1. Introduction.
Accounting for nonhuman, animal emotion is sometimes claimed to be a desideratum of any theory of emotion (Deigh Reference Deigh and Solomon2004). Thus, if one theory can account for, or explain, the phenomena of animal emotion more readily than another, then that theory gains an element of superiority over the deficient theory. Two general types of competing theories of emotion are the affect program theory and the cognitive theory. The affect program theory (APT) is the view that emotions are “pancultural syndromes inherited by biological capabilities” (DeLancey Reference DeLancey2002, 3). Cognitive theories of emotion emphasize the role of cognitions in emotional episodes. A typical assumption is that, since APT encompasses a biological approach to characterizing emotions, APT can easily account for animal emotion while cognitive theories cannot.
Here I argue that APT does not adequately capture the phenomena of animal emotion. The paper proceeds as follows. First, I describe the main elements and commitments of APT as expressed by some of its leading proponents in philosophy (DeLancey Reference DeLancey2002; Griffiths Reference Griffiths1997) and psychology (Ekman Reference Ekman, Dalgleish and Power1999). Second, I argue that, contrary to the assumption of APT, the theory does not satisfactorily account for two important cases of animal emotion: aggression in chimpanzees and fear in piping plovers. The basic version of APT fails to pay sufficient attention to the complex connections of emotional episodes with cognitive processes. My point is not to argue for a cognitive theory. Rather, I want to suggest that cases of animal emotion provide rich examples of detail that can inform theories of emotion. Paying attention to animal emotions helps the emotion theorist avoid the dichotomy between the extreme version of APT and of cognitive theories of emotion. I conclude that a proper examination of animal affect reveals that a theory of emotion should be more complex than APT.
2. The Affect Program Theory.
APT is influenced by many areas of research. It contains elements of Darwin's (Reference Darwin[1872] 1998) focus on the continuity of the expression of emotion in animals and humans, the physiological emphasis of James (Reference James1884), the emphasis on neural systems in modern neuroscience and the facial display research conducted by anthropologists and psychologists such as Ekman (Reference Ekman and Ekman1998).
The targets of explanation, emotions, are distinguished from moods in part by their duration. Emotions are brief episodes on the order of seconds and minutes while moods may last for hours or even days. The definition of emotion provided by APT is also typically taken to apply only to what are referred to as ‘basic emotions’. While the lists of basic emotions vary, the commonly accepted ones are anger, fear, surprise, disgust, joy and sadness. While most versions of APT (e.g., Griffiths Reference Griffiths1997; DeLancey Reference DeLancey2002)Footnote 1 allow that there may be higher cognitive emotions in humans that do not involve affect programs, this will not affect my criticism of APT. The examples I consider, aggression (anger) and fear, are prototypical basic emotions and thus fall under the scope of APT.
While there are particular differences among theorists, the main idea of APT is that emotions are “pancultural syndromes enabled by inherited biological capabilities” (DeLancey Reference DeLancey2002, 3). Syndromes are meant to refer to collections of biological responses. These responses are complex, coordinated and automated (Griffiths Reference Griffiths1997, 77). The complexity of an emotion is displayed by each emotion-type involving distinct physiological responses, stereotypical behaviors and relational behaviors. The bodily responses are things like unique autonomic nervous system (ANS) reactions, particular central nervous system activation, and endocrine system changes. Stereotypical behaviors include facial expressions, vocal expressions, and musculoskeletal responses. Relational behaviors include behaviors such as fleeing or fighting. These responses are thought to be coordinated because they regularly co-occur and act in patterns and sequences (Griffiths Reference Griffiths1997, 77). Finally, the automaticity of this coordination is evidenced by the lack of conscious direction they require.
The syndromes are claimed to be pancultural because there is strong evidence that many of the stereotypical behaviors involved in emotional responses are distributed across cultures. The behaviors, or bodily movements, typically studied in this respect are facial expressions (Eibl-Eibesfeldt Reference Eibl-Eibesfeldt, Cranach and Vine1973; Ekman Reference Ekman1992), though there are some studies of the universality of ANS activity in non-Western cultures (Levenson Reference Levenson1992). Since the structures that enable the expressions of emotion are universal, and presumably inherited, it makes it likely that the expressions are as well. Since similar expressions are found in other primates (Miller et al. Reference Miller, Caul and Mirsky1971; Chevalier-Skolnikoff Reference Chevalier-Skolnikoff and Ekman1973; Preuschoft and van Hoof Reference Preuschoft and van Hoof1995; Parr et al. Reference Parr, Preuschoft, de Waal, Ladygina-Kohts and de Waal2002), it is plausible that our expressions, and the emotions they help to constitute, are products of evolution.
