You are on page 1of 18

^ S 39:1 (2005) 106122

NOU

On Emotions and the Explanation of Behavior


ADAM KOVACH
Haverford College

CRAIG DE LANCEY
State University of New York at Oswego

We express our emotions in a remarkable range of ways. These expressions


include, at one extreme, involuntary changes of heart rate and skin tem-
perature, and at the other extreme highly calculated and reasoned actions,
such as the deliberate torture of a captive by hateful keepers. Between these
extremes lie complex, coordinated behavioral expressions of emotion which
are actions rather than mere bodily movements, because they are done
under voluntary control and in full awareness, but which are unlike proto-
typical rational actions in that they resist satisfactory explanation in terms
of the means-ends beliefs and desires of an agent. Such expressions include
caressing the object of ones affection, and performing a little dance out of
joy. Given this diversity, it is hard to see what unites the class of expressions
of emotion. Nonetheless, expressions of emotion do form commonsense
kinds, and this leads one to expect that there are some features which
unite these kinds.
We would like to take up two questions of prime interest for under-
standing how we express our emotions.

1. What, if anything, do expressions of emotion have in common?


2. Why is it that some complex coordinated behavioral expressions of
emotion, which we may call emotional actions, often prove resistant
to explanation in terms of an agents beliefs and desires?

Our purpose is to show that answers to these questions come from a


conception of some emotions as kinds of mental states, which necessarily

# 2005 Blackwell Publishing Inc., 350 Main Street, Malden, MA 02148, USA,
and P.O. Box 1354, 9600 Garsington Road, Oxford, OX4 2DQ, UK.

106
On Emotions and the Explanation of Behavior 107

involve motivational bodily states. This conception derives support both


from our commonsense practice of explaining emotional behavior, and from
research in the behavioral and brain sciences. In particular, it fits well with a
current neuropsychological theory of some emotions called the affect
program theory.
We proceed as follows. In section I, we present our view of the motiv-
ational role of some emotions from the perspective of common-sense folk
psychology. In sections II and III, we describe the affect program theory
and explain how it sharpens and lends evidence to the common sense view.
This allows us to answer the first question of what is common to expressions
of emotion in section IV. In section V, we answer the second question, by
distinguishing a particular type of explanation of behavior in terms of an
agents emotions, and showing how it differs from other types, notably
explanation in terms of an agents beliefs and desires. In section VI, we
discuss cases of emotional actions which are explained by our view but
resistant to belief-desire explanation.

I. Emotions and Motivation


We begin with a bit of folk psychology, a rough and ready common sense
view, which should be familiar, at least as a depiction of certain emotions, of
which anger and fear are paradigms. As we consider the case of someone
who flies into a state of rage and kicks a trash can, and does so without any
foresight or planning, say, shortly after receiving an insult, we are prepared
to draw certain conclusions. We are prepared to think that the kicking takes
place not only while the person is in a state of passion, but because that
person is in a state of passion. We are prepared to think that the insult has
elicited the persons rage. We are also prepared to think that the kicking of
the can is a response due to the rage, which has been elicited. None of this is
surprising, for our common sense grasp of emotion is replete with expect-
ations about how different emotions arise and about the results to which
they lead.
The observation that we understand certain behaviors as occurring
because someone is in an emotional state of mind (and not merely while
they are in such a state) exhibits what is perhaps most obvious to common
sense about some emotions, namely, that they motivate us. Thus, for
example, it is a normal and expected feature of anger that if unchecked it
causes people to perform certain characteristic aggressive behaviors or
actions. It is the nature of basic emotions to motivate by default. That is,
an emotional state of mind normally causes some sort of response, such as
leading to some sort of characteristic behavior, once some threshold of
arousal has been crossed. This default motivational character is reflected
in ordinary expectations about emotional behavior. When we consider the
case of an emotional state, which does not bring about any characteristic
108 NOUS

behavior, it is typically the absence of behavior, which strikes us as requiring


an explanation. Of course emotions do not always motivate to the point of
behavioral expression. It happens that arousal is weak, that countervailing
motivations check the expression of an emotion, or that external constraints
prevent it. Still, even where an emotion does not lead to a particular
behavioral expression, it can have motivational power by priming or pre-
paring one for action. We display an understanding of this type of motiv-
ational power, for example, when we exercise caution to avoid setting off
aggressive behavior on the part of someone whom we know to be angry.
Furthermore, some emotions have certain normal conditions of elicit-
ation. It is normal for someone to feel fear when faced with danger. It is
normal for a recipient of good news to be happy. Similarly, once elicited,
some emotions yield certain normal patterns of response. Grimacing and
fleeing in fear are normal responses, as is smiling with joy. Normal patterns
of emotional response can be complex. They can include a variety of
recognizable symptoms ranging from autonomic responses, such as changes
in skin temperature and levels of perspiration, to the occurrence of char-
acteristic facial expressions and postures, to complex, coordinated behav-
iors, such as fleeing a scene or kicking a can.
Common sense thus tells us that some emotions are states of mind, which
are (1) motivational by default, (2) triggered by certain characteristic con-
ditions of elicitation, and (3) cause certain characteristic patterns of
response, including full-blown behaviors and actions. What common sense
does not tell us is how the emotions possess this type of motivational power.
Our contention is that the best explanation of how emotions motivate
begins with the recognition that the occurrence of some emotional states
of mind involves the occurrence of certain distinct, specialized motivational
bodily states. Some remarks on the meaning of this claim follow. First, to
say that a state is motivational is to say that it has the normal function of
outputting certain behavioral responses, somewhat like machinery designed
for a purpose. As naturalists, we would seek an evolutionary account of
how these functions arose. Second, there are two ways in which the occur-
rence of an emotional state of mind can involve the occurrence of a motiv-
ational bodily state. The occurrence of a relevant motivational bodily state
may be a necessary condition for the occurrence of that emotional state of
mind, or it may be that the occurrence of the emotion includes a disposition
or tendency for the occurrence of a motivational bodily state.1
We mean to be cautious with the claim that some emotions involve
motivational bodily states. We do not make this claim about every state
of mind, which anyone might want to call an emotion. We focus our claim
on a particular class of emotions, which we call the M-emotions because
they are distinguished by a certain motivational structure. The obviously
motivational cases of anger and fear, of the familiar sort described at
the opening of this section, are the paradigm cases of M-emotions. The
On Emotions and the Explanation of Behavior 109

