Thoughts on the Relations Between Emotion and Cognition RICHARD S. LAZARUS University of California, Berkeley ABSTRACT: This paper argues that thought is a nec- the person and the environment. The appraisal essary condition of emotion. It therefore opposes the process gives rise to a particular emotion with •stance taken by Zajonc, which reflects two widespread greater or lesser intensity depending on how the misunderstandings about what is meant by cognitive relationship is evaluated with respect to the per- processes in emotion: (a) that a cognitive appraisal of son's well-being. Cognitive appraisal means that the significance of an encounter for one's well-being the way one interprets one's plight at any given must occur in fixed stages through the information moment is crucial to the emotional response. processing of initially meaningless inputs from the en- vironment; and (b) that such an appraisal is necessarily Cognition and emotion are usually fused in na- deliberate, rational, and conscious. Some of the phy- ture (Folkman, Schaefer, & Lazarus, 1979), al- logenetic and ontogenetic implications of a cognitive though they can be dissociated in certain unusual theory of emotion are also discussed briefly. or abnormal states. For example, cognitive coping processes (cf. Lazarus, 1981) such as isolation and Recent years have seen a major change in the way intellectualization (or detachment), which are psychologists view emotion—the rediscovery that aimed at regulating feelings, can create a disso- emotions are products of cognitive processes. The ciation between thoughts and feelings. Moreover, emotional response is elicited by an evaluative per- attack can occur without anger, and avoidance ception in lower animals, and in humans by a com- without fear. These latter conditions are also in- plex 'cognitive appraisal of the significance of stances in which the usual link between thought events for one's well-being. and feeling has been loosened or broken. Yet such Although there are many other issues concerning separations are less often a rule of living and more the relations between emotion and cognition, my often a product of coping under special circum- comments will focus on the role of thought in the stances. The full experience of emotion (as opposed emotional response. I will refer often to Zajonc's to sham rage, for example) normally includes three (1980) challenge to the assumption that cognition fused components: thoughts, action impulses, and occurs prior to emotion. I use his views to illustrate somatic disturbances. When these components are widespread misunderstandings of what it means dissociated we are left with something other than to speak of cognition as a causal antecedent of what we mean by a true emotional state. Our the- emotion; I also use his views as a point of departure ories of emotion must reflect the normal fusion, for rny argument that cognitive activity is a nec- and separating thoughts, action impulses, and so^ essary as well as sufficient condition of emotion. matic disturbances except under certain specifi- able conditions (as was done in the old days of faculty psychology—which treated cognition, Do Emotions Require Cognitive emotion, and motivation as independent entities) Mediation? distorts rather than clarifies the structure of the mind (cf. Lazarus, Coyne, & Folkman, 1982). My own position on this question is a variant of One bit of fallout from the above analysis is the a family of theories of emotion centered on the implication, often derived from statements of cog- concept of cognitive appraisal. Campos and Stern- nitive theory, that cognitive appraisal is a necessary berg (1981) state, for example, that "The recent history of the study of emotion has been dominated I wish to thank my research colleague, Susan Folkman, and my by approaches stressing cognitive factors. In the- secretary, Carol Carr, for providing substantial editorial advice ories of adult emotional response, cognitive ap- on this article. I appreciate their skill and judgment. Requests for reprints should be sent to Richard S. Lazarus, praisal now functions as the central construct" (p. Department of Psychology, University of California, Berkeley, 273). Its role is, to mediate the relationship between 4105 Tolman Hall, Berkeley, California 94720. Vol. 37, No. 9, 1019-1024 AMERICAN PSYCHOLOGIST • SEPTEMBER 1982 • 1019 Copyright 1982 by the American Psychological Association, Inc. 0003-066X/82/3709-1019$00.75 I as well as sufficient condition of emotion. Such a of a computer, proceeds by serially receiving, reg- position has been criticized trenchantly by Zajonc istering, encoding, storing for the short- or long- (1980). He writes that affect is erroneously re- run, and "retrieving meaningless bits—a transfor- garded in contemporary psychological theory as mation to meaning that is called "information pro- postcognitive, occurring only after extensive cog- cessing." Meanings and their associated emotions, nitive operations have taken place, and that in ac- or hot cognitions as Abelson (1963) referred to tuality