EMOTIONS AND THE EXPLANATION OF BEHAVIOR
Emotions mediate between our apprehension of the external world and
action. Cognition and perception can trigger emotions. Emotions in turn
can affect cognition and preferences and thus indirectly shape behavior.
Whether emotions can affect behavior directly is an unresolved issue.
The term “emotion” harbors various complexities.
First, it may refer to an occurrent event (a fit of rage) or to a standing disposition (irascibility or irritability). The disposition may be strong and yet rarely be triggered, if others learn about it and take care not to provoke it or if the agent is aware of it and avoids situations in which it might be triggered. Both occurrent
emotions and emotional dispositions may, therefore, enter into the explanation of behavior. In this chapter, the focus will be on occurrent emotional events. Second, one may be “in the grips” of an emotion, in the sense of displaying the
characteristic behavioral and physiological patterns of the emotion, and yet be unaware of it. If the culture lacks the concept of depression, a man whose woman-friend has just left him may state, non-self-deceptively, that he is “tired”
(Levy 1973). An emotion such as anger or love may “sneak up” on the agent without her being aware of it, until it suddenly bursts into consciousness. In such cases, we may talk of “proto-emotions”. Although they do affect behavior,
proto-emotions usually have less of a causal impact than the full-blown emotions that are conceptualized by society and acknowledged by the individual. The fact that animals can only have proto-emotions sets limits to the relevance of animal psychology for the study of human emotions.
Third, the term “emotion” should not be identified with the older term “passion” or with the more recent term “visceral factor” (Loewenstein 1996). Passions include most emotions, but hardly sadness, boredom (if that counts as an emotion) and other low-intensity states. At the same time the passions include strong non-emotional disturbances, such as madness, intoxication or addictive cravings. The ancients even thought of inspiration as a passion, because eof its involuntary and unbidden character (Dodds 1951). Visceral factors form an even broader class, including pain, intense thirst or hunger, and an overwhelming desire to sleep or to urinate. Some of the claims made about emotions in this chapter are also valid for these broader classes, whereas others are not.
Fourth, emotion theorists disagree on what emotions are and on what emotions there are. I shall not try to survey the controversy or offer my own solution.
Rather, I want to point to what may be an emerging consensus: the emotions do not form a natural kind. Two emotions, such as anger and love, may be related
as bats and birds or as sharks and whales rather than as bats and whales (both mammals). Although most of the states that we pre-analytically classify as emotions tend to have a number of features in common, they also exhibit many exceptions. For any alleged basic feature, such as sudden onset or rapid decay (Ekman 1992 a), one can provide counterexamples. The implication is that the appeal to emotions in explaining behavior is likely to take the form of mechanisms rather than laws (Elster 1999 a, Ch.I).
Finally, emotion theorists also disagree on what emotions are for. With regard to
the handful of emotions that are uncontroversially observed in all societies, it is tempting to assume that their universal character points to an evolutionary function. It is, however, a temptation to be resisted. Although it is valid to assume that a universally observed feature of human beings has an evolutionary explanation, the latter could well take the form of showing that the feature is a by-product of adaptation (pleiotropy) rather than adaptive in and of itself. It seems clear that fear is adaptive, but less obvious that envy or regret is. Many of the proposed evolutionary accounts are little more than just-so stories. An example is the idea that post-partum depression should be understood as a female bargaining tool in the battle of the sexes (Hagen 2003).
II. Rational choice and emotional choice
To understand the role of emotions in the explanation of behavior, it will be convenient to begin with a brief statement of the rational-choice theory of action. Emotions and rationality are not necessarily opposed to each other. A person may act rationally on desires that are shaped by emotion. Acting on beliefs that
are shaped by emotion will, however, typically be irrational. Although it is sometimes stated that emotions can enhance the rationality of beliefs, no plausible mechanism has been proposed.
What I take to be the standard model of rational choice is defined in terms of the relation among four elements: action, beliefs, desires (or preferences), and information (Fig.1).
The arrows in the diagram have a double interpretation: they stand for relations of causality as well as of optimality. The desires and the beliefs of a rational agent cause him or her to choose the course of behavior that they rationalize. In this Humean approach, no arrows lead to the desires: they are the unmoved movers of behavior. This is not to say, of course, that desires are uncaused, only that they are not the result of any optimizing operation.
The rational agent optimizes in three separate dimensions. He chooses the action that best realizes his desires, given his beliefs about what his options
are and about the consequences of choosing them. These beliefs are themselves inferred from the available evidence by the procedures that are most likely, in the long run and on average, to yield true beliefs. The blocked arrow reflects the fact that desires are not allowed to have any direct impact on beliefs, as they have in wishful thinking or in counterwishful thinking (see Section VI). Finally, prior to belief formation the agent gathers more evidence in an amount that is optimal in light of the agent’s desires and the
expected costs and benefits of gathering more information. The loop reflects the fact that the expected benefits may be modified by what is found in the search itself. We may note for future reference that the costs of information-gathering are of two kinds: direct costs (e.g. buying a catalogue) and opportunity costs (e.g. the risk of being bit by a snake while trying to decide whether it is a snake or a stick).
