68 Mood, Cognition, and Memory [Image not available in this electronic edition.] Figure 3.1 Attribution ratings made by subjects in a positive or negative mood for their performance in an earlier exam as a function of exam score (high vs low) and attribution type (internal vs stable) Source: Forgas, Bower, and Moylan, 1990 contrast, less salient, “peripheral” self-conceptions should require more time-consuming substantive processing and accordingly be influenced by an affect-priming effect The results supported these predictions, making Sedikides’s (1995) research the first to demonstrate differential mood-congruent effects for central versus peripheral conceptions of the self, a distinction that holds considerable promise for future research in the area of social cognition Affect also appears to have a greater congruent influence on self-related judgments made by subjects with low rather than high levels of self-esteem, presumably because the former have a less stable self-concept (Brown & Mankowski, 1993) In a similar vein, Smith and Petty (1995) observed stronger mood congruence in the self-related memories reported by low rather than high self-esteem individuals As predicted by the AIM, these findings suggest that low selfesteem people need to engage in more open and elaborate processing when thinking about themselves, increasing the tendency for their current mood to influence the outcome Affect intensity may be another moderator of mood congruence: One recent study showed that mood congruence is greater among people who score high on measures assessing openness to feelings as a personality trait (Ciarrochi & Forgas, 2000) However, other studies suggest that mood congruence in self-judgments can be spontaneously reversed as a result of motivated-processing strategies Sedikides (1994) observed that after mood induction, people initially generated self-statements in a mood-congruent manner However, with the passage of time, negative self-judgments spontaneously reversed, suggesting the operation of an “automatic” process of mood management Recent research by Forgas and Ciarrochi (in press) replicated these results and indicated further that the spontaneous reversal of negative self-judgments is particularly pronounced in people with high self-esteem In summary, moods have been shown to exert a strong congruent influence on self-related thoughts and judgments, but only when some degree of open and constructive processing is required and when there are no motivational forces to override mood congruence Research to date also indicates that the infusion of affect into self-judgments is especially likely when these judgments (a) relate to peripheral, as opposed to central, aspects of the self; (b) require extensive, timeconsuming processing; and (c) reflect the self-conceptions of individuals with low rather than high self-esteem Mood Congruence in Person Perception The AIM predicts that affect infusion and mood congruence should be greater when more extensive, constructive processing is required to deal with a task Paradoxically, the more people need to think in order to compute a response, the greater the likelihood that affectively primed ideas will influence the outcome Several experiments manipulated the complexity of the subjects’ task in order to create more or less demand for elaborate processing In one series of studies (Forgas, 1992), happy and sad subjects were asked to read and form impressions about fictional characters who were described as being rather typical or ordinary or as having an unusual or even odd combination of attributes (e.g., an avid surfer whose favorite music is Italian opera) The expectation was that when people have to form an impression about a complex, ambiguous, or atypical individual, they will need to engage in more constructive Mood Congruence 69 processing and rely more on their stored knowledge about the world in order to make sense of these stimuli Affectively primed associations should thus have a greater chance to infuse the judgmental outcome Consistent with this reasoning, the data indicated that, irrespective of current mood, subjects took longer to read about odd as opposed to ordinary characters Moreover, while the former targets were evaluated somewhat more positively by happy than by sad subjects, this difference was magnified (in a mood-congruent direction) in the impressions made of atypical targets Subsequent research, comparing ordinary versus odd couples rather than individuals, yielded similar results (e.g., Forgas, 1993) Do effects of a similar sort emerge in realistic interpersonal judgments? In several studies, the impact of mood on judgments and inferences about real-life interpersonal issues was investigated (Forgas, 1994) Partners in long-term, intimate relationships revealed clear evidence of mood congruence in their attributions for actual conflicts, especially complex and serious conflicts that demand careful thought These experiments provide direct evidence for the process dependence of affect infusion into social judgments and inferences Even judgments about highly familiar people are more