情绪心理学研究资料The Mobilization-Minimization Hypothesis

Psychological Bulletin 1991,Vol.110, No.l,67-85 Copyright 1991 by the American Psychological Association. Inc. 0033-2909/91/i3.00 Asymmetrical Effects of Positive and Negative Events: The Mobilization-Minimization Hypothesis Shelley E. Taylor University of California, Los Angeles Negative (adverse or threatening) events evoke strong and rapid physiological, cognitive, emotional, and social responses. This mobilization of the organism is followed by physiological, cognitive, and behavioral responses that damp down, minimize, and even erase the impact of that event. This pattern of mobilization-minimization appears to be greater for negative events than for neutral or positive events. Theoretical accounts of this response pattern are reviewed. It is concluded that no single theoretical mechanism can explain the mobilization-minimization pattern, but that a fam- ily of integrated process models, encompassing different classes of responses, may account for this pattern of parallel but disparatcly caused effects. In recent years, research on mood (eg, Isen, Daubman, & Gorgoglione, 1987), emotions (e.g., Frijda, 1988), and self-regu- lation (e.g. Carver & Scheier, 1990) has focused on the different origins and functions of positive and negative affect. Increas- ingly, researchers have argued that positive and negative affect cannot be considered endpoints of a single continuum, but rather must be thought of as qualitatively distinct phenomena (eg, Berscheid, 1983; Diener&Emmons, 1985; Isen, 1984; W&- son, Clark, & Tellegen, 1988). To date, however, there has been relatively little systematic investigation of the manifold ways in which positive and negative affect differ. Rather, suggestions that they are distinct phenomena have arisen from the lack of parallel effects in the literature. This article offers a framework for explicating a number of the asymmetrical effects of positive and negative events that have been observed. It begins with the observation that positive and negative events evoke different patterns of physiological, affective, cognitive, and behavioral activity at different points in their occurrence. Specifically, diverse literatures in psychology provide evi- dence that, other things being equal, negative events appear to elicit more physiological, affective, cognitive, and behavioral activity and prompt more cognitive analysis than neutral or positive events. Negative events tax individual resources, a re- sponse that appears to be mirrored at every level of responding. There is also evidence that, at every level, once the threat of the negative event has subsided, counteracting processes are initi- ated that reverse, minimize, or undo the responses elicited at the initial stage of responding. In essence, the organism re- Preparation of this article was supported by National Institute of Mental Health Grants MH 42258 and MH42152. Shelley E. Taylor was supported by a Research Scientist Development Award from the Na- tional Institute of Mental Health (MH 00311). 1 am grateful to Lisa Aspinwall, Jonathon Brown, Bram Buunk, Susan Fiske, Brett Pelham, Norbert Schwarz, William Swann, and three anonymous reviewers for their comments on earlier versions of this article. Correspondence concerning this article should be addressed to Shel- ley E. Taylor, Department of Psychology, University of California, Los Angeles, California 90024. spends to negative events with short-term mobilization and long-term minimization. This pattern seems to distinguish neg- ative events from positive or neutral ones. This article presents the evidence for the phenomenon and considers various expla- nations for its existence. First, a definition of the term negative event is required. A negative event is one that has the potential or actual ability to create adverse outcomes for the individual. Thus, the definition includes events that have not occurred but are perceived as po- tentially threatening, as well as those that have occurred and are perceived as harmful (cf. Lazarus & Folkman, 1984).' The first section of this article reviews evidence across dispa- rate classes of responses in support of the mobilization phase. It begins with more micro-level processes, such as physiological responses, through affective, cognitive, and judgmental re- sponses, to more macro-level processes, including social reac- tions to valenced events. The second major part of this article reviews evidence for the subsequent minimization phase across each of these classes of responses. In this sense, the sections on mobilization and minimization parallel each other. In these two sections, I show that physiological, affective, cognitive, and behavioral reactions to negative events show similar, though disparately caused, patterns of responding. The third major section of this article considers theoretical mechanisms that may account for the mobilization-minimiza- tion phenomenon. Given the