and cognitive psychologists had learned from the experimental mood research available at the time. One account, advanced by Isen, Shalker, Clark, and Karp (1978) and Bower (1981), held that moods increase the accessibility of moodcongruent information in memory. From this perspective, positive (negative) aspects of life are more likely to come to mind when we are in a happy (sad) mood, resulting in mood-congruent judgments. This approach was consistent with social psychology’s new adoption of the information processing paradigm and its emphasis on storage and retrieval processes. However, it didn't seem "quite right" introspectively: on good days, things just "felt" better and this did not seem to involve selective recall of past events of mood-congruent valence. Phenomenological analyses in the introspective tradition of German "armchair psychology" (e.g., Bollnow, 1956), which treated moods as an integrative reflection of one’s current situation, seemed closer to the mark – alas, such introspections are to be taken with a grain of salt (Nisbett & Wilson, 1977). A competing perspective, Zajonc’s (1980) "affective primacy" hypothesis, had the advantage of avoiding reliance on mood-congruent retrieval processes but lacked a process model specific enough to meet the developing criteria of social cognition research.
A conversation with Bob Wyer offered a different approach. Wyer and Carlston (1979) proposed that affect can serve informational functions, "for example, one's liking for a person may be based partly on the feelings of pleasantness when the person is around" (p. 192). In addition, they conjectured that affective states may direct our attention to information that is suitable to explain one’s feelings. While their conjectures were compatible with phenomenological approaches, their conceptualization emphasized the role of cognitive representations of experience at the expense of actual current experience itself, consistent with the information processing paradigm. Research into the influence of arousal (from Schachter and Singer’s, 1962, emotion research to Zillman's, 1978, arousal-transfer model and Zanna and Cooper's, 1976, dissonance studies) suggested, however, that the online experience itself may Feelings as information -- 4 play a crucial role. More important, this literature also suggested that misattribution manipulations would be suitable experimental tools to address the role of current experience in human judgment.
A post-doctoral year with Wyer and Clore at the University of Illinois provided the opportunity to pursue these issues. Clore and Byrne (1974) had proposed a reinforcement-affect model to account for affective influences on interpersonal attraction. Going beyond the learning theories of the time, their model assumed that rewards exert their influence through the positive affect they elicit. Supporting this notion, laboratory and field experiments showed that associating others with positive feelings is sufficient to increase interpersonal attraction, even when the feelings are incidental and due to an unrelated source (e.g., Griffitt &Veitch, 1971). By the late 1970’s Clore began to wonder why we “don’t all end up falling in love with the paymaster,” as he put it, if the mere co-occurrence of people with reward is sufficient to induce attraction. Does incidental affect only influence our judgments when we are not aware of its source, as Zillman’s (1978) arousal studies suggested?
Initial Evidence These converging interests resulted in a conceptually straightforward study (Schwarz & Clore, 1983, Experiment 1). We asked participants to vividly recall and describe a happy or sad event to induce a corresponding mood and crossed these mood inductions with a misattribution manipulation that took advantage of an somewhat bizarre little room, previously used for auditory research with monkeys (for the inside story see Schwarz & Clore, 2003). This allowed us to suggest to some participants that the room may induce elated feelings and to others that it may induce depressed feelings. Judgments of life-satisfaction served as the dependent variable.
Our procedure deliberately stacked the deck in favor of content-driven models: By inducing moods through the recall of a happy or sad event, mood-congruent recall would be facilitated both by the content of the recall task and the induced mood. The predictions were straightforward. If mood effects on judgment were a function of mood-congruent recall (Bower, 1981), participants should report higher life-satisfaction when in a happy rather than sad mood, independent of what we told them about the room. If the experience itself served informative functions, on the other hand, its impact should depend on the feeling’s perceived diagnosticity.
That is, mood effects on reported life-satisfaction should be attenuated when the mood is attributed to the influence of the room and hence considered uninformative for evaluating one's Feelings as information -- 5 life in general (a discounting effect in Kelley's, 1972, terms); however, it should be enhanced when one experiences the mood despite an allegedly opposing influence of the room (an augmentation effect).
