Nico Frijda. The International Handbook of Psychology. Editor: Kurt Pawlik & Mark R Rosenzweig. Sage Publications. 2000.

What is an Emotion?

Emotion is one of the three main areas of psychology, in its traditional division into cognition, conation, and emotion. Emotions are often considered to form the main source of action. Yet, for a very long time emotion was not a central topic in psychology. Only since about 1960 has it re-entered the interest of psychologists. This long neglect had several causes. One was the behaviorist distaste of subjective experience. Another was the lack of a consensual definition.

Defining Emotions

That there is no generally accepted definition of emotions is in part because emotions involve so many different component phenomena. More precisely, the concept ‘emotion’ is used to denote a large variety of phenomena, both in daily interaction and in scientific discourse.

These include feelings, evaluations of and cognitions about objects and events, the establishment or disruption of relations with them, physiological arousal, facial expressions, and shifts in the control of behavior and thought that sometimes cause the individual to act contrary to reason. Emotions are multicomponential phenomena. Each of the component phenomena can form the core of a definition of emotions, and actually have done so. It thus is not uncon-troversial to define emotions by one of them, the more so because the various components do not always all occur together. It is a basic fact about emotions that the intercorrelations between components are far from unity (Lang, 1984).

Emotions have been defined as feelings (e.g., Wundt, 1902), with the other components being viewed as caused by the them. In the early decades of the twentieth century emotions were often defined as modes of autonomic physiological reaction (or as the sensations coming from those reactions). During the 1960s, cognitions took a prominent place, and emotions were defined as a sort of judgments, or as cognitive attributions of felt autonomic arousal (Schachter & Singer, 1962).

The diversity of phenomena, and the fact that they do not always appear together has led to defining emotions instead as processes or dispositions underlying the phenomena. Emotions can be seen as a mental state ‘behind’ the phenomena. That mental state has sometimes been equated with the feeling, sometimes as an internal ‘nonpropositional signal’ (Oatley, 1992), and more often functionally, as the activated disposition to deal with certain contingencies (e.g., Ekman, 1982; Tomkins, 1962).

The dispositional notions represent an important shift in approaching emotions: a shift towards understanding them in terms of underlying mechanisms, rather than of patterns of phenomena. This is important in particular because the distinctions among the phenomena may not correspond to those among underlying processes. For instance, everything that can be grouped together as emotions may not derive from similar processes (LeDoux, 1996). Also, phenomena may not so much cause one another, but underlying processes may. These processes need not be thought of primarily in neurological terms. They can be conceived of in psychological, functional terms, that is, in terms of what they achieve or are meant to achieve.

These definitional issues are relevant for this major question: do creatures without conscious, reportable feelings have emotions? If emotions are defined by feelings, the answer of course is no. If they are defined by underlying processes, the answer is yes. The emotions of human adults, infants, and animals may differ in some respects, but still be basically similar. Indeed, the eliciting conditions of many behaviors and physiological reactions, their behaviors and physiological reactions themselves, and relevant brain structures and brain chemistries, are very similar (Panksepp, 1998). It is a plausible and useful assumption that animals and infants have emotions, and that one can learn very much from studying them for understanding the emotions of human adults. This is the standpoint taken in this chapter, and of most current theorists. Emotions are viewed as dispositional structures, to be defined functionally, and analyzed not only from conscious awareness or with respect to conscious awareness. The role and function of conscious feelings in emotions is an issue in its own right.

Emotions, Moods, and Well-Being

Definitions of emotion also diverge because some include all states that have one important phenomenon in common, for instance feelings of pleasure or displeasure, or the affective evaluation of objects, while others are more restrictive. It is indeed common to restrict the term emotion to that subclass in which the phenomena are elicited by a stimulus or event, and are of relatively short duration. Moods are usually split off by their longer duration or the absence of a clear object. Sentiments (e.g. ‘I hate that person’) are dispositions; the term sentiment is used for more or less permanent emotional attitudes, while emotion is usually reserved for acute reactions to an eliciting event. Other major concepts (e.g., well-being or happiness) are best seen as integrations of previous emotions (Kahneman, 1999).

Different Emotions

Emotion Taxonomy

What emotions are there, and how do they differ? All languages, of course, distinguish different emotions, but the differences differ in size. The emotion words in the given languages thus suggest a structure of more basic processes. When subjects are asked to rate the similarity between emotion words, the resulting similarity indices indeed show a hierarchical organization. Emotion words tend to divide into positive and negative ones (with often a group of neutral ones such as surprise). These large groups each consist of smaller clusters. The words in each of those clusters all appear to be variants of one ‘kind’ of emotion. Rage, irritation, anger, and hate, that themselves are part of the lowest level in the hierarchy, all can be seen as variants of anger.

The kinds of emotion are often considered to represent discrete emotion categories (Izard, 1977). They are often referred to as basic emotions. A typical set of such categories is the one found by Shaver, Wu, and Schwartz (1992): anger, disgust, sadness, fear, joy, surprise, love, pity. The sets of basic emotions found by such methods show considerable similarity between studies, and between languages (Shaver et al., 1992). Differences between studies and languages do, however, exist, and some words (e.g., hope, and jealousy) shift greatly from one cluster to another, and do not really belong in any one of them.

The findings on hierarchical structure of word similarities, and the notion of basic emotions, have led to the hypothesis that each emotion is a variant of one such basic emotion, or at most of the blend of two or three (e.g., Plutchik, 1980). The data do not support this idea: subjects cannot unambiguously assign all emotion words to one supposed basic emotion (Reisenzein, 1995). The supposition of an exhaustive hierarchy is unnecessary, even with a basic emotions concept. Some emotion words may refer to a particular component (e.g., excitement may refer to any emotion with pronounced auto-nomic arousal), and other to an eliciting circumstance (e.g., jealousy).