presence of parallel changes 1 Not all of the studies reviewed here involved events that were actu- ally or potentially personally threatening to the individual. Much of the research that has examined the impact of negative information and events has involved the provision of information about people or events that are hypothetical. The assumption of this research, one that I also adopt in this review, is that the processes invoked in simulated impres- sion formation or judgment conditions mirror what goes on in actual impression formation and judgment conditions. This is a quite conser- vative assumption, inasmuch as any differential effects of negative ver- sus positive information in these hypothetical settings would probably be weaker than actually occurs in real situations, thereby underesti- mating rather than overestimating any differential impact that nega- tive information may normally have. 67 68 SHELLEY E. TAYLOR across such different classes of response categories, it is unlikely that a single theoretical mechanism explains the pattern. Con- sequently, this third section reviews theoretical mechanisms that may account for changes within particular classes of re- sponses, focusing on the strengths and limitations of their scope for explaining the overall phenomenon. I then consider mecha- nisms whereby response-specific process models accounting for parts of the mobilization-minimization pattern may be re- lated and integrated with each other. The article concludes with a discussion of the implications of the mobilization-minimiza- tion pattern for the future study of affective processes and va- lenced events. Negative Events and Mobilization Event Valence and Physiological Arousal Do negative events evoke a stronger physiological response than positive ones?2 Although physiologists have not directly addressed this question, the assumption of such a difference is built into frameworks that examine arousal and its correlates. Animals and humans respond to the threat or reality of negative events with patterned, intense physiological changes. This re- sponse was first described by Walter Cannon (1932) as the fight- or-flight reaction. Cannon proposed that when the organism perceives a threat, the body is rapidly aroused and mobilized by the sympathetic nervous system and the endocrine system. This response is marked by the secretion of catecholamines leading to increases in heart rate, blood pressure, blood sugar, and respiration. According to Cannon, this concerted physic- logical response puts the organism in a state of readiness to attack the threat or to flee. Virtually all of the early work on physiological stress involved negative events arousing fear or, in the case of humans, anxiety, sadness, and anger (e.g, Mahl, 1952; Wolf & Wolff, 1947), the implicit assumption being that positive events do not evoke the same intensity of response. It should be noted that the overwhelming majority of current labo- ratory-based stress work continues to make use of negative slressors, such as electric shock, cold pressor tests, and the like, thereby perpetuating the assumption that negative events and physiological arousal are more clearly linked than positive events and physiological arousal. There is some evidence that arousal itself is more likely to be interpreted negatively than positively. When people find them- selves in a state of arousal that they cannot explain (as may occur when epinephrine has been administered without an ex- planation of its side effects), people are more likely to explain the resulting arousal negatively, for example, as feelings of un- ease or nervousness, rather than positively (Marshall & Zim- bardo, 1979; Maslach, 1979). Thus, arousal per se may be experi- enced as aversive unless otherwise labeled as positive. Research on attitudes shows that evidence in opposition to one's own opinions elicits physiological arousal. Studies that have exposed human subjects to opinions that disagree with their own find greater arousal than when opinions agree or are neutral with respect to the subjects' opinions (Burdick & Burnes, 1958; Clore & Gormry, 1974; Dickson & McGinnies, 1966; Gormly, 1971,1974; Steiner, 1966).3 Evidence suggestive of a greater role for negative events in evoking physiological activity is also implied by research on stressful life events. This research measures the number of stressful life events a person has encountered over a period of time and then relates it to subsequent illness. Although research initially suggested that both positive and negative events were capable of producing physical disorders because of their capac- ity to force the individual to make changes and accommoda- tions, research now indicates that negative events are substan- tially more potent in this regard than are positive ones (e.g, Suls & Mullen, 1981). That is, when the amount of change is con- trolled for, negative events are more strongly related to adverse health outcomes. It should