1 Introduction
The Affect Program Theory of Emotions
Abstract: The chapter develops a taxonomy of emotions and other affects and introduces and defends a version of the affect program theory. A general and primitive notion of affect as a motivational state is introduced. Affects are defined as real, occurrent states, functionally identified, and not well characterized by such bivalent features as positive/negative. Some of the states typically called "emotions," such as fear and anger, are found to be special kinds of affects, characterized primarily by the actions they cause or are associated with. These basic emotions are explained by a version of the affect program theory that takes many emotions to be evolved from action programs.
Keywords: action, affect, affect program theory, anger, basic emotion, emotion, fear, motivation
Craig Delancey
There probably is no scientifically appropriate class of things referred to by our term emotion. Such disparate phenomena—fear, guilt, shame, melancholy, and so on—are grouped under this term that it is dubious that they share anything but a family resemblance. But particular emotions are another matter altogether. There is good reason to believe that different sciences can make quite compelling sense of a more fine-grained differentiation of affects. My task in this book is to reveal some of the important and neglected lessons of some of the emotions for the philosophy and sciences of mind, and this task can be accomplished with just a working characterization of a few of these. More important, there is a compelling theory of some emotions that has far-reaching implications for the philosophy and sciences of mind. This is the affect program theory. Using a version of this theory as a guide to what phenomena we will be concerned with and to the nature of these phenomena will allow us to avoid fundamental confusions and to provide richer results.
The affect program theory is the view that some emotions are pancultural syndromes enabled by inherited biological capabilities. By calling them "syndromes," we mean to point out that they are coordinated collections of complex biological responses that occur together. These emotions will be characterized by several features, including at least physiological responses, such as autonomic body responses, and stereotypical associated behaviors, such as facial expressions but also relational behaviors. I will call the emotions that are taken to fall under the affect program theory "basic emotions," just so that we have some way to refer to them.1 This is a very general formulation of the affect program theory; however, with some small elaboration in this chapter, it will be sufficient to allow me to draw some very important lessons about the nature of mind.
This theory is meant to describe only some of the things that we might call an emotion. In part as a result of this, there is plenty of room for controversy regarding whether this is a proper theory of emotion. For example, some theorists have argued that conscious experience is a necessary element of an emotion (Clore, 1994), whereas this is not the case on the affect
end p.3
 
program theory. Thus, one might argue that the affect program theory does not properly describe the emotions as the normal speaker means to refer to them. The affect program theory is an empirical theory; it is not beholden to fit exactly our folk use of affect terms, or our folk theory about affects (see Griffiths, 1997). Ultimately, the defense of the affect program theory must rest on how well it (1) usefully defines and distinguishes the various affects, and (2) explains and predicts the relevant phenomena. Defending the theory's utility to explain and predict the relevant phenomena is done throughout this book, by way of applying the theory and showing how it can offer powerful new ways to think about some of the problems of mind. Defending the theory as a useful way to categorize the affects is something I will do in this chapter and the next. My approach will be to examine some of the features of affects that other scholars have singled out as necessary or sufficient or perhaps even just important to emotions and other affects. Our best scientific understanding of these features reveals that they are either consistent with the affect program theory, or are not appropriate ways to ground a theory of affect. This will also allow me to review the scientific evidence and theoretical reasons that lay the foundations for a view of mind that is quite different from most of those that characterize contemporary philosophy of mind.
Although scientists have tended to be more careful, and usually provide sufficient operational notions of the emotions and other affective states they study, until recently (e.g., Griffiths, 1997) there has been scandalously little concern among philosophers (even philosophers of emotion) for clarifying their taxonomic presuppositions. This oversight is not innocuous, since it fosters both an extremely error-prone armchair theorizing, sometimes even armchair neuropsychology, and also vagueness and confusions that can result in question-begging and pernicious ambiguities.
Most philosophy of emotion has proceeded in one of three ways. In recent years it has been most common for emotions to be investigated through the use of emotion terms. This is an approach which is sometimes taken to an extreme by those who endorse the position that the conceptual analysis of ordinary language is all that is needed to understand emotions, or by the social constructionists, who see culture—of which they take language to be the most important and revealing element—as the creator of emotions. Paul Griffiths (1997, 21ff.) has effectively criticized the former, pointing out that ordinary language analysis approach to emotion studies has been based upon philosophical presuppositions that are now largely debunked. I will criticize a strong social constructionist approach in chapter 4. A second method for philosophizing about emotions, more revealed in the lack of an explicit method, is to take emotions as primitives open to reliable introspection; not surprisingly, this approach usually yields the view that emotions are fundamentally cognitive. But taking emotions as having properties that are somehow obvious inevitably leads to begging all the important questions; emotions are introspected to have just the qualities needed to support whatever theory is at hand. I shall review some cases that show
end p.4
 
how psychologists and neural scientists have discovered some very surprising things about our everyday emotions, things which would certainly fail to be noticed by introspection. Introspection also results in subjective characterizations that are hard or impossible to pin down. Without some, even if rough, prior and objective (that is, third-person, open to observation) characterization of the things we are discussing, much of this work on emotions can be useless. A third approach is to simply define emotions and work with these definitions; this also has traditionally yielded cognitive approaches. Defining emotions up front in some cognitive form would be, of course, quite acceptable if this were not usually followed by sweeping generalizations that reach beyond the scope of the class of phenomena picked out by the definition. As it stands, all too often we find that a theorist starts with a definition of emotions that is strongly cognitive, then makes claims about all emotions, surreptitiously slipping in the assumption that all of what others call "emotions" fall under the definition of emotions as cognitive. We therefore either need to be extremely careful not to erroneously generalize from our definition, or we need to characterize (at least some) emotions in some sense that is guided by empirical data and allows us to formulate the core questions about emotions. I will take the latter route, beginning with a broad characterization of affects that is not by definition cognitive, and then exploring how we can build our way to a characterization of some emotions which will let us learn some lessons from them.
A General Notion of Affect
It will be useful to start with a more general characterization of affect. This will give us a chance to place the relevant emotions in relation to things like pleasure or mood. There is little agreement upon terminology for emotions and other affects in philosophy, psychology, or any other of the cognitive sciences. In general, terms like emotion and affect are used synonymously. However, for most of us (at least in the English-speaking world), paradigm emotions include fear, anger, joy, sadness, and disgust. At the same time, some people consider moods to be emotions, including thus long-term states that have motivational features very different from those of, say, terror. And philosophers will talk about the importance of emotions to rationality, seemingly grouping desire and other more general conative states together under the term "emotion." Given that such a disparate group of things can be labeled as emotions, we need to draw some distinctions among these phenomena. Here I shall try to avoid confusions by using "affect" as a general term, and desires, emotions, moods, and other states will classify as types of affects.
I still need to characterize affect in some positive way. The working definition I propose is: Affects are body states that are motivational. (Throughout this book, I will take body states to include neural states; when I want
end p.5
 
