Jump to ContentJump to Main Navigation
The Emotional Construction of Morals$

Jesse Prinz

Print publication date: 2009

Print ISBN-13: 9780199571543

Published to Oxford Scholarship Online: October 2011

DOI: 10.1093/acprof:oso/9780199571543.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2018. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a monograph in OSO for personal use (for details see http://www.oxfordscholarship.com/page/privacy-policy). Subscriber: null; date: 21 June 2018



(p.13) 1 Emotionism
The Emotional Construction of Morals

Jesse J. Prinz

Oxford University Press

Abstract and Keywords

Moral judgements are anything but indifferent and this is exemplified in this chapter through the comparison of asking opinion about capital punishment in which a common answer would be the feeling that it is unjustifiable. If it is turned into the question of one's feeling about trees, the answer would probably not be that one feels they photosynthesize. Ethical theorists, on the other hand, are prepared to reject the contrast as the claim that emotions figure into morality can be defined in various ways. There are two distinct emotionist theses and these are discussed in the present chapter arguing first that identifying moral properties cannot be made without referring to an emotion or class of emotions. Philosophical arguments are considered in favor of emotionism, although the cases presented are not intended to be demonstrative proof that the ontology and epistemology of morals can be implicated by emotions.

Keywords:   emotionism, metaphysical emotionism, moral properties, emotions, epistemic emotionism, essential relations, moral judgements, morality, moral development

1.1 Affective Morality

1.1.1 Two Species of Emotionism

Judging that something is right or wrong is not like judging that 3 is a prime number or that trees photosynthesize. We can form those latter judgments without the slightest stirring of passion. We can be utterly indifferent to them. But moral judgments are anything but indifferent. They ooze with sentiment. We are passionate about our values. Consider the questions, “How do you feel about capital punishment?” An appropriate answer might be, “I feel it is completely unjustifiable.” This figure of speech is awkward outside the evaluative domain. We would not ask, “How do you feel about trees,” and answer, “I feel they photosynthesize.” Rightness and wrongness, unlike primeness and photosynthesis, are things we feel.

Of course, many ethical theorists are prepared to reject this contrast. No one can deny that we feel strongly about our moral values, but one can reasonably doubt whether such strong feelings are constitutive of what it is to value or to be valuable. One can agree that moral judgments stir up our feelings while denying that something's status as a moral judgment depends on our having such feelings. One can admit that we feel strongly about moral facts while denying that those facts depend on our feelings. One can contend that there are things we ought to do and ought not to do, regardless of how we feel. The division between those theorists who think feelings are essential to morality and those who think emotions are incidental is perhaps the most fundamental rift in moral philosophy. I side with the members of the first camp. The claim that emotions figure into morality can be cashed out in various ways. I will use the term “emotionism” as an overarching label for any theory that says emotions are somehow essential. The term should not be confused with “emotivism,” which is a specific version of emotionism.

I want to distinguish two dissociable emotionist theses. According to the first, moral properties could not exist without emotions. In other words there is no way to specify the identity conditions of a moral property as such without reference to an emotion or class of emotions. More succinctly, we can say: (p.14)

Metaphysical Emotionism

Moral properties are essentially related to emotions

Defenders of this view are committed to moral realism, if we define moral realism as the view that there are moral facts. When a moral property is instantiated, there is a fact that consists in its instantiation. If one believes in moral properties, it follows that there are moral facts. The metaphysical emotionist embraces moral facts and claims that these facts depend on emotions. Some forms of utilitarianism qualify. Consider, especially, the classical utilitarianism of Bentham and Mill. They define the good as that which maximizes utility, and they define utility as happiness. There are moral facts, on this view, because there are actions that maximize happiness. And these facts are essentially emotional, because happiness is an emotion.

The term “realism” is sometimes reserved for a kind of mind-independence: the fact that a is F is real, on this interpretation, if a's being F does not depend on our regarding a as F. Utilitarians are realists in this strong sense, about good. Call this external realism. Internal realism, in contrast, is the view that a's being F is a fact, but that fact depends on our regarding a as F (see Putnam, 1980). Internal realism is factualism without mind-independence. Some metaphysical emotionists are internal realists. Consider the view that moral properties are secondary qualities. Secondary qualities are response-dependent properties. According to Locke, colors, tastes, and smells fit into this category. Lemons are tart—that's a fact—but they have this property only insofar as they cause a certain tart experience in us when we taste them. Accounts that develop the analogy between secondary qualities and morals have been dubbed “sensibility theories” (Darwall et al., 1992). The most influential recent versions we owe to McDowell (1985) and Wiggins (1987).

Sensibility theories descend from the “sentimentalist” theory of British moralists, such as Shaftesbury, Hutcheson, Hume, and Smith. Hutcheson tells us:

The word moral goodness…denotes our idea of some quality apprehended in actions, which procures approbation…Moral evil denotes our idea of a contrary quality, which excites condemnation or dislike.

(1738: 67)

Hume goes further, explicitly drawing an analogy between morals and secondary qualities:

Examine it in all lights, and see if you can find that matter of fact, or real existence, which you call vice. In which-ever way you take it, you find only certain passions, motives, volitions and thoughts. There is no other matter of fact in the case. The vice entirely escapes you, as long as you consider the object. You never can find it, till you turn your reflection into your own breast, and find a sentiment of disapprobation, which arises in you, toward this action. Here is a matter of fact; but 'tis the object of feeling, not of reason. It lies in yourself, not in the object. So that when you pronounce any action or character to be vicious, you mean nothing, but that from the constitution of your nature (p.15) you have a feeling or sentiment of blame from the contemplation of it. Vice and virtue, therefore, may be compar'd to sounds, colours, heat and cold, which, according to modern philosophy, are not qualities in objects, but perceptions in the mind.

(1739: III.i.i)

Hume's moral theory has features that distinguish it from modern sensibility theories. One difference is that, in this passage, Hume can be read as implying that moral properties do not exist (the question of whether that was Hume's considered view I leave to the scholars). It's easy to arrive at a skeptical view about moral properties if you begin with an antirealist conception of secondary qualities. The Lockean conception of secondary qualities is different. According to Locke, sounds and colors are real, but relational (powers to cause sensations in us). If one is a realist about secondary qualities, one can adopt a realist analogue of Hume's thesis. Contemporary sensibility theories tend to have that flavor. In modern parlance, sensibility theories are committed to perceptivism rather than projectivism (D'Arms and Jacobson, 2006). Perceptivists say that we perceive moral properties in virtue of having certain emotions, and projectivists say we do not perceive them, but instead project them onto the world. As perceptivists, sensibility theorists are committed to metaphysical emotionism.