Differences between Emotions

Language may not be the best way to approach the problems of discovering meaningful psychological distinctions. Not all emotion words point to distinct psychological phenomena; they may originate in cultural theory, social rules, or moral perspectives. The theoretical approach called social constructivism in fact finds emotion distinctions along these latter lines (see contributions in Harré & Parrott, 1996).

Moreover, analysis of words and their relationships forces one into a categorial view of emotions that may not be appropriate to the phenomena and processes. Emotion words do not all have sharply distinct meanings. Each refers to a ‘fuzzy class,’ characterized by a prototype around which individual instances cloud in irregular fashion (Fehr & Russell, 1984). Emotions themselves may not really fit into distinct, discrete classes. They can be viewed as mental states that vary along the continuous dimensions of pleasantness and activation (Wundt, 1902; Russell & Barrett, 1999). Categorial labels then might just refer to some ill-defined region in that two-dimensional space, perhaps further specified by prototype scripts (Russell & Barrett, 1999).

Different emotions may also, however, correspond to different variants of a particular component. One has often proposed that different emotions correspond to different feelings; the basic emotions might correspond to irreducibly different feelings, sometimes referred to as qualia (e.g., Oatley, 1992). This point of view is problematic, because the only criterion for different feelings consists in the application of different words.

Emotions may be distinguished by any other component; the major emotion categories may be defined in terms of patterns of autonomic response, for instance. More theoretically grounded is to distinguish them by supposed dispositions that, when activated, generate particular patterns of components. This is the most common view of basic emotions: as basic mechanisms or dispositions (Buck, 1999; Ekman, 1982; Izard, 1977; Plutchik, 1980; Tomkins, 1962). Authors do not fully agree as to which should be distinguished. Ekman (1982) distinguished joy, sadness, fear, anger, disgust, and surprise; he later added contempt; others include shame, guilt, and affection.

How to identify such dispositions? Ekman (1982) has taken universal facial expressions as the cue to biological dispositions; the cue has been considered a weak one (Russell, 1994). One may also take a functional perspective. One may assume a small set of dispositions to deal with different major adaptational challenges (e.g., Buck, 1999; Ekman, 1982; Plutchik, 1980; Tomkins, 1962). Fear can be viewed as the disposition to deal with environmental threat, anger to deal with social obstacles or power rivalry, surprise with unexpected events, joy with success, affection with potential mates. This approach appears particularly appropriate if the dispositions are understood as evolutionary products for dealing with these contingencies (Cosmides & Tooby, 2000; Plutchik, 1980), and as being based upon dedicated neural structures, for which there indeed exists some evidence (e.g., LeDoux, 1996; Panksepp, 1998; see below).

Investigators do not agree about which emotion categories are basic ones (Ortony & Turner, 1990), and whether the very notion of basic dispositions is meaningful (see relevant section in Ekman & Davidson, 1994). Emotions may be just bundles of response components mutually influencing each other (they are synchronized; Scherer, 2000), and jointly called up by a given event as appraised by the subject. Emotion categories may reflect frequently recurring contingencies (Scherer, 2000), or just social and linguistic habits (Mandler, 1984). Which of the approaches, the basic emotions approach, or the multicomponential bundle approach, best explains the phenomena is still unclear. Findings on brain mechanisms will probably decide.

Emotional Phenomena

The major emotional components will be briefly discussed.

Motivational Change

Certain behaviors and feelings involve ‘urges’ (Tomkins, 1962): strivings that appear to interrupt ongoing goals and voluntary behavior, and cause a shift in goal priorities. They thus manifest a change in control of behavior.

Motivational aspects, and the mode of control of behavior, have in fact been considered as defining emotions in early philosophical psychology. What we now call emotions were formerly called passions, from the Greek ψαθημα, implying passivity, and opposed to the concept of action. The same feature coined its Latin equivalent, affectus. Motivational change is one of the phenomena that defines emotions, bringing up the important ethical issue of responsibility for one’ emotions. Arnold (1960) defined emotional experience as experienced action tendency; Frijda (1986) defined emotions as states of action readiness, the latter defined as states of readiness to achieve a particular kind of subject—environment relationship.

Varieties of motivational state, as derived from behavior or from reports of experience, differentiate between emotions. Arnold (1960) defined different emotions as different action tendencies, and Frijda (1986) as different forms of action readiness (such as readiness to achieve proximity, hostile encounter, dominance, or general increase or loss of readiness to relate). Indeed, there are distinct relations between how one labels one’ emotion and felt state of action readiness (e.g., Frijda, 1986; Roseman, Wiest, & Swartz, 1994).


Feeling, subjective experience, has often been considered the central emotion component. Since the eighteenth or nineteenth century (but not before), emotions have been defined as feelings. But the nature of feelings is not immediately clear

It has been proposed that different emotions correspond to different qualia. As mentioned earlier, it has been proposed that all emotional feelings are variants of a few emotional qualia (e.g., Izard, 1977; Oatley, 1992). The hypothesis has not appeared tenable (Reisenzein, 1995).

A major core element of emotional feelings are the feelings of pleasure and pain. According to introspective studies by Wundt (1902), they are the only affective qualia, that is, experiences that cannot be reduced to body sensations and cognitions; the properly emotional in the feelings corresponds to the experiences of pleasure and pain, rather than body sensations such as those of autonomic arousal, as was proposed in theory of Schachter and Singer (1962).

Body sensations played an important role in the theory of emotional feelings of William James (1884), known as the James—Lange theory. Different feelings, presumably, correspond to different patterns of feedback from autonomic response. Such feedback returned in the theory of Schachter and Singer (1962). The distinctiveness of different feelings was thought to come from the feedback of autonomic arousal, complemented by cognitive attributions of their cause. The theory was not supported by the evidence. Other proposed distinctive body sensations are those coming from facial expression (facial feedback theory; Izard, 1977; Tomkins, 1962). Although none of these body sensations appear indispensable to characterize an experience as an emotional feeling they do contribute to the quality of emotional feelings, and may play a role in their impact (Damasio, 1994). Reported body feelings corresponds to different emotions, as labeled by the subjects.