to draw attention to the body independent of the central nervous system, I will use the term "extended body.") This is not in itself very enlightening, since motivation is not a little mysterious. But the principal feature of these motivations is that they are internal physical states of an organism that cause it to perform an action if the organism is not inhibited by different motivations or otherwise constrained. The relation of inhibition by other motivations, and also the notion of constraint, although both intuitively clear, are very hard to specify. Without a better account of what it is to inhibit or constrain a motivation, this characterization might be too vague if we meant to explore the nature of affect per se. But the claim that the affects are types of body states is sufficient to distinguish this notion of affect from many of the competing notions; in particular, it commits us to a realist theory of motivations (in contrast to, for example, ascriptivist notions of desire, such as I discuss below and in chapter 3). Furthermore, this is a claim for type-identity: the body states that motivate are instances of a recognizable type. Since it will be sufficient to have a working notion of just a certain class of emotions, I will take motivation as a primitive; however, this notion, as it is involved with the basic emotions that will be my concern here, will be developed at more length in the coming chapters. In the meantime, this definition makes it clear that I link affects to actions.
Affect Is Characterized in a Functional Way
Affects include desires, pleasures, emotions, and moods. We should note that these things are quite distinct in the physiological and, in particular, neural structures that underlie their function; we should not expect to find a single brain system for all motivation. Furthermore, when they are cognitive, affects can include significant input from not only subcortical brain areas but also from cortical polymodal and supramodal areas. More simply put: a lot of the brain, including areas seemingly dedicated to more abstract thought, can (but need not) become involved in the affect. Thus, as occurs with many biological functions, we should expect some of the brain and body substrates of affects to be distributed. All of these distinctions reveal that this notion of affects is a functional characterization that may not in any simple way reduce to a physical one.2 We may indeed find that the neural underpinnings, for example, of some particular affects can be quite clearly mapped out; but the concept of affects in general is unlikely to have such a common characterization.
Two other things should be noted about this characterization of affects. First, although I believe that they are necessarily motivational, pains are often understood in neuroscience as somatosensory phenomena that activate a motivational system. We could use "pain" in a broader sense to include the activation of the motivational systems that neuroscientists take the somatosensory aspects of pain to activate; but, given that nothing here depends on it, I will instead avoid expending effort on what could be a contentious issue. I will not require that pains be counted as affects. Second,
end p.6
 
moods pose special difficulties; since moods will only be a passing concern here, I will not try to characterize them at more length. As a working notion, we can think of moods as long-term affective states, perhaps even long-term emotions; as such, their motivational aspect is revealed more as a long-term and consistent alteration in motivation (relative to the subject when not in that mood).
Affects Are not All Bivalent/monodimensional States
Many have suggested that affects are states that are either negative or positive appraisals (of something, such as the organism's situation). It is extremely common in psychology to group emotions into groups with "negative" and "positive" valence. Similarly, some philosophers have defined emotions as belief states coupled with some bivalent feature or one-dimensional magnitude meant to capture the affective aspect of the emotion; Patricia Greenspan (1988) uses comfort/discomfort as this feature, while many others (e.g., Marks, 1982) assume desire is this feature. I will not respect these uses of the term "affect" because they are ultimately unhelpful; although they may be valuable when used to describe some affects, they fail as broad characterizations of all affects. For example, the notion that an appraisal or state is "positive" is too vague. What makes an appraisal positive? Ultimately, if the notion of a positive or negative appraisal is not to be vacuous, it must either yield some measurable feature of the body, or, better yet, it must reveal something about the kind of behavior that such an appraisal results in (such as approach or avoidance). One supposes that joy, for example, is positive (as per colloquial usage of "positive") and that it leads to approach (in some sense). But what about anger and fear? Colloquial usage would make them negative; but one can lead to approach of the emotion's object (in attack), the other to retreat from it (in flight). Given such distinct behaviors, the categories just do not explain anything. Similarly for comfort and discomfort. Suppose anger and fear are uncomfortable. What does this tell us about the behaviors that would result? That we seek to avoid them? But it seems, at least prima facie, that we sometimes seek these emotions, through art (revenge films include bad guys who are there specifically to raise our ire, and frightening movies garner audiences because they are frightening) or activities (like seeking fights or riding a roller coaster). Or does it mean that once we have the emotion we seek to get out of it? But, again, if a movie-goer or a mountain climber is even partly motivated by the thrill of fear, their behavior is inconsistent with such a supposition (they stay in the theater, or they keep climbing). Pleasure/displeasure, comfort/discomfort, positive/negative, and various degrees of satisfaction of a desire are all too crude to tell us anything interesting about many of the emotions and the behaviors that typify them.
Note that I am not arguing here against the use, by neuroscientists and others, of activation and inhibition (and cognate notions) of behaviors as
end p.7
 
general explanatory posits (e.g., Gray 1991); I am rejecting the use of (usually far more general) one-dimensional measures for taxonomizing emotions and other affects into, say, the positive group or the negative group. Another way of making the same point is to note that such monodimensional categorizing threatens to be far too impoverished for explaining data. It can result in such a reductive simplification that effects of the phenomena involved can be lost as they are pressed onto a single measure.3 One solution to this kind of simplification is to introduce a host of bivalent appraisals for each emotion; this is a strategy taken by Andrew Ortony, Gerald Clore, and Allan Collins (1988) in their discussion of the cognitive origins or causes of emotions. They argue (18) that emotions are bivalent reactions concerned with three aspects of the world: events, agents, or objects. But, of course, multiplying the number of dimensions in a model can distinguish any number of states; so before we accept a complex of bivalent appraisals or monodimensional features, we need some independent reason to accept the dimensions that are being offered. Here, we shall see that dropping the very notion of bivalent appraisals and related notions loses us nothing. The term affect will be used in a way that does not presuppose bivalent or monodimensional measures of this sort.
Affects Are Occurrent States, not Dispositions
Affect terms can all be used in a dispositional sense. If we say that Tony desires chocolate, or that Eric is angry at his landlord, we could mean at least two things in each case. We could mean that the person in question is in a particular body state, or we could mean that he tends to be in that body state, given the right conditions. The former I will call an occurrent affect, and the latter a disposition to affect.4 Thus, in ordinary discourse a sentence like "Eric is an angry person" can be ambiguous; it could mean that Eric is angry right now, or that Eric is the kind of person who is often angry. Similarly, one might say that Eric has been angry at his landlord for years, but of course it is not the case that anyone can be in an occurrent state of anger for that long a period of time. Instead, we mean that when reminded of his landlord or confronted with his landlord, Eric usually becomes angry. We might also mean that the beliefs and values Eric holds that cause him to be angry at his landlord—say, the belief that his landlord is charging him too much money, and the high value he places on being treated justly, and so on—are still held by Eric, which should have as a consequence that when he attends to these things he has an occurrent state of anger as a result. Or Tony can be said to have a disposition to desire chocolate if he desires chocolate often, or if he desires chocolate whenever he sees it. But Tony has only an occurrent desire for chocolate if he is actually in a state of desiring chocolate. Disposition to emotions and other affects are of particular importance to our normal discourse because we use them in attributions of temperament5 and other affective personality traits: a sybaritic person may be someone who has a disposition to desire to ingest chocolates and to
end p.8
 