I will have more to say about sensibility theories below, but I want to turn now to another feature of Hume's moral philosophy. Hume emphasizes the priority of character over action. Being right or wrong is a function of causing certain emotions, but we must distinguish the emotions that matter to moral evaluation from those that don't. The question of which emotions matter is analyzed by Hume as a question about whose emotions matter. Hume thinks that right and wrong are determined by the emotional responses of a person of character:

’Tis only when a character is considered in general, without reference to our particular interest, that it causes such a feeling or sentiment, as denominates it morally good or evil. 'Tis true, those sentiments, from interest and morals, are apt to be confounded, and naturally run into one another. It seldom happens, that we do not think an enemy vicious, and can distinguish betwixt his opposition to our interest and real villainy or baseness. But this hinders not, but that the sentiments are, in themselves, distinct; and a man of temper and judgment may preserve himself from these illusions.

(1739: III.i.ii)

In this respect, Hume's sensibility theory is also an example of another kind of theory: it is a virtue ethics. Some versions of virtue ethics qualify as forms of metaphysical emotionism. Consider the following view. An action is good if and only if it is that which a virtuous person would do. A virtuous person is a person who has certain character traits. Virtuous character traits are or include emotional dispositions. It follows that an action is good just in case it would be performed by an emotional agent.

Utilitarianism, sensibility theories, and virtue ethics all make metaphysical claims about the nature of moral properties. Many of their defenders are also committed to epistemic claims. To recognize a moral fact, one must grasp the corresponding moral concepts. If moral concepts refer to moral properties and (p.16) moral properties are constitutively related to emotions, then it is reasonable to think that grasping moral concepts involves emotions in some way. For example, utilitarians might say you cannot understand what the good is unless you possess the concept of happiness. Thus, for classical utilitarians, moral concepts may be essentially related to emotion concepts. Sensibility theories generally make an even stronger claim. They generally say that moral concepts must be defined, not in terms of emotion concepts, but in terms of emotions themselves. I will refer to this thesis as:

Epistemic Emotionism

Moral concepts are essentially related to emotions

To make this thesis plausible, it is important to draw a distinction between standard ways of possessing moral concepts, and deviant ways. Consider the analogy with color. There is a sense in which a congenitally blind person can grasp color concepts (see Crimmins, 1989). She might master the kinds of sentences that contain color words or she might even detect colors using a special apparatus that converts spectral information into another format. But this is not the way sighted people grasp color concepts. If colors are secondary qualities, we could say that a blind person is unable to grasp colors by their essential properties. A blind person cannot think about colors as such. Epistemic emotionism is supposed to be a thesis about our capacity to grasp moral properties in a standard way. The epistemic emotionist does not deny that there may be other ways of thinking about morality. A Martian without emotions could have deferential moral concepts, for example (“Wrong is what Earthlings call ‘wrong’”). I will have more to say about standard concepts below.

Another point of clarification is in order. In defining epistemic emotionism, I used the phrase “essentially related.” The most obvious form of essential relation is a constitution relation. Moral concepts are essentially related to emotions if they are constituted by emotions. On this approach, token instance of concepts such as WRONG and RIGHT are emotional states or have emotional states as component parts. This is what epistemic emotionists often have in mind. But some epistemic emotionists will want to allow for a dispositional relationship between moral concepts and emotions. They will want to allow that on some occasions people may make moral judgments without feeling anything, but they will insist that on such occasions, the people making those judgments are disposed to have emotional responses. By analogy, suppose you think the concept FUNNY is essentially related to amusement. On some occasions, you may judge, from memory for example, that someone is funny without actually feeling amused. On those occasions, however, you are being sincere only if you are disposed to feel amused when you are interacting with that person.

(p.17) One can be a metaphysical emotionist without being an epistemic emotionist. Classical utilitarians are a case in point. One can also be an epistemic emotionist without being a metaphysical emotionist. Consider those who deny that moral properties exist. If moral realism is false, then metaphysical emotionism cannot be true. But a moral antirealist can defend epistemic emotionism.

Emotivism is a theory of this kind. Emotivists maintain that moral judgments do not describe the world; rather, they express our attitudes. Ayer (1952) says that the sentence “stealing is wrong” is equivalent to saying “stealing!” with a tone of horror. It does not ascribe any property to stealing money; it merely communicates a feeling. Stevenson (1937) defends a slightly different version of emotivism. He says that “stealing is wrong” does assert something, namely that I don't like stealing, but does not merely assert that fact; it asserts it in a “dynamic” way that expresses my dislike emotionally and thereby enjoins you to share in that attitude. Thus, even though Stevenson admits that moral terms express facts (likes and dislikes), their primary function is to express and commend emotions. Emotivism has sometimes been dubbed the boo/hurrah theory, because its defenders sometimes compare moral terms to expletives. Saying that stealing is wrong is somewhat like saying “boo to stealing!” because both “wrong” and “boo” are principally used to convey and prescribe feelings rather than to report facts.

Recent authors have defended more sophisticated expressivist theories. Blackburn (1984) is close to traditional emotivism, but he emphasizes the projective nature of moral judgments. We talk about moral properties as if they were in the world, but do not take on any serious ontological commitment in so doing. Blackburn and the classical emotivists are epistemic emotionists, but they reject metaphysical emotionism. Blackburn's account is often compared to another theory, called norm expressivism, which has been advanced by Gibbard (1990), but the two are importantly different. Gibbard claims that moral judgments express our acceptance of emotional norms. To say that stealing is wrong is to express acceptance of a norm that mandates feeling guilty when I steal and angry if someone else steals. Gibbard's view is different from emotivism because moral judgments do not express emotions directly; rather they express norms that commit us to the appropriateness of emotions. Thus, Gibbard is not strictly an epistemic emotionist; on his view, one might say that moral judgments mention emotions (they express the attitude that I have the right to be angry), but they don't use emotions (they don't express anger).

Epistemic emotionism is a psychological thesis. It is a thesis about moral concepts. The label “epistemic” adverts to the fact that concepts are the psychological tools by which we come to understand morality. But psychology has another dimension. It is the locus of action. And it is in this domain that emotionism shows another face. In order to act, we must be motivated. Emotions and motivation are linked. Emotions exert motivating force. There is clinical evidence that, without emotions, people feel no inclination to act. Damasio (p.18) and Van Hoesen (1983) describe a condition called akinetic mutism, in which patients who have sustained injuries to emotional areas of the brain lie motionless in bed; upon recovery, they report that they were fully conscious, but they felt no emotions, and hence, no inclination to act. Moral emotions may be especially important in motivating decent behavior. For example, there is evidence that guilt promotes helping. In one study, McMillen and Austin (1971) induced some subjects to cheat in an exam, and then they asked those subjects to help score some questionnaires; subjects who hadn't been induced to cheat helped for only 2 minutes, but the cheaters helped for 63 minutes. If you feel guilty about doing something, you will try to make up for it, and if you anticipate feeling guilty about doing something, there's a good chance you'll resist the temptation to doing it. Therefore, if moral concepts contain emotions, then moral judgments will promote behavior that aligns with those judgments.