emotions often called basic emotions in the literature, such as disgust,


care, erotic love, joy and sadness are significantly similar.2 But a number of
other emotions, including instances of affection, hatred and jealousy also
appear to fit the picture. The picture does not prima facie fit some cases that
can be classed as emotions, like a contemplative feeling of awe. In order to
draw the limit of the family circle more distinctly, however, we need a more
detailed account of the motivational bodily states that underlie the
M-emotions. As a matter of empirical fact, the bodily states involved in
our M-emotions are primarily states of the central nervous system. Our
claims about the motivational structure of M-emotions need not be taken
solely on the basis of faith in our intuitions. There is empirical support for
this perspective. It is in agreement with an ascendant research program in
the behavioral and brain sciences called the affect program theory.

II. The Affect Program Theory


In recent years, a powerful kind of empirical theory of some emotions,
typically including anger, fear, joy, sadness, disgust, surprise, erotic love/
lust, and agapic love/care, has been developed and a growing body of
evidence has been gathered to support it. Here, we can only describe this
theory in a very limited and general way. The theory is called the affect
program theory, and we may call the emotions that come under its scope
the AP-emotions.3 The affect program theory is the view that some of the
states of mind that we call emotions manifest themselves as syndromes, or
collections of features, which include physiological responses, stereotypical
actions, and perhaps even normal cognitive roles. Instead of reductively
explaining emotions in terms of one of these types of features, the affect
program theory holds that these elements occur together, caused by and to
some degree controlled by a central program. The psychologist Paul Ekman
defines the central concept of the theory as follows: 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. . . .
Thus, the affect programs cause, and to varying degrees control, the
skeletal, facial, vocal, autonomic, and central nervous system changes that
occur initially and quickly for one or another emotion. . . . (1980, p. 82)
One source of evidence for the affect program theory is neuroscience.
Recent advances in the neuroanatomy of emotions show that there are
dedicated neural systems underlying the functioning of AP-emotions, and
reveal relevant facts about the nature of these systems. (cf. Panksepp, 1998)
Three findings are of special importance to our concerns here. First, it is
known that these systems are to varying degrees independent of the abilities
that we typically take to constitute higher cognition, in that they can and
often do operate independently of the systems underlying higher cognition.
Second, these systems are not unique to human beings. They have close
110 NOUS

homologues in nonhuman animal speciesfor some systems, even in all


mammals. Third, direct stimulation of the relevant neural systems can be
sufficient to cause distinctly emotional behavior, and it has this effect largely
independently of corollary stimulation of brain structures thought to enable
our cognitive capabilities. (King, 1961; Frijda, 1986; Panksepp, 1998) Thus,
neuroanatomy shows that at least some emotional responses, including
some complex behaviors, are primarily consequences of these neural pro-
grams, and do not require, nor are necessarily coordinated by, cognitive
abilities.4
The affect program theory is also in part an evolutionary theory, accord-
ing to which the affect programs necessary for the basic emotions are
heritable, and their normal eliciting conditions and responses are to be
explained on the basis of hypotheses concerning the evolutionary function
of those emotions. As Panksepp puts it, The underlying circuits are genet-
ically predetermined and designed to respond unconditionally to stimuli
arising from major life-challenging circumstances. . . . These circuits organize
diverse behaviors by activating or inhibiting motor subroutines and con-
current autonomic-hormonal changes that have proved adaptive in the face
of such life-challenging circumstances during the evolutionary history of the
species. (1998, p. 49)
The affect program theory strongly suggests that the proper way to
consider all the AP-emotions is as natural, biological kinds. Each of these
kinds exists in virtue of a physiological basis and an evolutionarily-
grounded functional basis. The actual determination of which emotions
have affect programs, of course, is an empirical matter, but we take it as a
partially confirmed hypothesis that at least anger, fear, joy, sadness and
sexual attraction are such natural kinds, since plausible candidates for both
the neural systems and the functional roles of each of these have been
identified.
Since different AP-emotions manifest themselves as different syndromes,
the underlying neural machinery, normal eliciting conditions, and normal
responses of each one must be studied and delineated separately. As an
example, we can briefly consider fear. It is known that fear is elicited by,
among other things, the expectation of painful stimuli. Fear results in a
variety of responses, which prepare the body for action, including the
release of adrenalin, increase in temperature, heart rate, blood pressure,
and muscle tension. (Ekman et al. 1983, Frijda 1986) Fear also reliably
generates behavioral responses, which fall into a handful of distinct func-
tional categories (with some variations in specific species), including flight in
response to powerful stimulation (e.g., the expectation of very noxious
stimuli), and freezing in response to more mild stimulation. These responses
can be reliably and predictably generated in laboratory conditions. The
neural systems enabling fear are extremely complex, but great strides have
been made recently in understanding these systems. We now know that the
On Emotions and the Explanation of Behavior 111

