Core Affect and Psychological Construction of Emotion: Expanding the Emotional Inquiry - P, Apuntes de Psicología

¡Descarga Core Affect and Psychological Construction of Emotion: Expanding the Emotional Inquiry - P y más Apuntes en PDF de Psicología solo en Docsity! Emotion, core affect, and psychological construction James A. Russell Boston College, Chestnut Hill, MA, USA As an alternative to using the concepts of emotion, fear, anger, and the like as scientific tools, this article advocates an approach based on the concepts of core affect and psychological construction, expanding the domain of inquiry beyond ‘‘emotion’’. Core affect is a neurophysiological state that underlies simply feeling good or bad, drowsy or energised. Psychological construction is not one process but an umbrella term for the various processes that produce: (a) a particular emotional episode’s ‘‘components’’ (such as facial movement, vocal tone, peripheral nervous system change, appraisal, attribution, behaviour, subjective experience, and emotion regulation); (b) associations among the components; and (c) the categorisation of the pattern of components as a specific emotion. Keywords: Emotion; Core affect; Psychological construction; Facial expression; Emotional behaviour. In fear, your heart races, your palms sweat, your face broadcasts fear, you scream, and you flee. But, does this happen in all cases of fear? Most? In the realm of the emotions, reflex-like consistency is the exception rather than the rule. The rule is differences*both between and within individuals and situations. As Barrett (this issue) describes, there is more variety within fear and other categories of emotion than our standard theories suggest*indeed, more than commonsense supposes. To understand these differences in emotion requires a conceptual framework that anticipates differences and that supports a more idiographic approach to research on them. It requires a new way of thinking about emotion. This article outlines one such approach. Correspondence should be addressed to: James A. Russell, Department of Psychology, McGuinn Hall, 140 Commonwealth Avenue, Boston College, Chestnut Hill, MA 02467, USA. E-mail: james.russell@bc.edu This work was supported by NSF grant # 5000590. This article is based on a talk given at the Symposium on Emotions and Individual Differences, Leuven, 30 May to 1 June 2007. I thank attendees at the symposium and Giovanna Colombetti for their comments on the ideas presented here. I especially thank Kristen Lindquist for an insightful reading of a draft of this article. COGNITION AND EMOTION 2009, 23 (7), 12591283 # 2009 Psychology Press, an imprint of the Taylor & Francis Group, an Informa business www.psypress.com/cogemotion DOI: 10.1080/02699930902809375 D ow nl oa de d by [ B os to n C ol le ge ] at 1 0: 04 3 1 M ar ch 2 01 2 The major barrier to progress is not our ignorance but our illusion of knowledge. Increasingly, our traditional ‘‘knowledge’’ about emotion is being questioned at the most fundamental level. The traditional assumption that emotion is a unitary event is challenged by the idea that emotion is a multi-component process, with no one component identified with the emotion (Scherer, 2001). A similar challenge is Clore and Ortony’s (2008) argument that an emotion is an emergent construction rather than a latent entity and their call for a shift from discussing ‘‘emotion’’ to discussing ‘‘affective processes’’. Minsky (2008) called for wholesale remodelling of this field. Kagan (2007) challenged the language used in this field and its methods for gathering knowledge. There are many exciting avenues to explore and new ways to think about emotion. My version has been presented at length with supporting evidence elsewhere (Russell, 2003, 2006). Here I clarify, develop and defend that approach. First, however, I briefly summarise some problems with the traditional way of thinking about this topic. THE STRANGLEHOLD OF COMMON SENSE Citing Tomkins (1962, 1963) and his followers (Ekman, 1984; Izard, 1977), much of the psychology of emotion has been guided by a highly productive research program that I call basic emotion theory. The central idea of basic emotion theory is that human nature includes a small number of qualitatively distinct kinds of emotion, each of which produces (or is) a distinct, tightly organised and recurring pattern of manifest components. Basic emotion theory is a plausible approach that has stimulated the gathering of much evidence. As often happens in science, however, the evidence has revealed surprising problems, which I summarise in Table 1. No one or two problems are fatal, but, altogether, the problems point to a poor prognosis for this type of theory. Rather than elaborate on the problems listed in Table 1, let me instead suggest a diagnosis of their source. The problems encountered by basic emotion theory are not philosophical but empirical. They are problems encountered by most analyses of emotion, modern and ancient, and in diverse disciplines. They stem from the preconceptions that underlie not only much of scientific thinking but much of our everyday thinking about emotion. The source is deeper than simply the answers provided by basic emotion theory, which