The notion of an affect program is what is intended to capture each distinct set of complex and coordinated responses for each emotion-type. Emotion-types are general families of emotions like anger, fear, and disgust. The family is defined by its particular affect program and three other components of an emotion: the elicitor, appraiser and the response system (Ekman Reference Ekman, Dalgleish and Power1999; DeLancey Reference DeLancey2002). The general process of an emotional state starts with some elicitor (e.g., a snake) being perceived and appraised as important (e.g., as a threat). This appraisal is conceived of as an unconscious, quick and dirty, process. The appraisal mechanism operates independently of what are typically referred to as cognitive processes and is heavily biased by fitness and survival contingencies of our ancestors (however, the mechanisms are malleable enough to respond to important features in the life of an individual); see Ekman Reference Ekman, Dalgleish and Power1999. In so doing, the mechanism makes fast appraisals that lead to organisms acting as if the information represented is true and important. After the appraisal mechanism is triggered, the appropriate affect program kicks into gear. This involves the coordination of all of the relevant bodily responses.Footnote 2 These responses are integrally tied to the response system which is the activator of the behaviors triggered by the affect program. While most emotions do involve an elicitor, appraisal, affect program and response system, the latter two are necessary for an emotional response while the former two are not (DeLancey Reference DeLancey2002, 24). The necessity of these two components emphasizes the centrality to APT of the emotion's connection with action. All variants of the APT view emotion as essentially a mechanism to act. In this mechanism, the affect program provides the main impetus in creating behavior through its inclusion of specific motor programs.
An element common to variants of APT, and implied by the types of processes that constitute an emotional response, is a conception of the emotional system as being one of a modular nature. Griffiths explicitly compares the processing in APT with Fodor's (Reference Fodor1983) description of modular cognitive systems (Griffiths Reference Griffiths1997, 93). The shared properties include (1) the passive, involuntary operation of the emotion processes; (2) the inability to become aware of the processes that lead to ones emotional responses; and (3) the informationally encapsulated nature of the emotion system in APT. This last feature is the claim that the information in the emotional system is separated from information in other cognitive systems. The emotion system cannot take advantage of information in other systems and cannot be influenced by that distant information source. This explains how, for example, one can be afraid of something like a spider that one knows to be harmless. The modularity of APT is apparent in the quick appraisals triggering emotional responses and in the involuntary, stereotypical behaviors that result from an emotional appraisal.
In sum, the model of emotion according to APT is the following, as expressed by Griffiths.
The affect program system creates brief, highly stereotyped emotional reactions. It has only limited involvement with the cognitive processes which control longer-term action. A stimulus appraisal which initiates an affect program reaction is to a large extent informationally encapsulated. The subsequent complex set of actions unfolds automatically, and it is difficult to interfere with these actions voluntarily. (Griffiths Reference Griffiths1997, 100)
While these features are what characterize emotional responses according to APT, they are the same features that ultimately prevent APT from being able to adequately account for the complexity of some animal emotional behaviors.
3. Trouble Cases for APT.
The appeal to emotions in animals plays at least two roles in establishing the affect program theory of emotion. One, it serves to articulate certain features of the affect programs that constitute the basic emotions, and two, it serves to support APT by both acting as a reason to reject its chief rival, the cognitive theory of emotions, and as a reason to think that there are such things as affect programs. The importance of exploring APT's ability to account for emotions in animals rests in challenging the strength, or completeness, of APT as a theory of basic emotions. According to at least one proponent of APT, DeLancey (Reference DeLancey2002), the strength of a theory of emotion lies in its ability to adequately define emotions and account for emotional phenomena. If APT fails at this for animals, then pointing to their emotions does little in establishing the theory and does not go far in articulating what the components of emotional states really are.