In philosophical jargon, this means that epistemic emotionism may entail motivational internalism. Motivational internalists believe that there is a necessary connection between moral judgments and the motivation to act in accordance with those judgments (Brink, 1989): if one believes that stealing is wrong, one is thereby motivated to act in a certain way (e.g., to refrain from stealing or work to prevent others from stealing) even if, under some circumstances, those motivations get swamped out by other motivational demands on action. As the name “internalism” implies, motivational internalists think that moral judgments carry motivational force on their own, with no need for help from the outside. For example, if you believe that stealing is wrong, you don't need an overarching desire to avoid the wrong in order to be motivated not to steal. But how might a moral judgment be intrinsically motivating? The answer is clear on an epistemic emotionist picture. Moral judgments contain moral concepts, and, epistemic emotionists claim that there is a necessary connection between moral concepts and emotions. Suppose that the necessary connection is such that tokening moral concepts always results in an emotional state. Empirical evidence demonstrates that emotions have motivational force. Thus, if this version of epistemic emotionism is correct, then moral judgments cannot occur without motivation.

Motivational internalism is a controversial doctrine. My point here is that epistemic emotionists have an explanation of how it could be true. It must also be noted that different forms of epistemic emotionalism would entail different forms of motivational internalism. On the form that I hinted at in my example, motivational internalists claim that moral judgments are always intrinsically motivating. There are also weaker forms of motivational internalism. For example, one might claim that moral judgments are ordinarily motivating, or capable of being intrinsically motivating, or dispositionally linked to motivation. Likewise, epistemic emotionism might come in different varieties. One might have the view that one cannot token a moral concept without tokening an emotion. Or one might have the view that tokening moral concepts disposes us to emotions. And so on. Each version of epistemic emotionism seems to entail (p.19) a corresponding form or motivational internalism. In each case, there is a link between moral concepts and states that are motivating.

1.1.2 Essential Relations

In the definitions just presented, I said that emotionists postulate an “essential relation” between emotions and things in the moral domain. What is it to be essentially related? I chose this phrase, rather than “necessarily related” because there can be some leeway between necessity and essence. Something A belongs to the essence of another thing B if one cannot specify what it is to be B without mentioning A. This formulation does not invoke necessity in a strong modal sense. It does not say that all Bs are necessarily As. One might construe essential relations in this strong way. It is not uncommon for philosophers to think of essences as necessary and sufficient for membership in a category. The kind of essentialism associated with modern philosophy of language has this tone. When Kripke (1980) says that “water” refers to H2O, he means water is H2O in every possible world. This might give the impression that emotionists are committed to the view that emotions are present every time moral judgments or properties are present, just as oxygen in present in every sample of water. That impression is misleading.

Fist of all, there are other ways of construing essences. For example, Boyd (1988) defines an essence as a homeostatic property cluster: a collection of properties that tend to co-occur and promote each other's occurrence. On this view, some particular property could be part of the essence of some kind of thing even though it didn't always occur in every instance of that kind.

Second of all, even on a Kripkean view of essences, the phrase “essentially related to emotions” does not entail that emotions are active whenever there is a moral property or judgment instantiated. Suppose, for example, that moral concepts are constituted in part by dispositions to have emotions. Suppose, further, that such dispositions are essential to moral concepts in a Kripkean sense (in every token of a moral concept in every world, that token is constituted in part by an emotional disposition). It would follow that moral concepts are essentially related to emotions, because they are essentially related to emotional dispositions, and an emotional disposition is a relation to emotion. Essential relations are transitive. As long as the relations in question are not constitution relations, the emotionist can say that there is a strong modal connection between morality and emotion while conceding that emotions and morals are not always co-instantiated.

1.1.3 Strong Emotionism

The two species of emotionism that I have described can be accepted together or separately. I have already mentioned some of the theories that take on one or another species without embracing all of them. A partial breakdown is presented in the Table 1.1.


Table 1.1. Species of emotionism


Kantians reject both forms of emotionism. Morally bad actions are those that I could not will as a universal law. This is not intended as an axiom about my passions or tastes. The bad is not that which I detest. Universalizability is a rational requirement on morality. Certain forms of conduct cannot coherently be universalizable. Kant (1785) gives lying promises as an example. If everyone lied when promising, the whole construct of promising would collapse. Promising makes sense only against a background where promises are generally reliable and honest. Kant sees a similar rational foundation to positive prescriptions. Helping the needy is morally required because one cannot universalize a lack of help for the needy. Everyone is needy or potentially needy some time, so it would be irrational for any one to will a world where no one helps the needy.

Kantians also reject epistemic emotionism, because conceptualizing something as right or wrong is a matter of forming a judgment about what is rational. Generally speaking, one can do that without being in any emotional state. Kant thinks, in making successful moral judgments, we would generally do well to ignore our passions.

Classical utilitarians agree with Kantians in denying epistemic emotionalism. They deny that moral concepts are essentially related to emotions. One could token a moral concept without having any disposition to experience an emotion. On the other hand, utilitarians think that metaphysical emotionism is true. The good is defined in terms of happiness. Emotivism is, in this respect, the inverse of utilitarianism. Emotivists claim that emotions are essential to moral concepts, but they reject the metaphysical thesis. They claim that there are no moral facts.

Utilitarianism and emotivism can be called weak emotionist theories, because they entail one emotionist thesis and not the other. A strong emotionist theory would entail both. Sensibility theory is the most salient instance. Here is a schematic statement of the view:

(S1) Metaphysical Thesis: An action has the property of being morally right (wrong) just in case it causes feelings of approbation (disapprobation) in normal observers under certain conditions.

(p.21) (S2) Epistemic Thesis: The disposition to feel the emotions mentioned in S1 is a possession condition on the normal concept RIGHT (WRONG).

I think a theory of this kind can be defended. I endorse strong emotionism. Much of this book will be dedicated to justifying and elucidating that endorsement.

1.2 Might Emotionism be True?

Evidence from a variety of sources suggests that emotions are central to morality. In the remainder of this chapter, I will focus on evidence for epistemic emotionism, though I will offer some support for the metaphysical thesis at the end. I will add further arguments and responses to objections in the chapters that follow.