is, after all, just a more explicit and systematised version of common knowledge. The problem already exists in the questions that this family of emotion theories attempts to answer. Questions such as: What is an emotion? How many emotions are there? What are they? How 1260 RUSSELL D ow nl oa de d by [ B os to n C ol le ge ] at 1 0: 04 3 1 M ar ch 2 01 2 developed, but the very questions that motivated our work and the optimism that we will soon succeed. Loss of vocabulary and questions leaves theorists with a sense of vertigo. Critiques of current practices have then been dismissed as anti-emotion theories. Zachar (2006) characterised my alter- native as Eliminativist. Colombetti (in press) suggested that my alternative implies that ‘‘emotions have no existence independent of our categorization’’ (p. 2). For many readers it might seem as if I am proposing that we toss out everything of importance. So, what remains? In abandoning the geocentric theory of the solar system, some facts survived the move to a heliocentric system: the earth and sun remain, their relative positions remain, motion remains, observed time between events remain. Similarly, in abandoning an emotion-centric view, many of our observations about events called emotion remain. Indeed, all the observable phenomena remain in my account: faces move, prosody changes, hearts speed up and slow down, people aggress and they flee*all such events really occur, independent of our categorisation. Some people see anger in faces of others and they feel anger in themselves1*these events really occur, but do imply categorisation via the folk concept of anger. Emotion’s manifestations (although on my account the phrase is a misnomer) are real and important and cannot be eliminated. A scientific account of these manifestations can be had without the concepts of emotion, anger, etc., used in a scientific role and without assuming them to be manifestations of something. Even the concepts of emotion, fear, anger, and the rest, remain*although as everyday folk concepts rather than as scientific ones. In that role, they are important topics, much as are all other folk concepts from angel to the zodiac. Anger, disease, emotion, marriage, meal, money, poker, tax, tornado, soap and many other folk concepts play a role in human affairs that must be understood. On my account, events now called emotion (in human adults and infants and non-human animals) bear a family resemblance to one another. Although it does not follow that emotion/not-emotion divides nature at its joints, the word emotion does point in an interesting direction, and the events it refers to are real and important. The same can be said for fear, anger, and the like. In short, on my account, much remains but is understood in a different framework. My account does not allow the concept of emotion to determine boundaries of the domain to be explained. My account postulates no single 1 To be clear: when I write of ‘‘seeing’’ anger or another emotion, of ‘‘seeing’’ two lines in the MüllerLyer figure as of unequal length, or of ‘‘seeing’’ the sun set, I mean to describe what the perception is like for the perceiver. In the realm of the emotions, we see faces move, which they really do, hear the voice change, which it really does, see palms sweat, which they really do, and so on, all in some context. We then name the combination of these events as anger, thereby inferring anger as their cause. CORE AFFECT AND PSYCHOLOGICAL CONSTRUCTION 1263 D ow nl oa de d by [ B os to n C ol le ge ] at 1 0: 04 3 1 M ar ch 2 01 2 causal entity*the emotion of anger, fear, or their neural equivalent, an affect programme*to explain the occurrence of the manifest components of that emotion or to explain the pattern or correlations among these components. In the remainder of the article, I summarise my account by clarifying two key concepts: Core affect and psychological construction. Core affect is part of, but not the whole of, what are called moods and emotions. Psychological construction is an umbrella term for the set of processes that produce a token emotional episode’s manifestations and its categorisation as an emotion.2 CORE AFFECT Core affect is a pre-conceptual primitive process, a neurophysiological state, accessible to consciousness as a simple non-reflective feeling: feeling good or bad, feeling lethargic or energised. There is something it is like to feel core affect. Its presence in consciousness varies from being focal to peripheral to out of sight. A structural description of core affect is an empirical matter, and I have proposed a circumplex with the two underlying dimensions of pleasuredispleasure and activationdeactivation (Russell, 1980, 2005; Yik, Russell, & Steiger, 2008). Although two-dimensional, core affect is, subjectively, a single feeling. That is, the two dimensions combine in an integral fashion to form one unified feeling. Pleasure and arousal combine to form the single feeling of ecstasy, for example. An analogy is the sensation of a specific colour, such as the red of the autumn leaf outside my window. The dimensions of hue, saturation, and brightness combine in an integral fashion to form one unified sensation of any particular colour. It may help to say what core affect