In this section, I present two different cases of animal emotional behavior. If the phenomenon of animal emotion supports APT, in both justifying it and articulating it, then it should provide a satisfactory account of animal emotion. If it fails to do so, then what is commonly taken for granted, viz. that APT's strong point is its ability to explain the phenomena of animal emotional behavior, is mistaken. I argue that APT does not provide a satisfactory account of animal emotion. Because of this, APT loses part of its justification and articulation as a theory of emotion. Furthermore, the examples of animal emotions suggest that a theory of emotion ought to provide for both simplicity and complexity in natural emotional episodes. The first case involves chimpanzees. Since APT focuses on homologous emotional responses, it should be able to deal with examples of emotion in primates, especially ones as closely related to humans as chimpanzees. The second case involves piping plovers. These birds are much more distantly related and so APT might be less applicable in this case. However, it is often argued that affect programs are evolutionarily ancient and are enabled by very old subcortical structures in the brain. These structures are now thought to be homologous to structures in the avian brain (Jarvis et al. Reference Jarvis2005), so accounting for plover emotional behavior should be within the domain of APT.
3.1. Chimpanzee Aggression.
Animals that live in social groups are prime candidates for episodes of aggression. In chimpanzee culture, outbreaks of aggression are not uncommon. APT appears perfectly suited for these emotional behaviors. They involve distinct displays such as the bulging-lips face, raised hair, and upright walking. Chimps are also known to use facial expressions of emotional information to guide their own behavior (Russell et al. Reference Russell, Bard and Adamson1997; Parr et al. Reference Parr, Preuschoft, de Waal, Ladygina-Kohts and de Waal2002). Further, studies with some primates indicate that there are physiological changes (e.g., heart rate) when around dominant individuals that are nonexistent around subordinates and kin. Interestingly, reconciliation behaviors bring increased physiological measures back to normal rates (Aureli et al. Reference Aureli, Preston and de Waal1999). While these factors in aggression match up well with APT, there are other important elements involved in chimpanzee aggression that do not.
Aggression is deeply embedded in a social context for primates. It is intimately connected to dominance relations and thus arises in many particular situations. Some of these, for chimps and capuchins, include failures to observe practices of sharing food, inappropriate mating behavior, violating the proper order in which group members eat and the violation of territorial behaviors (de Waal Reference De Waal1996). Many of the occasions of aggression that arise in these contexts seem to be ones that are directed at norm-violating actions, that is, they are aggressive social reactions triggered by the violation of group norms. Cases of aggression are not just reactions of distress or frustration over the experience of some negative outcome. It is fundamentally a relationship between individuals with particular concerns expressed as the object of the aggression. As a theory of expressions and physiological activity, APT is hard-pressed to account for the directedness of emotions, especially when the objects of emotions are things like norms. This is in part due to the general difficulty in explaining how such physiological activity can have intentional objects. However, even if that problem were overcome, the complexity of representing and responding to norms, if that is indeed what chimps are doing, is a sophisticated cognitive capacity that is at odds with the primitive, subcognitive forms of information processing involved in affect programs. The social nature of aggression is troublesome because APT treats emotions as mostly individualistic and misses the importance of relations within a group in determining the nature of the emotional behavior of aggression.
Another way APT seems inadequate in accounting for aggression in chimps is in the inadequacy of some of its key explanatory elements. While particular bouts of aggression do happen in brief spurts, they can also continue for long periods of time. This is seen in border-raiding parties that chimps engage in with other communities of chimpanzees. It is also seen in a case documented by de Waal (Reference De Waal1996) at the Arnhem Colony where two chimps, Nikkie and Luit, vied for power in complex ways over many years, recruiting the assistance of other chimps in their power struggles. While granted APT is only focused on brief emotional behaviors, aggression can be an extended type of emotional behavior that still embodies intuitively emotional attitudes. It is unclear how APT's picture of emotion is supposed to illuminate these extended emotions. Do the behaviors continue after the emotion ceases? Do episodes of anger keep recurring for as long as the behaviors continue? The first possibility seems odd since APT identifies the behaviors as an important constituent of the emotion. The second possibility seems ad hoc.