core fear system extends from the temporal lobe through the anterior and
medial hypothalamus to the lower brain and specific autonomic compon-
ents of the brain stem and spinal cord. (Panksepp, 1998, p. 213) Direct
stimulation of parts of this system can initiate fear responses. The role of a
brain structure called the amygdala in fear conditioning and related kinds of
learning and memory formation has been well studied, revealing that fear
conditioning normally integrates but does not require higher cortical func-
tions, and that these higher cortical functions can occur without causing
fear in situations where some elements of the subcortical neural systems of
fear are lesioned. (LeDoux, 1996) This supports the view that fear evolved
(likely prior to the evolution of higher cognitive abilities) as a way to
respond to threats in the environment, and that the behaviors that fear
causes can occur independently of higher cognitive capabilities because they
need not be generated nor controlled by those functions. This growing body
of research provides reason to think that fear is a natural kind, identifiable
in the neural systems that enable it and because of the important functional
roles that it playsroles which are essential explanatory elements in
a number of linked sciences (including neuroscience, psychology, and
ethology).

III. Affect Programs and Emotional Motivation


For our purposes, what matters most is that the affect program theory
grounds our particular philosophical conception of emotion in the biology
of emotion. We have claimed that understanding how emotions motivate
begins with the recognition that the occurrence of some emotional states of
mind (the M-emotions) involves the occurrence of certain distinct, special-
ized motivational bodily states. Affect programs are, or include, motiv-
ational bodily states of just the sort that this conception calls for. If affect
programs play significant roles in motivating emotional behaviors and
actions, then the class of M-emotions is not empty, because the AP-emo-
tions must count as M-emotions. The converse does not hold. Even if there
were no affect programs, or if the neural circuitry identified with the affect
programs turned out not to play the roles attributed to them, there might
still be M-emotions. The philosophical conception calls only for the exist-
ence of some relevant motivational bodily states. Given the present state of
the art in the science of emotion, affect programs are likely candidates.
Although it is a partially confirmed empirical hypothesis that there are
unique bodily statesdedicated neural systems, which can become acti-
vatedcorresponding to each of the AP-emotions, it is not necessary that
there be an analogous distinct bodily state, one for each type of M-emotion.
Rather, we delimit the class of M-emotions as follows. (1) Some M-
emotions involve motivational bodily states in the strong sense that the
occurrence of a motivational bodily state is a necessary condition for their
112 NOUS

occurrence. AP-emotions appear to be of this type. (2) Some M-emotions


involve motivational bodily states, in that their occurrence includes the
occurrence of, or a tendency or disposition for the occurrence of, an
M-emotion of the first type. For example, if a particular instance of hatred
includes active anger, or a disposition to become angry, it too counts as an
M-emotion.5
Next, we must ward off some objections that might be considered reasons
for resisting the perspective we have outlined so far. First, we have heard the
complaint that some emotions are intentional, but the affect program theory
does not explain the intentionality of emotions. The fact that the affect
program theory is consistent with the intentionality of the emotions should
be sufficient to forestall this criticism. We accept the intentionality of
emotions. Our concern here, however, is that some emotions are curiously
motivational, and being told that they are intentional does not explain that.6
According to reductive cognitivist theories of emotion, emotions are com-
plexes of beliefs, and perhaps other elements such as desires. It is sometimes
claimed as an advantage of such theories that they explain the intentionality
of the emotions, as if the intentionality of beliefs and desires were itself
unproblematic. In fact, the intentionality of emotions is no more or less
problematic than that of other intentional states of mind. We need not take
sides in the current debate about theoretical approaches to intentionality in
order to recognize that M-emotions, and AP-emotions, are intentional.
A second reason for resistance is that emotions have variable durations.
Short, rapid conflagrations of anger come and go in moments, but a
persons anger may also smolder through the years. It may be thought
that because affect programs are dedicated to the production of relatively
short-term, patterned emotional episodes, the affect program theory can
only help explain what goes on in some cases of occurrent, hot emotion,
but that it is irrelevant to understanding potentially long term emotional
states of mind. The reply to this objection is that long-term emotional states
of mind can be dispositional states of mind. Since dispositions must be
dispositions to something, dispositional properties are grounded in non-
dispositional properties. Long-term anger involves a disposition to enter
occurrent angry states of mind. Thus, if the affect program theory helps
explain occurrent anger, it helps explain dispositional anger as well.
Furthermore, there need be no disagreement, even with a critic who insists
that there are some cases of anger completely unlike the ones we have
described, which do not involve any disposition to enter occurrent affective
states of mind. It is not our intention to capture every state of mind that one
might call anger under one theory. We only maintain that there are cases of
anger and other emotions that fall under the conception of emotion outlined
here.
Here we must avert a potential misconception about the nature of affect
programs. Affect programs are involved in the production of short-term,
On Emotions and the Explanation of Behavior 113

physiological responses. The misconception is that this is all they do.