is not. So, here is a series of conceptual distinctions. The idea is to point to ostensibly different phenomena, without precluding the possibility that core affect is contingently related to them. Core affect, although it may be empirically involved, is not a judgement that something is morally good or bad or tactically advantageous or harmful. Core affect has been shown empirically to be related to changes in the autonomic nervous system, facial and vocal behaviour, instrumental behaviour, cognitive processes, reflexes, and a host of other things (Russell, 2003). Still, core affect is conceptually distinct from them. 2 The influence of the concept of emotion is pervasive and subtle. For example, critics of my account (Russell, 2003) have mistakenly assumed that mine is a ‘‘dimensional’’ rather than ‘‘categorical’’ account of emotion, i.e., that the domain of emotion is to be subdivided by dimensions of core affect rather than by discrete categories such as fear and anger. Another false interpretation is that core affect carries the same assumptions that emotion does, i.e., that core affect must account for facial and vocal expressions, autonomic changes, emotional behaviour and the like. 1264 RUSSELL D ow nl oa de d by [ B os to n C ol le ge ] at 1 0: 04 3 1 M ar ch 2 01 2 Core affect is not a substitute term for emotion, nor is it the essence of emotion; it is not a necessary feature of emotion.3 Prototypical emotional episodes are said to begin and then, after a short time, end. In contrast, one is always in some state of core affect, which simply varies over time, sometimes slowly, sometimes rapidly, without beginning or end. Prototypical emotional episodes are directed at something (one is angry with, afraid of, or sad about something). In contrast, core affect is not necessarily directed at anything. (In this regard, core affect is closer to the English word mood than to emotion, although mood prototypically is long lasting and mild.) Core affect per se can be free floating (as in feeling down but not knowing why), but it can come to be directed at something. The full experience of core affect can thus become intentional in the philosophical sense, in much the same way that the full experience of a pain can become intentional (Searle, 1992, p. 251). Core affect is also distinguishable from the affective properties we perceive in objects, events, and features. We perceive objects to be beautiful, ugly, awful, soothing, and so on. We perceive how pleasant or unpleasant or how energising or soporific something promises to be. Typically, perceiving something to be beautiful is a pleasant experience, perceiving something ugly unpleasant. And, such perceptions are logically related to core affect: to perceive something as pleasant is to judge it capable of producing pleasure. Nevertheless, I want to allow the possibility that the perception of an affective quality can occur even when we’re not personally moved by the object or event. That is to say, such perceptions are not, by definition, accompanied by changes in core affect. Distinguishing the two thus leaves as an empirical hypothesis the idea that core affect is involved in the perception of affective quality. Even when non-intentional, even when we have no clue why we feel as we do, core affect is caused. Individuals vary greatly in how their core affect is experienced (Barrett, this issue) and changes over time (Kuppens, Van Mechelen, Nezlek, Dossche, & Timmermans, 2007). Core affect changes in response to many simultaneous influences. Sometimes the influence is a single powerful and obvious external event, as in James’s prototypical 3 Core affect does provide the pleasant or unpleasant hedonic tone to those token emotional events that have that tone. Core affect is also a central feature of the mental prototypes of some emotions but not a necessary feature of all cases of those emotions. For example, cases of fear can be found without a core affect of unpleasant arousal, but they are mediocre or borderline rather than prototypical cases. (Fear without arousal: contemplating a distant danger; fear without displeasure: thrill seeking.) Charland (2005) underscored the distinction between core affect and emotion by observing that, on Panksepp’s theory, some emotions (lust, seeking) can, depending on context, involve very different values of core affect prototypically. Of course, the concepts of emotion, fear, and so on, could be redefined for scientific purposes such that core affect is a necessary feature, but I’m not advocating that route. CORE AFFECT AND PSYCHOLOGICAL CONSTRUCTION 1265 D ow nl oa de d by [ B os to n C ol le ge ] at 1 0: 04 3 1 M ar ch 2 01 2 together at a psychological level at the time of its occurrence. There is no one general mechanism. Put differently, the traditional assumption is that a ‘‘theory of emotion’’ will differ from a theory of cognition, or behaviour, or conation. My claim, in contrast, is that any theory that explains all cases called emotion will be close to the whole of psychology, a theory that of course will not be limited to emotion but will extend to all psychological processes. My proposal raises three questions: (1) how to explain any given emotional episode, each token event; (2) how