What role do the emotions play in cognition?
Towards a new alternative to cognitive theories of emotion'
Jason L. Megill
Louisiana State University
This paper has two aims: (1) to point the way towards a novel alternative to cognitive theories of emotion, and (2) to delineate a number of different functions that the emotions play in cognition, functions that become visible from outside the framework of cognitive theories. First, 1 hold that the Higher Order Representational (HOR) theories of consciousness — as generally formulated — are inadequate insofar as they fail to account for selective attention. After posing this dilemma, I resolve it in such a manner thai the following thesis arises: the emotions play a key role in shaping selective attention. This thesis is in accord with A. Damasio's (1994) noteworthy neuroscientine work on emotion. I then begin to formulate an alternative to cognitive theories of emotion, and 1 show how this new account has implications for the following issues: face recognition, two brain disorders (Capgras' and Fregoli syndrome), the frame problem in A.I., and the research program of affective computing.
Keywords: emotion, cognition, cognitive theories of emotion, consciousness, selective attention, face recognition, the frame problem
Introduction
This paper has two principle — and interrelated — aims. The first aim is to offer what is at least a partial answer to the question, 'what role do the emotions play in cognition?' As we'll see, my answer to this question is directly related to this paper's second aim: to begin to sketch an alternative to cognitive theories
Consciousness & Lmotion 4:1 (2003), 81-100.
issn 1566-5836 / k-issn 1569-9706 O John Benjamins Publishing Company
82    Jason L. Mcgill
of emotion; if the emotions are to play the role in cognition that I assign them, then cognitive theories can't be correct.
The paper consists of three sections. In section one, I criticize the Higher Order Representational (HOR) Theories of Consciousness and cognitive theories of emotion. After offering a brief introduction to the HOR theories of consciousness, I argue that the theories — as they are generally formulated — are inadequate insofar as they fail to account for selective attention. I propose a solution to this problem that not only allows for an explication of selective attention, but also (1) explains why we don't suffer from one aspect of the frame problem (as diagnosed by Dennett (1984)), and (2) suggests that the emotions play a key role in determining what we pay attention to. In short, I offer an alternative to the HOR theories that overcomes the deficiency I diagnose. Finally, 1 discuss cognitive theories of emotion, rehearsing some of the criticisms these theories face.
In section two, 1 point the way towards an alternative to cognitive theories of emotion. Admittedly, the view that I sketch, in order to be considered a full-tledged theory in its own right, would need further development and justification. Nevertheless, as we'll see, there are certain factors that suggest that the view I sketch is a promising possible direction that theories of emotion may take in the future.
Roughly, with important qualifications, I claim that the emotions play a key role in determining the meaning of thoughts, i.e. an affective state associated with a certain thought plays a role in shaping the 'Tregean sense" of the thought. This view has two immediate implications that I explore: (1) it leads to the thesis that — perhaps not surprisingly — the emotions play a key role in face recognition, and (2) this view can make sense of two perplexing brain disorders (Capgras' and Fregoli syndromes). Further, this view would seem to mesh well with the work of two prominent neuroscientists: A. Damasio and V.S. Ramachandran.
In part three, I conclude by briefly remarking that the claims made in this paper, and any claim to the effect that the emotions play a key role in shaping our intelligent behavior, provides further motivation for the artificial intelligence research program of affective computing.
l.   The higher order representational theories of consciousness and cognitive theories of emotion
Here, I first offer a brief introduction to the Higher Order Representational (HOR) theories of consciousness. I argue that these theories are inadequate as
What role do the emotions play in cognition? 83
presently formulated insofar as they fail to account for selective attention. I propose an alternative theory to the HOR approach that can overcome this deficiency. I then suggest that the emotions play a key role in determining what we pay attention to. I also remark on some of the common criticisms of the HOR approach, implying that perhaps they are insurmountable, which further motivates the acceptance of my alternative. Finally, I briefly discuss cognitive theories of emotion, suggesting that, like the HOR theories, this approach is flawed.
The higher order representational theories of consciousness
The HOR theories of consciousness — (1) the Higher Order Thought (HOT) theory and (2i the Higher Order Perception (HOP) theory — seek to explain the phenomenon of conscious awareness. Lycan (2000) writes, "Occurrent menial or psychological states fall roughly into three categories: those whose subjects are not aware of being in them, but could have been had they taken notice, those whose subjects are aware of being in them, and those, such as language processing states, which are entirely subterranean and inaccessible to introspection." The HOR theories seek to explain this tiered structure of consciousness and conscious awareness by holding that "a subject's awareness of her/his own mental state consists in a representing of that state itself (Lycan, 2000). A thought or state that we are not aware of— but potentially could be — is called a "first-order representation", a thought or state that represents a first-order representation (thereby making us aware of the first-order representation) is called a "second-order representation", while some thoughts or states can never become the objects of higher order representation, and are therefore always entirely subterranean (Lycan, 2000).
The HOP theorists (Armstrong (1981), Lycan (1987, 1996, 2000)) posit an internal attention mechanism that scans or monitors first-order thoughts. If the attention mechanism focuses upon a given first-order thought and represents it as a second-order thought, we become aware of the first-order thought; if the mechanism fails to take notice of a first-order thought, then the thought hovers below conscious awareness, while some first-order thoughts can not be scanned by the attention mechanism, and are therefore always subterranean. Lycan (2000) states that "when we deliberately introspect and thereby become aware of a first-order mental state that we had not realized we were in, the awareness is quasi-perceptual"; "it feels as thought we are "looking at" a particular sector of our cognitive or phenomenal field".