be noted that these findings are not necessarily evidence for the greater physiological impact of neg- ative over positive events. It is possible that negative events exert their adverse effects on physical health through mechanisms other than direct physiological impact. Far example, in the case of health outcomes, stressful negative life events may under- mine the effective practice of healthful behaviors, leaving peo- ple more vulnerable to illness. The evidence is, then, merely suggestive. Moreover, positive events can produce physiological arousal just as negative events can (Levi, 1965; Patkai, 1971; see Frankenhaeuser, 1975, for a review). In the physiological literature, then, there is an implicit as- sumption of and some evidence for the belief that negative events elicit greater physiological arousal than comparable posi- tive events. However, the kind of research evidence that would support the point most clearly is not generally available. A clearer test would involve calibrating positive and negative events for their affective equivalency and then assessing their impact on physiological functioning. At present, then, the evi- dence is suggestive, not conclusive. 2 There is an issue of calibration involved in comparing negative and positive events: How does one know that the negative stimuli (events, trait words, and the like) are as negative as the positive stimuli are positive? The strongest case can be made in studies in which the posi- tive and negative stimuli involved occur on the same interval scale (e.g., the costs or gains in dollars of a wager): Any inequivalency of the positive and negative stimuli is psychological, and therefore is part of the phenomenon, not a confound. A less strong but defensible case can tie made when the positive and negative stimuli are rendered equiva- lent on some scale related to the inference to be drawn. Positive and negative trait adjectives may be matched on evaluative extremity, for example, or positive and negative life events may be matched as to the change or the disruption they produce. A weaker form of inference exists in studies that sample a range of positive and negative stimuli on the assumption that meaningful differences in intensity will random- ize out. One can assume that in the absence of calibrating positive and negative stimuli, their impacts on relevant responses would be as likely to favor positive as negative events and that any systematic finding that negative events are more potent than positive ones would constitute an informative difference. 3 In shadowing experiments involving a dichotic listening task, sub- jects were aroused by words presented in the unattended channel that had previously been paired with shock without being aware that they had heard them. The control words were affectively neutral, however, raising the possibility that words paired with reinforcement might evoke the same response (Corteen & Dunn, 1974; Corteen & Wood, 1972; von Wright, Anderson, & Stenman, 1975). THE MOBILIZATION-MINIMIZATION HYPOTHESIS 69 Event Valence and Affect As in the research on physiological responses, studies of emotion have not directly investigated the hypothesis that nega- tive events evoke stronger emotional reactions than do positive events. However, several lines of work are consistent with such an argument. Negative events appear to be more potent determinants of mood than positive events. In a series of six investigations ex- ploring the determinants of mood across situations as varied as driving, test-taking, somatic symptoms, and pregnancy, Pers- son and his colleagues (Appel, Blomkvist, Persson, & Sjoberg, 1980; Persson, 1988a, 1988b, 1988c; Persson & Sjoberg, 1985, 1987; Sjoberg, Persson, & Svensson, 1982) found that expecta- tions of future negative events were the strongest determinant of mood. Moreover, the negative mood evoked by such expecta- tions dominated and suppressed the influence of positive ex- pectations on mood.4 Research on stressful life events also suggests a greater role for negative over positive events in evoking emotional reactions. When the change and disruption of positive and negative life events is equated, negative events are associated with more dis- tress, and they predict depression better than do positive events (Myers, Lindenthal, Pepper, & Ostrander, 1972; Paykel, 1974; Vinokur & Selzer, 1975). Positive stressful events (such as having a baby) tend to evoke a mix of responses, including positive emotions in response to the valence of the event but distress in response to the changes that positive life events can produce. Handler's (1975, 1984) theory of emotion accords negative events a central, though implicit, role. He argued that emotion occurs whenever an organism's goals are interrupted. The emo- tion that results is likely to be labeled negatively, because in- terruption can produce feelings of helplessness and loss of con- trol. Mandler argued that positive emotions are rarely