to explain the patterning of emotion’s components; and (3) how to explain our compelling perceptions that particular episodes are instances of a general category, such as fear or anger. Explaining emotion’s manifest components So, how specifically does psychological construction explain any given event called emotion? The first step is to recognise that each token consists of manifest components (Scherer, 2001, this issue), such as facial and vocal expression, changes in the autonomic nervous system, subjective experience, and so on. Let us begin by examining each component, with an eye toward how we might begin to explain it without positing an emotion as its cause. Doing so also explains why there are individual and situational differences that occur even within a given category of emotion. Facial and vocal expressions. For the face or voice to ‘‘express’’ emotion implies, among other things, that the emotion causes the facial or vocal changes. Explanations of facial and vocal changes in terms of emotion have dominated the research agenda, but alternative explanations are starting to appear. Faces move as part of perception: to look away, to stare, to smell, etc.: We wrinkle our noses at smells, gag at noxious tastes. Faces move as part of cognitive processes such as attention and appraisal (Ortony & Turner, 1990; Scherer & Ellgring, 2007; Smith & Scott, 1997). Faces move as part of action, including speech and other goal-directed actions: We turn our faces in the direction we are heading, and we make faces as we tell stories. Some facial signals evolved (either genetically or epigenetically) to guide social interactions, such as threat, greeting, or submission (Fridlund, 1994). The face and voice change with general arousal and positive versus negative core affect (Lang, Greenwald, Bradley, & Hamm, 1993). My hypothesis is that facial and vocal behaviour during an emotional episode can be explained through such non-emotional processes (Russell, Bachorowski, & Fernandez Dols, 2003). 1268 RUSSELL D ow nl oa de d by [ B os to n C ol le ge ] at 1 0: 04 3 1 M ar ch 2 01 2 Autonomic nervous system (ANS). The search for an ANS signature for each emotion is based on the commonsense assumption that our heart races because of fear, and so on. Such signatures have not been found (Larsen, Bernstson, Poehlmann, Ito, & Cacioppo, 2008), and, more generally, ANS changes can be accounted for without attributing them to an emotion. ANS activity is on going, with both general features (degrees of arousal) and specific chores. Perhaps one can also detect broad response patterns, mobilisation for action versus relaxation for homeostasis or perhaps preparation for approach versus avoidance (Larsen et al., 2008). Long ago, Lacey (1950, 1967) found individual- and context-specific patterns of autonomic response. Emotional behaviour. In the traditional way of thinking, emotions cause behaviour: we strike because are angry, we flee because we are frightened, and so on. James (1884) turned the causal arrow around: we are angry (or, more precisely, perceive ourselves to be angry) because we strike and are frightened because we flee. Baumeister, Vohs, DeWall, and Zhang (2007) reviewed evidence that questioned whether there is any causal arrow between anger and aggression, or between fear and flight, or, more generally, between emotion and behaviour. Scherer (this issue) cites as a feature of emotion the looseness of its link to behaviour. Fight, flight, and other emotional behaviour can be accounted for without emotion. Core affect, even when outside consciousness, influences behaviour (Winkielman, Berridge, & Wilbarger, 2005). Appraisal of the current situation can lead to formation of goals and plans and their execution (a process that can be implicit as well as explicit, and quick and ill advised as well as slow and wise). Strack and Deutsch (2004) brought together various lines of evidence and theorising to propose a two-system account of behaviour. An ‘‘impulsive system’’ influences behaviour through associative and motivational mechanisms, whereas a ‘‘reflective system’’ influences behaviour through a knowledge-based mechanism. On their account, emotional behaviour is not thought of as resulting from an emotion, but as the joint product of these two interacting systems. Experience of having an emotion. Persons sometimes feel afraid, angry, sad, and so on. Such conscious feelings are real, but we need not then go on to assume that such feelings are veridical detections of (caused by) an emotion. On my alternative account, emotion is not their cause. Rather, as James (1884) proposed, such experiences are perceptions: to feel afraid is to perceive oneself as afraid (see Prinz, 2004). More technically, I characterised this feeling as a meta-experience because the raw data on which it relies include other experiences: core affect, somatosensory feedback, appraisal of the eliciting event, attribution, beliefs, desires, plans, and behaviour. CORE AFFECT AND PSYCHOLOGICAL CONSTRUCTION 1269 D ow nl oa de d by [ B os to n C ol le ge ] at 1 0: 04 3 1 M ar ch 2 01 2 Similarly, Lambie and Marcel (2002) characterised the experience of having an emotion as a second-order experience*that is, an experience that