In the rational-choice model of action, the role of beliefs is limited to that of informing the agent about how best to realize his or her desires. The emotion-based model relies on the fact that beliefs can also generate emotions that have consequences for behavior. Among the couple of dozen emotions that can be robustly identified, each has a specific set of cognitive antecedents that tend to trigger it. Examples (spelled out more fully in Elster 1999 a) include the following:
BELIEF OF EMOTION
A imposed an unjust harm on B B feels anger towards A A imposed an unjust harm on C B feels “Cartesian indignation”
in the presence of B towards A
A is evil B feels hatred towards A
A is weak or inferior B feels contempt towards A
B feels contempt towards A A feels shame
A has behaved unjustly or immorally A feels guilt A has something that B lacks and desires B feels envy A is faced with impending danger A feels fear ??? B loves A
A suffers unmerited distress B feels pity towards A
A has helped B B feels gratitude towards A
An important distinction is between the emotions that are triggered by beliefs about actions (anger, Cartesian indignation, guilt, gratitude) and those that are triggered by beliefs about character (hatred, contempt, shame).
Another important distinction is between interaction-based emotions (e.g.
anger) and comparison-based emotions (e.g. envy).
The third-party emotion that I refer to as “Cartesian indignation” was first identified by Descartes (1985, Art. 195), with the important proviso that if B loves C the indignation is transformed into anger (ibid. Art. 201). Descartes
was also the first to identify (in a letter to Princess Elisabeth of Bohemia from January 1648) what we might called “third-party gratitude”, that is, B’s
positive feeling towards A caused by A’s helpful behavior towards C.
As noted in the table, the cognitive antecedents of love are unclear. According to Stendhal (1980, p.279) B’s belief that A may (but also may not)
love B is a necessary condition for B’s loving A. It is clearly not a sufficient condition. In some cases, at least, B’s love is triggered by B’s perception of
A as beautiful or gracious rather than by B’s beliefs about A. Although B
may come to believe that A has all sorts of wonderful qualities, the belief tends to be an effect of the emotion rather than its cause (ibid., p.287).
Sexual desire, if that counts as an emotion, often requires no cognitive triggers.
It is in fact an important truth about emotions that they can be triggered by perceptions with no propositional content. In a standard example, the mere sight of a snake-like shape on the path may cause fear through a pathway that goes directly from the sensory apparatus to the amygdala, without passing through the cortex (LeDoux 1996). From the point of view of the
social sciences, however, the more important effect of perception occurs when it is added to cognition rather than substituting for it. Abstract knowledge gains in vividness and motivating power when it is accompanied by visual cues. Some ex-smokers who are afraid of relapsing keep color photographs of smokers’ lungs on the walls of their apartment. The sight of a beggar in the street may trigger more generosity than the knowledge of starving children in Africa. Propaganda often relies more on images than on words. Petersen (2005) illustrates this effect with a picture showing a
graphic image of a happy and grinning Serb cutting the throat of a young Albanian boy. The caption urges the reader not to let Serbs return to Kosovo. The poster provided no new information, since everyone knew that the Serbs committed atrocities; it only made the information more vivid and present. As will be shown below, memory of past injustice has more power to trigger current emotion if the harm has left visible traces in the present. There may be some indeterminacy in the triggering and the targeting of emotions. In one case, A intentionally causes B to kill C. Will the relatives or neighbors of C feel anger towards A or towards B? Assume that A is a resistance movement in a German-occupied country, B is the occupational force and C is the as-yet uncommitted population. If the Italian resistance movement kills a German officer and the Germans respond by killing a hundred civilians chosen at random, the anger of the population may be directed towards the resistance and not, as the resistance leaders may have hoped, toward the Germans (Cappelletto 2003). In another case, A offers a privilege to B but not to C. Will C feel anger towards A or envy towards B? Assume that A is the French king before 1789, B the French nobility and C the French bourgeoisie. Tocqueville argued that the bourgeoisie
predominantly felt envy towards the nobility (see Elster 1999 a, pp.192-99 for references and discussion). In both cases, the emotion experienced by C and/or its target are subject to indeterminacy, which may of course be lifted if the relevant features of the situation are identified.
Fig.2 offers a way of thinking about “emotional choice”. Here, the thick arrows represent only causal effects of the emotions, without implying optimality.
The feedback relation between emotion and belief is clear from the diagram. In some cases, the effect of emotion is to strengthen the belief that triggered it, which in turn strengthens the emotion itself. Consider the proverb, “Fear increases the danger”, that is, the perceived danger. Ekman (1992 b) refers to this mechanism as an emotional “wildfire”. An historical example is the
Great Fear that developed independently in seven different parts of France in the spring and early summer of 1789, a process in which the people scared itself, “se faisait peur à lui-même” (Lefebvre 1988, p.56). From an
evolutionary point of view, such behaviors are puzzling: why would the tendency to exaggerate the danger enhance the fitness of the organism? Or to take another example, why would natural selection favor the tendency to perceive non-existent virtues in the object of one’s affection, thereby causing the affection to gain strength? As these remarks suggest, it may be wiser to stick to the search for proximate causes and to remain agnostic about ultimate ones.
References to fear can be ambiguous (Gordon 1987, p.77). Sometimes they refer to prudential fear, as when I stay at home because I fear it’s going to rain . In this case my decision to stay home is explained by the simple belief-desire model in Fig.1. Americans who fear an attack by terrorists may take appropriate (rational) precautions. After the first anthrax cases in 2001, for instance, it may have been quite rational to take precautions when opening letters of unknown origin. In other cases talk about fear clearly refers to visceral fear, in which fear as an emotion is involved in one of the ways shown in Fig. 2. The idea of hope is also open to these two interpretations. Fleeing danger out of visceral fear can be irrational, if it causes us to go from the frying pan into the fire. A plausible example is provided by the estimated 350 excess deaths caused after 9-11 by Americans using their car instead of flying to wherever they were going (Gigerenzer 2004). By contrast, it appears that no excess deaths were caused by people switching from train to car after the attacks in Madrid on March 11 2004. One reason may be that the Spanish were habituated to terror by decades of ETA actions,