pursue the experience of various other pleasures; a choleric person is someone who has a disposition to be angry. However, the concept of disposition to affects is (at least as I am using the term here) derived from the concept of occurrent affect, and does not admit of many of the features that occurrent emotions have (for example, there is no sense in arguing whether a disposition to affect is a propositional attitude—this could at best mean that the occurrent affect for which one has a disposition is itself a propositional attitude). I shall hereafter mean an occurrent affect by any affect term.
Affects Are Real Physical States, not Ascribed Explanations
There is a related notion of affect which can be held by someone who denies that there are occurrent affects, and holds that talk about affects and about disposition to affects are both just a convenient gloss for dispositions to behavior. On such a view, attributions of affects may not correspond to an actual body state but rather might just be a kind of logical construction relating actions and beliefs.6 Say, Adam always ascend the steps to his front door in a single leap. It may be that there is no significant sense in which Adam has a kind of body state that corresponds to the desire to leap up to the door; rather, he may just do it out of habit, without any need to choose between this option and the option of taking the steps one at a time. However, one might still say that Adam "desires" to leap the three steps in a single bound and simply mean by this that Adam believes (if he were queried) that he can get to the door that way, and furthermore he does get to the door that way. We then might understand the "desire" as a kind of relation between the relevant belief or beliefs and the relevant action. One who is very skeptical about affects being actual body states in any significant sense might advocate the view that all or many such affects are just kinds of logical attributions. There are measurable occurrent states that seem to correspond to instances of desirelike states (though it is dubious that there is any generic motivational state like the philosopher's notion of desire), but I need not defend this claim here, since my goal is to develop a theory of some of the emotions—emotions for which it is uncontroversial that there are strongly related physiological and brain states. We need only note, then, that affect terms as they are used here will not be meant as mere logical relations between belief and action or between any other mental states or actions; what they stand for must necessarily include actual (in principle measurable) body states that are best identified as states resulting from or constituting the affect.
Distinguishing Features of Affects
Ultimately, we shall do best to fix a theory of basic emotions and other affects on a developed scientific understanding of the neural systems that
end p.9
 
enable those affects. Thus, our best criteria to identify affects will include those such as Jaak Panksepp (1998) uses. He writes: "The most compelling evidence for the existence of such systems is our ability to evoke discrete emotional behaviors and states using localized electrical and chemical stimulation of the brain" (52). Such evidence often reveals quite definite neural structures, some of which offer very compelling neuroanatomical evidence in favor of the affect program theory. I shall refer to some of the relevant neuroscientific evidence throughout my discussions of the basic emotions and other affects. However, since my task here is in part to relate the affect program theory to commonsense notions of emotion, including the kinds of features that have traditionally come in for much conceptual analysis and therefore have been of concern to philosophers, I will begin with a number of observable or introspectable features; these features are also a good starting place because some of them are likely essentially linked to the functional role of the relevant affects.
Such possible distinguishing features of occurrent affects that have interested scientists and philosophers are their physiological state, conscious experience, associated actions, and relations to cognitive content.7 We might also add to this list the relative temporal duration of the affect: Generally, it seems that affects that are not moods or emotions do not last as long as emotions, and that emotions last less long than moods. One might hold that two affects can be indistinguishable as to their physiology, but can be distinguished according to duration (sadness and depression, for example, might be such a case). There is a significant body of literature on stress that is concerned with duration of some affects. For my purposes here, however, this research will not be taken to be sufficient to characterize the emotions.8 Here I will remain agnostic about all the possible meanings of differences in duration. Instead, I will turn next to the first three of these four features. Since in the next chapter I will discuss the cognitivist theories of emotion (the view that emotions are in some part constituted by, or at least require, beliefs or other propositional attitudes), I will leave a discussion about cognitive content for that chapter.
Physiological State
Affects, especially some emotions, have noticeable and measurable physiological correlates. For example, a large body of research reveals that some forms of decision making (and thus, presumably, very basic forms of affects) result in often very subtle autonomic changes measurable by electrodermal recordings of skin conductance (e.g., Damasio, 1994). For emotions, many more measurable physiological changes occur. Depending upon the intensity of the emotion, these can include changes in autonomic functions, such as heart rate, blood pressure, respiration, sweating, trembling, and other features; hormonal changes; changes in body temperature; and of course changes in neural function as measurable by EEG (Frijda, 1986, pp. 124-175).
end p.10
 
For a long time, it has been controversial to suppose that some of these changes were distinct for particular emotions. It has often been seen as an important element of a cognitive theory of emotion to hold that the physiological changes accompanying an emotion amount to a kind of undifferentiated excitation, and that cognitive contents were needed to distinguish anger from fear, happiness from sadness, and so on (a source often cited in support of this view is Schachter & Singer, 1962).9 However, much of the previous failure to clearly establish distinguishing physiological profiles for emotions or other affects appear now to largely have arisen because of the inadequacy of past measuring techniques. Although the claim remains controversial, evidence is growing for the view that autonomic activity distinguishes among at least some emotions. Paul Ekman, Robert Levenson, and Wallace Friesen have found, for example, that discrimination of a number of emotions (fear, anger, sadness, disgust, surprise, and happiness) was possible just by observing temperature and heart rate changes from baseline measurements (that is, measurements of the subject when presumably not experiencing the emotion). Since these are measurements from a baseline, this study (1983) does not establish that we can actually identify one of these emotions in a subject on first observation, but it does at least show that we can distinguish the emotion from some others when several measurements are available. These experiments were done with actors, but later found to work with normal subjects (Levenson, Ekman, & Friesen 1990). They also worked not only for directed facial action (asking subjects to form the expression of an emotion) but for reliving (that is, recalling, thinking through) an emotional experience; and results from many other researchers is consistent with these findings (see Levenson 1992 and 1994 for a review). More research is needed in this area as some outstanding questions remain,10 and the experimental difficulties are great (generating fear, sadness, joy and so on in laboratory conditions is not easy), but these results are substantial and encouraging: they show that a significant number of the emotions may be distinguishable from each other by these autonomic features alone.
These results do not yet allow us to identify emotions by their physiological effects or constituents. But these kinds of investigations at least provide compelling evidence that there are reliable physiological changes that accompany some affects. For the emotions that we will be concerned with here, there is sufficient evidence that these affects necessarily include physiological responses such as changes in temperature, heart rate, and other features—even when the subject is having a relatively weak emotional experience, and even when the subject may be unaware of any such changes. Many cognitivists will deny that emotions necessarily have these correlates. In such a case, we can just be disagreeing about the semantics of our terms: these cognitivists take emotions to be mental contents, perhaps social relations, and these other features are incidental. But, as I will show in the next chapter, such a position is inconsistent with the scientific evidence, and it leaves us unable to distinguish emotions from other kinds of
end p.11
 