Further differentiation of feelings derives from associated cognitions. Different emotional experiences, say, of fear or joy, do not only differ in pleasantness, activation, and body feelings. They obtain their specificity largely from several other aspects. One of the major ones is appraisal awareness, that is, awareness of what the emotion is about. This ‘aboutness’ of feelings, also called their intentional character, is a general aspect of emotional feelings (Frijda, 1986; Oatley, 1992). It constitutes the cognitive component of emotional feelings, and may well be what distinguishes emotional feelings from moods. Appraisal awareness refers to the individual’ awareness of what the emotional event may do or offer to him or her, or what he or she could or could not do to cope with it (Lazarus, 1991). Different emotions correspond to different forms of appraisal awareness, as has been abundantly shown in self-report studies (see the contributions and references in Schorr, Scherer, & Johnstone, 2000).

As just mentioned, emotional feelings also include awareness of motivational state, or state of action readiness. The various self-reported feeling aspects correspond to actual appraised emotion-eliciting aspects of events, actual states of action readiness (as inferred from behavior), and actual autonomic reactions only to a limited extent. Much in self-reports may well derive from preconceptions and post-hoc constructions (Parkinson, 1995; Rimé, Phillipot, & Cisamolo, 1990).

An important additional aspect of emotional feelings is the significance of the emotion to the individual (Frijda, 1986): his or her acceptance or rejection of the emotion, and its felt implications for self-esteem and the likely reactions of others. Emotion significance differs between emotions, between individuals, and between cultures (Mesquita, Frijda, & Scherer, 1997). It is what determines emotion control.


In this chapter, I use the term for the feelings of pleasure and pain, and for the processes underlying them. Those processes may be manifest in the perception of liked and disliked objects, rather than feelings (Zajonc, 1994). They may also operate nonconsciously, without the feelings, particularly at low process intensities (Zajonc, 1994). Subliminal exposure of pleasant or unpleasant stimuli (smiling face, spider) appear to influence later affect ratings of neutral stimuli (e.g., Murphy & Zajonc, 1993), or to facilitate conditioning (ö hman, 2000).

Pleasure and pain are opposite poles of one continuum of feeling (Russell & Barrett, 1999). The underlying mechanisms, however, are probably separate and to some extent independent (Ito & Cacioppo, 1999). The two processes can be simultaneously active, resulting in ambivalence and in the often only moderate negative correlations between separate positive and negative affect ratings. The two mechanisms also have different properties. For instance, positive affect tends to increase sociability and creativity (Isen, 1999), whereas negative affect does not do the opposite; negative affect shows negativity bias, a steeper increase of intensity with stimulus magnitude than positive affect (Ito & Cacioppo, 1999). Positive and negative mood have different effects on thinking strategies (Fiedler & Bless, 2000). Measurements of pleasure and pain (e.g., by rating scales) usually reflect some form of integration of feelings over time that do not relate in a simple manner to pleasure at specific moments (Kahneman, 1999).

Autonomic Reactions

Emotions have sometimes been defined as patterns of autonomic physiological responses (changes in heart rate, blood pressure, skin temperature, electrodermal or psychogalvanic skin responses). According to the James—Lange theory, different emotions should correspond to different patterns of autonomic response. Empirical research, by and large, has not supported that theory. All ‘excited’ emotions are, it has been argued, accompanied by an overall autonomic arousal pattern that serves energy mobilization; this has been called the emergency response, in the classic analysis by Cannon (1927). Different response patterns do in fact occur (for instance during active and passive coping), but do not correspond to different emotions like anger and fear (Cacioppo, Bernston, Larsen, Poehlman, & Ito, 2000). Ekman, Leven-son, and Friesen (1983) reported findings suggesting that certain patterns do differentiate between basic emotions, but these findings may have been spurious (Cacioppo et al., 2000), or reflect differences in muscular effort.

Emotional Behavior

A major subject for research has been expressive behavior, defined as behavior suggesting an emotion in a person, that is by and large unlearned and universal in the species concerned. The designation ‘expressive behavior’ or ‘expression’ is misleading. It suggests that the behavior serves to express feelings, which is a theoretical interpretation that may not be correct (Frijda & Tcherkassof, 1997).

The most studied is facial expression (see Russell & Fernandez-Dols, 1997). Important descriptive analysis and evolutionary interpretations were given by Darwin (1872). Precise scoring methods have been developed (MAX by Izard, 1971; FACS by Ekman and Friesen, in Feldman & Rimé, 1991). High degrees of interobserver agreement are obtained when selected posed facial expression photographs are rated in terms of a small set of emotion categories (such as Ekman,’ 1982, categories). The agreements (and correspondence with the categories the expressions were meant to convey) range between 98% (for joy) and 56% (for contempt) in Western cultures; in illiterate cultures, they were appreciably lower, but still far above chance (Ekman, 1994; also, Izard, 1971). The amounts of agreement are probably influenced by the research methods used (Russell, 1994), but still sufficient to demonstrate similarity in identifying emotions from the face that is largely culture-independent (Ekman, 1994). The evidence has been interpreted as showing that facial expressions are parts of the neural dispositions for basic emotions (Ekman, 1982, 1994). This view has been contested by Fridlund (1994). Facial expressions do not strictly correspond to particular emotions. According to Frid-lund, facial expressions have nothing much to do with emotions; they are also shown in the absence of emotion, and are sensitive to context, such as the presence of an audience. Their function is to influence other individuals, and not to express emotions. This function of expressions is, however, not incompatible with their usually resulting from emotions. Expressive behaviors may well be understood as ‘coping behaviors’ (Lazarus, 1991) for dealing with the emotional events, which includes influencing others by threats or calls for help (Frijda & Tcherkassof, 1997).

There are many other forms of expression: posture, voice intonation and cries, whole-body movements (Feldman & Rimé, 1991). Sophisticated coding systems for postures and body movements have been developed, but so far little research has been devoted to their specific relationships with emotions. Laughter and crying of course are whole-body reactions of a complex and ill-understood nature (Ruch, 1993).