experi- enced as intensely as negative emotions because they occur when people feel in control. In negative emotions, the degree of arousal is higher. Davitz (1969) concurred that the degree of activation involved seems to be less for positive emotions than for negative emotions. Schwarz (1990) suggested that negative emotions signal that action needs to be taken, whereas positive emotions do not, a point that may account for the apparent greater activation associated with negative emotions (see Frijda, 1988; Kanouse & Hanson, 1972). Event faience and Attention Negative affective states lead people to narrow and focus their attention (e.g., Broadbent, 1971; Easterbrook, 1959; Ey- senck, 1976), particularly to features that elicited the negative state (Schwarz, 1990; Wegner & Vallacher, 1986), and they ap- pear to do so to a greater degree than positive events and infor- mation (see Peeters & Czapinski, 1990, for a review). For exam- ple, in a study of person perception, Fiske (1980) presented subjects with sentences describing a person about whom they were told to form an impression. Subjects attended dispropor- tionately to negative information by looking at it longer than was true for positive or neutral information. This effect was independent of the unexpectedness of negative information, although unexpected, as opposed to expected, information also engaged attention more. In a similar vein, Hansen and Hansen (1988) showed an asymmetry in the processing of facial infor- mation. They found that threatening faces "pop out" of crowds, in comparison to faces with more positive expressions. Analyses of what people think about spontaneously also show a negativity bias. Klinger, Barda, and Maxeiner (1980) asked college student subjects to list up to seven things they thought about a lot and up to seven things they thought about very little. The items these students reported thinking about most were a threatened relationship, the challenge of some forthcoming event, and unexpected difficulties in pursuit of a goal. Thus, negative events, particularly unresolved ones, ap- pear to be focal in consciousness, at least among college stu- dents. Weighting ofValenced Information in Judgments A wide variety of research has suggested that negative aspects of an object, event, or choice are weighted more heavily than positive aspects in judgments (Kahneman & Tversky, 1984; see Czapinski & Peeters, 1990; Peeters & Czapinski, 1990; Skowronski & Carlston, 1989, for reviews). In tasks that involve forming impressions of others from trait adjectives, sentence descriptions, or moral and immoral behavior descriptions, neg- ative information tends to be given more weight than positive information (eg, Anderson, 1965,1968,1974; Birnbaum, 1972, 1973, 1974; Dreben, Fiske, & Hastie, 1979; Feldman, 1966; Fiske, 1980; Hamilton & Huffman, 1971; Hodges, 1974; Ka- rouse & Hanson, 1972; Lampel & Anderson, 1968; Oden & Anderson, 1971; Reeder & Coovert, 1986; van der Plight & Eiser, 1980; Warr, 1974; Wyer, 1974; Wyer& Watson, 1969; see Fiske & Taylor, 1984,1991; Kanouse & Hanson, 1972; for re- views). Negative information is also weighted more heavily in the attribution of evaluations to others (Abelson & Kanouse, 1966). The disproportionate effects of negative information oc- cur when the positive and negative stimuli are equally polarized on a good-bad evaluation scale (e.g, Anderson, 1966; Feldman, 1966). Most of the impression formation studies involve trait or sen- tence descriptions of hypothetical others. Similar effects, how- ever, have been observed in more naturalistic situations involv- ing more meaningful impressions. For example, in a study of husbands' and wives' perceptions of each other, Weiss, Hops, and Patterson (1973) found that unpleasant events accounted for more variance in ratings than did pleasurable events. Because negative events tend to be unexpected, unexpected- ness provides an alternative account for the impact of negative information on impressions. Studies that have empirically dis- entangled frequency from negativity, however, have found large and independent effects of negativity (Abelson & Kanouse, 1966; Feldman, 1966; Fiske, 1980). 4 It should be noted that longitudinal studies have generally not found a relationship between negative events and later mood (Stone & Neale, 1984; see also Eckenrode, 1984). Stone and Neale suggested that it may be because people actively attempt to manage and undo the stress associated with negative events, an explanation that is compati- ble with the idea of long-term minimization of the impact of negative events. 70 SHELLEY E. TAYLOR A few studies have found a reversal of the usual effect of weighting negative information more heavily than positive in- formation (Skowronski & Carlston, 1987; see Skowronski & Carlston, 1989, for a review). These studies have found that when subjects are making judgments about another's ability they tend to weight positive information more he