The HOT theory (advocated by Rosenthal (1990, 93), Gennaro (1995) and
84    Jason L. Megill
Carruthers (2000)), by contrast, states that we become consciously aware of a first-order thought when we have a second-order thought about the first-order thought. Rosenthal (1990), for instance, argues that the process through which we become aware of a thought "comes in the form of a cognitive, rather than a perceptual, state" (Guzeldere 1995, p. 336).
So, the HOR theories of consciousness both hold that awareness of a thought occurs when the thought is re-represented as a second-order representation, whether an internal mechanism scans and represents our thoughts to ourselves, or whether the mechanism of awareness is a more cognitive affair, i.e. whether the attention mechanism is replaced with "thoughts about thoughts".
At this stage, a question naturally arises: is there something about a given first-order thought selected for awareness that differentiates it from the other merely potentially conscious first-order thoughts not singled out for awareness? Is there something about a first-order thought that draws the attention of the attention mechanism to it so as to ensure that it will be represented, and thereby enter our conscious awareness, or does the Higher Order mechanism or process simply choose first-order thoughts for representation/awareness at random? In short, is the Higher Order process guided or directed, or is it random? Insofar as HOR theories remain silent on this question, it appears that while HOR theories may have a plausible explanation of conscious awareness, they fail to account for selective attention.
To clarify, it seems implausible to state that the Higher Order mechanism singles thoughts out of the pool of potentially conscious first-order thoughts at random. Saying that the Higher Order mechanism is random would be a denial of the very existence of selective attention. More often than not, our attention or awareness is directed at those aspects of our environment that are salient or relevant, that is, we do not systematically get lost in the trivia of our immediate surroundings. If the Higher Order mechanism selected first-order thoughts for awareness at random, we would suffer from the frame problem. Like Dennett's (1984) robot, we would get lost in the trivial aspects of our environment while a bomb ticked away under the table; we would stare at the wallpaper while the room was on fire.
It seems more plausible to hold that the Higher Order mechanism is guided or directed, and that there is some aspect of the Higher Order process that causes the awareness mechanism to select some first-order thoughts (i.e. the more relevant or salient thoughts) for awareness over others (i.e. trivial thoughts). Rather than being a denial of the phenomenon of selective attention, this latter path may provide for the possibility of an explication of the
What role do the emotions play in cognition? 85
phenomenon of selective attention.
So, if there is something about a first-order thought (some aspect or property of the thought that is intrinsic to the thought) that singles it out from other first-order thoughts, and draws the attention of the internal attention mechanism or higher order thought process to it, then what could this something be? I suggest that one answer to this question is emotion; in short, if a given thought has an affective flavor — or if a given thought has emotional connotations — then this makes it more likely that the higher order mechanism will single it out for awareness. Given two thoughts, one with emotional import and one without, the Higher Order mechanism will select for awareness the one with emotional import. This is not a complete answer, i.e. not every thought that we become aware of has emotional connotations, but I believe it to a partial answer. As we'll sec below, this answer is in accord with recent work from neuroscience.
In sum, it seems the HOR approach to conscious awareness is inadequate as it fails to account for selective attention. I formulated an alternative in which certain properties of thoughts make them more likely to be re-represented, thereby paving the way for an account of selective attention. I suggested that emotional connotation is one specific property a thought may have that ensures it enters awareness ahead of other thoughts.
Now, 1 mention several common criticisms of the Higher Order approach. Lycan (2000), for example, points out that it is commonly objected that the HOR theories lead to an inlinite regress. Critics feel that "if a second-order representation is to confer consciousness on the first-order state, it must itself be a conscious state; so there must be a third-order representation of it", ad infinitum (Lycan 2000). Lycan (2000) argues that this objection is misguided insofar as "HOP and HOT theorists reject the opening conditional premise", i.e. HOR theorists deny that for a second-order state to confer awareness on a first-order state, the second-order state has to enter conscious awareness itself. The second order state can enter our conscious awareness through the introduction of a third-order state, i.e. I can become aware that I am aware of P, but it needn't necessarily do so to make us aware of P (Lycan 2000).
A further, more serious worry is that some see little reason for supposing that a thought about a thought (or a perception of a thought) is likely to make us aware of the initial thought; that is, one may wonder why a re-representation of a first-order thought as a second-order thought makes us aware of the first-order thought. Or, what difference does it make — as regards awareness — if a given thought happens to be registered in the brain twice, i.e. in its initial uncon-
86    Jason L. Megill
scious form and as the object of an awareness conferring second-order thought?
Lycan (2000) remarks, "in general, to be aware of any thing or state of affairs is to represent that item in some way, and to be unaware or unconscious of it is to fail to represent it". If one accepts Lycan's (2000) claim that there is a fundamental relationship between awareness and representation, insofar as awareness can be equated with representation, then the above worry doesn't arise. That is, HOR theorists, often implicitly, assume that awareness simply is representation, so to them, the claim that we become aware of a thought by re-representing it, or by registering it twice, is unproblematic. However, it seems that one could reject the equation "awareness = representation;" this notion, while perhaps a widespread assumption, doesn't appear to be an obvious tautology by any means, in which case the HOR theories do seem to be problematic, or at least insufftcicntJy justified.