affective judgments are fairly independent them, are built through such processing. As Erdelyi of, and even precede, the perceptual and cognitive (1974) and others (e.g., Neisser, 1967) have sug- activities on which they are said to depend. Zajonc gested, however, emotion can influence the process argues that not only can affect occur without ex- at any of its stages. With this in mind, it is not tensive perceptual and cognitive encoding—and surprising that Zajonc might be troubled by the even before—but that affect and cognition are con- implication that emotion lies at the end of a tor- trolled by separate and partially independent neu- tuous cognitive chain of information processing, ral systems (see also Tomkins, 1981). Zajonc thus and therefore find it necessary to suggest an in- seems to be saying two things contrary to what I dependent system making possible rapid, nonre- have argued: first, that the proposed directionality flective emotional reactions. in which cognition determines affect is wrorig and As many have argued (Folkman et al., 1979; that the actual direction is affect to cognition; and Wrubel, Benner, & Lazarus, 1981), humans are second, that cognition and affect should be re- meaning-oriented, meaning-creating creatures who garded as relatively independent subsystems rather constantly evaluate events from the perspective of than as fused and highly interdependent. their well-being and react emotionally to some of Building his argument, Zajonc cites a stanza of these evaluations. Zajonc is therefore correct in poetry from e. e. cummings (1973): asserting that meanings are immediately inherent since feeling is first in emotionally laden transactions without lengthy who pays any attention or sequential processing, but for the wrong reasons. to the syntax of things In my view, the concept of meaning defined by will never wholly kiss you. (p. 160) the traditional information processing approach He also cites Wundt's (1907) concept of affective subscribed to by Zajonc has a perfectly reason- primacy, and Bartlett (1932), Ittelson (1973), Os- able-—and better—alternative. good (1962), and Premack (1976) as having adopted We do not always have to await revelation from the view that feelings come first. He states, for information processing to unravel the environ- example: . < mental code. As was argued in the New Look In fact, it is entirely possible that the very first stage of movement in perception, personal factors such as the organism's reaction to stimuli and the very first ele- beliefs, expectations, and motives or commitments ments in retrieval are affective. It is further possible that influence attention and appraisal at the very outset we can like something or be afraid of it before we know of any encounter. Concern with individual differ- precisely what it is and perhaps even without knowing ences leads inevitably to concern with personal what it is. (p. 154) meanings and to the factors that shape such mean- The most serious mistake in Zajonc's analysis lies ings. We actively select and shape experience and in his approach to cognition, which is characteristic in some degree mold it to our own requirements of much of present-day cognitive psychology. In (see also Rychlak, 1981). Information processing this approach information and meaning stem from as an exclusive model of cognition is insufficiently the conception of mind as an analogue to a com- concerned with the person as a source of meaning. puter (Shannon & Weaver, 1962), a view illus- The history of debate about the phenomenon of trated also by the work of Newell and Simon (1961) subception is instructive (see Eriksen, 1956, 1960, and Weiner (I960). This conception has been re- 1962; Lazarus, 1956; Lazarus & McCleary, 1951). butted by Dreyfus (1972), Polanyi (1958, 1966), In a controversial experiment, McCleary and I and others, although the rebuttal has not affected showed that by associating a set of nonsense syl- the mainstream of cognitive psychology. The lables to the threat of a painful electric shock, sub- mainstream stance is that meanings for decision jects would later react with a galvanic skin response and action are built up from essentially meaning- selectively to the shock-associated syllables, even less stimulus display elements or bits and that sys- when they had misperceived and misreported, tematic scanning of this display generates infor- them. We referred to this phenomenon as "auto- mation. Thus, human cognition, like the operations nomic discrimination without awareness," or "sub- 1020 • SEPTEMBER 1982 • AMERICAN PSYCHOLOGIST ception," arguing that subjects somehow sensed the basis of these affective judgments. Affective threat without consciously recognizing the sylla- reactions can occur without extensive perceptual bles. and cognitive encoding" (p. 151, emphasis added), The debate sparked by this interpretation and he refers to "affective judgments" (p. 157), touched on many complex issues, but it mainly implying that cognitive judgment is indeed in- centered on a claim by Bricker and Chapanis volved in emotion.