The results, shown in Table 1, were consistent with these predictions, although only under sad mood conditions. The latter observation was compatible with Wyer and Carlston’s (1979) suggestion that sad moods may require more explanation than happy ones, which would render them more susceptible to attributional manipulations. More important, the obtained augmentation and discounting effects made it unlikely that the influence of moods was driven by mood-congruent recall. After all, we had induced moods by having participants recall happy or sad events, thus adding semantic priming to the assumed affective activation of valenced material (Bower, 1981). Nevertheless, the accessible semantic content had little impact when participants discounted the accompanying negative feelings, assigning a crucial role to the subjective experience itself. Finally, the obtained attributional effects highlighted that the path from feelings to judgment was inferential, in contrast to Zajonc’s (1980) assertion that “preferences need no inferences.” A subsequent, more naturalistic study took advantage of sunny and rainy weather as a mood manipulation (Schwarz & Clore, 1983, Experiment 2). As daily experience suggests, participants reported higher life-satisfaction (and a more positive mood) when they were called on sunny than on rainy days. However, the negative influence of bad weather was eliminated when the interviewer, who pretended to call from out of town, first inquired about the weather at respondents’ place of residence. This discounting effect was not obtained under sunny weather conditions, again suggesting that sad moods are more likely to be explained than happy moods.
In combination, these studies provided first evidence for several assumptions that became core themes in the development of feelings-as-information theory.
Core Themes First, our findings showed that people attend to their momentary feelings as a source of information in forming judgments, essentially asking themselves, “How do I feel about this?” Later research extended this “informative function” (Wyer & Carlston, 1979) of affective states to other feelings, including non-affective feelings, like the metacognitive experience of ease of recall (Schwarz et al., 1991), and bodily sensations (Stepper & Strack, 1993).
Second, the observed discounting and augmentation effects highlighted that people use Feelings as information -- 6 their feelings like any other source of information. They do not rely on them when they become aware that their feelings may be due to an unrelated source, thus undermining their informational value for the judgment at hand. Conversely, they consider their feelings particularly informative when they experience them despite opposing forces. Later research, much of it conducted by Michel Pham and his colleagues, identified additional variables that influence how much weight we give to our feelings (see Pham, 2004).
Third, our initial studies documented more positive judgments under happy than sad moods. While this is true for the bulk of mood research (Schwarz & Clore, 2007), Leonard Martin and colleagues (e.g., Martin, Abend, Sedikides, & Green, 1997) demonstrated that positive feelings can result in negative evaluations. For example, when we feel happy while reading a sad story, we may conclude that it is not a “good sad story” after all, or else it would make us feel sad. Such findings illustrate that the influence of feelings depends on the specific question on which the feeling is brought to bear. This theme proved particularly important in later research on metacognitive experiences (Schwarz, 2004).
Finally, the observation that misattribution effects only emerged under sad moods (Table
1) proved more puzzling. Because most people feel mildly positive most of the time (Matlin & Stang, 1979), we initially suggested that sad moods are deviations from one’s usual state and hence more likely to require explanation. This, in turn, would direct attention to possible sources of one’s mood (Wyer and Carlston’s, 1979), rendering sad moods more susceptible to (mis)attribution manipulations. If so, being in an unexplained sad mood should interfere with other cognitive tasks, due to the competing demands of explaining one's mood. Testing this prediction, Bless, Bohner, Schwarz, and Strack (1990) exposed participants in happy or sad moods to strong or weak persuasive arguments and assumed that sad moods would reduce systematic message elaboration.
To our surprise, we found the opposite: sad participants engaged in message elaboration, whereas happy participants did not, by now a familiar and frequently replicated finding (for a review see Schwarz, Bless, & Bohner, 1991). Similarly, Sinclair (1988) reported strong evidence that being in a sad mood reduced halo effects in impression formation. Clearly, sad moods did not pose an explanation problem that interfered with other processing demands; to the contrary, sad moods increased, and happy moods decreased, systematic processing in these studies.