emerges out of first-order experiences. Perceiving oneself as having an emotion is no different in kind from other perceptions. Percepts are often compelling, but they are not simple. Nor are they infallible. Like other acts of perception, an emotional experience is not entirely ‘‘bottom-up’’, not entirely data driven. A percept is the end product of a complex process involving raw data, concepts, learning, and context. Smith and Neumann (2005) developed a sophisticated account along these lines by drawing on dual-process models of perception. To perceive oneself as afraid is to categorise oneself by means of the concept of fear. It is to establish the meaning of one’s state via the concept of fear. In turn, the concept of fear can be unpacked as a script laying out a series of subevents (the components) in a temporal and causal order (Fehr & Russell, 1984). To perceive oneself as afraid is to see a resemblance between one’s current state and a mental script (the concept) for that emotion. A separate hypothesis is that there are both similarities and differences in emotion concepts across cultures and languages (Russell, 1991; Wierzbicka, 1999). Although some of the raw data on which emotional experience depends are universal (core affect, somatosensory feedback, attributions, appraisals, etc.), where there are differences in emotion concepts, the corresponding emotional experiences would then vary as well. Evidence from a surprising source supports this prediction. In their effort to find a universal ANS signature for each basic emotion, Levenson, Ekman, Heider, and Friesen (1992) studied the Minankabau of West Sumatra. Participants were instructed to contract facial muscles into the prototypical configura- tions hypothesised for basic emotions. Doing so, in turn, alters ANS activity. For North Americans, this alteration of facial muscles and ANS activity resulted in reports of the experience of specific emotions. For the Minankabau, however, the same procedure failed to produce the emotional experience, presumably because their emotion concepts differed from those of North Americans. Appraisal. On more traditional accounts, emotion is separated from thought and reason, as in theories in which emotion is encapsulated from cognition (Griffiths, 1997; Prinz, 2004). In other accounts, an appraisal is thought of as a part of the emotion or as an event that mediates between an antecedent event and the emotion. The dichotomy between rational thought and irrational emotion was long ago undermined by research showing, on the one hand, that cognitive processes emphasise economy and speed as much as rationality and, on the other, that cognitive processes are relevant to emotion. 1270 RUSSELL D ow nl oa de d by [ B os to n C ol le ge ] at 1 0: 04 3 1 M ar ch 2 01 2 with similar results (Reisenzein, 2000, 2007). There is much less patterning to be accounted for than is traditionally assumed. Many of the components are continuous psychological processes: core affect is always present; the ANS is continuously active; people are always behaving, always perceiving, always appraising, and so on. Because emotion’s components are on-going processes, they necessarily occur in combination. The weakness of the correlations among components implies a large number of possible combinations. The question has not to my knowledge been raised which of these combinations constitute organised patterns and which mere combinations. For now, I won’t pursue this distinction and will call all combinations patterns. Explaining the patterns that do exist. Correlations found so far among components are weak, but not zero. Besides, patterns could exist that have not yet been established. Correlations that are found among components thus require explanation. Traditionally, theorists have sought a single central mechanism to account for what correlations exist*a fear-producing mechanism such as an affect programme, for example, to account for what associations occur among fear’s components. A single central mechanism, however, is not warranted until simpler explanations are exhausted. Some writers have recently acknowledged the growing evidence on the weakness of the correlations among emotion’s manifest components and have begun exploring dynamic systems theory as a conceptual framework that accommodates the tremendous variation across situations and indivi- duals within a given category of emotion (Camras, 2000; Colombetti, in press; Fogel et al., 1992; Lewis, 2000; Scherer, 2000). I believe a much simpler approach is possible, one that relies on already known sources of correlation. There are at least three, not mutually exclusive, alternatives. First, features in the environment have a correlational structure. When two correlated environmental features each elicit a separate response, then those two responses will be correlated. For example, suppose that novel events are more likely than familiar events to block a goal. (Perhaps problems presented by familiar events have been previously encountered and solutions found.) Suppose further that goal blockage elicits an ANS pattern of cardiac acceleration and that novel events elicit frowns. If so, cardiac acceleration and frowning will be correlated even if no internal process