Panksepps description (quoted above) makes it clear that affect programs
are responsible for the production of a broad range of emotional responses,
including complex coordinated behaviors with significant duration, involv-
ing the integration of perceptual and motor capacities. This point is absent
from Ekmans definition (also quoted above). The reason for this is that
Ekmans research focuses on the role of affect programs in generating
emotional facial expressions, which occur rapidly at the very onset of
emotional response. This role is compatible with additional roles in the
generation of more complex, coordinated and sustained behaviors. Ekmans
claim that affect programs cause and control responses that occur initially
and quickly for emotions should not be taken to imply that affect pro-
grams are only operative briefly at the onset of emotion. The point is rather
that because affect programs activate rapidly, some emotional responses,
such as facial expression, are not under voluntary control at the onset, yet
they can come under voluntary control shortly thereafter, as when a facial
expression is suppressed or modified. When a genuine smile endures under
awareness and voluntary control, we know of no reason to doubt that
the underlying affect program is causally operative well past the onset of
emotion.7
Finally, our position is not that all expressions of emotion are innate, in
any of the senses of being unlearned, inflexible, or invariable from person to
person or culture to culture. While some emotional responses, such as the
basic facial expressions, are pan-cultural and even present in newborns, the
specific ways in which people express their emotions bear the influence of
culture and personality. That is, the ways in which we express our emotions
are determined in part by our individual learning histories.

IV. What Expressions of Emotion Have in Common


The best way we know of to approach the question of what the variety of
expressions of emotion have in common is to look for a shared causal basis.
As a likely matter of empirical fact, the diverse expressions of a particular
AP-emotion have a shared causal basis. For each type of AP-emotion, there
appears to be a particular bodily state, or collection of statesthose
involved in the activity of the particular affect programwhich operates
when one is in the grip of that emotion. These states, at least partly, cause
the many expressions of that emotion. For instance, whenever one is angry,
one enters a motivational state capable of producing, or contributing to, a
range of effects such as gritting ones teeth, raising ones voice, stomping
from the room, or attacking a newspaper vending machine. We know of no
reason, however, to think that there is anything significant that unites the
class of expressions of emotion as a whole. Even if we limit our view to the
114 NOUS

AP-emotions, we know of nothing, for example, which expressions of anger


have in common with expressions of joy or sadness.

V. Explanatory Consequences of the View


There are two, closely related, consequences of the conception of emotions
as involving motivational bodily states for the explanation of behavior.
First, attributions of emotional states are a major source of these explan-
ations. For a wide range of behaviors, such explanations are sufficient. For
certain behaviors, including some emotional actions, given some (often
minimal) background knowledge about a persons situation and the knowl-
edge that the person is in the grip of an M-emotion, nothing further is
needed to explain why a person engages in a behavior. Thus, that one is
angry can suffice as an explanation of why one stomps or shouts. Stomping
and shouting simply fall within a normal range of expectable outcomes.
Admittedly, we cannot normally predict with great accuracy the outcome of
a persons anger. So, we may not know precisely whether a person will
stomp or shout. However, when nothing in the expectable range comes of a
persons anger, that is unexpected, and it does require further explanation.
This is a consequence of the default motivational character of such
emotions.
A second consequence is explanatory independence between attributions
of emotions on the one hand, and attributions of beliefs and desires on the
other. M-emotions have independent motivational power. Because they
involve distinct motivational bodily states, they do not need the assistance
of beliefs and desires in order to intelligibly move a person to action. Thus,
some emotional actions can be explained without reference to an agents
beliefs and desires, and even where the explanation of an emotional action
makes reference to an agents beliefs or desires, a correct explanation may
require additional reference to the agents M-emotions as independent
causal factors.
This point about explanatory independence is consistent with the obser-
vation that as fully cognitive beings we may typically believe and desire at
the same time as we emote. Our emotions may cause us to have beliefs and
desires, and vice versa. Our emotions may cause behavior in combination
with our beliefs and desires. It may even be granted that beliefs and desires
are constituents of some emotions. The commonness of these types of
connections between cognition and emotion make it possible to overlook
explanatory independence, but these cognitive relations do not by them-
selves constitute or cause the motivational power of the M-emotions. For
this reason, a common way in which explanations of some emotional
actions can go wrong is by failing to take into account the M-emotions
independent motivational power.
On Emotions and the Explanation of Behavior 115