1.2.1 Moral Judgments are Accompanied by Emotions

The most obvious reason for taking emotionism seriously stems from the mundane observation that moral judgments are often accompanied by emotions. It is hard to remain dispassionate when you read newspaper stories about child molesters, atrocities of war, or institutionalized racism. The intensity of our emotions is often a very reliable guide to the strength of our moral judgments. For example, crimes against children are often deemed worse than crimes against adults and they also seem to stir up stronger emotional responses.

The emotional impact of moral judgment is apparent from the fact that we tend to avoid bad behavior. Violating moral rules is often advantageous. If we steal things, we get to have them for free. If we cheat on our lovers, we can multiply our pleasures. Even killing can be advantageous; if you enter an essay contest, there is no better way to increase your chances of winning than to kill off the best writers in the competition. As it happens, we don't make a habit of doing these things, even when we can get away with them. Why not? The obvious answer is that doing bad things makes us feel bad.

This is poignantly illustrated by an experiment that Stanley Milgram conducted in the early 1970s. He asked his graduate students to board a New York City subway train and ask strangers to give up their seats. This violates a norm. We ordinarily obey a rule according to which anyone who finds an empty seat first is entitled to that seat. If you found the last free seat at 14th Street, and I board at 23rd Street, I have no right to your seat; it would be wrong of me to ask for it unless I was old, injured, or otherwise incapable of standing without risk. Milgrim asked his students to violate this norm, because he wanted to know how people would react. He had a general interest in obedience. But almost all of his students refused. He could only coax one student into performing the study. That student dutifully boarded the subway and asked people to give up their seats. When he came back, he said that the experience was incredibly difficult, (p.22) and that he could not collect as much data as Milgrim had requested. Rather than asking twenty people for their seats, he stopped at fourteen. The difficulty had nothing to do with the fact that people were uncooperative. On the contrary, the majority of people willingly gave up their seats. The assignment was difficult because it was emotionally painful to break a norm.

Milgrim discovered this for himself after losing patience with his reluctant graduate students and performing the study himself. This is how he describes the experience in a 1974 interview:

The words seemed lodged in my trachea and would simply not emerge. Retreating, I berated myself: “What kind of craven coward are you?” Finally after several unsuccessful tries, I went up to a passenger and choked out the request, “Excuse me sir, may I have your seat?” A moment of stark anomic panic overcame me. But the man got right up and gave me the seat. A second blow was yet to come. Taking the man's seat, I was overwhelmed by the need to behave in a way that would justify my request. My head sank between my knees, and I could feel my face blanching. I was not role-playing. I actually felt as if I were going to perish.

(Quoted in Blass, 2004: 174)

This anecdote illustrates an important fact about moral norms. When we do things that violate moral values, we incur emotional costs.

There is now abundant empirical evidence that emotions occur when we make moral judgments. It is of particular interest that every neuroimaging study of moral cognition seems to implicate brain areas associated with emotion (Greene and Haidt, 2002). Consider some examples. Heekeren et al. (2003) asked subjects to evaluate whether sentences are morally incorrect (such as, “A steals B's car”) or semantically incorrect (such as, “A drinks the newspaper”). In the moral judgment condition, subjects showed significantly more activation in emotion areas. In a similar study, Moll et al. (2003) had subjects make “right” or “wrong” judgments about both moral sentences such as, “They hung an innocent person,” and factual sentences such as, “Stones are made of water.” Once again, emotion areas were more active for the moral judgments. Moll et al. (2002) also found emotional activation when subjects listened to morally offensive sentences as opposed to neutral sentences (e.g., “The elderly are useless” versus “The elderly sleep more at night”). Sanfey et al. (2003) asked subjects to play an “ultimatum game” in which one player was asked to divide a monetary sum with another player. When the second player judged a division to be unfair, emotional regions of the brain were active. Singer et al. (2006) had subjects watch as electric shocks were administered to people (actually experimental confederates) who had played either fairly or unfairly in a prior prisoner's dilemma game. Areas associated with negative emotions and vicarious distress were more active when subjects watched fair people being shocked. Berthoz et al. (2002) gave subjects stories in which social rules were broken and contrasted these with cases of situations that are merely socially awkward. For example, subjects either heard about a person who rudely spits food into a napkin at a dinner party or about a person (p.23) who innocently spits out food while choking at a dinner party. The social rule violations were associated with greater emotional activation.

The structures that are implicated in these studies include the insula, anterior cigulate cortex, the temporal pole, the medial frontal gyrus, and oribitofrontal cortex, which are all regular players in emotion studies (Phan et al., 2002). Moral judgments and emotions seem to coincide in the brain, just as epistemic emotionism predicts. A natural explanation of these findings is that moral judgments are constituted by emotional responses.

It must be conceded, however, that this is not the only explanation. The Milgrim anecdote and neuroimaging studies show that moral judgments have emotional costs, but that is consistent with two different models of how emotions relate to moral judgments: a causal model and a constitution model. The causal model says that moral judgments can have emotional effects. This is uncontroversial. Anyone who thinks we care about morality might be willing to say that moral judgments cause emotions. Music, sporting events, and sunny weather all cause emotions too, but they are not constituted by emotions. On a causal model, moral judgments occur prior to emotions, and are hence independent of emotions. On the constitution model, concepts such as RIGHT and WRONG literally contain emotions as component parts. This is what epistemic emotionists have in mind. The evidence so far cannot decide between these two possibilities. To support the constitution model, further evidence is needed.

1.2.2 Emotions Influence Moral Judgments

The emotionist can make progress showing that emotions actually influence our moral judgments. If moral judgments comprise emotions, then this influence can be explained. If the judgment that something is wrong contains indignation, then becoming indignant would promote that judgment. By analogy, suppose that the judgment that something is amusing contains amusement. More specifically, imagine that when we judge something to be amusing we are making a judgment of the form “that thing causes this state,” where “this state” is an inner demonstrative pointing to amusement. Becoming amused promotes the judgment that something is amusing by furnishing us with one of its constituent parts.

There are various ways to show that emotions promote and influence moral judgments. Consider, for example, moral intuitions about killing and letting die. We tend to think killing is worse. Why is that? One answer is that killing arouses stronger negative emotions. Think about this from the first-person perspective. If your actions allow someone to die, and this is not your primary intention, you can focus away from the victim and concentrate on whatever your primary intention happens to be. When you imagine deliberately taking a life, you cannot focus away from the victim, so negative feelings brought out by sympathy with the victim are likely be strong and ineluctable. It may be that killing seems worse as a result of these stronger emotions.