links the two. Second, one component process can influence another. For example, suppose that forming the face into a threat expression (the ‘‘anger face’’) alters breathing and muscle tension, which in turn alters ANS activity, perhaps cardiac acceleration. The consequence would be that the threat face is correlated with cardiac acceleration. The component of emotional experience uses the other components as raw data in the formation of the CORE AFFECT AND PSYCHOLOGICAL CONSTRUCTION 1273 D ow nl oa de d by [ B os to n C ol le ge ] at 1 0: 04 3 1 M ar ch 2 01 2 percept. This hypothesis predicts that emotional experience is correlated with other components*more highly than they are with each other. Appraisal of current situation and attribution lead to emotional behaviour. And, third, two components will be correlated when they are both influenced by a central mechanism other than emotion. For example, suppose that focused attention produces both muscle tension in the face and cardiac acceleration. If so, muscle tension in the face will be correlated with cardiac acceleration. The central mechanisms responsible for emotional behaviour (on Strack & Deutch’s account, impulsive and reflective systems) influence other components such as facial and vocal behaviour and ANS activity. In short, if my thesis pans out, the explanation for the pattern of manifest components that occurs in each instance of emotion is the explanation for the individual components plus well-known and simple explanations for the correlation among those components. No additional emotion-producing mechanism is needed. As an analogy, consider a fair game of poker. When cards are dealt fairly, each hand consists of a pattern of cards (a pair or straight, or whatever, although most patterns have no name within the rules of poker). Shuffling and dealing are the mechanisms that produce patterns; there is no additional pair-producing or straight-producing mechanism (other than cheating). Of course, in poker and in emotion, we see certain patterns, and doing so can be very important. Which brings us to the next topic. Perception of emotions As outlined so far, psychological construction is a research agenda that hopes to explain each particular emotional episode by explaining its manifest components and the correlations among the components. To many readers, this programme, even if it were successful, would still leave something out. The available empirical evidence on correlations among components would not lead scientists to hypothesise fear or other types of emotion as their explanation. But then most of us do not hypothesise fear* we see it. We see discrete emotions in others and experience them in ourselves. We remember prototypical blue-ribbon cases of anger, fear, etc.* each with all the components in the right pattern. These compelling observations, experiences and memories of emotions do not entail the traditional view of emotions, but they do raise a question: How to account for these perceptions and memories? My account has two parts: the event perceived and the perceiver. Consider the event perceived. Someone is going about his or her daily life, encountering various situations and pursuing various goals. He or she will necessarily be undergoing some of the components listed above, since they 1274 RUSSELL D ow nl oa de d by [ B os to n C ol le ge ] at 1 0: 04 3 1 M ar ch 2 01 2 are continuous on-going processes. Some component processes are therefore always present and always in some pattern. As plans unfold and situations occur, the components change and hence form new patterns. Therefore, even if the component processes were completely independent of one another, certain nameable patterns would form from time to time. That the components are somewhat correlated with each other means that certain patterns are more likely to form than others. Now consider the perceiver observing this person. The perceiver does not simply register an external reality. Rather, the perceiver brings to the task a set of concepts embedded in a set of implicit assumptions and inherited from our linguistic ancestors*concepts for English speakers such as emotion, fear, anger, and so on. A concept such as fear is a mental script that specifies a temporal and causal pattern among various components. The perceiver observes the other’s face, voice, behaviour, signs of physiological state, current situation, and so on. Perhaps the perceived person reports a subjective experience. The perceiver notes the similarity between the pattern of observed components and the script defining a category such as fear or anger. Occasionally, there is sufficient resemblance for that pattern to count as a member of the emotion category. The key here is that membership in the emotion category does not require a set of necessary and sufficient features. Resemblance is a matter of degree. Occasionally, perceiving a single feature of the script will be enough for the instance to count as a member. Different observers may arrive at different categorisations of the same event. Although the script specifies a temporal and causal pattern, these aspects of the concept are not necessary either. Thus, uncoordinated combinations of components may still count as an