There are further problems with the HOR theories. Shoemaker (1994) worries about the fact that HOR theories posit a cognitive mechanism that makes us aware of our thoughts. Any mechanism can go awry, in which case "the privileged access" to our mental states that many believe we have would be violated. Lycan, (2000), in turn, claims that such a worry arises only when HOR theories are considered as theories of lirst-person privilege and/or self-knowledge, which HOR theories never claimed to be. I add that it is by no means settled that we do in fact have privileged access, i.e. it is plausible to claim that perhaps we don't, and the assumption that we do — which goes back to Descartes' (1993) Meditations— is unwarranted; in which case, who is to say that this cognitive mechanism doesn't go awry?
Related to Shoemaker's (1994) concern is the possibility of "false positives:" it seems (hat the Higher Order mechanism could misfire in such a manner that someone would believe they were in excruciating pain, "while actually having no pain at all" (Lycan 2000). Both Lycan (1998) and Neander (1998) have responded to this worry. Further, it appears that this worry doesn't arise, or at least seems less plausible, on the novel view I sketched above. Recall that I held that we become aware of a thought when there is something — some intrinsic part, aspect or property of the thought — that triggers or attracts the attention of the Higher Order mechanism; so, in order for us to become aware of a thought, the thought must have some intrinsic property. But, if a thought doesn't exist (i.e. if it is a false positive), then how can it have the property that triggers the attention mechanism in the first place? In short, a thought that doesn't exist can't have the property thoughts need to trigger the awareness mechanism, and thereby enter awareness in the first place, so, how can false
What role do tin- emotions play in cognition? 87
positives occur? I suppose it is possible that the mechanism can go so awry as to falsely activate, and then register a thought that doesn't (or didn't previously) exist, but it seems unlikely.
There are a couple of other worries. Carruthers (2000) worries that it seems that a Higher Order mechanism would demand a large amount of cognitive resources (as a person's awareness can be quite complex at times), in which case it's difficult to see how animals, or perhaps even small children, can have the resources needed for awareness. Further, Rev (1983) claims that if all that is needed for awareness is the monitoring of internal states, then many technological devices qualify for awareness, as they, too, monitor their own states, and it seems absurd to confer awareness on a relatively simplistic technological device. White (1987) and Lycan (1996) have responded to this worry.
So, as we see, there are serious difficulties the HOR approach faces in addition to the difficulty it faces in accounting for selective attention.
Cognitive theories of emotion
Now 1 briefly discuss cognitive theories of emotion, emphasizing their central claims, broad theoretical and methodological commitments, and some of the difficulties these theories face.
R Griffiths (1997, p.2), in his influential work What Emotions Really Are, delineates what he calls "propositional attitude theories" of emotion. Although such theories take numerous different forms (so making broad statements about the propositional attitude approach is difficult, insofar as exceptions can usually be found), in general they hold that:
emotional phenomena can be dealt with by a psychology whose main theoretical entities are beliefs and desires that feature in everyday explanation of peoples actions. These theoretical entities are the 'propositional attitudes' of folk psychology. (Griffiths 1997, p.2)
In short, although propositional attitudes (PAs) are not necessarily the sole, or even the only important, theoretical entity in many PA theories (in fact, many PA theories allow for, or demand, a discussion of other factors of emotion such as facial expressions, action tendencies, and so on), they are generally the most important.
As Griffiths (1997, p.2) points out, for the past 30 years, philosophical work on the emotions "has been dominated" by the PA approach. Adherents of the PA approach 'often refer to their view as the "cognitive theory" of emotion",
88    lason L. Mcgill
generally out of a desire to differentiate their views from "feeling theories" of emotion, which claim that "emotions are introspective experiences characterized by a quality and intensity of sensation" (Griffiths 1997, p.2), i.e. on "feeling theories" of emotion, "emotions are simply assimilated to certain kinds of felt experience" (de Sousa 1987, p.37). Griffiths (1997), a critic of cognitive theories and the PA approach (see also Griffiths 1989 and 90), deems the label of "cognitive" inappropriate insofar as it suggests that the PA approach concerns itself with the empirical findings of cognitive psychology, whereas most "cognitive" theories are formulated a priori through conceptual analysis (an approach Griffiths (1997) eschews in the study of emotion; below I will follow Griffiths in this methodological point).
Again, whereas "feeling theories" of emotion equate emotions with "certain kinds of felt experience" (de Sousa 1987, p.37), or with "emotional qualia", so to speak, cognitive theories eschew this aspect of emotion, a move that they have been criticized for. As de Sousa (1987, p.37) points out, the feeling theory seems to mesh well with common sense (although de Sousa himself does not endorse the feeling theory), and more importantly, such a view can account for the affective component of emotion (or what Nash (1989) calls the "passive nature" of emotion, i.e. the degree to which we feel afraid, or angry, etc). Cognitive theories have often been faulted for not accounting for this aspect of emotion (in fact, Nash's (1989) view, which I discuss below, is in part an attempt to account for this aspect of emotion from within the cognitive perspective, and hence an attempt to overcome this perceived inadequacy of cognitive views).
To continue, here is another description of cognitivism, one offered by VV. Lyons (1980, p.33):
In general, a cognitive theory of emotion is one that makes some aspect of a thought, usually a belief, central to the concept of emotion and, at least in some cognitive theories, essential to distinguishing different emotions from one another.