(p.24) This idea can explain intuitions about trolley cases (Thomson, 1976). In these thought experiments, we are typically asked to compare two scenarios. In both, a trolley is heading toward five people who have been tied down to the tracks. You are not close enough to free them, but you can save them. In one scenario, you can do this by pushing a person off a footbridge into the trolley's path, killing him, and causing the trolley to stop. In the other scenario, you can save the five by pulling a lever that switches the trolley to another track where you know that one person is tied down, instead of five. In both cases, your intervention would result in there being one death instead of five. Many people have the intuition that it is morally impermissible to intervene in the first case, and morally permissible to intervene in the second (Mikhail, 2000). Pushing someone in front of a trolley seems wrong, but it seems okay to divert a trolley away from five people and toward one. Why is this? A popular answer among philosophers is that killing a person is morally worse than letting someone die. In the pushing case, we are killing someone, but in the lever case we are merely allowing someone to die. Another explanation is that killing just stirs up more intense emotions. We don't want to push anyone into the trolley tracks because doing so fills us with horror, and the negative feeling causes us to think that the action is wrong.

The philosophical answer is compatible with the emotional answer. On the philosophical story, we have two rules: one that says we should not kill and another that says we should save lives, and the former is stronger than the latter. We don't have a rule against letting people die, or at least not a very strong rule. Thus, saving trumps letting die, and killing trumps saving. But what exactly are these rules, psychologically speaking? One answer is that they are grounded in emotions. We have negative feelings about killing, and positive feelings about saving lives, and few feelings about letting die. When considering dilemmas, the stronger feeling wins. This story makes two key predictions. One is that emotions should come on line when considering moral dilemmas, and the other is that our intuitions about what's right should be influenced by changes in the emotional content of the scenarios we consider. If moral rules are grounded in emotion, then factors that alter our emotions should affect our application of those rules.

Greene et al. (2001) have used functional magnetic resonance imaging to measure brain activity as subjects consider trolley cases. They showed significant activation in emotional areas of the brain when subjects were asked whether it is appropriate to push someone off a footbridge into the path of a trolley. Emotion activations were lower when subjects were asked whether it is appropriate to pull a lever that would divert a trolley away from five people toward one person. Greene et al. also note that, in the lever-pulling scenarios, subjects also show brain activations in areas associated with working memory. On this basis, the authors suggest that moral reasoning is driven by two dissociable processes: a cool rational process and an emotional process. I interpret their data differently. I suspect that emotions are involved in both cases. On the emotionist account that I just sketched, we have an emotion-backed rule that it's bad to kill, and (p.25) a somewhat weaker emotion-backed rule that it's good to save lives. In the pushing case, we imagine killing in a very vivid way, and the emotional wallop packed by the “don't kill!” rule overwhelms the weaker emotions associated with the “save lives!” rule. In the lever-pulling case, we don't imagine the harm we are causing very vividly, so the “save lives!” rule can guide our actions. Here the numbers matter. We calculate that pulling the lever will result in more lives saved, and that results in an emotional preference for pulling the lever. The activations in working memory areas result from the fact that our decision depends on thinking about the numbers. We can coolly calculate which course of action will save more lives, but once we figure out that it's morally best to pull the lever, that judgment may be backed by an emotional response. This is consistent with the data. Greene et al. found that emotions are active during both the pushing scenario and the lever scenario. Emotions are more intense in the pushing case, but that's no surprise: pushing someone to his death is a very evocative activity.

If I am right, we deliberate about moral dilemmas by pitting emotions against emotions. Conflicting rules have different emotional strength, and the stronger emotions win out. If that's right, then it should be possible to alter intuitions about trolley cases by changing the scenarios in emotionally significant ways. Here's a prediction. When subjects say it is morally acceptable to pull the lever to save five people and kill one, they are imagining that the lever is far away from the tracks. Now suppose we tell subjects that the lever is just a few inches away from the person who would be killed if the lever were pulled. Imagine yourself in that situation. A man is tied down to the tracks right next to you. You cannot free him. He is writhing around and howling in terror. You know that there are five people on another track, which is some distance away, and you know that the trolley is heading that way. Would you sacrifice the person at your feet? Would that be morally acceptable? Here, I think intuitions would change. This is more like the pushing case. People who had not been exposed to many of these examples would, by default, have serious moral misgiving about sacrificing the life of someone inches away. The strong emotions elicited by proximity to the victim would, I predict, influence the judgment.

Conversely, we can imagine an emotionally attenuated variant on the footbridge case. Now you are located in a control room, and learn that a trolley is heading toward five people. By pulling a lever, you can open a trap door, causing a person standing on a footbridge to fall in the trolley's path and derail it. In this scenario, no physical contact with the victim is required. This has recently been tested by Greene et al. (forthcoming), and they found that most people think it is permissible to kill the man on the footbridge in this variant. If subjects are told that they have to push the man off the bridge, only 31 percent say it is permissible, and if they are told they just need to pull a lever that opens a trap door, 63 percent think it's permissible. Diminishing the emotional intensity of the method of killing doubles the approval rating.

(p.26) These examples suggest that we are not slaves to a principle that killing is worse than letting die. We normally adhere to such a principle, but a change in emotional intensity can lead us to endorse clear violations of it. Moreover, the principle itself may be partially underwritten by the fact that killing is usually more emotionally charged than letting die. Killing usually involves physically contacting another person and perceptually experiencing that person's suffering. We can let someone die without any contact (as we so often do with distant crises around the world). I am not claiming that there is no moral difference between these cases (for that view, see Kagan, 1989). My point is that our moral intuitions about such cases are influenced by emotions.

Consider one more trolley case (for a more complete survey, see chapter 7 below and Prinz, forthcoming a). When you refuse to push a person in front of a speeding trolley to save five lives, you are making a deontological moral judgment. You are siding with those moral philosophers who claim that intentionally killing a person is wrong regardless of the consequences. You must obey the principle of humanity: you cannot use a human being as a means, rather than as an end. But such deontological intuitions can, famously, be overridden by changing the numbers. Suppose that, instead of five people tied to the track, the trolley is filled with powerful explosives and heading toward a village where it will detonate, killing five hundred people. Now it seems that pushing the person into the tracks and causing the trolley to derail would be morally commendable. We shift from being deontologists to being consequentialists. This switch in intuitions is an embarrassment for philosophers who think that deontological theories and consequentialism are in competition. But suppose that neither theory is right. Suppose that the concept of the good is not the concept of bringing about the best consequences or the concept of strictly following rules that obey the principle of humanity. Suppose instead that the concept of the good is the concept of that which causes strong emotions of approbation. In some cases, the action proscribed by deontological principles causes approbation, and in other cases, we approve of the consequentialist demand. In the present example, that shift is explained by the fact that imagining five hundred deaths fills us with an acute sense of horror. The scale of the loss pulls on our heartstrings. The emotional difference between five lives lost and one is big, but not enormous. It is not big enough to outweigh the revulsion we would feel pushing a person into the path of a speeding trolley. But the enormity of loss in the explosives case trumps the revulsion of killing a single individual. I think our emotions are influencing our judgments.