The thesis of explanatory independence highlights the existence of a


certain type of explanation, which applies to certain emotional actions,
and which is compatible with the existence of other types of explanation
for other behaviors, including other emotional actions. Explanations of
behavior in terms of an agents M-emotions can be distinguished from
two other classes of explanation. The first class consists of explanations of
behavior that make reference to behavioral dispositions such as drunken-
ness, clumsiness, and drives such as sleepiness and thirst. The second class
consists of those that make reference solely to an agents beliefs and desires.
Explanations in terms of emotions are like explanations in terms of dispos-
itions and drives in that they can be independent of an agents beliefs and
desires. Anger, like clumsiness or fatigue, can be a sufficient explanation of
why an agent lets a cup drop on the table with a clang. However, emotions
are unlike these kinds of dispositions and drives, and like beliefs and desires,
in that they are intentional states of mind. One cannot be clumsy, sleepy, or
thirsty about anything.8
It is a slight difficulty for advancing the explanatory independence thesis
that the concept of desire is a slippery fish. One can adopt a very broad
usage, according to which any intentional, motivational state of mind is
classed as a desire. If this is what one means by desire, then one might
claim that M-emotions are simply a subspecies of desire. It would be a
mistake, however, to oppose the explanatory independence thesis on the
grounds that M-emotions are a type of desire. In advancing this thesis, we
use the word desire in its more narrow and more common sense. By a
desire, we mean a type of evaluative state of mind, which represents a goal,
or preferred outcome for an agent. Thus the ascription of a desire, together
with some means-ends beliefs, reveals what John McDowell aptly calls the
favorable light in which an agent sees a course of action. It is a mistake to
think that M-emotions are desires of this type. The concept of an
M-emotion is not the concept of a desire to achieve a goal or outcome
state. It is rather the concept of a mental state involving a specialized
motivational bodily state that has the function of producing certain
behavioral responses. Thus, as an M-emotion, fear does not motivate an
agent by representing the goal of avoiding danger. Rather, fear motivates by
producing a behavioral response, such as flight. The most compelling reason
to respect this distinction between the motivational structures of M-emo-
tions and belief-desire complexes is that the failure to distinguish them
comes at the cost of poor explanations of behavior. One cannot generally
substitute attributions of desires for attributions of emotions in explana-
tions of behavior. Where one tries to do this, misleading explanations are
often the result, as we shall show in the next section.
116 NOUS

VI. Emotions and Rationalizing Explanations of Behavior


On the view that we have outlined, sometimes a sufficient, complete explan-
ation of an action is just that the agent is in a particular emotional state of
mind. Sometimes, even though an agents beliefs and desires are relevant to
a complete explanation of the agents behavior, the correct explanation
cannot rest with the beliefs and desires alone, but must also take into
account the independent motivational power of the agents emotions.
Explanations of this type are different from rationalizing explanations of
actions. Rationalizing explanations work through the attribution of appro-
priate desires and means-ends beliefs to an agent. They take the general
form, S does f because S desires j and S believes that doing f will bring
about j. In two ways, the view outlined here challenges philosophies of
action, which give primacy of place to rationalizing explanations. First, it
contradicts the strong claim that all actions are to be explained by ration-
alizing explanations. Second, because many explanations of behavior mix
attributions of beliefs, desires and emotions, the view challenges the idea
that there is a sharp distinction to be drawn between rationalizing explan-
ations of action and design level explanations of behavior.
The extent of disagreement with those who would offer rationalizing
explanations of all emotional action can be clarified by comparing our
view to Rosalind Hursthouses. (1991) Hursthouse has argued convincingly
that emotions sometimes cause arational actions. Arational actions are
intentional in that they are done in full awareness of what one is doing,
and in that they are voluntary, since one can prevent them from being done,
if one chooses to. Yet, arational actions are not appropriately explained by
rationalizing explanations in terms of the agents desires and means-end
beliefs.9 Hursthouse illustrates this point with the example of a grieving man
who presses himself to the clothes of his deceased wife. The action is
appropriately explained by citing the mans grief and not by citing any
means-end beliefs and desires, which the man actually has. On our view,
arational actions are perfectly intelligible. If emotions have motivational
power independently of beliefs and desires, there is no reason to expect
that rationalizing explanations should always be available for emotional
actions.
Our view has this further consequence. Because they involve distinct
motivational states, emotions can motivate independently, even in the pres-
ence of rationalizing beliefs and desires. Even where a rationalizing explan-
ation is available, sometimes it is superfluous. It arrives too late on the
scene. In some other cases, a rationalizing explanation is only a partial
explanation, since it leaves out some of the operative causes. For example,
in a case of over-determination, one may flee because one is afraid of a
dachshund and also because one wants to avoid being bitten and one
believes that by fleeing one can avoid being bitten. In such a case, fear
On Emotions and the Explanation of Behavior 117