instance of the category. The border between fear and not-fear is fuzzy. The upshot is that members of the fear category resemble one another along different dimensions, and the set of all cases of fear is much more heterogeneous than most of our current scientific theories assume. This hypothesis is consistent with the general trend in the study of semantic categories. Traditionally, it was assumed that all the instances of a category must possess the same features, in this case the recurring pattern of defining components. Classically, the features were assumed individually necessary and collectively sufficient to determine membership in the category. Wittgenstein pointed out in his analysis of the concept of game that these assumptions are not necessarily so. We neither have nor need classical definitions. Analyses of the concepts of emotion, anger, and the like have similarly challenged the classical assumption (Fehr & Russell, 1984; Russell & Fehr, 1994). Further, memory is biased toward the more prototypical exemplars, which are first to spring to mind at mention of the category name. CORE AFFECT AND PSYCHOLOGICAL CONSTRUCTION 1275 D ow nl oa de d by [ B os to n C ol le ge ] at 1 0: 04 3 1 M ar ch 2 01 2 One could of course argue that all the behaviours that occur in dangerous situations are ‘‘avoidance behaviours’’ or suppose the existence of an action tendency to avoid. Doing so, however, encounters three problems: First, there are counterexamples (thrill-seeking and bravery). Second, labelling a variety of different behaviours as ‘‘avoidance behaviours’’ adds nothing to their explanation, for the particular behaviour that actually occurs (jumping back from the spider, rushing to the hospital, etc.) remains to be explained. And, third, one cannot classify an isolated behavioural act as avoidance except in the context of an interpretation of the situation as dangerous. For example, to interpret running as fleeing rather than as approaching something else or as running for exercise requires knowledge of the context. Thus, the association between danger and a behaviour being avoidance borders on circularity. Colombetti (personal communication, 28 August 2008) replied that, ‘‘These examples all support the view that there are common behaviors! Not across these examples, but within . . . there seem to be recurrent behaviors in specific contexts’’. Commonsense supports Colombetti’s assumption, and many would assume she’s correct. But, her claim is an empirical one, and I predict that it will not be supported. Although not much evidence is available, what is available is not encouraging for her claim. The first context I mentioned, fear in phobics, has been studied. Indeed, this was the context that initially exposed the lack of consistent behaviour in fear (Lang, 1968, 1979, 1988; Lang et al., 1993; Rachman & Hodgson, 1974) and led Rachman (1984) to declare that ‘‘fear is not a lump’’. In experimental tests, individuals are all subjected to an identical situation, and yet individual differences prevail. I predict that within each of the contexts that I listed*being afraid while having a sick child, watching a film, riding a roller coaster, and contemplating global warming*a variety of behaviours will be found. The defensive behaviour of rodents in dangerous situations illustrates my view. Rats show a range of behaviours including active investigation of the threat stimulus, alarm vocalisations, escape, freezing, attack, avoidance, information seeking and risk assessment. Mother rats move pups to safer ground. As Blanchard, Hebert, and Blanchard (2007) summarised, ‘‘Defen- sive behaviours are modulated by features of both the threat stimulus and the situation in which it is presented’’ (p. 653). Perhaps a cage environment can be so constrained that all rats show the same behaviour, but maybe not. For example, two rats in a small empty cage typically leap into the air when the floor delivers mild electric shock, but occasionally they attack one another. Finally, even if Colombetti’s claim were correct (behaviour is consistent within although not between situations), it would not justify postulating fear as an entity. In that case, the behaviour would be explained by the situation and not by fear. For the hypothesis of fear to have a scientific use requires 1278 RUSSELL D ow nl oa de d by [ B os to n C ol le ge ] at 1 0: 04 3 1 M ar ch 2 01 2 evidence of a similar pattern across the various contexts in which fear is said to occur. FUTURE DIRECTIONS I have laid out an ambitious research agenda, which, needless to say, is subject to revision as data are gathered. I end with an even broader agenda. Core affect has been criticised because it does not capture all there is in emotion. My phrase ‘‘psychological construction’’ seems naturally to imply the psychological construction of emotion. I have been asked what evidence I can offer that core affect plays a central role in emotion, more central, for example, than appraisal or somatosensory feedback. Implicitly, the domain for our field is limited to emotion and the questions asked implicitly assume that emotions are what are to be explained. It is just such preconceptions that must be challenged. Admittedly, I presented psychological