These examples suggest that moral judgments are linked in an essential way to emotions. If emotions were merely concomitants of moral judgments, then they should not influence those judgments. The fact that we are influenced by our emotions is predicted by the hypothesis that emotions are the basis of our judgments and, perhaps, constituent parts.

One might respond to this line of argument by pointing out that, while emotions can guide moral judgments, they need not. A dedicated deontologist (p.27) might say, “It would fill me with unspeakable anguish to allow the decimation of a village, but it is still wrong for me to prevent that outcome by taking a human life.” Moral judgments and emotions seem to be dissociable in this way. Doesn't this undermine the emotionist claim?

I will postpone serious discussion of this kind of objection until chapter 3. For now, I will mention four ways in which an emotionist theory could accommodate the deontologist who insists that it's okay to decimate the village. First of all, the deontologist might be quite passionate about the principle of humanity. Her emotional investment in the principle that it is wrong to use one person as a means to save others might be strong enough to trump countervailing considerations. Second, moral judgments may depend on particular kinds of emotions, and not others. When the deontologist says it is right not to push the person into the tracks, she may be recognizing that she would feel guilty if she did. If she lets the villagers die, she might feel intense sadness but not guilt. The sadness may be more intense than the guilt she would feel if she pushed the person into the tracks, but it would be the wrong emotion. Non-moral emotions can fuel moral emotions, but careful deliberators can keep these apart. Third, the deontologist may be judging that our emotions are misplaced in this case. By analogy, imagine the anguished victim of a crime who condemns a falsely accused suspect. The anguish causes the condemnation, but it is directed toward the wrong person. Likewise, when we imagine five hundred villagers dying, the anguish causes us to look for a perpetrator, and we may condemn a person whose actions or inactions would seem blameless if we considered the scenario in a cooler moment, with all the facts in. Finally, the deontologist might be self-deceived. Suppose she allows five hundred people to die, and then feels intense guilt. She might continue to insist that she doesn't believe the action was wrong, but we can challenge her self-assessment. We can say, “Clearly, you have moral misgivings about this action; clearly, it seems wrong to you.” We can claim that she is merely mouthing the words when she says her inaction was right. Or one might suppose that she correctly recognizes that the action was right in a non-moral sense (she did as reason demanded), while painfully recognizing that her inaction conflicted with her basic moral values.

Intuitions about trolley cases do not prove that moral judgments involve emotions necessarily, but they suggest that emotions can exert a serious influence on moral judgments. This conclusion gains further support from research on the effects of emotion induction. In one study, Wheatley and Haidt (2005) hypnotized subjects to feel a pang of disgust when they hear either the word “take” or the word “often.” They are then asked to evaluate morally the protagonist of various stories containing one of these two words. For example, they hear about a congressman who “is often bribed” or “takes bribes.” The wrongness evaluations go up when the word choice corresponds to the word that triggers disgust in the subject. In fact, when the trigger word is used in neutral stories, subjects tend to condemn the protagonist as well. For example, (p.28) they hear about a student in charge of scheduling discussions in school, who often picks interesting topics. Subjects who are disgusted when they here the word “often” find this student morally suspect, though they can't say why (“He seems like he's up to something”). In another study, Schnall et al. (2005) asked subjects to make moral evaluations of stories while sitting at a desk that was either tidy or filthy. The filthy desk has an old greasy pizza carton next to it, a chewed up pencil, used tissues, and a dirty beverage cup. Subjects who are good at introspecting their emotions (as measured by a body self-awareness scale that is correlated with emotion awareness) responded differently. Those seated at the filthy desk judged the scenarios to be worse than subjects seated at the clean desk. For example, they gave higher wrongness ratings to a scenario describing a person who accidentally kills his pet dog and then eats it. These effects do not depend on disgust. Lerner et al. (1998) showed subjects film clips that were either neutral or evocative of anger. They were then asked to consider some unrelated vignettes that describe people who perpetrate relatively minor transgressions, such as selling a used car without disclosing a defect. Subjects who viewed the anger-inducing clips recommended harsher penalties for the perpetrators in these vignettes. In addition, some studies have shown that induction of sad moods can lead to more negative appraisals of people (Fogas and Bower, 1987). Conversely, physical attractiveness, which is known to induce positive affect, can promote positive appraisals of people, including appraisals of honesty and integrity (Dion et al., 1972). It has also been shown in jury studies that attractive or smiling defendants are treated more leniently (Darby and Jeffers, 1988).

Together such findings support the case for epistemic emotionism. They suggest that emotions can influence moral evaluations even when the emotions are induced by morally irrelevant factors. This is just what epistemic emotionism predicts. Epistemic emotionism provides a natural explanation of the phenomenon: if moral concepts have an emotional component, then induction of emotions should influence application of those concepts. Compare: if the concept FUNNY contains the emotion amusement, then covertly tickling people should increase their tendency to think that a joke is funny. If the tickling is too obvious, they will attribute amusement to the tickling and not the joke, but, if the tickling is subtle, they may rate the joke as more amusing than they otherwise would.

I am not suggesting that this is a knock-down argument for emotionism. The fact that emotions influence moral judgments does not entail that moral judgments contain emotions. Emotions might influence moral judgments in another way. For example, empathy for the victim of a crime could instill the desire for punishment, and that could lead us to weigh evidence selectively in assigning blame to a suspect. On this view, the assignment of blame would not need to be an emotional judgment in its own right, even though emotions played a role in bringing it about. Emotions play a causal role, here, but they are not constitutive.

I think epistemic emotionism offers a better explanation of how emotions influence moral judgments. In the controlled experiments, there is no question (p.29) about whether the characters described in various vignettes are guilty. The guy who eats his pet dog is clearly responsible for his actions. So emotions cannot be affecting wrongness judgments by influencing the way people weigh evidence. Moreover, there are dissociations between the desire for punishment and judgments of wrongness. Compare a crazed axe murderer to a calculating murderer who uses a gun. Because of his insanity, we may think the axe murderer should be less harshly punished than the gun murderer, but, because axe murders are more gruesome, we may judge that his crimes are more wrong. This is just my intuition, but it would be easy to test.