mental states. The claim that measurable physiological changes are necessary—leaving open whether they are sufficient to identify the relevant emotions—is important because such changes are sufficient to distinguish emotions from some other states with which some like to conflate emotion, such as belief. Furthermore, the autonomic patterns and related physiological changes are surely part of the phenomenal experience of some emotions. And these physiological responses are probably also essentially connected to relational actions and other affective behaviors. At the very least, we must explain or take into consideration these physiological features if we are to have a satisfactory theory of emotions.
If the physiological changes accompanying an emotion are necessary but perhaps not sufficient to identify that emotion, we must turn next to the three features of conscious experience, associated actions, and relations to content in order to get a more complete understanding.
Conscious Experience of Affects
Affects like anger, fear, despair, pleasure, and many others can have distinct conscious experiences. It might then seem that affects all are necessarily accompanied by a conscious experience; and many scientists and philosophers assert that emotions must be conscious. There is ambiguity in the term conscious here, one that has recently come under much analysis by philosophers (I will return to this in chapter 9). However, in this section I am concerned with the notions of consciousness that scientists tend to use; intuitively, a process is conscious if the subject is aware of it, in some sense reflecting upon it, and can use that awareness in directing or performing some action. I will call this sense of consciousness working consciousness whenever there is a threat of ambiguity.11 What it is to be aware of a state is not clear, and there certainly are mental states of which the subject is not aware but which influence working-conscious action. This lack of clarity alone casts grave doubts upon the idea that we can gain any definitive understanding of emotions by asserting that they are conscious, or by otherwise finding a role for consciousness in them. Thus, in order to try to ground my discussion of consciousness and emotions, I will have to find some criteria for something's being conscious. One sign of working consciousness is that the agent can, barring any deficiencies (such as brain damage that makes speech impossible, etc.), report on the state. This criterion is too-strong, and it does not get to what the notion of working consciousness seems to be aiming for (that is, I grant that the ability to report on a state is not the same as being aware of it). However, it is at least relatively clear. Furthermore, it comes close to capturing what I believe is really motivating many who insist that emotions must be working conscious: a notion that emotions play a part in our rational and deliberative control of our activities. So for the sake of clarity, I shall use in this section the very strong criterion that a process is working conscious if a subject can report upon that process (I am leaving vague what counts as a report; this should
end p.12
 
be acceptable since the examples discussed below are clearly instances of inability to give different kinds of report).
If we are to retain the idea that motivation is the quintessential feature of affects, then not all affects are working conscious. Instead, we have strong evidence that there are affective states which are effectively motivating but of which the agent is not able to report—not even indirectly. One of the most interesting examples is found in the mere exposure effect, as primarily championed by R. B. Zajonc. Much research has established that people tend to prefer familiar stimuli, even when they fail to properly recognize those stimuli (see Zajonc 1968, 1980). What Zajonc and his colleague found was that subjects could form preferences for certain stimuli to which they were subjected for extremely short durations (e.g., tens of milliseconds), making it extremely unlikely that they performed the kind of complex cognitive processing necessary for categorizing and memorizing the stimuli for later recognition of a kind sufficient for a declarative report (Kunst-Wilson and Zajonc 1980). When subjects were shown pictures of shapes, then were asked to pick out, among a number of shapes, those they had seen before, they chose randomly. When asked to choose the shapes they preferred, however, they showed a significant preference for the stimuli to which they had first been exposed. Mere exposure effects on preferences can also be shown for some nonhuman mammals (Hill 1978).
An everyday example with similar import might be the use of polygraph machines, the so-called lie detectors. Lie detectors measure skin conductance response, which changes as a result of activity in sweat glands and which appears to be well correlated with other physiological changes. As we have noted, physiological body activity, including autonomic activity, is one of the distinctive features of at least some affects. What is interesting is that normal subjects show measurable galvanic skin conductance changes to certain kinds of plausibly affective situations—such as to a situation in which they want to deceive and be undetected and are, perhaps as a result of their awareness of their potential loss from being detected, experiencing some affective reaction of which they need not be conscious nor over which they have any conscious control.
But one might argue that the affective states seen in mere exposure and in subtle skin conductance differences are not emotions, and that although some affects can fail to be conscious, emotions are always conscious. However, it seems possible that emotions are capable of being unconscious. First, an emotion might be unconscious in the sense that one has an impaired ability to understand or describe the emotion. Such seems to occur in some cases of alexithymia (Sifneos 1972). Alexithymics show an impairment in both the verbal and nonverbal recognition of emotions (Lane et al. 1996), and this impairment can extend to their own emotions. Sometimes these subjects report that they are experiencing some kind of an emotion, show some of the stereotypical behavior of an emotion (e.g., weeping), but are unable to say what caused the behavior (Nemiah and Sifneos 1970) or to properly categorize it. There is also some evidence that some alexithymics
end p.13
 
can have abnormally large autonomic responses to emotion-generating stimuli (Martin and Pihl 1985; Papciak, Feuerstein, and Spiegel 1985). The best explanations of alexithymia are of the form that an emotion is occurring, but that the individual is not properly aware of it (perhaps in a way analogous to blindsight; see Lane et al. 1997) or is unable to properly categorize it (perhaps because of a failure to have developed a cognitive skill to recognize emotions; see Lane and Schwartz 1987).12 If by an emotion being conscious we mean that the subject can recognize and properly categorize the occurrence of an emotion in herself, then the alexithymic subject has an unconscious emotion.
Second, there is evidence supporting the hypothesis that some phobias arise because some individuals have an inherited predisposition to fear certain stimuli (including concrete objects), and that this predisposition allows for fear reactions that are unconscious (see Seligman 1971). Thus, results similar to the mere exposure effect have been found for fear by Arne Öhman (1988) and his colleagues (Öhman, Dimberg, and Esteves 1989; Öhman and Soares 1993, 1994). In these experiments, subjects have demonstrated skin conductance responses for fear-conditioned stimuli that are presented so quickly, and with masking, that they are not consciously recognized. For example, in Öhman and Soares 1993, subjects were tested with fear-relevant images (snakes and spiders), along with neutral images (flowers or mushrooms), and some of each were used in fear conditioning; following the conditioning, these stimuli were shown for short durations and followed with a mask (a neutral stimulus which interferes with any conscious memory of the initial stimulus); skin conductance responses were then shown to be strong only for the conditioned fearful stimuli. This strongly suggests that unconscious processing of some kind is sufficient to cause fear responses. These results were also shown to be independent of where in the visual field the stimulus was presented, which is consistent with the process being subcortical since no lateralization (as occurs with many cognitive, cortical processes) of the ability is observed. Similar results were found with images of angry or happy faces, using aversive conditioning only for the former (Esteves, Dimberg, and Öhman 1994) (attention can have effects on these results, but the subjects are not conscious of
Fear Conditioning
In fear conditioning, a conditioned stimulus (for example, a sound) is paired with an unconditioned stimulus (such as a painful shock), and after some short training period, fear reactions (such as autonomic changes, and a startle) are shown for the conditioned stimulus alone. In most organisms, once fear conditioning has occurred, the response can be suppressed but apparently never unlearned; it can spontaneously reappear, or reappear under stress, or be retrained in significantly less time.
end p.14
 