prone to affect infusion when a more substantive processing strategy is used Recent research has also shown that individual characteristics, such as trait anxiety, can influence processing styles and thereby significantly moderate the influence of negative mood on intergroup judgments (Ciarrochi & Forgas, 1999) Low trait-anxious Whites in the United States reacted more negatively to a threatening Black out-group when experiencing negative affect Surprisingly, high trait-anxious individuals showed the opposite pattern: They went out of their way to control their negative tendencies when feeling bad, and produced more positive judgments Put another way, it appeared that low trait-anxious people processed information about the out-group automatically and allowed affect to influence their judgments, whereas high trait anxiety combined with aversive mood triggered a more controlled, motivated processing strategy designed to eliminate socially undesirable intergroup judgments To the extent that affect influences thinking and judgments, there should also be a corresponding influence on subsequent social behaviors Positive affect should prime positive information and produce more confident, friendly, and cooperative “approach” behaviors, whereas negative affect should prime negative memories and produce avoidant, defensive, or unfriendly attitudes and behaviors Mood Congruence in Social Behaviors Mood Congruence in Self-Disclosure In this section we discuss research that speaks to a related question: If affect can influence thinking and judgments, can it also influence actual social behaviors? Most interpersonal behaviors require some degree of substantive, generative processing as people need to evaluate and plan their behaviors in inherently complex and uncertain social situations (Heider, 1958) Self-disclosure is one of the most important communicative tasks people undertake in everyday life, influencing the development and maintenance of intimate relationships Self-disclosure is also critical to mental health and social adjustment Do temporary mood states influence people’s self-disclosure strategies? Several lines of evidence suggest Mood Congruence in Responding to Requests A recent field experiment by Forgas (1998) investigated affective influences on responses to an impromptu request Folders marked “please open and consider this” were left on several empty desks in a large university library, each folder containing an assortment of materials (pictures as well as narratives) that were positive or negative in emotional tone Students who (eventually) took a seat at these desks were surreptitiously observed to ensure that they did indeed open the folders and examine their contents carefully Soon afterwards, the students were approached by another student (in fact, a confederate) and received an unexpected polite or impolite request for several sheets of paper needed to complete an essay Their responses were noted, and a short time later they were asked to complete a brief questionnaire assessing their attitudes toward the request and the requester The results revealed a clear mood-congruent pattern in attitudes and in responses to the requester: Negative mood resulted in a more critical, negative attitude to the request and the requester, as well as less compliance, than did positive mood These effects were greater when the request was impolite rather than polite, presumably because impolite, unconventional requests are likely to require more elaborate and substantive processing on the part of the recipient This explanation was supported by evidence for enhanced longterm recall for these messages On the other hand, more routine, polite, and conventional requests were processed less substantively, were less influenced by mood, and were also remembered less accurately later on These results confirm that affect infusion can have a significant effect on determining attitudes and behavioral responses to people encountered in realistic everyday situations 70 Mood, Cognition, and Memory an affirmative answer: As positive mood primes more positive and optimistic inferences about interpersonal situations, self-disclosure intimacy may also be higher when people feel good In a series of recent studies (Forgas, 2001), subjects first watched a videotape that was intended to put them into either a happy or a sad mood Next, subjects were asked to exchange e-mails with an individual who was in a nearby room, with a view to getting to know the correspondent and forming an overall impression of him or her In reality, the correspondent was a computer that had been preprogrammed to generate messages that conveyed consistently high or low levels of self-disclosure As one might expect, the subjects’ overall impression of the purported correspondent was higher if they were in a happy than in a sad mood More interestingly, the extent to which the subjects related their own interests, aspirations, and other personal matters to the correspondent was markedly affected by their current mood Happy subjects disclosed more than did sad subjects, but only if the correspondent reciprocated