Of course, the opponent of emotionism could devise other explanations for why emotions influence moral judgments if the desire for punishment doesn't accommodate all the data. I think the main reason for preferring the emotionist explanation is that it fits better with findings that I am about to describe. If emotions were not constituents of moral judgments, but merely exerted a causal influence, then emotions would be neither necessary nor sufficient for regarding something as wrong. The wrongness concept would be something above and beyond the emotions, and hence independent of them. On the causal influence account, there should be cases in which people moralize without having emotions, and there should not be cases in which emotions alone are, on reflection, the sole basis of moral judgment. As we will see, these predictions of the causal influence account are incorrect. The view that emotions constitute our moral judgments fits better with the data.

1.2.3 Dumbfounding

The evidence adduced so far shows that emotions can sway our moral judgment. But epistemic emotionists make a stronger claim. They say that having a moral attitude is a matter of having an emotional disposition. If this is right, then someone should be able to have a moral attitude in the absence of any rational justification. Emotional attitudes should be sufficient for moral attitudes. There is empirical evidence supporting this prediction. People's reflective moral judgments seem to have an emotional foundation. If we ask people why they hold a particular moral view, they may offer some reasons, but those reasons are often superficial and post hoc. If the reasons are successfully challenged, the moral judgment often remains. When pressed, people's deepest moral values are based not on decisive arguments that they discovered while pondering moral questions, but on deeply inculcated sentiments.

This conclusion has been compellingly defended by Jonathan Haidt and his collaborators. Haidt defends a “social intuitionist” account of moral decision-making, according to which we usually arrive at a moral judgment by introspecting our sentiments. Arguments for that judgment are usually contrived after the judgment is made, and play no essential role in arriving at the judgment or in sustaining the judgment.

(p.30) To support this model, Murphy et al. (2000) studied moral attitudes toward consensual incest. They asked American college students to consider a case in which a brother and sister have sex. In the scenario, the siblings consent to intercourse, use contraception, enjoy the experience, and keep it a secret. Eighty percent of the subjects judged that the behavior was morally wrong, but they had great difficulty explaining why. Each time they came up with an argument to show that the siblings had done something immoral, the experimenters explained why the argument fails. Many subjects worried that the couple would have deformed children, but the experimenters reminded them that contraception was used. Some were worried about the effects on the community, but that worry is inapplicable, because the couple in the scenario did not tell anyone what they had done. Some subjects might have complained that the couple would be traumatized, but the scenario specifies that they actually enjoyed the experience and it strengthened their relationship. A few subjects suggested that incest is condemned in the Bible, but none could recall where (certainly not in the story of Lot and his daughters!). Subjects were presented with decisive counterarguments to every argument that they gave against consensual incest. They tended to concede that the counterarguments were successful, but only 17 percent changed their initial moral judgments. The others typically bottomed out in unsupported declarations and emotional exclamations. Incest is nasty! Incest is just wrong: it's gross! Reasons fell by the wayside, but moral convictions and moral emotions were recalcitrant.

Murphy et al. (2000) found the same pattern of responses when they presented subjects with a scenario involving cannibalism. A woman working alone late one night in a medical pathology lab decides to cook and eat a discarded piece of a human cadaver that was donated to the lab for medical research. Once again, subjects say this is wrong, but they cannot articulate reasons sufficient to support that conclusion. They say that their moral appraisal of the case is based on a “gut feeing.”

These dumbfounding results can be interpreted in several ways. One possibility is that subjects have good reasons for their views about incest and cannibalism, but these reasons operate unconsciously. After all, a lot of problem-solving is done unconsciously, and people have limited insight into how they arrive at judgments in other domains (Nisbett and Wilson, 1977; Moscovitch, 1995). Call this the hidden reasons interpretation.

This interpretation strikes me as highly unlikely. We may arrive at our moral assessment of incest unconsciously, but there is no evidence that much reasoning is taking place. It is very hard to know what those reasons would be. It's one thing to say that the reasons are not accessible to consciousness when we initially arrive at our judgments, and another to say that extensive careful reflection cannot gain access to them. Coming up with arguments against consensual incest is hard, and, since there is little public discussion of incest, it is difficult to believe that subjects have internalized arguments from earlier (p.31) reflection or education. Moreover, there is a straightforward explanation of how people arrive at their moral judgments that does not require postulation of hidden reasons. Incest and cannibalism have been emotionally tagged as repulsive and taboo. That fully explains the knee-jerk moral condemnations of perpetrators. And our tenacity in denouncing incest and cannibalism derives from the fact that reasons do not easily override the deeply entrenched emotional responses.

According to a second interpretation of the dumbfounding results, people always base their moral judgments on reasons, but those reasons are sometimes bad. After all, people do offer arguments against incest and cannibalism. It just turns out that their arguments are flawed. Murphy et al.'s results are consistent with the hypothesis that moral judgments derive from reason, rather than passion.

The problem with this interpretation is that people usually don't revise their moral assessment when their reasons are debunked. They recognize that the reasons are flawed but they dig in their heels about the wrongness of consensual incest and cannibalism. This suggests that the reasons they offer did not play a very central role in the formation or maintenance of their moral judgments.

Another possibility is that subjects do not really regard incest and cannibalism as immoral. Perhaps they are just saying these things are wrong because they recognize that to be the prevailing view. Endorsing a taboo behavior in public has serious social consequences, so subjects have good reason to make it appear as if they find incest and cannibalism bad.

This interpretation is also unconvincing. If taboos are powerful enough to make people say that they categorically oppose incest and cannibalism, then they should be strong enough to instill the corresponding beliefs. The issue could be tested by having people answer questions about consensual incest and cannibalism on an anonymous questionnaire. I would predict that subjects would continue to condemn.

A fourth possibility is that subjects have no reasons for their moral judgments. They simply have a gut reaction that consensual incest and laboratory cannibalism are wrong, and a few post hoc rationalizations, which play no important role in driving those reactions.

I think this proposal is almost right, but it's a bit misleading. Usually, if you have no reason for a belief, you are rationally required to give it up, but I don't think that people regard their moral attitudes as subject to this requirement. Values can be basic in a way that places them outside the reason-giving game. People tend to express their views about incest and cannibalism by saying, “It's just wrong!” My guess is that they would say the same thing about killing or inflicting harm on an innocent person. Consider the question, “Why is it wrong to rape a toddler who will never remember the incident?” This is an odd question. It is difficult to answer. It's just wrong to do that. Very wrong. Fundamentally wrong. And morally monstrous. When we say, “It's just wrong” we are not obviating reason; we are implicitly giving one. The “just” in “just wrong” signals that this is a basic value. We have hit rock bottom. Someone who sincerely asserts (p.32) that he does not regard it as wrong to rape a toddler doesn't understand what we mean by “wrong.” He is using the word differently. Compare someone who insists that strawberries are not red.