the stimuli in that they cannot identify, even in forced-choice tests, the stimuli after exposure). These results provide strong evidence for at least fear conditioning and for fearful or phobic responses occurring unconsciously.
Third, there is a significant body of psychoanalytic literature dedicated to the idea that emotions can be unconscious, and that they can still play an important role in shaping actions by, and in the psychoanalytic explanation of the behavior of, the subject. It is unclear, however, to what degree and in what senses these emotions are unconscious. Are they dispositions to emotions, which lead to occasional occurrent emotions of which the subject is unaware? Are they, as Freud apparently held, not emotions but emotionally relevant unconscious beliefs? Or is it that the subject sometimes has occurrent emotions and it is the cause of these that he or she is unaware of? It will not be my place to try to answer these questions here, but only to note that some of these kinds of explanations require that unconscious emotions be possible. If any proves to be a powerful explanatory tool, that is reason to posit unconscious emotions.
Fourth, for some theorists, the underlying notion of an emotion being working conscious seems to be that we somehow know why we are having the emotion, that we are aware not only of the emotion but also of its cause or at least its object (and, on some views, its cause should be its object). We can refine our criterion in such a case to include that the subjects can report not only that he is having an emotion, but also why he is having it; or at least that when having an emotion the subject is aware of the object and cause of it. If this is required for one's notion of what it is for an emotion to be conscious, then the view that emotions can be unconscious has some valuable supporting evidence to be found in neuropsychology. Working with split-brain patients in the 1970s, Michael Gazzaniga and Joseph LeDoux were able to show a very clear sense in which emotions were not, in this sense, conscious. These split-brain patients are people who have had a commissurotomy, a surgical procedure to cut the commissure, a bundle of nerves that connects the two neocortical hemispheres of the brain. This procedure is used as a last resort treatment for some forms of epilepsy. But neuropsychologists have long known that the two hemispheres of the brain have specialized functions. What happens if you separate one of the primary links between them? Gazzaniga was able to show that subtle deficits can be revealed under controlled conditions. A stimulus can be shown to one side of the brain, leaving the other side of the brain in some senses unaware. For example, the language centers of most people are in the left hemisphere. Showing a figure just to the right side of the brain (done by placing it only in the left-hand side of the field of vision) can result in the object being (in some senses, defined in the respective experiments) recognized, but with the split-brain subject being unable to say what the thing is. LeDoux and Gazzaniga used this same approach to study affects. They could show the right side of the brain an affective stimulus. Though the subject was unable to verbally identify the stimulus, the affective import of
end p.15
 
the stimulus seemed to somehow "leak" to the left neocortex. The subject, wholly unconscious of what the stimulus had been (in the sense of being unable to report on it), could properly categorize it under some value terms as "bad" and "good." This at least shows that affective characterizations or related value judgments can be made in a way that is unavailable for report. In one case, for example, a word was shown to the right hemisphere and the subject was told to perform the action described by the word. Gazzaniga and Le Doux report:
His reaction to the word kiss proved revealing. Although the left hemisphere of this adolescent boy did not see the word, immediately after kiss was exposed to the mute right hemisphere, the left blurted out, "Hey, no way, no way. You've got to be kidding." When asked what it was that he was not going to do, he was unable to tell us. Later, we presented kiss to the left hemisphere and a similar response occurred: "No way. I'm not going to kiss you guys." However, this time the speaking half-brain knew what the word was. In both instances, the command kiss elicited an emotional reaction that was detected by the verbal system of the left hemisphere, and the overt verbal response of the left hemisphere was basically the same, regardless of whether the command was presented to the right or left half-brain. (1978, 151)
The researchers conclude that this result "is inconsistent with the currently accepted cognitive theory of emotion" because "the left hemisphere appeared to have experienced a directionally specific emotion in the absence of a cognition" (152). That is, the affective reaction was significantly directed—it resulted in or was a withdrawal from a suggested course of action—and the subject is clearly aware of something. However, the subject is not aware of the affect in a way that enables him to identify its cause or object; it would appear, at least, that the kind of conscious awareness that a cognitive theory of emotion requires was not present. Using my terminology, the behavior here is not necessarily revealing a basic emotion: it may require only what we are calling "affect." But it does at least show that some strong affective reactions, plausibly related to emotions, are unconscious in this sense. The next case is more relevant to emotions.
Gazzaniga also found spontaneous emotional reactions of laughter unavailable to report. In the following passage, "the machine" is the apparatus used to ensure that visual stimuli are seen only in one side of the visual field and thus get only to the opposite hemisphere:
When a pin-up was flashed without warning to the right hemisphere of [the subject], amongst a series of more routine stimuli, she first said, upon being asked by the examiner, that she saw nothing, but then broke into a hearty grin and chuckle. When queried as to what was funny, she said that she didn't know, that the "machine was funny, or something." When the picture was flashed at the left hemisphere she laughed too,
end p.16
 
and quickly reported the picture as being a nude woman. Using a different modality (olfaction), Gordon and Sperry (1968) recently confirmed this kind of result.
Neither hemisphere in [another subject] found the nude overtly funny (he was 51 at the time of testing), but did find other testing situations humorous. In one test of tactile learning capacity, using the left hand, [this subject] broke out laughing when feeling one member of the stimulus pair. The particular stimulus consisted of a tack nailed into the middle of a wooden square block. Every time he felt it, he would pick it up and twirl the block about the axis and would chuckle heartily when doing so. When asked what was funny he would say, "I don't know, something in my left hand I guess." (1970, 105-106)
If laughter is properly an expression of an emotion, then that emotion is occurrent but unavailable to the relevant kind of introspection for these patients. Note that I do not endorse, and these observations do not require us to conclude, that emotions are cortical (that is, that the neural centers that underlie their function are in the neocortex) and lateralized (that is, that the underlying neocortical center is specific to one side of the brain); rather, for my concerns here the point is that the kinds of capabilities that constitute working consciousness in this stronger sense, or at least that offer criteria for its existence, are cortical and lateralized phenomena. These split-brain studies show failures in working consciousness that differentiate some of its features from emotions or other affects.
Defining affects in terms of their conscious role is therefore unlikely to be a strategy that succeeds well in identifying them or otherwise helping us to understand them, since some of them can be unconscious and still have behavioral effects other than conscious reports. Nonetheless, given that some affects are sometimes not working conscious, it remains that certain affects seem distinguishable from other affects by the nature of their phenomenal experience when there is such a working-conscious awareness of the experience. This is particularly true of the common emotions: rage, joy, sadness, fear, and shame—to pick just a few examples—seem to have feelings (when they are conscious) specific to the emotion (or, at least, specific enough to distinguish the emotion from other kinds of states, like belief), so that we may find it convenient to use their conscious experience as one of their distinguishing features. Should we then define some affects, such as emotions, in terms of their phenomenal experience? This strategy has several stumbling blocks. There is the problem, already observed, that some affects can be unconscious. But, supposedly the suppressed affect has effects on the subject, and these are usually the kind of effects one would expect of that affect. If we are able to identify unconscious occurrent anger with a working-conscious instance of occurrent anger, or any suppressed affect with its working-conscious counterpart, then the common element must be something other than the phenomenal experience of the affect, since those properties per se are just freely spinning wheels without the
end p.17
 
working-conscious awareness of them. If there are unconscious emotions we thus cannot have as a defining feature of affects their phenomenal features alone. Another problem is that, like the "feeling theory of affect" which has long been in disrepute, treating emotions as conscious states characterized by one's awareness of the experience does not explain much. Ultimately, philosophers and scientists want to understand how emotions relate to behavior, and what role they play in our mental lives, and specifying how they "feel" does little to further this goal. A related problem is that reference to phenomenal experience does not give us any objective (that is, third-person) criteria with which to distinguish these emotions. But we certainly do properly recognize emotions in others, and if we are to study affects in a scientific way we will need some objective criteria with which to work.
The conscious experience of an affect, although important, cannot be a fundamental feature used to define emotions.
Associated Actions
Another approach to understanding and categorizing emotions is to look at the kinds of behaviors that they cause, or with which they are associated. This is not to say that emotion concepts are nothing but useful ways of grouping together disparate classes of behaviors; on the contrary, looking at emotional behavior has also provided evidence that many of them are highly associated with stereotypical, pancultural behaviors; and this in turn should be viewed as evidence that the behaviors are caused by biologically based, inheritable capabilities.
Some of the most compelling evidence for pancultural human emotions has come from studies of facial expression. It was Darwin who first argued at length that facial expressions of emotions are evolved emotion-expressing behaviors. In more recent times strong evidence has been gathered that Darwin was correct. Irenäus Eibl-Eibesfeldt studied the facial expressions of children born deaf and blind, some with extensive brain damage. He discovered that these children showed spontaneous signs of emotions such as smiling when playing or sitting in the sun, laughing when playing, and crying when in an unfamiliar environment (1973; see also Fulcher 1942). Some of these children had severe cognitive deficits, and none was able to see or hear the emotional expressions of others, so it is highly implausible that they learned these behaviors.
Cross-cultural studies of facial expressions have found evidence of high cross-cultural correlations. These kinds of studies were made in a thorough manner by Ekman and colleagues (Ekman, Sorenson, and Friesen 1969), who sought to get as pure a cross-cultural study as was possible. They created a set of thirty photographs of facial expressions that they felt expressed six emotions that other psychologists had proposed as basic (happiness, surprise, fear, anger, disgust, and sadness). They then showed the photographs to college students in the United States, Brazil, and Japan, and
end p.18
 