As Griffiths (1997, pgs 21-22) points out, this statement, while seemingly unproblemalic, houses a number of common assumptions and tenets of cognitivism that are questionable.
Again, while many cognitivists allow for additional factors (aside from PAs) when explaining what gives rise to an emotion, factors such as facial expression, often, "the details of these other factors are of no importance in distinguishing one emotion type from another" (Griffiths 1997, p.22). That is, as regards the
What role do the emotions play in cognition? 89
question of individuating emotion types (fear, anger, sadness) from one another, a common doctrine of" cognitive views is that "the identity of particular emotions depend upon the kind of propositional attitude they involve" (Griffiths 1990, p.517), as opposed to, say, biological or physiological factors. In short, many cognitive theorists individuate emotion types with propositional attitudes, and deemphasize biology and physiology.
So, we have met several commonplace aspects of cognitive theories: (1) its methodology is conceptual analysis (and by and large, it ignores empirical research), (2) it ignores the sensation or feeling aspect of emotion, (3) it deemphasizes the biological and/or physiological aspects of emotion, and (4) cognitive theories differentiate emotion types with propositional attitudes. Now, I briefly look at two specific cognitive theories: R. Solomons (1976; 83) and R.A. Nash's (1989).
Solomon's (1983) view identifies the emotions with (intense) judgments, i.e. with mental acts of a certain sort. "An emotion is a judgment (or a set of judgments), something we do" (Solomon 1983, p. 185). Some have objected that it seems that the emotions simply are not judgments, as many cognitivists hold. First, it seems that emotions can be triggered by subliminal stimuli (Zajonc 1980, 1985), in which case, where is the judgment (Griffiths 1997)? Further, some charge that "identifying judgments with emotions gives us too many emotions", and that "there are whole classes of evaluative judgments that are never likely to be the content of emotions", i.e. "an account is needed as to why only some evaluative judgments are emotions" (Griffiths, 1997, p. 29).
R.A. Nash's (1989) "New (Pure) Cognitive Theory of Emotion" is an interesting case as regards the claims of this paper as Nash's (1989) view also argues that the emotions shape selective attention. There are two fundamental notions to Nash's theory: (1) focus of attention and (2) overvaluation. Imagine the following scenario (which deals with fear): Fat is in a crowded movie theater when someone yells "Fire!" Pat has a dispositional state E "that is activated by this situation" (Nash 1989, p.510). "One output of E is an intensification of attention on the content of certain states: typically, the beliefs and desires which trigger E", i.e. in the state E, Pat focuses his attention on the fire, which triggered E (Nash 1989, p. 510). Pat also "overvalues" the fire; his being completely preoccupied with the fire causes him to neglect other factors of the situation he would otherwise pay attention to. For instance, in his fear, he tramples people on his way to the exit. In short, the state E is characterized by a focus of attention on the cause of E and an overvaluation of E, "i.e. in a disproportionate weighting of the facts" (Nash 1989, p. 511).
90    Jason I.. Megill
Before concluding this section, a word should be said on "hybrid theories" of emotion. Hybrid theories attempt to account for the physiological aspects of emotion while also remaining true to the spirit of cognitivism to a large degree (see Lyons (1980) and W.C Davis (1988)).
To summarize: cognitive theories of emotion, like the HOR theories of consciousness, face a number of serious criticisms, which suggests that it would be fruitful to look for alternatives.
2.  Towards an alternative to cognitive theories
Thus far, I have offered primarily negative claims: I pointed out that the HOR theories are inadequate, and that cognitive theories of emotion may be inadequate as well. In this section, I begin to sketch a possible direction that theories of emotion might take in the future. As we'll see, my view can be thought of as a "complicated" sensationalist theory, or perhaps as a hybrid of sensationalist and physiological views. Again, I refrain from calling my position a full-fledged theory, as it still stands in need of further development and justification; the main goal of this section is to suggest what direction a satisfactory theory of the emotions may take.
In section one, I suggested that the emotional connotations or affective components of thoughts play a role in shaping selective attention; if a thought has an affective flavor, it is more likely to enter conscious awareness. Here, I put forth another claim about the emotions: emotions shape the meaning or "Fregean sense" of a thought. After developing this claim, I go on to offer reasons for its acceptance. As we'll see, the claim that the emotions shape the meaning of thoughts, as well as the claim that the emotions guide selective attention, commits me to a rejection of cognitive theories and an endorsement of a sensationalist/physiological view. Finally, I show how this theory of emotion drastically differs from cognitive theories of emotion.
The emotions shape the meaning of thoughts
Take any propositional attitude: "5 believes that P", "S doubts that P", "5 remembers that P" and so on. As propositional attitudes are commonly understood, in each case "P" corresponds to a proposition or a thought, such as "Sally went to Hawaii". That is, S can believe that "Sally went to Hawaii", or S can doubt that "Sally went to Hawaii" and so on.
What rote do the emotions play in cognition? 91
Now, I wish to claim that the emotions are most readily associated with "P", i.e. with the thought or proposition that is the object of the propositional attitude (and that the rest of the propositional attitude can be ignored for the time being). Or, emotions are a part of, or an aspect of, P. Specifically, the emotions play a role in shaping or determining the meaning of P, or the "Fregean sense'" of P, or, if one prefers, the "narrow content" of P (see Stalnaker 1991 and Block 1991 for a discussion of narrow content). In short, the emotions help determine what a thought means, and are therefore most readily associated with P, i.e. with a thought or proposition that can serve as the object of a propositional attitude.