Vocal expression of emotion has been studied by using standard or fake sentences, with varying intonations. They can be recognized about as well as facial expression photographs, and that in different linguistic groups (Johnstone & Scherer, 2000).

Emotions also induce more general behaviors, such as angry aggression, fearful flight, desirous approach. Some of these behaviors are innate, species-specific behaviors (modes of threat and attack, of fear responses such as freezing, running, and hiding), others are learned. Expressive behaviors rather generally are part of more encompassing behavioral patterns, and some expressive patterns also are whole-body responses (crying and laughter are notable examples).

Both innate and learned behaviors can largely be understood as coping behaviors; the surprised facial expression can be seen as an orienting response, and laughter as a play response signaling playful interaction (Darwin, 1872; Frijda & Tcherkassof, 1997; Ruch, 1993). The behaviors that co-occur or follow each other in given situations tend to have similar functions, such as self-defense, hostility, play (nonserious interaction), affinity, as Van Hooff (1982) found in analysis of chimpanzee behavior. Such functional equivalence is the cue for interpreting emotions as motivational states, action tendencies (Arnold, 1960) or states of action readiness (Frijda, 1986).

Cognitive Changes

Emotions can strongly influence cognitive processes. First, there is the arousal of attention, and the effects of the distribution of attention, as these are notable, for instance, in memory of emotional incidents (Christianson, 1992). Attention allocation also helps in producing the inverted U-curve, that is, improvement of performance with moderate emotions and disturbing it with strong ones. Positive mood states tend to improve recall of positive memories, and negative moods of negative memories (Clore & Gasper, in press). Positive emotions and moods tend to improve cognitive flexibility and originality (Isen, 1999); for instance, after giving subjects an unexpected gift, they score better on tests of cognitive flexibility. Social judgments are influenced by mood states, positive moods making them more favorable, and negative ones, more negative (Forgas & Vargas, in press). Emotions may make judgments more tenaciously held, more resistant to change by incompatible information (see contributions in Frijda, Man-stead, & Bem, in press).

Emotion Regulation

Practically all emotional reactions are being controlled to some degree, as evident from the relatively few instances in which this is not so (blind panic or anger, alcohol disinhibition, disinhibiting group influences). Emotion regulation appears to be so ubiquitous that it can be counted among the emotion mechanisms.

Emotion regulation is not only caused by social norms, as regulation in animals shows. According to Gray (1987), behavioral inhibition is elicited by aversive response consequences, or their anticipation. These aversive consequences may come from social censure, but also from undesired disruption of group harmony or from the harm that one might inflict upon others by unrestrained response. The latter controlling influences are not unique to humans but are also observed in primates (De Waal, 1996). Further sources of emotion control are the negative effects of emotions upon motor and cognitive performance. These various sources of regulation determine various other emotional manifestations, such as consideration for other people (or animals), and the whole level of social emotions that make smooth social interaction possible. Studies of patients with orbitofrontal damage illustrate how much the gamut of emotional reactions is impoverished when regulatory emotional reactions are interfered with (Damasio, 1994).

Emotion regulation proceeds along a number of different roads, such as appraisal change, response suppression, seeking distraction. The different procedures are being extensively discussed and studied in the literature (e.g., Gross, 1999).

Emotion Elicitation

General Framework

Emotion elicitation can be described in a general way as due to the appearance or termination of events that are intrinsically pleasant or unpleasant, or to the appearance or termination of events that signal or announce such pleasant or unpleasant events. This generalization comes from behaviorist psychology (Mowrer, 1960), where it was phrased in terms of the actual or signaled increase and decrease of positively and negatively reinforcing stimuli. The two times four contingencies (those for actual events, and those for signaled ones) can account for the elicitation of different emotions. The anticipation of an unpleasant event, for instance, would cover the antecedents of fear (fear being defined by escape or avoidance tendency, or fearful feeling), and the termination of a pleasant event, as well as the advent of an unpleasant one, that of any negative emotion. Further refinements were introduced by Gray (1987), by adding omission of an expected event and, more importantly, the availability or non-availability of coping behavior. Anticipation of the termination of a pleasant event would lead to fear if escape or avoidance appears possible, anger if hostile behavior belongs to the subject’ repertoire, and sadness or despair if there is nothing he or she can do about it. Additional contingencies, and thus the antecedent conditions for further emotions, are based upon the likelihood of future events, and distinctions among who or what caused them (Scherer, 2000). In addition, so-called non-hedonic or cognitive emotions (surprise, interest, curiosity) are evoked by unexpected events (Meyer, Schützwohl, & Reisenzein, 1993).

Motivational Relevance

The same events can be described at a somewhat more integrative level. According to a large number of theorists, emotions are evoked by events that are relevant to the individual’ concerns, motives, major goals, or well-being (Frijda, 1986; Lazarus, 1991; Mandler, 1984; Oatley, 1992; Stein & Trabasso, 1992). Events that appear favorable for reaching the goal or the concern’ satisfaction generate pleasant emotions, and those that appear to obstruct its achievement generate negative emotions. An event of which the relevance is unclear may evoke surprise or curiosity. Different contingencies of goal achievement or non-achievement (loss of a goal or definitive impossibility of achieving it) cause sadness, someone obstructing progress causes anger, and so forth. The approach finds its support mostly in self-report studies on goals achievement contingencies and emotions (e.g., Oatley, 1992; Stein & Trabasso, 1992). In fact, the various contingencies mentioned in the previous subsection closely match the goal-relevant ones.

Many emotions result from motivational relevance in a more specific sense: from, innately or on the basis of learning, recognizing persons and objects as fit for satisfying certain motives. Falling in love, lust, curiosity, and enjoyments can be understood in this way (Kubovy, 1999; Rozin, 1999).


It is evident that emotions are only rarely elicited by intrinsically affective stimuli alone. Most emotions result from the meaning of events, that is, by their links to future stimulus events or to conditions for which the events are relevant. Psychological processes thus intervene between the actual eliciting events and the emotional reaction.