to volunteers in New Guinea and Borneo. The six emotion terms were translated into the appropriate languages and then the subjects were asked to group the pictures under the terms. A very significant degree of agreement was usually found—higher for some emotions than others, and for some cultures than for others, but in general there was an unmistakably significant degree of agreement. Ekman and Friesen re-created this experiment (1971) working with the Fore of New Guinea, a cultural group relatively isolated from the rest of the world, and found agreements again ranging from 64 percent (for fear) to 92 percent (for happiness). This work and related work (Izard 1971) support the view that human facial expression of some emotions is pancultural.
These results have some interesting supporting evidence in neuroanatomy. There are two distinct neural pathways that control facial movements. One is through the pyramidal tract, and the other through the phylogenetically older extrapyramidal tract. It seems that emotional facial expressions are controlled by the older, extrapyramidal tract. This is evident when damage to the motor cortex that impairs motor control of the face (as often occurs in hemiparalysis) is sometimes (when the damage is localized to the motor control area) spontaneously overcome in the expression of emotion by the unfortunate subject. For example, a stroke victim might be unable to smile on the paralyzed side of the face when so commanded, but might smile involuntarily and normally at a joke. Conversely, damage to the extrapyramidal tract could leave voluntary control intact but result in the loss of all spontaneous emotional facial expression (Rinn 1984).
These findings suggest that emotional facial expression is pancultural because of inheritable, evolved neural structures that are shared by all, or at least many, human beings. There is also interesting evidence available for this view in studies of nonhuman primates. Research by R. E. Miller, W. F. Caul, and I. R. Mirsky has shown that the facial expressions of rhesus monkeys can transmit significant information to other monkeys, and though monkeys raised in isolation do not recognize the meaning of the facial cues of other monkeys as readily as do the monkeys raised in a social setting, these isolated monkeys show facial affective cues that other monkeys recognize and properly understand (1971). This research is consistent with the view that our near evolutionary cousins share with us the having of innate facial expressions of affect, and that the innate expressions are therefore highly likely to have evolved in a common ancestor.13
These results all find surprising support in some of the studies by Ekman of facial expressions among Japanese and American college students (1980). In the experiments, each student was left alone to watch films, some of which were stressful, and some of which were not. Their facial expressions were recorded, and these recordings measured by researchers who did not know what films the subjects were watching. When the students were alone, both Japanese and American students showed significantly similar facial expressions. In some cases, however, someone dressed in a
end p.19
 
lab coat was put into the room with the subjects. In these cases, as expected, Japanese students altered their expressions much more, smiling more and showing less stress. This is consistent with facial expressions being pancultural but amenable to different display rules. Even when researchers were in the room with the subjects, the pancultural element was not wholly eliminated: "Examining these videotapes in slow motion it was possible to observe sometimes the actual sequencing in which one movement (a smile, for example) would be superimposed over another muscle action (such as a nose wrinkle, or lower lip depressor)" (94). In other words, the evidence suggests that the pancultural facial expression is being generated but then promptly suppressed. Note that this is also very suggestive of a two-track view of these emotional expressions: a potentially subcognitive emotion causes the facial expression, perhaps primarily through the extrapyramidal tract, and a slower, secondary, cognitive appraisal suppresses it.
This is consistent with the use of surface electromyographic recordings (EMGs) in studies of emotion (see Tassinary and Cacioppo 1992). Such recordings measure muscle action potentials in, for example, the face—that is, neural stimulation of facial muscles. They can detect these muscle action potentials even if they fail to result in any change in facial expression, for example if they are too weak or too short in duration to cause a muscular action. This technology provides a tool for psychophysiologists to measure facial reactions to emotion-eliciting stimuli even when no observable change in facial expression occurs. The underlying method is guided by the belief that emotions can cause muscle action potentials that are not under conscious control or awareness of the subject.
Emotional behaviors are much more than just facial expressions. Perhaps one of the most compelling accounts of the use of emotion concepts is found in D. O. Hebb's classic 1946 article on the recognition of emotion. Hebb reviews an experiment at a primate laboratory where for two years the scientists working with the primates were not allowed to use emotion terms to describe the animals' behaviors. Instead, they had to keep records which described only what the animals did at one time or another. What Hebb discovered is that describing different chimpanzees without using emotion terms left people unable to really convey the sense of the character of the different primates. One could not tell, just from looking over the records of past events—described painstakingly without "anthropomorphicterms"—what the animal was like or how it would behave:
All that resulted was an almost endless series of specific acts in which no order or meaning could be found. On the other hand, by the use of frankly anthropomorphic concepts of emotion and attitude one could quickly and easily describe the peculiarities of the individual animals, and with this information a newcomer to the staff could handle the animals as he could not safely otherwise. Whatever the anthropomorphic terminology may seem to imply about conscious states in the chimpanzee, it provides an intelligible and practical guide to behavior.
end p.20
 
The objective categorization therefore missed something in the behavior of the chimpanzees that the ill-defined categories of emotion and the like did not—some order, or relationship between the isolated acts that is essential to comprehension of the behavior. (88)
A pragmatist should be satisfied on this observation alone that emotions are genuine scientific entities. Someone of a more realistic bent might rightly argue that Hebb's conclusion is true because some emotions lead to, or are in some way linked to, actions that are specific to and explicable by these emotions.
Hebb's observations should remind us of the strategy of the ethologist. The ethologist looks to find patterns of behavior in animals. If there are patterns that occur again and again, and if these patterns can be found in isolated groups and even in closely related but different species, then this is some evidence for a homologous behavior. The ethologist is not therefore much distinct from the evolutionary biologist, utilizing the concept of homology for behaviors as well as for anatomical structures (where homologous behaviors would presumably arise from, and ultimately be explained by reference to, homologous structures). The ethologist's method applies to humans as well (see Eibl-Eibesfeldt 1989). Evidence that some emotional expressions are pancultural, that the structures allowing for the expression are inheritable, and that certain patterns of reoccurring behavior are inexplicable (not regularly predictable) without emotion concepts all point toward the primary thesis that some emotions can be identified via their homologous associated actions.
This makes sense of the presence of emotions in other nonhuman animals. Our primary means of recognizing fear in a rat, anger in a dog, surprise in a cat, and so on, is through the behaviors that they show in such states. Scientists regularly use these criteria (and others, such as autonomic responses) to study emotions in nonhuman animals. It is difficult to see how else we are going to understand these claims except through the identification of shared kinds of behavior.
Some cognitivists about emotion have argued that observations of behavior fail to provide any evidence for emotions in nonhuman animals, and therefore fail to support theories like the affect program theory. Ortony, Clore, and Collins have claimed:
It is tempting to suppose that animals experience fear. However, such attributions are typically based on observations of behaviors (aggressive behavior or avoidance behavior), which turn out to be dissociated from the emotional states to which they are presumed to be linked. . . . It would be a relatively straightforward matter to program a robot to exhibit aggressive or avoidance behavior toward certain objects or classes of objects, yet, if having done so one were to claim that one had produced the emotions of anger or fear in the machine, one would be scoffed at by the scientific community, and rightly so. (1988: 27-28)
end p.21
 