may be a contributing cause to the action; it may also be the main cause of
the action.
To see the advantage of our view, it is helpful to consider how advocates
of rationalizing explanations have treated some relevant cases of apparently
arational action. Michael Smith (1998) maintains that all actions can be
given rationalizing explanations in terms of agents beliefs and desires.
Emotions enter into explanations of actions, but where they do, they merely
supplement and presuppose rationalizing explanations. Smith argues
that the grieving mans action of pressing himself to the clothes of his
deceased wife is, in fact, explicable in terms of the mans beliefs and desires.
To give such an explanation, Smith says, we look among the mans beliefs
and desires to find an appropriate pair, which will give an answer to our
question of why he is doing what he is doing. What Smith finds is that the
man is doing what he is doing because he desires to press himself to his dead
wifes clothes and believes that he can do so by doing just what he is
doing. (p. 22) Smith admits that this explanation attributes a relatively
bizarre belief-desire pair to the grieving man, and that we will not find it
completely satisfactory on its own. The explanation can be supplemented
with the information that the man is grieving, which adds further intelli-
gibility because grief at the loss of a loved one is by definition, a state in
which we are disposed to think, and to desire and to do all sorts of things.
There is no basis for Smiths claim that an explanation in terms of grief
supplements and presupposes a rationalizing explanation. There are, how-
ever, two strong objections to Smiths explanation. First, the rationalizing
explanation is a bad one. It is bad because, so long as it can be assumed that
one is aware of performing a certain action, it is always possible to explain
that action in this way, by attributing a desire to perform that action and a
belief that one can achieve it by doing whatever one does. Any explanation
of a type, which is not so generally available would be preferable. Second,
the rationalizing explanation is superfluous. Given that Smith accepts that
grief is by definition a state in which we are disposed to desire and to do
all sorts of things, why not admit that this could be a case in which
someone just does one of those things, without appealing to the elements of
the rationalizing explanation? That the man grieves for his wife is already
sufficient explanation. This is exactly why the rationalizing explanation
sounds so bizarre; it comes along too late.
Hursthouses original account of the case as involving an action that can
be explained in terms of grief alone is adequate, and preferable to Smiths
revision. In arguing for this point, however, it is no use to contest Smiths
explanation by insisting that one must be wrong in attributing to the man
the desire to press himself to his wifes clothes, and the means-end belief that
he can accomplish this by doing whatever it is that he is doing. Belief-desire
explanations are notoriously easy to forgehence, Daniel Dennetts sugges-
tion that one might get away with taking the intentional stance to a
118 NOUS

stationary rock by claiming that it desires to stay put and believes that it can
stay put by doing whatever it is doing, provided that one can say this with a
straight face. The point here is rather that the case of the grieving man is a
plausible example of an emotional action, which need not be given a
rationalizing explanation, and which can be given an adequate, simpler,
and better explanation of the action in terms of an agents emotions.
It might be objected that the mans beliefs must enter into the explanation
of some aspects of his behavior. Surely, it is relevant to the mans action that
the man believes that these are his wifes clothes. It is not our intention,
however, to deny the plausiblity of the claim that the man has such beliefs, or
to deny that these beliefs are relevant to explaining some of his actions. This
particular belief certainly can enter into an explanation along with the mans
grief, serving to explain, for example, why he presses himself against these
clothes and not some others. But this belief is not a means-ends belief which
might rationalize his action. So it is not the basis for a rationalizing explana-
tion, which might compete with the emotional explanation of the mans
action. It is fully consistent with our view that an agents beliefs can explain
some aspects of the behavior, which is motivated by the agents emotions.
The advocate of rationalizing explanations may respond that there are
more plausible ways than Smiths to explain the case of the grieving man by
attributing beliefs and desires to him. The flaccidity of the framework of
belief-desire attribution ensures that with just a little ingenuity one will be
able to come up with any number of alternative explanations. For example,
perhaps the man desires to maintain a connection with his lost wife, or
perhaps he desires to have an experience that makes his memory of her more
vivid. Although clearly people can have desires like these, and they can act
on them, it would be a mistake to insist that one of these explanations must
be correct, and that we cannot make do with an explanation in terms of
grief alone. For the emotional explanation is the simpler one, and the
rationalizing explanations come with a cost. It is not just that the choice
between such explanations appears to be arbitrary in light of what we know
about the case. In attributing such complex desires in order to explain the
case, we attribute a specific interpretation of the significance of the mans
action to the man himself. Yet the man may or may not make any such
interpretation. If he does not, the rationalizing explanation is misleading.
The very indiscipline of the belief-desire framework, which makes it so easy
to invent alternatives to explanations in terms of an agents emotions,
creates this risk.
The difference between emotional explanations and belief-desire explan-
ations is further illuminated in another of Hursthouses cases of arational
action, in which hateful Jane scratches out the eyes of a photograph of Joan.
According to Hursthouse, Janes hatred for Joan explains her action as well
as anything could. Peter Goldie detects a difference between this case and
the case of the grieving man. (2000, p. 28) While the grieving mans action is
On Emotions and the Explanation of Behavior 119