The processes are collectively referred to as appraisal processes. They form the core of appraisal theory, or cognitive emotion theory (Arnold, 1960; Frijda, 1986; Lazarus, 1991; Oatley, 1992; Scherer, 1999; Smith & Ellsworth, 1985; see also the contributions in Schorr, Scherer, & Johnstone, 2000). Appraisal theory makes a number of different points. First, there are processes that turn a perceived event into one with hedonic value. These processes may be disturbed, for instance by damage to the brain region named the amygdala (see below), and under the influence of drugs (many antidepress-ants render the individual more indifferent). Second, different emotions correspond to different contingencies as appraised, and they correspond to appraised contingencies rather than to different objective stimulus contingencies. Without adequate appraisal, a personal loss may not evoke sadness. Third, many aspects of these contingencies are subject-dependent anyway, such as the role of the available coping repertoire.

A large amount of research has sought to verify which appraisals correspond to which emotions. Most of that research consists of asking subjects to recall an incident of a number of emotions as identified by name (‘please recall an incident when you were angry’), and to fill out a questionnaire asking to rate the role of each of a number of appraisal components (e.g., was the event favorable or harmful to your goals? how controllable did the event appear? was someone responsible for that event?). On the whole strong correspondence was found between particular emotions and particular appraisal patterns. It is, however, unclear to what extent this provides evidence that these appraisals are the emotions causal antecedents. Self-report results may report emotional experiences rather than their antecedents, or be due to post-hoc constructions (Parkinson, 1995, and contributions in Schorr, Scherer, & Johnstone, 2000). Experimental research to find support for this aspect of the theory also exists, showing, for instance, that fear depends both upon outcome uncertainty and anticipated danger, but it is still modest.

There are more important points. Different individuals may appraise the same events differently, and thus differ in their emotions. The differences may be due to different expectations, different cognitive schemas that steer inferences from actual events, and different things that individuals desire or strive for. Individuals may also differ in the extent to which the features of different contingencies are actually picked up, or are expected to be present when in fact they are not. Appraisal theory thus tries to account for individual differences in emotions, and changes in emotion from moment to moment.

By including cognitive variables among emotion antecedents, appraisal theory substitutes the notion of situation for that of stimulus, in describing emotion antecedents. Stimuli rarely occur in isolation, and their effects usually depend upon a wide context that includes the individual’ history as well as situational context (e.g., effects of social support, ways out under stress, awareness of coping repertoire called self-confidence).

Recognition that situations rather than stimuli elicit emotions explains that given events may have multiple emotional meanings, that are relevant to different concerns and allow different appraisals. Events may simultaneously or successively evoke various and even opposite emotions. Event—emotion links are therefore best described as sets of mechanisms, the outcomes of which may conflict (e.g., ‘oppression causes anger’ and ‘oppression causes submission’), the appraisal details allowing one or the other are often hard to specify in advance (Elster, 1999).

Intrinsically Affective Stimuli and Conditions

Both accounts of emotion elicitation include the basic assumption that particular stimuli or conditions are intrinsically pleasant or unpleasant. The term is used for innately liked or disliked stimuli and conditions (like sweet taste and physical pain), as well as for stimuli that are liked or disliked due to early habit formation (like most food preferences; see Rozin, Haidt, & McCauley, 2000). Further emotion elicitors derive from these intrinsically affective stimuli and conditions by association, other forms of learning, or inference. Some stimuli do not properly innately elicit affect or emotion, but may innately facilitate the formation of liking or dislike, or facilitate the formation of conditioned avoidance or approach reactions. This is the preparedness hypothesis (see ö hman, 2000); the hypothesis is illustrated by the easily acquired fear of snakes, spiders, heights, and the like.

Intrinsically affective stimuli are those that behaviorist theory considers primary reinforcers. They include more than tastes and pain and other bodily discomfort. When near-universality among humans (and, in relevant cases, com-munality with primates and other higher animal species) is taken as evidence, they include sexual visual stimuli (young female forms) for human males, and friendly, distressed, and angry facial and vocal expressions (Fridlund, 1994). Other intrinsic affective stimuli are actually constellations. For instance, novel stimuli tend to be liked and evoke interest (Ito & Cacioppo, 1999), but ‘novelty’ of course is not an elementary stimulus attribute. The things people strive for, and that form the satisfaction conditions of concerns, all are complex, and related in obscure ways to elementary stimuli. These conditions include the presence of particular individuals that offer occasion for clinging (bases for ‘love’ Bowlby, 1969), familiar stimuli (Zajonc, 1968), being accepted or liked by familiar conspecifics (the basis of social bonding; Baumeister & Leary, 1995). They also include the states that define suprapersonal values (e.g., justice, personal freedom, interpersonal consideration), that may form motives and give rise to vehement emotions when threatened or achieved. Likewise, the elementary causes or implications of such an important concern as that for self-esteem are far from clear. There also are certain types of action that intrinsically produce positive affect, but cannot be properly characterized as ‘events.’ Examples are unimpeded functioning, play in the young of the species, and performing beyond habitual levels of functioning (Kubovy, 1999; Rozin, 1999).

General Conditions for Emotion Arousal

Emotions in the sense of reactions involving more than just feelings, but including behavioral inclinations or physiological change, require more involved conditions than just the presence of the stimuli or conditions just mentioned. Emotions tend not to result from mere presence of something pleasant or unpleasant, but from change over what was already present or what was expected. Emotion arousal, and emotional intensity, depend upon current level of adaptation, that itself depends upon previous and expected events (Parducci, 1995). Comparison processes play a very important role. Sometimes, it is the comparison with what could have been that counts; this is so, for instance, in the sharp emotions of regret; when you just missed the bus it feels worse than when it had left ten minutes ago (Landman, 1993). In other cases, it is the comparison with what others have that counts, as in envy.