construction as a way of accounting for emotion. But doing so is meant to be a bridge between current concerns and a new approach. Psychological construction is an umbrella term for a host of mechanisms that explain not only those events called emotion, but a much larger set of events. Core affect is important in psychology; its role in those events called emotion is an empirical question and range from central to nonexistent in individual cases. Core affect extends beyond the domain of emotion. When the general concept of emotion is treated as the folk concept that it is, without authority to determine scientific boundaries, then more real and important events become more visible and present themselves to be explained. Once the concept of emotion is relegated to a chapter title, with no real scientific work to do, then the domain of our inquiry is revealed to be much wider. Some of the events are what are now called emotions, some moods, some feelings, and many have no name. Blue-ribbon emotional episodes (those that fit scripts for specific emotions closely) are included, but have no special status. Borders between blue-ribbon cases, mediocre cases, borderline cases, and non-cases are fuzzy and serve no scientific purpose. Instead, let us be open to a wider range of events. Feeling good that the sun is shining or feeling bad that the weather is turning too warm may not qualify as emotions, but they are frequent events, influence other behaviours, and require explanation. More generally, even without emotion, anger, and the rest as scientific concepts, we are left with each of the components discussed here. The task is to explore each component and the relations among them. The reader may find the analysis outlined here pessimistic. I find it optimistic. I believe that progress in the science of emotion has been blocked by unwarranted but hidden assumptions. On my suggested strategy, we will CORE AFFECT AND PSYCHOLOGICAL CONSTRUCTION 1279 D ow nl oa de d by [ B os to n C ol le ge ] at 1 0: 04 3 1 M ar ch 2 01 2 have no grand theory of emotion, but we will have mini-theories of individual components. Research in practice already does just this. Freed from these roadblocks, progress may be rapid. Indeed, much has already been achieved, but not recognised because of these hidden assumptions. For example, consider the various theories of emotion proposed over the last century or so. Each theory has been found wanting, because it has been asked to perform an impossible task: account for all cases of emotion. Freed from this goal, these theories complement one another and point to important processes. Each may solve part of the puzzle. Cumulatively, they may together provide a viable account. The analysis of this article is aimed encouraging the development of separate accounts of each ‘‘compo- nent’’ of emotion and the empirical search for links and patterns among those components. REFERENCES Arnold, M. B. (1960). Emotion and personality (Vol. 1). New York: Columbia University Press. Barrett, L. F. (2009). Variety is the spice of life: A psychological construction approach to understanding variability in emotion. Cognition and Emotion, 23(7), 12841306. Baumeister, R. F., Vohs, K. D., DeWall, C. N., & Zhang, L. (2007). How emotion shapes behavior: Feedback, anticipation, and reflection, rather than direct causation. Personality and Social Psychology Review, 11, 167203. Blanchard, D. C., Hebert, M., & Blanchard, R. J. (2007). Defensive behaviors. In G. Fink (Ed.), Encyclopedia of stress (Vol. 1, 2nd ed., pp. 722726). Oxford, UK: Academic Press. Camras, L. A. (2000). Surprise! Facial expressions can be coordinative motor structures. In M. D. Lewis & I. Granic (Eds.), Emotion, development and self-organisation: Dynamic systems approaches to emotional development (pp. 100124). Cambridge, UK: Cambridge University Press. Carroll, J. M., & Russell, J. A. (1997). Facial expressions in Hollywood’s portrayal of emotion. Journal of Personality and Social Psychology, 72, 164176. Charland, L. C. (2005). Emotion experience and the indeterminacy of valence. In L. F. Barrett, P. M. Niedenthal, & P. Winkielman (Eds.), Emotion and human consciousness (pp. 231254). New York: Guilford. Clore, G. L., & Ortony, A. (2008). Appraisal theories: How cognition shapes affect into emotion. In M. Lewis, J. M. Haviland-Jones, & L. F. Barrett (Eds.), Handbook of emotions (3rd ed., pp. 628642). New York: Guilford Press. Colombetti, G. (in press). From affect programs to dynamical discrete emotions. Philosophical Psychology. Ekman, P. (1984). Expression and the nature of emotion. In K. Scherer & P. Ekman (Eds.), Approaches to emotion (pp. 319343). New York: Erlbaum. Ellsworth, P. C., & Scherer, K. R. (2003). Appraisal processes in emotion. In R. J. Davidson, K. R. Scherer, & H. H. Goldsmith (Eds.), Handbook of affective sciences (pp. 572595). New York: Oxford University Press. Faucher, L., & Tappolet, C. (2006). The modularity of emotions. Canadian Journal of Philosophy, Supplement Volume, 32. Fehr, B., & Russell, J. A. (1984). Concept of emotion viewed from a prototype perspective. Journal of Experimental Psychology: General, 113, 464486. 1280 RUSSELL D ow nl oa de d by [ B os to n C ol le ge ] at 1 0: 04 3 1 M ar ch 2 01 2