Before clarifying and qualifying this claim, I point out that if the emotions are to play the role in cognition I assigned them in section one, i.e. if the emotions are to help shape selective attention, then they must be associated with P, as opposed to, say, the propositional attitude itself (i.e. believes, doubts, remembers). If the emotions are to guide or direct intentionality, then they can't be a type of intentionality themselves, i.e. they can't be associated with, say, the propositional attitudes (believes, etc). In short, the emotions must be an intrinsic aspect of a thought (P) if they are to serve as a means of attracting the attention of the attention mechanism to a thought, i.e. if they are to guide intentionality (insofar as they guide selective attention).
Now, the claim that the emotions shape the meaning of a thought needs to be clarified and qualified. For the purposes of exposition, I first discuss my view in terms of individual concepts (i.e. I argue that the emotions shape the meaning of individual concepts), before suggesting below that the emotions can only be associated with entire thoughts or propositions, as opposed to the individual concepts that comprise thoughts and propositions.
lake my "mother" concept as an example. For me, this concept is associated with a certain emotional or affective component and, in claiming that emotions shape or determine meaning, I'm claiming that if this affective component were removed, my concept would literally take on a different meaning.
To make a qualification: this affective component of my "mother" concept — and the affective components of all concepts (that have an affective component) — needs to be differentiated from "objective" components of meaning. That is, each person may associate a different affective component with their mother concept, and while there may be a broad overlapping of most of the individual affective components associated with all of these individual mother concepts (i.e. it may be that many people have the same positive emotions associated with their mothers), there will be subjective variations. It may even
92    Jason L. Megill
be that a given person is completely indifferent to their mother; in short, if, in fact, the emotions help shape the meaning of concepts, then they do so in a manner that varies across individuals, and insofar as this is the case, the emotional component of the concept is "subjective".
Alternatively, "red" is a part of my — and many peoples — "apple" concept, and is an "objective" component of meaning (most people can look and see that apples are red and so on). But, imagine that S loves apples (perhaps 5 grew up on an apple orchard); S's love of apples is not shared by all, or even most, and hence, is different than "red" in this sense. (As regards the notion that affective components shape meaning, perhaps there are better examples than "mother" and "red?" Some concepts, such as "frightening" or "disgusting", seem to demand that we feel the affective component associated with the concept (i.e. we feel fright or disgust) to grasp the meaning of the concept; and this goes for all of the concepts that are associated with concepts such as "fright" as well, like "ghost".)
I've developed the claim that emotions help shape meaning in terms of individual concepts, but as 1 now show, this is problematic, as perhaps the notion that emotions shape meaning only makes sense when applied to entire thoughts or propositions, as opposed to individual concepts? The problem arises due to the fact that the emotions are highly contextual phenomena. For example, I'm not afraid of the polar bear at the zoo, i.e. I don't associate fear with the "polar bear" concept in this context. However, if a polar bear somehow made its way into my apartment, I probably would associate fear with the "polar bear" concept. This is to say: I associate fear with the proposition "A polar bear is in my apartment", but not with "A polar bear is in the zoo", so it seems that emotions, if they do play a role in shaping meaning, do so only as regards entire propositions, thoughts, or states of affairs, and not as regards individual concepts. It is possible thai each concept has an individual affective component, and these affective components intermix (when placed together in the same thought or proposition) to produce a given emotion, but, insofar as the claim that the emotions shape the meaning of entire thoughts (as opposed to individual concepts) is simpler, it seems more plausible.
To summarize: I claimed that the emotions shape, determine or fix meaning, or Fregean sense. I first developed this claim in terms of individual concepts, before going on to claim that since emotions are highly contextual phenomena, perhaps it is better to hold that they shape the meaning of entire thoughts or propositions, as opposed to individual concepts? As regards propositional attitudes, since the emotions shape the meaning of P, they are
What role do the emotions play in cognition? 93
most readily associated with P, i.e. with the object of a propositional attitude (and insofar as this is the case, the remainder of the propositional attitude can be ignored). Further, I differentiated between "subjective" and "objective" components of meaning, holding that the emotions are subjective (in that the emotional connotations associated with various thoughts can vary across individuals). As 1 now show, there are good (though probably not conclusive) reasons for believing that the claims set out thus far (in this section and section one) are true.
Damasio (1994) has argued that the emotions play a key role in our cognitive behavior. An important notion in Damasio's (1994) work is the "Somatic Marker hypothesis". Imagine that you are placed in a situation where you must pick a course of action. There are a large number of possible actions, a large number of alternatives. These alternatives present themselves as mental images. Option A has a bad outcome associated with it; upon thinking of said outcome, you have a visceral physiological response, an unpleasant gut feeling. In a sense, the image or alternative is marked by the emotion, i.e. is marked by a "somatic marker". As a result of the unpleasant emotional reaction to As outcome, A is quickly discarded from your list of possible actions. This process of deleting undesirable options from your list of options continues until a more manageable number of possible actions remain (see Damasio 1994, especially pp. 173-74).
So, in a sense (in a neurological and physiological sense) — under the Somatic Marker hypothesis — the emotional import of a thought or mental image — the somatic marker — is a component part of the thought (i.e. fear is associated with — or is an aspect of — the thought that "a polar bear is in my apartment").