Furthermore, whether an emotion is aroused by a given contingency (say, threat, or loss, or goal achievement) depends upon the availability or non-availability of possibilities for coping. Emotion is absent (or at least, most emotion components are) when the subject possesses a routine way to cope with the contingency.

Danger does not evoke fear if the subject knows how to deal with it, as does a well-trained mountain climber. Felt emotion, autonomic arousal, and behavioral disturbance occur only when no routine way is available, or when there was uncertainty about coping prior to succeeding. Joy and pride result from success to the degree that success was uncertain before (Weiner, 1985). In learning avoidance behavior in response to signals that electric shock will come, such signs are present in early phases of learning but disappear, once the animal has learned to escape smoothly and in time (Strongman, 1987). Some uncertainty about how to cope successfully and on time appears one of the conditions for emotions.

Appraisal of a given stimulus event may vary vastly, due to assessed meanings, the individual’ history and cognitive schemata, and the specific context. This implies that a given event may tend to elicit conflicting emotions. A sexual stimulus may activate lust, and its context (or the very response of lust) may activate interpersonal consideration, or fear of consequences.

Cognition and Emotion

Appraisal processes involve cognitive processes. Thus, cognitions influence emotion arousal, and modifying cognitions may modify emotions. Lazarus (1991) reports earlier experimental studies illustrating this: different task instructions, when viewing emotional material, changed emotional response. This notion has been taken up in cognitive behavioral therapy (e.g., Beck, 1976).

Their role in emotion elicitation has given rise to vivid controversies and debate, in part because emotion and cognition have been traditionally opposed, and in part because the very notion of cognition is unclear and ambiguous. It is sometimes taken to refer to conscious thought and deliberation, and sometimes to the processing of information, whether conscious or unconscious (Leventhal & Scherer, 1987).

Different levels of cognitive processes can be distinguished. Teasdale and Barnard (1993) distinguished propositional and implicational (or schematic) levels, the first referring to mere factual information representation, as in a verbal statement, and the second to information representation including mental images and links to feelings and actions. Propositional information does not influence emotions; only implicational information does. The distinction seeks to account for the fact that not all relevant knowledge has emotional consequences; the abstract knowledge that spiders do no harm, and verbal warnings that smoking causes cancer, often have none. Power and Dalgleish (1997) further, in their SPAARS model added associative and analogical representations. Associations (as between a conditioned stimulus and the unconditioned electric shock) and analogical representations (raw affective stimuli and their images), however, are not what most investigators would call ‘cognitive.’ That is to say that, although most interesting emotions have cognitive antecedents, some elementary emotions do not. This is so in particular because the effects of conditioning (notably traumatic conditioning, such as severe electric shock after a light flash) may become fully independent from recall of the original unconditioned affective stimulus; they cannot be corrected by information about harmlessness of the conditioned stimulus (LeDoux, 1996).

Even in arousing emotions that heavily depend upon cognitive antecedents, noncognitive elements play a key role. Potential emotion antecedents evoke emotions only when and because they have links to intrinsically affective stimuli or conditions—by associations, images or earlier feelings (Frijda, 1988); it probably is what makes them ‘implicational.’

Functions of Emotions

Evolutionary considerations suggest that emotions have a function, or had so in the evolutionary past. Different emotions apply to different adaptational contingencies: fear to physical or other threats, anger to social conflicts of interest or power conflicts, disgust to coping with the dangers of inappropriate food intake or contamination (Plutchik, 1980; Rozin, Haidt, & McCauley, 2000; Tomkins, 1962). That type of functional analysis is current in ethology and in evolutionary perspectives on emotions (e.g., Cosmides & Tooby, 2000).

In general, emotions can be said to have the function of detecting hedonic or concern-relevant events, and to modify or maintain the individual’ relationship with the environment. Detection is ensured by the various appraisal mechanisms, as well as by their end result, affect. Affect is the ‘common currency’ in terms of which the impact of different events can be compared, and response resources can be distributed (Ito & Cacioppo, 1999).

Furthermore, emotions are functional because they allow flexible adaptation to the various adaptational contingencies. Emotions involve behavioral flexibility, rather than fixed action patterns, which is a major difference with reflex behavior. The motivational states of readiness command different behaviors, as circumstances require and the behavior repertoire allows. Flexibility is the mark of emotions because emotions may profit from innate repertoires, learning, and cognition (Ekman, 1982; Frijda, 1986; LeDoux, 1996; Panksepp, 1998; Rolls, 1999; Scherer, 1999; Tomkins, 1962).

In addition to dealing with adaptational dilemmas, an important domain of emotion function is the regulation and even the constitution of interpersonal relationships and social harmony (Harré & Parrott, 1996). Many emotions serve to control interaction within groups. Shame enforces behavior in agreement with social habits, and guilt feelings motivate reestablishment of social equilibrium after having inflicted harm on someone else (Baumeister, Stillwell, & Heatherton, 1994); these interpretations are based both upon the conditions under which shame and guilt occur, and upon their consequences. Shame displays, for instance, tend to appease ill-will after transgressions, as was shown when experimental stooges upset supermarket displays and did, or did not, manifest embarrassment (Keltner & Buswell, 1997).

Emotions like sympathy, liking, and affection also establish and maintain social bonds. They enhance social cohesion in groups and support social adaptation. For instance, prior to feeding time chimpanzees often indulge in mutual grooming and kissing; when this happens, fights for the food are less frequent than otherwise (De Waal, 1996). Social emotions (loyalty, empathy, liking for social harmony) may sometimes dominate self-directed emotions, particularly in so-called interdependent or collectivist cultures (Markus & Kitayama, 1991). Moreover, emotions appear to function in social cohesion by the almost universal tendency to share one’ emotions with others, who then may share that with third parties (Rimé, Finkenauer, Luminet, Zech, & Philippot, 1997).