Perhaps most importantly, for my purposes, is the cognitive element that seems to exist. Individual bouts of aggression do not show limited involvement with cognitive processes as APT claims and do not involve a stimulus appraisal that is informationally encapsulated. Chimp appraisals of competitors in struggles over dominance require complex information processing that underlies representation of rank, affiliations with other members of the group of both participants (since alliances can significantly affect whether aggression escalates) and the likely consequences of engaging in further aggression (de Waal Reference De Waal1996). Because of the sensitivity of episodes of aggression to these features, they do not just unfold automatically and are not genuinely involuntary. The individual's representations of complex knowledge play crucial roles in determining the emotional behaviors engaged in. Again, APT's focus on the modular nature of affects, their relative passivity and their stereotypical action performance misses the complexity and variation of chimpanzee aggression.
3.2. Piping Plover Fear.
Plovers display interesting behaviors when confronted with intruders who come into close proximity with their nest or young. They are known to exhibit a variety of distraction behaviors including walking in front of the intruder and emitting a peeping sound, showing off their underside and bright wings while flying overhead of an intruder, falsely appearing to sit upon a nest, and feigning injury (Ristau Reference Ristau and Ristau1991, 94). Ristau's studies focus on the latter type of behavior because it is more easily observed. This is the broken-wing display (BWD) which, though there are degrees of severity, primarily involves the plover arching its wing or wings, dragging them and walking awkwardly (Ristau Reference Ristau and Ristau1991, 94).
Anti-predatory behavior by plovers is a quintessential context in which to find emotional behavior, in particular that of fear and perhaps maternal care. The situation is life-threatening for both the mother and her young. It involves the invasion of a danger that is incongruent with the birds’ goals, demands a quick and effective response, and is of high importance to the birds involved, especially the mother. Since APT picks out emotional contexts by their similarity to ones we recognize as emotional, and since APT relies on intuitive judgments of animal emotion in the animal argument against cognitive theories, APT should grant the emotional context of Ristau's plovers. Thus, APT should be applicable to plover fear insofar as it is homologous to primate fear or insofar as APT strives to be a general theory of basic emotion.
Ristau (Reference Ristau and Ristau1991) presents three sets of experiments conducted with plovers: (1) plover response to intruders; (2) plover response to gaze; and (3) plover discrimination between safe and dangerous intruders. Though I do not have the space here to go into detail about these experiments, I can report Ristau's basic results. When the plovers engage in a BWD, the display moves such that the intruder, upon following the bird, moves away from the nest. The plovers also situate themselves in the visual field of the intruder, continuously monitoring their movements. If the intruder stops following the birds, they reorient themselves to the intruder's visual field. Plovers also seem to be responsive to the attention of an intruder. Plovers react more strongly to those who walk near the nest and gaze in the direction of it as opposed to those who walk near the nest and gaze away from the nest. Plovers also appear to discriminate between intruders who exhibit “dangerous” behavior (gaze and search for the eggs) and intruders who exhibit “safe” behavior. The learned discrimination of dangerous and safe individuals results in different behavioral responses toward those individuals so labeled.
How do we explain the plover behavior? Ristau suggests a belief-desire account, but this is more than is needed for the point I wish to make. It seems plausible to locate this behavior in the context of an emotional response for reasons mentioned above. The problem, though, with applying APT to this case is that it seems inadequate in dealing with the complexity of the piping plover's behavior. The features of APT mentioned in Section 2 do not fit well here. The emotional behavior is not brief but lasts for as long as the plover is engaged in satisfying the goal of removing the danger to its young. The behaviors are somewhat stereotyped, but they are still flexible. When one behavioral strategy is not working, the plover changes its behavior to one that seems to fit the situation more appropriately. Plus, the behavior is performed such that it almost always moves the intruder away from the nest, even if that involves recalibrating to the intruder's changing behavior. This ability shows that the emotional behavior is not a fixed routine lacking in flexibility and that it does not have limited involvement with cognitive processes. Monitoring the intruder's behavior and gaze and learning to distinguish the dangerous and safe walkers-by implies connections of the affective state with perceptual, control, motor and memory systems. Likewise, learning the discriminations shows that the stimulus appraisal is not informationally encapsulated. And finally, the emotional behavior is not automatically unfolding or passively had. The plover seems to be actively engaged in acting in a goal-directed manner. The flexibility of the behavior speaks to this aspect.