There are at least two errors in this argument. First, it is not the case that we posit that there are emotions in nonhuman animals just because we observe simple behaviors. An ethologist who sees a bird leaving a branch to fly to another is not about to claim it fled the oak in fear in order to attack a maple out of anger. We posit that there are emotions in nonhuman animals because it is the best explanation for a very large body of evidence. This evidence includes, but is not limited to, behavioral patterns which are in particular ways both flexible and inflexible, so that they reveal the pursuit of a kind of action (see chapter 3); which are observed again and again; which can be best described as fulfilling the functions that we suppose in our theories that these emotions fulfill, or even that we ascertain the emotions in ourselves fulfill; which can reliably be described as being elicited by conditions consistent with that function; which are reliably accompanied by expressive behaviors; which include autonomic and other physiological changes that are special to the emotion and perhaps even shared by us; and which (most important!) in some animals are caused or constituted by neural structures which have homologs in the human neuroanatomy of emotion. Thus, behavior is a crucial element, but it does not stand alone.
Second, to say that it is a "straightforward matter to program a robot to exhibit aggressive or avoidance behavior toward certain objects or classes of objects" is a patent falsehood. It is a major accomplishment to get a robot to navigate a small, unchanging, and extremely simplified environment. To get a robot to actually recognize, effectively track, and pursue a resistant (say, a moving or even fleeing) object so that the robot could effectively attack it is truly the kind of engineering beyond, or at least at the very limit of, contemporary AI and robotics. Of course, it is a straightforward matter to program a robot that on a flat surface in a featureless environment moves toward or away from a light, for example (the kind of "behaviors" sometimes referred to as "Braitenberg behaviors"; see Braitenberg 1984). But this cannot be the kind of thing that is meant by "aggressive or avoidance behavior," because the very thing at issue here is the attribution to nonhuman animals of emotions had by humans; and so no respectable argument for the presence of these emotions in nonhuman animals would depend on counting such simplistic "behaviors" as examples of aggression or avoidance (and, as noted, the attack or avoidance behaviors of most animals are extremely sophisticated). Furthermore, this kind of reasoning may well be backward; to program a robot that can exhibit effective behaviors like aggression and avoidance with the kind of flexibility that even a relatively simpler animal (such as an insect) reveals—to actually engineer an autonomous agent—may best be accomplished by working with a robust theory of affects and then attempting to engineer an affective agent (I shall indulge in some speculations in this direction in chapter 12). Finally, to suppose that it is a simple matter to program these behaviors may be an instance of a common fallacy—what I will call the cognitive autonomy fallacy—that what is not cognitive is simple and inflexible, while what is cognitive is
end p.22
 
complex and flexible and the wellspring of autonomy. I will return to this point several times.
The view that some emotions can be identified through the actions with which they are associated is perhaps merely a consequence of my definition: since affects are motivations, then the principal method we have for discerning and distinguishing them is through the behavior they motivate. We can always keep in mind, however, a realist (as a philosopher would call it) criterion: when we identify an affect, we are identifying a genuine physical state of an individual organism, and if it later turns out that there is no such significant (that is, measurable) state, or the behavior was best explained in some other way, then we were wrong to so identify the state. In the cases of things like preferences, the motivation is very general (let us assume, for a moment, that there is a state corresponding to "preference"). If a subject S prefers to do some action A, then we are saying little more than that S is in a motivational state which has as an effect that she will A, ceteris paribus (when it is possible, when she is not constrained, and when there is no stronger motivation to do something inconsistent with A). But other affective states are much more structured. We can understand fear by supposing that if subject S fears some object O, then S will flee from O—with the same ceteris paribus clause. Some emotions, it seems, are characterized specifically by the complex behavior that they have as a consequence—what psychologists sometimes call "relational actions," since they are explicitly concerned with relations to other things (Frijda 1986, 14-24).
The Affect Program Theory
Some of the things that we call emotions appear to be a collection of things: physiological responses, stereotypical actions, and perhaps even normal cognitive roles. Instead of reductively explaining these emotions in terms of one of these features, I will adopt the naturalistic theory that tries to respect all of them: the affect program theory. This theory is not favored by philosophers or by psychologists who work on the social end of their discipline, but in various forms it is quite common to psychobiologists, neuropsychologists, and others who concern themselves with the biology of emotion. I adapt the notion from Ekman, who took the term from Silvan Tomkins:
For there to be such complexity and organization in various response systems, there must be some central direction. The term affect program refers to a mechanism that stores the patterns for these complex organized responses, and which when set off directs their occurrence. . . . The organization of response systems dictated by the affect program has a genetic basis but is influenced also by experience. The skeletal, facial, vocal, autonomic, and central nervous system changes that occur
end p.23
 
initially and quickly for one or another emotion, we presume to be in largest part given, not acquired. (Ekman 1980; 82)
By "affect program," Ekman means to refer to only some aspects of the emotions in question. He argues that an emotion is made of an affect program along with a response system, an appraiser, and elicitors (86-87). In a sense, this is of course correct, and a weak form of cognitivism about emotions is tantamount to the view that all of these things are normally present in emotions but they need not all be. I will therefore here just use the term "affect program theory" to refer to the whole syndrome, recognizing that the cognitive elements are in humans quite common, but unnecessary, and that the physiological and behavioral consequences are necessary.
The idea of emotions as affect programs best explained by reference to our evolutionary heritage is perhaps most indebted to the research of Paul MacLean. MacLean introduced the "triune brain" hypothesis (1990), in which the brain is seen as having three systems, hierarchically arranged, each of which is to some degree independent of the others and which corresponds to a definite stage of evolutionary development. These systems are the "reptilian brain," the paleomammalian or limbic brain, and the neomammalian neocortex. On this model, many affects are reptilian or limbic system adaptive programs that in humans can operate to varying degrees independently of our neocortical systems.
The neuroscientist Jaak Panksepp also offers a compelling approach to the basic emotions that is consistent with the affect program theory. He has offered six criteria that distinguish the basic emotional systems:
1.  
The underlying circuits are genetically predetermined and designed to respond unconditionally to stimuli arising from major life-challenging circumstances.
2.  
These circuits organize diverse behaviors by activating or inhibiting motor subroutines and concurrent autonomic-hormonal changes that have proved adaptive in the face of such life-challenging circumstances during the evolutionary history of the species.
3.  
Emotive circuits change the sensitivities of sensory systems that are relevant for the behavioral sequences that have been aroused.
4.  
Neural activity of emotive systems outlasts the precipitating circumstances.
5.  
Emotive circuits can come under the conditional control of emotionally neutral environmental stimuli.
6.  
Emotive circuits have reciprocal interactions with the brain mechanisms that elaborate higher decision-making processes and consciousness. (1998, 49)
What these various approaches share is a common recognition that some emotions are complex, coordinated events that include motor programs or
end p.24
 