The various behaviors, including facial expressions, can be understood as being instrumental to the various adaptational goals. The surprise expression can be viewed as an orientation response, fear expressions as self-protective responses, and crying as an alarm call. This functional perspective can be applied in subtle fashion. For instance, submission can be viewed as a non-aggressive solution to conflicts of power and interest. The solution can be achieved by signaling one’ submission by, for instance, glance avoidance and making oneself look small. Several emotions can be interpreted as forms of submission, and as involving submission displays. Embarrassment, shame, guilt, respect, and humility, can all be viewed from that angle (Harré & Parrott, 1996).

The examples suggest that emotions were not merely functional in the distant evolutionary past. Theory has to account for the fact that emotions often are more of a nuisance than a help, such as the sufferings of grief and the behavioral interferences caused by nervousness. One of the efforts at explanation is the inverted-U-curve hypothesis: emotions improve performance and adaptation at moderate levels of intensity, but may be detrimental at high levels.

Brain Mechanisms


A hundred years ago, it was shown that emotions specifically depend upon lower brain centers. Removal of a dog’ cortex did not eliminate anger responses (it in fact enhanced them), but transection of the brain at the level of the thalamus did. Later, it appeared that the responses involved the hypothalamus rather than the thalamus. Somewhat later, Klüver and Bucy (1937) discovered severe interference with emotional sensitivity after damage to the temporal lobes. The various discoveries led MacLean (1990) to his triune brain hypothesis: the ‘reptilian brain’ (brainstem core and related structures) ensured basic biological functions, such as feeding and sex; the paleomammalian brain or limbic system contained the provisions for emotions and social feeling; the neomammalian brain or newer cerebral cortex was related to emotions only for guidance and control. The tripartite partition of the brain is not generally accepted, but formed a heuristic in research and functional description. I will give a brief overview of some of the findings in the analysis of brain mechanisms in emotions; it will include only a fraction of what has been discovered over the last decades. I will discuss the findings from a functional point of view, organizing them by major functions that brain research has helped to outline. Important sources are Buck (1999), LeDoux (1996), MacLean (1990), and Panksepp (1998).

Affect Mechanisms

Throughout the brain, there are sites sensitive to opioid substances, including the endogenous opioids such as endorphins and enkephalins. Opioids are closely linked to pleasure and to behaviors linked to stimulus acceptance (attention, facial reactions in the case of sensory stimuli). Opioid production is enhanced by rough-and-tumble playing, grooming in primates, sexual satisfactions. Opioid antagonists tend to decrease several of these activities.

Different kinds of liking (e.g., food tastes and smells, sexual pleasures) probably are linked to different sites. Taste and smell liking, for instance, appear to involve mechanisms in the orbitofrontal cortex (Rolls, 1999). A number of different neuropeptides are also involved.

Serotonin and periventricular (PVG) pathways probably are relevant in pain and other unpleasantness.

Activation Mechanisms

Attentional arousal and readiness to process relevant information is an essential aspect of emotions. Hence it is useful to mention the role of cortical arousal (Lindsley, 1951), determined or influenced by what was then called the ascending reticular activating system (ARAS), and that was a neural correlate of attentional arousal. Cortical arousal, EEG desynchronization, corresponds with attentional arousal or wakefulness.

More central to emotion are the phenomena of tonic activation or action readiness. Such readiness is evident, for instance, in intracranial self-stimulation (ICSS): animals may work hard, and for extended periods, to deliver themselves electrical stimulation at particular brain sites. The ICSS phenomena led to the hypothesis that these sites represented pleasure and pain ‘centers.’ There were, however, a number of puzzling findings, incompatible with that hypotheses; for instance, in humans, only in some places does stimulation evoke feelings of pleasure or pain. This led to the hypothesis of a Seeking System (Panksepp, 1998) or Behavioral Activation System (BAS; Gray, 1987), or to a ‘reward system’ (Rolls, 1999) that links stimuli for which it is sensitized to interest in and efforts towards obtaining those stimuli—a wanting rather than a liking system (Berridge, 1999). Although closely linked, pleasure and the activation of the seeking or wanting system appear to be distinct, and can be separately activated (Berridge, 1999).

The ICSS brain sites, too, are found throughout the brain, most concentrated in the medial forebrain bundle that traverses the hypothalamus towards the nucleus accumbens, septum, frontal cortex and the amygdala, and in those latter structures. The system is activated by dopamine (the mesolimbic and mesocortical dopamine systems). Dopamine generally plays an important role in interest and activation, as illustrated in the film Awakenings, and in the effects of cocaine and amphetamine.

Appraisal Mechanisms

As mentioned, temporal lobe damage severely impairs emotional discrimination. Animals lose their usual fears of humans and novel objects; they do not avoid objects that they had learned to avoid. They may respond sexually to animals from other species, and to inanimate objects. They try to eat inedible or unusual objects. The deficits appear to be caused specifically by damage to the amygdala (Weiskrantz, 1956).

LeDoux (1996) extensively showed that fear conditioning cannot be established after amygdala ablation (at least, it becomes considerably more difficult to do so), nor is established conditioning maintained. Other findings suggest that the amygdala may more generally be essential for emotional appraisal (Rolls, 1999). It appears to be a key point in linking perceived events to their affective value, both in a positive and in a negative sense, at least with respect to learned or conditioned stimuli. The amygdala thus may be essential for establishing the links between events and their affective implications. They may or may not be involved in affect arousal by unconditioned stimuli.

Discrimination Mechanisms

Appraisal of a total situation may lead to activation of a different emotional response from that which would be activated by the major or core aspects of that situation. Evidently, the brain is capable of on the spot modification of its learned or unlearned emotional reactions. That this is so appears from the effects of damage to the orbito-frontal cortex. Such effects are vividly described and discussed by Damasio (1994). Patients with such damage behave like psychopaths, and lack consideration as well as emotional foresight. On the basis of such and other findings, Rolls (1999) interprets the main role of the orbito-frontal area in emotion to be the rapid change of emotional reactions from what given stimulus events usually or primarily evoke. For the moment, the experimental evidence (on response to reward reversals) is mainly restricted to gustatory rewards, however.