3.3 Summary and Qualifications.
Each case above provides similar reasons for rejecting the basic version of APT, or any theory of emotion, that reduces natural occurrences of emotion to simplistic, reflex-like processes. The examples point out the complicated way in which emotions are infused with rich information processing. The type of information used by animals to assess a situation and another organism's actions in that situation is much more than a simple and automatic triggering mechanism. The process requires detailed knowledge of the situation's likely consequences, the properties of the other organism, and how these factors are related to the individual's goals or needs. For example, chimpanzees assess the relative rank of an opponent and the opponent's allies. Differences in these factors lead to differences in aggressive behavior. Plovers assess the characteristics of an intruder, including their past transgressions and their present behavior, in relation to the bird's behavior and circumstances. This seems to be an evaluation of the danger an intruder presents. The ensuing emotions then, if engaged in, do not simply unfold either. They are intimately related to other sources of information as they occur. The chimp can terminate aggression if it becomes a bad strategy at the time, and the plover can heighten its displays if the intruder is not being distracted. This exposes the dynamic nature of the emotion process and the complicated ways in which such processes can be affected and guided by complex cognitive resources. All of these features of an emotional episode are glossed over by a theory of emotion that focuses on inherited, subcognitive syndromes. And yet, these complexities of emotions are found in two widely distinct species. For this reason, and because it seems prudent that cognition be available to emotional actions, it is likely that these complex features of affective states are common to many other species as well. Leaving the cognitive complexities out of naturally occurring emotions misses a very important aspect of them.
The critical remarks above are intended to apply to the features of APT, or any such theory of emotion, that are discussed in Section 2 of this paper. Those features are: modularity, automaticity, and passivity of emotions. While such versions of APT have been presented and defended in the philosophical literature, there are more complex theories of emotion that build on the basic insights of APT (e.g., Panksepp Reference Panksepp1998). A proponent of one of these more complex views might object that I have unfairly focused on a strawman of APT. These theories allow for more extensive connections between cognitive and affective processes, but it is unclear whether they can be incorporated into the typical APT picture. I agree that a more sophisticated theory of emotion than APT is required to capture emotional phenomena, and I think that cases of animal emotion show this. My point of criticism here is that the assumption that the basic version of APT does account for animal emotion is mistaken. APT is insufficient in this regard.
I need to make another qualification as well. There is a question as to whether APT is just a search for the essential, or necessary, elements of basic emotions, or whether it is intended to account for the basic emotions as they occur naturally. If APT is only meant to give the necessary conditions for basic emotions, then my criticism that it cannot account for all examples of animal emotion misses the mark. This is because those cases may involve more than just the necessary elements of such emotions. I think part of the goal of APT is to articulate the necessary elements of basic emotions, however, I do not think this is all that APT strives to accomplish. For one thing, it would be an extremely meager goal for a theory of emotion. Furthermore, if this were the only goal of APT, the use of animals, as both an articulation of affect programs and as an argument in favor of the theory, would be less than convincing. This is because many cases of animal emotion involve more than the bare-bones emotional episodes APT suggests. The use of animals as support for their theory seems to suggest that APT is intended to account for the basic emotions as they occur naturally. Thus, an investigation as to whether it does so is certainly warranted.
4. Conclusion.
The subject of animals and their emotions plays many important roles in debates over theories of emotion. One of these roles is as a form of justification for APT. Another role is as an articulation of central elements of APT. In many ways the two roles are intertwined since the success of the justificatory role is dependent on being able to make sense of animal emotions on APT's terms. This, in turn, will rely on the adequacy of fit between animal emotions and APT and how well APT is articulated by those cases. The first role has largely been accepted as successful without much critical evaluation. In this paper, I have questioned this assumption by exploring some of the key components of APT and applying them to cases of intuitively emotional behavior in chimpanzees and piping plovers. The conceptual tools of APT seem less than satisfying in matching up with and explaining those behaviors. APT does, however, provide many important insights into baseline elements of animal emotion such as subcognitive information processing, physiological activity, and bodily expression. To be an attractive theory of emotion, though, APT needs to be elaborated in a way that can capture the complexity and flexibility of many emotional states and behaviors. A hybrid view of emotion, one that mixes elements of cognitive and of subcognitive theories, is probably best suited for such a task. Paying attention to the lives of animals can help illuminate such cases and inform theories of emotion.