primitively intelligible given his emotional state, what Jane does is not
primitively intelligible. Nor is it primitively intelligible that Jane desires to
scratch out the eyes of a photograph because she hates Joan.
Goldie suggests an explanation of hateful Janes action as expressive of
a wish. In a wish, one desires something and also imagines, or is disposed
to imagine, the desire to be satisfied. So, hateful Jane desires to scratch out
Joans eyes, and imagines herself doing this as she scratches out the eyes of
the photograph. Goldie is careful to distinguish this type of explanation
from a rationalizing explanation, since we should not attribute any bizarre
belief to Joan about how scratching the photograph would satisfy any of her
desires. Still, there is a rather cognitive slant to this explanation. People may
have wishes of this type, and this may be a partial explanation of the case.
However, the wish-explanation is incomplete, because it does not explain
why the desire to harm Joan together with some imagining should result in
any action at all. Not all imaginings are cases of active pretending, as this
one appears to be. Why does this particular pairing of desire and imagin-
ation bring about an action? The motivational power of Janes hatred of
Joan (or at least of the anger involved in her hatred) is surely at the root of a
satisfactory answer.
We would like to suggest an alternative explanation of Janes action as
the result of a redirected emotion. The concept of redirection has an
established role in explanation in ethology.10 (Tinbergen, 1961) An example
is the case of the lioness. Irritated upon having her tail bitten by a cub, she
turns to a tree trunk and gives it a good scratching. What has happened is
that an aggressive response to the cub has been redirected at another target.
What works to spare the cub also spares the top of many a child whose
exasperated parent pounds a tabletop instead. An explanation in terms of
redirected anger makes what the lioness does intelligible. This suggests that
there may be no need to invent cognitive epicycles to obtain an explanation
of Janes behavior. Unlike the lioness, Jane surely can make wishes, but like
the lioness, she neednt. An agents beliefs, desires and powers of imagin-
ation may well constitute some of the restraints, which account for the
redirection of emotional behavior, but sometimes the best explanation
requires less cognitive apparatus.11
The ease with which we find ourselves introducing rationalizing explan-
ations for behavior is sometimes a source of confusion and error. During
the war in Bosnia there were reports of drunken prison guards who beat
captives to death and then continued to beat them until their corpses were
fully disintegrated. One report speculated about the guards motives.
Perhaps they were motivated by the desire to demonstrate complete power
over their enemiespower sufficient to determine the very forms of their
bodies. Perhaps it was the desire to deprive their victims of everything down
to their last recognizable features. Perhaps, but what if the guards had less
imaginative motives? The guards were surely in the grip of a terrible received
120 NOUS

ideology, which gave them some relevant means-ends beliefs. This may have
contributed to what the guards did, and may have fueled some of the more
proximate causes such as anger and hatred. Nevertheless, at the time of the
beatings, the guards may have been motivated primarily by intense rage,
which did not abate until long after their victims had died. Perhaps they
were also motivated by joy or sexual arousal at what they did, and this too
would not fade until long after the beating had lost any other point. The
tendency to over-intellectualize the motives behind terrible actions is famil-
iar. Where an action is particularly terrible, some people prefer an explan-
ation, which appeals to a motive that is particularly thoughtful, imaginative
or elaborate. Such preferences, however, may not track the truth.
Although we may sometimes make rationalizing explanations of the
belief-desire variety when they are inappropriate, it is our contention that
this is far less common than defenders of reductive cognitive theories of the
emotions might lead us to believe. Our common sense folk methods of
explaining behavior include alternatives to rationalizing explanations. We
often describe the actions of others through attributions of emotions. We
also understand ourselves in terms of emotions. Thus, it would be an
explanation both for us, and for the grieving husband, that he acts the
way he does because he is grieving, just as it would be an explanation for
us and for Jane to say that she is consumed with hatred for Jane. Because
people understand themselves as emotional beings and know about the
regular motivational roles of their emotions, peoples actions can be intel-
ligible to them as appropriate responses to their emotions. Thus, their
emotions can make sense of their actions from their own points of view.
That the nature of some emotions is in part inherited, so that their motiv-
ational roles may in part be illuminated by biological and psychological
theories does not vitiate this point.
Giving primacy of place to rationalizing explanations leads to intellec-
tualizing the explanation of behavior. This can be an attractive way of
making ourselves intelligible because it brings the explanation of behavior
into an idealized space of reasons. It is widely held that this is where
justification happens, and justification is a fine form of intelligibility. Of
course, we dont live solely in such a space of reasons. Since we are
passionate creatures, trying to locate our behaviors in this space alone will
mislead us and make us less intelligible to ourselves. It is worth considering
what follows from this admission. It may be thought that there is no great
consequence of recognizing that some complex behaviors are not amenable
to rationalizing explanations, since we already require different means for
the explanation of various kinds of bodily changes, reflex actions and the
like. These are to be explained in a distinct way, by taking what Daniel
Dennett calls the design stance. Reflection about the behavioral expres-
sions of emotions should call this into question. If some emotions are
grounded in design-level features, for example, in the affect programs, and
On Emotions and the Explanation of Behavior 121

yet we often accept explanations of behavior, which combine belief, desire


and emotion as causes, then perhaps we should rethink the distinction
between the intentional stance and the design stance, between rationalizing
explanations and explanations in terms of mere dispositions or drives. For if
emotions involve motivational bodily states, there are explanations of
behavior that do not fit neatly into either of these categories.12