Further, Damasio (1994) argues that somatic markers play a role in shaping attention. More precisely, a somatic marker boosts attention: if we come across a mental image B that has the appropriate somatic marker, we will hold B in our attention to the exclusion of other thoughts or images. I've already claimed that the emotional import of a thought helps determine what enters our attention or conscious awareness in the first place (section one); Damasio (1994) argues that the emotional import of a thought or mental image — the somatic marker — plays a role in determining how long a given thought is kept in conscious awareness after it is permitted entry. In short, there is scientific evidence behind Damasio's (1994) hypothesis, and the hypothesis seems to lend plausibility to some of the claims made thus far in this paper; specifically, the emotions guide attention and, the emotions are a part of, or aspect of, thoughts.
94    Jason L. Megill
Patients suffering from Capgras' Syndrome "believe that important people in their lives (family members, etc) have been replaced by exact doubles" (Berson 1983, p. 969). For instance, a Capgras' patient may look at her mother, see that the person's outward appearance is exactly like that of her mother, yet outright deny that the person is in fact her mother. The neuroscientist V.S Ramachandran advances the following explanation for Capgras': the "connection between the area (of the brain) which causes emotion to the area that senses emotion" (Cheung 1998) is severed, so while the patient can look and see that the person has the appearance of her mother, the patient feels none of the characteristic emotions marked by the appearance of her mother (Cheung 1998). The lack of the expected emotion "causes great psychological disparity", which leads to the patient's denying that the person in front of her is in fact her mother (Cheung 1998).
So, here we see evidence that the emotional component of a thought is not only an aspect of the thought, but also helps shape the meaning or sense of the thought; without the expected emotion, a Capgras' patient is unable to assign the proper sense ("that is my mother") to reference (mother). In short, Capgras' patients can't match up sense with reference in the proper manner because the emotions are part of sense, and Capgras' patients lack normal emotional reactions.
Fregoli syndrome — a "cousin" of Capgras' syndrome — is characterized by the delusion that friends or family members continually alter their appearance to look entirely like someone else (Mojtabai 1994, p.458). For example, one suffering from Fregoli syndrome may mistake a series of strangers for their mother; they believe the strangers are their mother, who has somehow completely altered her appearance. The following would seem to be a plausible account of what is going on here: the specific emotional import or connotation associated with the thought "that is my mother" somehow comes to be associated with — or is attributed to — complete strangers. That is, since the emotional import of a thought helps fix the meaning of the thought, and since Fregoli patients can't match the appropriate emotional connotation with the appropriate thought, patients can't match up sense with reference in the appropriate manner. In short, here we see further evidence for the claim that the emotions help shape or fix the meaning of thoughts.
Interestingly, these brain disorders (Capgras' and Fregoli) can be reinterpreted as the inability of patients to properly match sense with reference as a result of the loss of access to the emotional connotations of their thoughts, and this lends plausibility to the claim that the emotions are not only a part or aspect of thoughts (i.e. emotions should be associated with P), but also to the
What role do the emotions play in cognition? 95
claim that the specific role the emotions play as relates to thoughts is that they fix the meaning or sense of thoughts.
Finally, Fregoli syndrome seems to suggest that the emotions play a vital role in face recognition, perhaps not surprisingly. That is, in Fregoli syndrome, the emotional component of the thought "that is my mother" is attributed to a stranger, and this leads to the patient's claim that the stranger is their mother, i.e. leads to inaccurate face recognition. In effect, these inabilities to match sense with reference in the proper manner can also be interpreted as issues concerned with "face recognition".
Now, in the interest of clarity, I pause to summarize my entire view, some aspects of which are still implicit. The emotions play the following role in cognition: they help fix the meaning or sense, and hence the reference, of thoughts. They also help shape selective attention, and play a role in face recognition. All of these roles imply that the emotions are most readily associated with "P" (as in "S believes that P"). The precise manner that emotions are connected to thoughts (in a neurobiological sense) is the following: a given thought is associated with what Damasio calls a somatic marker, which are feelings ("unpleasant gut feelings", for example) that arise through physiological changes (so I emphasized both an affective emotional experience and physiological changes, specifically, the affective emotional experience is an experience of these physiological changes).
To give an example: 1 walk into a crowded room in which an old friend — who I haven't seen in awhile— is present. Physiological changes occur, changes that manifest themselves as a felt emotive experience, and that correspond to a somatic marker that marks the thought "That is my friend A". This somatic-marker marks the thought in such a way that I become aware of the thought (I become aware of the presence of my friend, as opposed to the presence of a stranger beside him etc). Further, this somatic marker plays a role in my recognition of my friend, i.e. it plays a role in face recognition. Also, if the emotion or emotive experience wasn't present, I might not realize that A is indeed my friend, i.e. emotions help fix the meaning, and hence the reference, of the thought "That is A".
On the differences between my view and cognitive theories
In this section, I show how my view differs from the cognitive approach.
Recall Griffiths' (1997, p. 2) worry that the label of "cognitive" in "cognitive theory of emotion" is inappropriate as it suggests that the PA approach concerns
96    Jason L. Mcgill
itself with the empirical findings of cognitive psychology, when in fact most "cognitive" theories are formulated a priori through conceptual analysis. Clearly, in utilizing empirical evidence from neuroscience to formulate and justify my view above, I departed from PA approaches — at least in general, and as they are described by Griffiths (1997) — in this important methodological aspect.