with a high degree of disclosure These results suggest that mood congruence is likely to occur in many unscripted and unpredictable social encounters, where people need to rely on constructive processing to guide their interpersonal strategies Synopsis Evidence from many sources suggests that people tend to perceive themselves, and the world around them, in a manner that is congruent with their current mood Over the past 25 years, explanations of mood congruence have gradually evolved from earlier psychodynamic and conditioning approaches to more recent cognitive accounts, such as the concept of affect priming, which Bower (1981; Bower & Cohen, 1982) first formalized in his well-known network theory of emotion With accumulating empirical evidence, however, it has also become clear that although mood congruence is a robust and reliable phenomenon, it is not universal In fact, in many circumstances mood either has no effect or even has an incongruent effect on cognition How are such divergent results to be understood? The affect infusion model offers an answer As discussed earlier, the model implies, and the literature indicates, that mood congruence is unlikely to occur whenever a cognitive task can be performed via a simple, well-rehearsed direct access strategy or a highly motivated strategy In these conditions there is little need or opportunity for cognition to be influenced or infused by affect Although the odds of demonstrating mood congruence are improved when subjects engage in heuristic processing of the kind identified with the AAI model, such processing is appropriate only under special circumstances (e.g., when the subjects’ cognitive resources are limited and there are no situational or motivational pressures for more detailed analysis) According to the AIM, it is more common for mood congruence to occur when individuals engage in substantive, constructive processing to integrate the available information with preexisting and affectively primed knowledge structures Consistent with this claim, the research reviewed here shows that mood-congruent effects are magnified when people engage in constructive processing to compute judgments about peripheral rather than central conceptions of the self, atypical rather than typical characters, and complex rather than simple personal conflicts As we will see in the next section, the concept of affect infusion in general, and the idea of constructive processing in particular, may be keys to understanding not only mood congruence, but mood dependence as well MOOD DEPENDENCE Our purpose in this second half of the chapter is to pursue the problem of mood-dependent memory (MDM) from two points of view Before delineating these perspectives, we should begin by describing what MDM means and why it is a problem Conceptually, mood dependence refers to the idea that what has been learned in a certain state of affect or mood is most expressible in that state Empirically, MDM is often investigated within the context of a two-by-two design, where one factor is the mood—typically either happy or sad—in which a person encodes a collection of to-be-remembered or target events, and the other factor is the mood—again, happy versus sad—in which retention of the targets is tested If these two factors are found to interact, such that more events are remembered when encoding and retrieval moods match than when they mismatch, then mood dependence is said to occur Why is MDM gingerly introduced here as “the problem”? The answer is implied by two quotations from Gordon Bower, foremost figure in the area In an oft-cited review of the mood and memory literature, Bower (1981) remarked that mood dependence “is a genuine phenomenon whether the mood swings are created experimentally or by endogenous factors in a clinical population” (p 134) Yet just eight Mood Dependence years later, in an article written with John Mayer, Bower came to a very different conclusion, claiming that MDM is an “unreliable, chance event, possibly due to subtle experimental demand” (Bower & Mayer, 1989, p 145) What happened? How is it possible that in less than a decade, mood dependence could go from being a “genuine phenomenon” to an “unreliable, chance event”? What happened was that, although several early studies secured strong evidence of MDM, several later ones showed no sign whatsoever of the phenomenon (see Blaney, 1986; Bower, 1987; Eich, 1989; Ucros, 1989) Moreover, attempts to replicate positive results rarely succeeded, even when undertaken by the same researcher using similar materials, tasks, and mood-modification techniques (see Bower & Mayer, 1989; Singer & Salovey, 1988) This accounts not only for Bower’s change of opinion, but also for Ellis and Hunt’s (1989) claim that “mood-state dependency in memory presents more puzzles than solutions” (p 280) and for Kihlstrom’s (1989) comment that MDM “has proved to have the qualities of a will-o’-the-wisp” (p 26) Plainly, any effect as erratic as MDM appears to be must be considered a problem Despite decades of dedicated research, it remains unclear whether mood dependence is a real, reliable phenomenon