Notes
1
In the case of some emotions, it might further be argued that occurrence of a motiv-
ational bodily state is sufficient for the occurrence of the emotion, but we do not make any
claim of this type in what follows. In claiming that the occurrence of a motivational bodily
state is necessary for the occurrence of some emotions, our view contradicts the extreme
cognitivist views that all emotions are fundamentally judgments of value, or that they are
completely reducible to belief-desire complexes. We hold that this is an advantage, because
such accounts of the emotions are empirically false. This is not the place to criticize cogniti-
vism about emotion, so much as to develop an alternative way of thinking about some
emotions. For discussions of problems with cognitivism see (Deigh, 1994; Griffiths, 1997;
and DeLancey, 2002).
2
Here we cannot discuss the frequently used distinction between basic and non-basic
emotions. Our characterization of the M-emotions will by-pass this distinction. In particular,
we want to remain neutral on the question of whether we should think of an emotion like
hatred as a complex emotion, which can be given a compositional analysis in terms of basic
emotions such as anger, fear or disgust, and other mental states.
3
The precise scope of the affect program theory is not yet settled. The psychologist Ekman
hypothesizes affect programs underlying anger, fear, joy, sadness, disgust and surprise, emo-
tions for which there are characteristic, pancultural facial expressions. (1980) The neuroscientist
Panksepp claims that there is at present good biological evidence for seven innate emotional
systems: fear, anger, sorrow, anticipatory eagerness, play, sexual lust, and maternal nurturance.
(1998, p. 47) Our list is a merger of Ekmans and Panksepps lists. Griffiths (1997) gives an
introduction to a version of the affect program theory for philosophers.
4
This is consistent with the observation that emotions can be influenced to a great degree
by higher cognition. Clearly, for normal human beings, they are.
5
There is a third way in which the class of M-emotions might be extended. One might
argue that certain other mental states have motivational structures significantly analogous to
that of the paradigm M-emotions. We do not do that here.
6
Chapter 5 of (DeLancey, 2002) discusses the intentionality of the basic emotions. The
structure of an affect program allows that the mechanisms that initiate and guide an emotional
response involve contentful mental representations. This may be the basis of an explanation of
the intentionality of some emotional states of mind.
7
Griffiths argues that the class of emotions as a whole is heterogenous on the grounds that
basic emotions grounded in the affect programs induce certain characteristic short-term
physiological responses, e.g. facial expression, while other higher cognitive emotions do
not. This argument is consistent with the claim that AP-emotions can also motivate complex
coordinated behaviors.
8
The case of thirst can be a little misleading. Since we do talk of a person being thirsty for
a particular type of drink, the word thirst is sometimes used to mean a desire-like intentional
state of mind. We can accept this point, while still distinguishing drives, affects and desires. In
addition to their intentionality, emotions differ from drives in that they are more flexible and
122 NOUS

display greater dependence on learning, especially with respect to the conditions that trigger
them.
9
Some would not call intentional any action which fails to be explicable in terms of the
agents rationalizing reasons. This looks to us like a maneuver to preserve a theory. Though we
maintain that an action can be intentional if it is done in awareness and under a sufficient
degree of voluntary control, we would like a better characterization. This much is apparent:
given the absence of rationalizing beliefs and desires, the excuse I was angry doesnt always
establish that one acted unintentionally.
10
Tinbergen discusses redirected affective behaviors under the title displacement activ-
ities. The phenomena of redirection are surely complex and require further investigation.
Goldie describes symbolic relations between the objects which elicits an emotion and the object
to which emotional responses are directed. (2000, p. 30) These are surely relevant for some
explanations of redirected behaviors.
11
Might not the case of the grieving man be best understood in a similar way? Whatever
else it includes by way of an agents beliefs and desires, grief involves sorrow over the loss of
someone one cares about. A plausible explanation of the grieving mans action is that it is the
result of redirected care for his wife.
12
Earlier versions of this paper were presented to a symposium session at a meeting of the
American Philosophical Association Eastern Division and to the Department of Philosophy at
SUNY Fredonia. We thank our commentator, Laura Sizer, members of the audiences at those
presentations, and especially the anonymous referees of this journal for helpful comments.

References
DeLancey, Craig. (2002) Passionate Engines: What Emotions Reveal about Mind and Artificial
Intelligence, New York: Oxford University Press.
Deigh, John. (1994) Cognitivism in the Theory of Emotions. Ethics, 104 4: 824854.
Ekman, Paul. (1980) Biological and Cultural Contributions to Body and Facial Movement in
the Expression of Emotions. In Amelie Oksenberg Rorty (ed.), Explaining Emotions,
Berkeley: University of California Press, 1980, pp. 3772.
Ekman, Paul, Robert Levenson, and Wallace Friesen. (1983) Autonomic Nervous Activity
Distinguishes Among Emotions. Science, 221, pp. 120810.
Frijda, Nico. (1986) The Emotions, New York: Cambridge University Press.
Goldie, Peter. (2000) Explaining Expressions of Emotion. Mind, 109, pp. 2537.
Griffiths, Paul. (1997) What Emotions Really Are, Chicago: University of Chicago Press.
Hursthouse, Rosalind. (1991) Arational Actions. The Journal of Philosophy, 88, pp. 5768.
King, H. E. (1961) Psychological Effects of Excitation in the Limbic System. In D. E. Sheer
(ed.), Electrical Stimulation of the Brain. Austin: University of Texas Press. pp. 47786.
LeDoux, Joseph. (1996) The Emotional Brain: The Mysterious Underpinnings of Emotional Life,
New York: Simon & Schuster.
Nussbaum, Martha. (2002) Upheavals of Thought: The Intelligence of Emotions, New York:
Cambridge University Press.
Panksepp, Jaak. (1998) Affective Neuroscience: The Foundations of Human and Animal Emo-
tions, New York: Oxford University Press.
Plutchick, Robert. (1994) The Psychology and Biology of Emotion, New York: Harper Collins.
Smith, Michael. (1998) The Possibility of Philosophy of Action. In J. Bransen (ed.), Human
Action, Deliberation and Causation. Dordrecht: Kluwer Academic Publishers, 1998,
pp. 1741.
Tinbergen, Niko. (1961) The Herring Gulls World, New York: Basic Books.

You might also like