But, my view makes even more drastic departure:* from cognitive theories; that is, while formulating my view above, I implicitly committed myself to a view that greatly differs from the cognitive approach, and I now bring these commitments to the surface. Recall that 1 associated an emotion with a thought in the following manner: the affective component, or the introspective emotive experience or sensation associated with a thought, helps determine the thought's meaning. In support of this view, I cited Ramachandran's explanation of Capgras' syndrome, which claimed that the patient's lack of the expected emotive experience associated with person A led the patient to deny that A was in fact A. Here, to a certain degree, I implicitly endorsed a "feeling theory" of emotion, insofar as I characterized emotion as an affective, introspective experience, or as a certain sensation. (But, as we saw above, and as we'll see again below, this affective experience is not the sole aspect of emotion that I discussed, so I surpass "feeling theories" of emotion (at least in their simplest form, i.e. in the guise of a view that completely subsumes emotion into the experience of emotion, and neglects other factors that contribute to emotion.))
So, in short, insofar as I emphasized the emotive or affective experience of emotion in my view, insofar as I equated the emotions with such experiences or sensations, 1 threw my support behind one of cognitivism's main rivals, the feeling theory.
Thus far, we've seen that my view departs from cognitive theories (as they are generally formulated) in a methodological aspect (in utilizing empirical evidence as opposed to conceptual analysis), and in emphasizing the emotive or affective experience of emotion, just as cognitivism's main rival (the feeling theory) does. But, the differences between my view and cognitive theories (still understood in general terms) do not stop there; my view makes further significant departures from PA approaches. Recall W. Lyons' (1980, p.33) description of cognitivism:
In general, a cognitive theory of emotion is one that makes some aspect of a thought, usually a belief, central to the concept of emotion and, at least in some cognitive theories, essential to distinguishing different emotions from one another.
What role do the emotions play in cognition? 97
Again, while many cognitivists allow for additional factors (aside from PAs) when explaining what gives rise to an emotion, often, "the details of these other factors are of no importance in distinguishing one emotion type from another" (Griffiths 1997, p. 22). That is, as regards the question of individuating emotion types (fear, anger, sadness) from one another, a common doctrine of cognitive views is that "the identity of particular emotions depend upon the kind of propositional attitude they involve" (Griffiths 1990, p.517), as opposed to biological factors. In endorsing Damasio's (1984) "lamesian" somatic marker view above, where somatic markers are physiological or biological changes that can be equated with emotion, and that not only become associated with mental images or thoughts, but also can be used to individuate the various emotion types (insofar as each emotion type is also a characteristic physiological state), I also rejected the cognitive theory notion that propositional attitudes are what differentiate emotion types. In short, while many cognitive theorists individuate emotion types with propositional attitudes, I individuated emotion types by means of physiological states.
Perhaps even more significantly, I elevated these physiological aspects of emotion (along with the emotive or affective experiences of emotion discussed above) to the status of principle theoretical entity, as opposed to cognitive views, which hold that PAs are the most important theoretical entities. (Note that, here, insofar as I'm endorsing physiological views, I'm also going beyond mere feeling theory accounts of emotion in their simplest form. That is, in emphasizing physiological aspects as well as the affective experiences of these physiological aspects, I'm going beyond simplistic feeling theories of emotion that merely emphasize the emotive or passive aspect of emotion.)
To summarize, the sketch of my view of emotions departed from cognitive theories (as they are generally formulated) in a number of important ways. I emphasized the affective experience of emotion, as well as the physiological aspect. Propositional attitudes are not my main theoretical entity. Further, I individuated emotion types with physiological states, and not PAs. Also, I departed from PA views in one methodological aspect, as they are often formulated through conceptual analysis, while I utilized empirical findings in neuroscience.
I reiterate that, in order to be considered a full-fledged theory, my view needs further development and justification (one particular worry is that much of my view's justification comes from (overly) speculative interpretations of ill understood brain disorders (Capgras' and Fregoli syndromes)). However, if philosophical theories of emotion are ever going to be able to take advantage of work in
98    Jason L Megill
neuroscience, then it seems the cognitive approach will have to be abandoned; I offer one possible alternative direction theories can take in the future.
3.   Concluding remarks
Finally, I briefly discuss some implications this paper has for the frame problem, affective computing and A. I. in general. Any argument that holds that the emotions play an important role in cognition has implications for A. I.; if the emotions do play a role in our cognitive capacities, then it seems any would-be A.I. needs some analogue to the emotions to function intelligently.
Also, any argument that holds that the emotions play an important role in cognition provides further motivation for the research program of affective computing — which tries to imbue computers with some analogue to our emotions (see Picard (1997) for an introduction to affective computing). More specifically, I alluded to the frame problem above (section one): one aspect of the frame problem is the question of how we know what to pay attention to in our environment, i.e. we seem to naturally pay attention to the more salient aspects of our environment. If the emotions play a role in selective attention, then this paper may not only offer a partial answer as to why we don't suffer from this aspect of the frame problem, but also may suggest strategies for combating this aspect of the frame problem in A.I.
Note
* I'd like to thank Ralph Ellis, and an anonymous referee, for extremely helpful comments on an earlier draft of this paper.