of memory But is MDM a problem worth worrying about, and is it important enough to pursue? Many researchers maintain that it is, for the concept has significant implications for both cognitive and clinical psychology With respect to cognitive implications, Bower has allowed that when he began working on MDM, he was “occasionally chided by research friends for even bothering to demonstrate such an ‘obvious’ triviality as that one’s emotional state could serve as a context for learning” (Bower & Mayer, 1989, p 152) Although the criticism seems ironic today, it was incisive at the time, for many theories strongly suggested that memory should be mood dependent These theories included the early drive-as-stimulus views held by Hull (1943) and Miller (1950), as well as such later ideas as Baddeley’s (1982) distinction between independent and interactive contexts, Bower’s (1981) network model of emotion, and Tulving’s (1983) encoding specificity principle (also see the chapter by Roediger & Marsh in this volume) Thus, the frequent failure to demonstrate MDM reflects badly on many classic and contemporary theories of memory, and it blocks understanding of the basic issue of how context influences learning and remembering With respect to clinical implications, a key proposition in the prologue to Breuer and Freud’s (1895/1957) Studies on Hysteria states that “hysterics suffer mainly from 71 reminiscences” (p 7) Breuer and Freud believed, as did many of their contemporaries (most notably Janet, 1889), that the grand-mal seizures, sleepwalking episodes, and other bizarre symptoms shown by hysteric patients were the behavioral by-products of earlier traumatic experiences, experiences that were now shielded behind a dense amnesic barrier, rendering them impervious to deliberate, conscious recall In later sections of the Studies, Freud argued that the hysteric’s amnesia was the result of repression: motivated forgetting meant to protect the ego, or the act of keeping something—in this case, traumatic recollections—out of awareness (see Erdelyi & Goldberg, 1979) Breuer, however, saw the matter differently, and in terms that can be understood today as an extreme example of mood dependence Breuer maintained that traumatic events, by virtue of their intense emotionality, are experienced in an altered or “hypnoid” state of consciousness that is intrinsically different from the individual’s normal state On this view, amnesia occurs not because hysteric patients not want to remember their traumatic experiences, but rather, because they cannot remember, owing to the discontinuity between their hypnoid and normal states of consciousness Although Breuer did not deny the importance of repression, he was quick to cite ideas that concurred with his hypnoid hypothesis, including Delboeuf’s claim that “We can now explain how the hypnotist promotes cure [of hysteria] He puts the subject back into the state in which his trouble first appeared and uses words to combat that trouble, as it now makes fresh emergence” (Breuer & Freud, 1895/1957, p 7, fn 1) Since cases of full-blown hysteria are seldom seen today, it is easy to dismiss the work of Breuer, Janet, and their contemporaries as quaint and outmoded Indeed, even in its own era, the concept of hypnoid states received short shrift: Breuer himself did little to promote the idea, and Freud was busy carving repression into “the foundation-stone on which the whole structure of psychoanalysis rests” (Freud, 1914/1957, p 16) Nonetheless, vestiges of the hypnoid hypothesis can be seen in a number of contemporary clinical accounts For instance, Weingartner and his colleagues have conjectured that mood dependence is a causal factor in the memory deficits displayed by psychiatric patients who cycle between states of mania and normal mood (Weingartner, 1978; Weingartner, Miller, & Murphy, 1977) In addition to bipolar illness, MDM has been implicated in such diverse disorders as alcoholic blackout, chronic depression, psychogenic amnesia, and multiple personality disorder (see Goodwin, 1974; Nissen, Ross, Willingham, MacKenzie, & Schacter, 1988; Reus, Weingartner, & Post, 1979) 72 Mood, Cognition, and Memory Given that mood dependence is indeed a problem worth pursuing, how might some leverage on it be gained? Two approaches seem promising: one cognitive in orientation, the other, clinical The former features laboratory studies involving experimentally induced moods in normal subjects, and aims to identify factors or variables that play pivotal roles in the occurrence of MDM This approach is called cognitive because it focuses on factors—internally versus externally generated events, cued versus uncued tests of explicit retention, or real versus simulated moods—that are familiar to researchers in the areas of mainstream cognitive psychology, social cognition, or allied fields The alternative approach concentrates on clinical studies involving naturally occurring moods Here the question of interest is whether it is possible to demonstrate MDM in people who experience marked shifts in mood state as a consequence of a psychopathological