subroutines, that evolved and are recognizable in homologous form in related organisms, and that are fundamentally enabled in neural circuits. For my purposes here, one of the most fruitful features of the basic emotions, as understood in the affect program theory, is the action or motor programs that in part constitute some of them.
The Central Role of Action and the Parsimony of the Affect Program Theory
The linking of emotions to actions is widely accepted. Nico H. Frijda writes, "Emotions are changes in readiness for action as such . . . or changes in cognitive readiness . . . or changes in readiness for modifying or establishing relationships with the environment . . . or changes in readiness for specific concern-satisfying activities" (1986, 466). More strongly, he says: "It will be clear that 'action tendency' and 'emotion' are one and the same thing" (71). The psychobiologist Robert Plutchik has argued that "an emotion is a patterned bodily reaction of either protection, destruction, reproduction, deprivation, incorporation, rejection, exploration or orientation, or some combination of these, which is brought about by a stimulus" (1980, 12). More recently, he added that "emotions are complex chains of events with stabilizing loops that tend to produce some kind of behavioral homeostasis. . . . [The] physiological changes [that accompany an emotion] have the character of anticipatory reactions associated with various types of exertions or impulses, such as the urge to explore, to attack, to retreat, or to mate" (1994, 100). So that
From an evolutionary point of view one can conceptualize emotions as certain types of adaptive behaviors that can be identified in lower animals as well as in human. These adaptive patterns have evolved to deal with basic survival issues in all organisms, such as dealing with prey and predator, potential mate and stranger, nourishing objects and toxins. Such patterns involve approach or avoidance reactions, fight and flight reactions, attachment and loss reactions, and riddance or ejection reactions. (229)
Silvan Tomkins claims that emotions are innately patterned responses and that these affect programs are stored in subcortical brain centers (1962, 1963). Richard Lazarus argues that emotions result from primary appraisal of a situation, and a secondary appraisal results in a coping action (1991). And, as noted above, Panksepp advocates a psychobiological theory of some emotions in which they arise from neural circuits and enable adaptive behaviors; these neural circuits "are genetically hard-wired and designed to respond unconditionally to stimuli arising from major life-challenging circumstances" and they "organize behavior by activating or inhibiting classes of related actions (and concurrent autonomic/hormonal changes) that have proved adaptive in the face of those types of life-challenging
end p.25
 
circumstances during the evolutionary history of the species" (1982, 411). Howard Leventhal has presented a perceptual motor theory of emotions, in which "there is a basic set of stimulus-sensitive expressive-motor templates, each of which generates a different emotional experience and expressive-motor behavior" (1984, 127). I advocate, and will assume here, the hypothesis that basic emotions have as an essential element a motor program.
What is the motor program that is part of the affect program of some emotions? This is an empirical question, but here I can clarify the notion, draw some likely conclusions about its evolution, and warn off likely misunderstandings of the term "program." The program need only be functionally specified for my purposes, but it surely is (primarily) instantiated in a neural system. Once activated, this action program will, if not actively inhibited, result in the emotional behavior. Strictly speaking, the functional definition of the action program therefore has the action as a consequence—much as a functional definition of motor cortex activity, for example, can have motor activity as a consequence.14 Thus, on this view, given an occurrent basic emotion, it is not the emotional action but the common lack of it, or the modification of it, that requires additional theoretical posits. This is all consistent with the compelling working hypothesis that some emotions evolved from innate behavioral responses—that is, what ultimately amounts to motor programs—in ancestors of the emoting agent. The term "program" is perhaps unfortunate, but I use it because I know of no clear alternative. The motor program is not meant to be a simplistically deterministic list of discrete symbolic instructions, such as a computer program written in Java, for example. It is rather a dynamic capability. A rat running from a fearful stimulus might take a different path each time it flees—but it still may always consistently flee. Many brain systems are perhaps best thought of as dynamical systems (see Port and van Gelder 1995), and like many dynamical systems result in output that is most conveniently described in terms of a range of possible continuous trajectories moving through a state space—which, compared to a computer program, has the flavor of a kind of qualitative, as opposed to quantitative, description.
With this general notion of motor programs in place, the affect program theory yields a bonus of increased parsimony in our theorizing. As we saw, many theories of emotion (including some cognitive theories) share the supposition that an essential feature of emotions is that they have some kind of significant relation to action; the most widespread agreement is that the emotions are at least some kind of disposition or tendency. Although "disposition" takes on the sound of a substantial and well-placed primitive concept in much action theory, it is a mysterious entity and provides not a proper part of a theory but rather a debt to be discharged. Present understanding of the human mind and brain are not sufficient to expect a successful theory of all our disposition talk, and many or most of our disposition concepts and related concepts are merely placeholders for the possibility of the relevant action. However, I have suggested an inversion
end p.26
 
of the usual explanations: we should take the emotional action as primary, and either the failure to act, or the cognitive guidance of action, as secondary. Since we do have general theories of how inhibitions can work,15 and since cognition is already a problem, there is some theoretical gain in this approach. Every debt we can pay off is, after all, a net gain in our theoretical finances.
Evolution, Innateness, and Inheritability
The affect program theory will ultimately be verified and fully developed as the relevant neural systems are identified and understood. However, from a functional and from a psychoevolutionary perspective, the most distinguishing feature of an affect program is the behavior that, at least in part, constitutes it. Presumably, like the facial expressions that accompany and express some basic emotions, the more complex relational action patterns that characterize some basic emotions started as motor programs that evolved into inheritable patterns of behavior. As some of the species having these motor programs evolved ("toward" us, for example), some of these behaviors remained, although they became subject to alteration and inhibition via new capabilities that accrued to the species involved. In ourselves, these action programs can be occurrent—one might say, "running"—but result in diverse or even no overt behavior. Thus, the program that makes up an occurrent basic emotion, I claim, is in part the occurrence of the relevant behaviors (in the broad neuroscientific sense); and for at least some of the basic emotions, this includes some relational action. The relational action of a basic emotion is a consequence of the occurrent action program if the action program is not inhibited. Similarly, most other features of an affect program can also best be explained by reference to their role in the behavior of the emotion.
But I have been rough with the evolutionary claim about the affect program. This is partly because the conclusions I aim to draw in this book are largely independent of the variations that I gloss over. Thus, how "universal" the relevant affects are is a concern I hope to pass over in the interest of avoiding a set of important but distinct philosophical problems. For my purposes here, any significant portion of the relevant populations having some of these features is going to be sufficient. Thus, I will hold only that the basic emotions are biologically based capabilities (that is, the structures which allow them to occur can be described by a biological science—above all, neuroscience), that they are pancultural (that they arise in every culture, even if not in every individual), and that they are inheritable (the reason they occur in individuals in every culture is because some people inherit this capability). Maintaining only these presuppositions should allow me to avoid such issues as, for example, the degree to which the inheritability of the basic emotions is "innate" or a result of the inheritance of common environments. It is fair to say that no feature of our neuroanatomy is not shaped by learning, and I certainly would deny a claim that basic
end p.27