Specific Emotional Mechanisms

There are a number of neural dispositions for particular modes of dealing with the environment or for dealing with particular contingencies. They play a key role in the arousal of particular emotions, as defined behaviorally and by feeling. These dispositions are sensitive to neuro-chemical influences, some of which tend to be different for different systems. The following is borrowed mostly from Panksepp (1998). The identification of these neural dispositions supports the hypothesis that emotions have a biological basis, as well as the notion of basic emotions.

For instance, some brain circuits are of central importance in anger. Anger, one may infer, involves the activation of that circuit. Electrical stimulation of sites on that circuit, in cats, induces attack towards an animate object in its field of vision, combined with signs of excitement (hissing, piloerection). The attack and excitement die down within seconds after stimulation stops. Humans that are stimulated in equivalent locations, or that suffer from irritation in those areas, report violent rage attacks. The circuit includes sites in the medial amygdala, the ventro-lateral hypothalamus, and down to places in the midbrain (the periacquaductal gray). Activity of the circuit is supported by the neuropeptide called substance P. In normal animals, the circuit receives its information from elsewhere, for instance from regions in the orbitofrontal cortex where different sorts of information are integrated (for instance, the information on expectations and their nonfulfillment, which constitute frustration).

A major circuit has likewise been established that is essential for fear (or at least, for major forms of fear). Upon electrical stimulation, the animal may freeze, or make efforts to escape; it also may vocalize in ways as it does when faced with danger. The circuit includes the central and lateral amygdala, the anterior-medial hypothalamus, and midbrain periacquaductal gray, so again including evolutionary very old brain regions. In humans, stimulation is accompanied by a sense of foreboding. The circuit is sensitized by various neurochemical agents, among which the peptide corticotropin release factor (CRF) is pre-eminent.

Panksepp (1998) likewise describes circuits that are responsible for maternal care and, presumably, for the various urges and emotions that motivate and attend it; in humans, we call them affection, love, and tenderness. Damage to this care system (in the preoptic area, the hypothalamus, and midbrain) essentially eliminates maternal behavior in female rats, and injections of the peptides oxytocin and prolactin in some locations induce it. Interestingly, oxytocin is also involved in delivery and milk secretion. Neurochemically, the physical and emotional aspects of maternal caring belong together. Also interestingly, oxytocin plays a stimulating role in both male and female sexual activity, as well as in response of infants to their mothers, in responses to stroking, and in the attenuation of various forms of aggression.

As already mentioned, social bonds and various social pleasures are accompanied by increases in brain opioids. Opioids have a corresponding important role in the brain system linked to sadness or separation distress; Panksepp (1998) labels it the PANIC system. From the ancient midbrain periacqaductal grey (PAG) upwards to the medial thalamus, the septal area and the preoptic areas (just in front of the hypothalamus), up to the anterior cingulate cortex are circuits responsible for crying, or its animal counterpart, the distress call. Lesions in all the relevant circuits diminish or abolish distress calls, stimulation in those circuits induce them, whereas opioid antagonists enhance them and opioids (including morphine) again attenuate. The midbrain PAG circuits for separation distress intermingle with those that mediate physical pain, which make it plausible that, at least at that level, the two feelings resemble each other, and perhaps sensitivity to the cues of separation from caretakers or other attachment figures.

Culture and Emotions

The preceding sections contain ample evidence that emotions are based in biological, wired-in mechanisms of affective sensitivity, appraisal, and action readiness change. At the same time they wear the stamp of culture. There is wide cultural diversity.

Emotion taxonomies differ from culture to culture, or at least from one language to another. Numbers of emotion words vary between seven (for the Chewong in Malaysia), and two thousand (for English; Russell & Barrett, 1999). Many words are untranslatable between languages, and when they are, meanings may differ substantially (Russell, 1994; Wierzbicka, 1994). Although the structures of the emotion word sets in different language tend to yield similar basic concepts, there also are important differences (Shaver, Wu, & Schwartz, 1992). Basic emotions distinguished in cultural lore may be very different. In Hindu philosophy, for instance, the basic emotions are joy, sorrow, anger, sexual passion, energy or heroism, disgust, and amazement (Shweder & Haidt, 2000).

Differences exist in the relative importance of the various dimensions of cognitive appraisal and action readiness. For instance, magical influence is an important appraisal aspect in African cultures, and negligible in Western ones (Mesquita et al., 1997, for a review). The focality of given emotions (e.g., shame) varies as a function of differences in the prominence of concerns for social propriety and self-esteem (Mesquita et al., 1997).

Although basic mechanisms and emotional sensitivities may well be universal, the actual phenomenology of emotion may differ, and their roles in society may make the ‘same’ emotions (say, the emotion of aggressive self-assertion, in English called anger) in practice to be quite different ones (Harré & Parrott, 1996). There is no contradiction between the two conclusions of universality of basic mechanisms and cultural variability of specific emotions. As we saw, almost all aspects of emotion are influenced by cognition or other forms of acquired information. We discussed this in connection with emotion elicitors, with the sources of emotion referred to as motives or concerns, and in the forms taken by response repertoires. Culture as such is built-in in emotion in people’ sensitivity for what others do or might do, and in the processing of symbolic information.

On the other hand, when methods of research allow such comparison, the variance of emotions and their manifestations due to culture appears much smaller than the variance due to types of emotions (see discussion of the work of Scherer in Mesquita et al., 1997).

Concluding Remarks

Emotion psychology has shown considerable progress over the last several decades. The progress is in various domains. One domain is that of theory. There has been a significant shift from emphasis upon phenomena to underlying mechanisms. The old oppositions between emotion and reason, and nature and nurture, have for all practical purposes been resolved as theoretical issues. Progress can also be seen in the field of empirical research. Cognitive influences have come to be a focus of interest, as also has research into cognitive and noncognitive determinants, and their respective roles. Finally, the advances in techniques, findings, and conclusions in the neuropsychology of emotion has been tremendous. Emotion has found a place in the brain, as it has in psychology as a whole.