References
Armstrong, D. M. (1981). What is consciousness? In Tfie Nature of Mind. Ithaca, NY: Cornell University Press.
Ben-Zeev, A. (1987). The nature of emotions. Philosophical Studies 52: 393-409. Ben-Ze'ev, A. (2000). The Subtlety of Emotions. MIT Press: Cambridge, Mass. Berson, R. J. (1983). Capgras Syndrome. American Journal of Psychiatry* 140:969. Block, N. (1991). What narrow content is not. In B. Loewer and G. Rey (Eds) Meaning in Mind: Fodor and His Critics. Blackwell Publishers.
What role do the emotions play in cognition? 99
Carruthers, P. (2000). Phenomenal Consciousness. Cambridge: Cambridge University Press.
Cheung, R. (1998). The architecture of mind: V.S. Ramachandran. Notes from a CSSA sponsored lecture by Ramachandran, September 25, 1998. http://www.csua.berkcley.edu/ -reeser/cssa_ra m a. h t m 1
Clarke, S.G. {1986). Emotions: Rationality without cognitivism. Dialogue25: 663-74.
Damasio, A. (1994). Descartes' Error (Emotion, Reason, and the Human Brain). Grosset/ Putnam. CP. Putnam's Sons. New York: NY.
Davis, W.C. (1988). A causal theory of experiential fear. Canadian Journal of Philosophy 18:459-83.
Dennett, D. (1984). Cognitive Wheels: The Frame Problem of Artificial Intelligence In Hookway (ed.) Mind, Machines and Evolution. Thyssen Grove Volume. Cambridge U. P.
de Sousa, R. (1987; 1991). The Rationality of Emotion. Cambridge: MIT Press.
Gennaro, R. (1995). Consciousness and Self-Consciousness. Philadelphia: lohn Benjamins.
Griffiths, P (1989). The degeneration of the cognitive theory of emotion. Philosophical Psychology 2,1989.
Griffiths, P. (1990). Modularity and the psychocvolutionary theory of emotion. Riologv and Philosophy, 5 1990. Reprinted in Lycan (F.d) Mind and Cognition. Blackwell Publishers Inc. Massachusetts.
Griffiths, P. (1997). What Emotions Really Are. The University of Chicago Press: Chicago and London.
Guzeldere, G. (1995). Is Consciousness the Perception of What "Passes in One's Own Mind?'- In
Mctzinger (ed.) Conscious Experience. Imprint Academic. Schöning Paderborn. James, W. (1892). Psychology. New York: Henry Holt.
Lycan, W.G. (1987). Consciousness. Cambridge, MA: Bradford Books/MIT Press. Lycan, W.G. (1996). Consciousness and Experience. Cambridge, MA: Bradford Books/MIT Press.
Lycan, W.G. (1998). In defense of the representational theory of qualia. (Replies to Neander,
Rey and Tye). In Tomberlin 1998. Lycan, W.G. (ed) (1999). Mind and Cognition (An Anthology). Blackwell Publishers Inc.
Massachusetts.
Lycan, W.G. (2000). Representational theories of consciousness. Tl\e Stanford Encyclopedia of Philosophy http://plato.stanford.edu/entries/consciousness-representational/
Lyons, W. (1980). Emotion. Cambridge: Cambridge University Press.
Mojtabai, R. (1994). Fregoli Syndrome. Australian/New Zealand Journal of Psychiatry, September, 28:3, 458-62.
Nash, R.A. (1989). Cognitive theories of emotion. Nous, 23. Nous Publications, Blackwell Publishers Inc.
Neander (1998). The division of phenomenal labor: A problem for representational theories
of consciousness. In Tomberlin l998.Picard, R.W. (1997). Affective Computing. MIT
Press, Cambridge, Mass. Rey, G. (1983). A reason for doubting the existence of consciousness. In R. Davidson, G.E.
Schwartz and D. Shapiro (Eds) Consciousness and Self Regulation, Vol. II. New York:
Plenum Press.
Rosenthal, D. (1990). Explaining consciousness. Manuscript. New York, CUNY, Graduate School.
joo  Jason L. Megill
Rosenthal, I). (1993). Thinking that one thinks. In Daviesand Humphreys (eds.) Consciousness. Oxford: Basil Blackwell.
Shoemaker, S. (1994). Self-knowledge and inner sense: Lecture II: The Broad Perceptual Model. Philosophy and Phenomenological Research 54: 271-90.
Solomon, R.C. (1976; 83). The Passions. New York: Doubleday.
Stalnaker, R. (1991). Narrow content. In C. Anderson and J. Owens (Eds) Prepositional
Attitudes. Stanford University Press. Tomberlin, J.E. (Ed.) (1998) Language, Mind ami Ontology (Philosophical Perspectives) vol.
12. Ridgeview Publishing: Atascadero, CA. Zajonc, R.B. 11980). Feeling and thinking: preferences need no inferences. American
Psychologist 35: 151-75. Zajonc, R.B. (1984). The interaction of affect and cognition. In Scherer and Ekman (Eds)
(1984) Approaches to Emotion. Hillsdale, N.J.: Erlbaum.
Author's address Jason Megill
4618 Alvin Dark Ave. #17 Baton Rouge LA 70820 jasonmegill@hotmail.com (225)766-4723
Copyright of Consciousness & Emotion is the property of John Benjamins Publishing Co. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.