condition, such as bipolar illness In the remainder of this chapter, we review recent research that has been done on both of these fronts Cognitive Perspectives on Mood Dependence Although mood dependence is widely regarded as a nowyou-see-it, now-you-don’t effect, many researchers maintain that the problem of unreliability lies not with the phenomenon itself, but rather with the experimental methods meant to detect it (see Bower, 1992; Eich, 1995a; Kenealy, 1997) On this view, it should indeed be possible to obtain robust and reliable evidence of MDM, but only if certain conditions are met and certain factors are in effect What might these conditions and factors be? Several promising candidates are considered as follows Nature of the Encoding Task Intuitively, it seems reasonable to suppose that how strongly memory is mood dependent will depend on how the to-beremembered or target events are encoded To clarify, consider two hypothetical situations suggested by Eich, Macaulay, and Ryan (1994) In Scenario 1, two individuals—one happy, one sad—are shown, say, a rose and are asked to identify and describe what they see Both individuals are apt to say much the same thing and to encode the rose event in much the same manner After all, and with all due respect to Gertrude Stein, a rose is a rose is a rose, regardless of whether it is seen through a happy or sad eye The implication, then, is that the perceivers will encode the rose event in a way that is largely unrelated to their mood If true, then when retrieval of the event is later assessed via nominally noncued or spontaneous recall, it should make little difference whether or not the subjects are in the same mood they had experienced earlier In short, memory for the rose event should not appear to be mood dependent under these circumstances Now imagine a different situation, Scenario Instead of identifying and describing the rose, the subjects are asked to recall an episode, from any time in their personal past, that the object calls to mind Instead of involving the relatively automatic or data-driven perception of an external stimulus, the task now requires the subjects to engage in internal mental processes such as reasoning, reflection, and cotemporal thought, “the sort of elaborative and associative processes that augment, bridge, or embellish ongoing perceptual experience but that are not necessarily part of the veridical representation of perceptual experience” (Johnson & Raye, 1981, p 70) Furthermore, even though the stimulus object is itself affectively neutral, the autobiographical memories it triggers are apt to be strongly influenced by the subjects’ mood Thus, for example, whereas the happy subject may recollect receiving a dozen roses from a secret admirer, the sad subject may remember the flowers that adorned his father’s coffin In effect, the rose event becomes closely associated with or deeply collared by the subject’s mood, thereby making mood a potentially potent cue for retrieving the event Thus, when later asked to spontaneously recall the gist of the episode they had recounted earlier, the subjects should be more likely to remember having related a vignette involving roses if they are in the same mood they had experienced earlier In this situation, then, memory for the rose event should appear to be mood dependent These intuitions accord well with the results of actual research Many of the earliest experiments on MDM used a simple list-learning paradigm—analogous to the situation sketched in Scenario 1—in which subjects memorized unrelated words while they were in a particular mood, typically either happiness or sadness, induced via hypnotic suggestions, guided imagery, mood-appropriate music, or some other means (see Martin, 1990) As Bower (1992) has observed, the assumption was that the words would become associated, by virtue of temporal contiguity, to the subjects’ current mood as well as to the list-context; hence, reinstatement of the same mood would be expected to enhance performance on a later test of word retention Although a few list-learning studies succeeded in demonstrating MDM, several others failed to so (see Blaney, 1986; Bower, 1987) In contrast to list-learning experiments, studies involving autobiographical memory—including those modeled after Scenario 2—have revealed robust and reliable evidence of mood dependence (see Bower, 1992; Eich, 1995a; Fiedler, 1990) An example is Experiment by Eich et al (1994) During the encoding session of this study, undergraduates ... early studies secured strong evidence of MDM, several later ones showed no sign whatsoever of the phenomenon (see Blaney, 198 6; Bower, 198 7; Eich, 198 9; Ucros, 198 9) Moreover, attempts to replicate... (194 3) and Miller (195 0), as well as such later ideas as Baddeley’s (198 2) distinction between independent and interactive contexts, Bower’s (198 1) network model of emotion, and Tulving’s (198 3)... foundation-stone on which the whole structure of psychoanalysis rests” (Freud, 191 4 /195 7, p 16) Nonetheless, vestiges of the hypnoid hypothesis can be seen in a number of contemporary clinical accounts For