This reveals something about the practice of reason-giving in morality. When we provide a reason for thinking that some behavior is wrong, we imply that its wrongness consists in the fact that it has a particular property that makes it wrong. But suppose we iterate the why-question. Why is drunk driving wrong? The answer is that it endangers innocent lives. Why is it wrong to endanger? Because danger is risk of harm, and harming an innocent person is wrong. Why is it wrong to harm an innocent person? Here the question becomes odd. Trained philosophers might have views about this, and others may be able to come up with reasons, but it is unlikely that those reasons are the source of the moral intuition. If one could come up with some feature that makes killing wrong, we could ask what makes that feature wrong. At some point, we grasp for straws. At some point the why-question looks misplaced, bizarre, or even depraved. We might say that people have no reasons for their basic values, but it would be better to say that basic values are implemented in our psychology in a way that puts them outside certain practices of justification. Basic values provide reasons, but they are not based on reasons.

I return to basic values in chapter 3 under the label “grounding norms.” I present them here as a way of explaining the Murphy et al. dumbfounding results. People get flustered when asked to explain their condemnation of incest, because this is a basic value. Moreover, basic values seem to be implemented in an emotional way. When we get down to basic values, passions rule. People say incest and cannibalism are disgusting. Murder is abhorrent. Stealing is unconscionable. A typical member of this culture would endure a considerable emotional penalty for committing any of these acts.

1.2.4 Moral Development

If this interpretation of Haidt's findings is right, normal adults have values that are not maintained by a network of carefully thought-out reasons. They are implemented by gut feelings. This picture gains further support from research on moral development.

The most widely discussed theory of moral development has been propounded by Laurence Kohlberg (1984). Kohlberg asks subjects to resolve moral dilemmas. For example, he tells them about a man named Heinz who cannot afford to pay for a drug needed to save his wife from cancer. After unsuccessfully pursuing legal means to get the drug, should Heinz break into a lab and steal it? Kohlberg assesses moral understanding by looking at how subjects justify their responses to such cases. On the basis of this research, he concluded that children go through a progression of stages in moral development. Kohlberg identifies six stages, grouped into three levels. In the first stage, children focus on obedience and (p.33) punishment. They justify moral judgments by appealing to the punitive responses of authorities. After that, children begin to think instrumentally about morality; they think about the benefits to the moral agent. Kohlberg calls these two stages preconventional morality; they are characterized by an egoistic orientation. This is followed by a two-stage conventional level of moral thinking, in which children begin to think about conformity to a group. In the first stage of conventional morality—stage three of the overall sequence—children adopt a “good boy/nice girl” orientation. They begin to focus on how they will be regarded by others. In stage four, there is a focus on law and order. At this stage, there is a focus on duty to fixed rules and the maintenance of social order. Kohlberg thinks that conventional morality can be followed by postconventional morality, but he recognized, empirically, that this final level of development is rarely attained, even among adults. Postconventional morality begins with a fifth stage of moral development in which people focus on social contracts. At this stage, people continue to think in terms of law and order, but now they justify laws by appeal to broadly utilitarian principles. There is a potential sixth stage after that, which Kohlberg characterizes in terms of universal moral principles. These rules are abstract and categorical (like the Golden Rule), rather than concrete and particular (like the Ten Commandments). In other words, Kohlberg thinks that moral development should, but rarely does, bring us ultimately to a Kantian conception of morality. In a longitudinal study in the United States, Colby et al. (1983) found little evidence for reasoning at stages five and six. Most adults reason at stage four most of the time. In a review of cross-cultural research, Snarey (1985) found that stage four was the highest stage exhibited in rural and village societies.

Kohlberg's findings are consistent with the view that emotions are essential to moral judgment. First of all, the relative absence of reasoning at the fifth and sixth levels suggests that ordinary people are neither utilitarians nor Kantians. Standard moral concepts do not seem to be grounded in the kinds of principles that dominate philosophical ethics. This raises some doubts about philosophical accounts in the utilitarian and Kantian traditions. These programs may be better construed as revisionist, rather than as accurate analyses of how ordinary people understand moral concepts (see chapters 3 and 4). Second of all, Kohlberg's first three stages of moral development implicate emotions quite explicitly. In stage one, people express fear of punishment. In stage two, people appeal to hedonic gains. In stage three, people express the desire to be liked by others. Simplifying, one can say that the concept WRONG is sequentially linked to fear, frustration, and ultimately, the anticipated sadness of social rejection.

What are we to say about Kohlberg's fourth level, which is dominant in Western industrialized societies? Reasoning at this level is not explicitly emotional, but that does not mean it lacks an emotional foundation. At the fourth level people appeal to law and order. Appeals to order may have an emotional undertone. People say that moral rules are justified by the fact that society would fall (p.34) apart without them. This justification lacks force if one is neutral about societal collapse. It is natural to suppose that thoughts of societal collapse evoke fear and concern for loved ones. Those who jeopardize social stability pose a threat to well-being, and are thus viewed with contempt or anger. If so, appeals to order may reflect an emotional attitude. Turn now from order to law. Sometimes, instead of raising worries about societal collapse, people try to justify their moral judgments simply by citing the existence of a law or policy. In effect, they say that something is wrong because there is a rule against it. Such people treat rules as if they had intrinsic value. This pattern of justification is actually predicted by epistemic emotionism. If we ground norms in emotional reactions, then our moral convictions lack a rational foundation. Now suppose you give subjects a reasoning task, in which they are asked to justify their belief that φ-ing is wrong, and they find themselves unable to articulate any reasons because the rule is grounded in emotions. At this point, justifications will begin to sound circular. Why is φ-ing wrong? Well, it just is. Put differently, φ-ing is wrong because that's the rule.

The emotionist can explain this kind of rule fetishism. In the course of moral development, we are conditioned to have a strong emotional reaction to the violation of certain rules. Merely thinking about someone violating those rules elicits negative feelings. Thus, the rules take on a kind of obviousness. They are immediately compelling to us, and we assume that they are obvious to others. Evocation of a rule that has been conditioned in this way feels sufficient for purposes of justification. Thus, stage four is like the earlier three stages of moral development, in that all of them make implicit appeal to emotions. The main change in development is that, by the time we reach the fourth stage, we assign emotional significance directly to rules, rather than derivatively. In earlier levels, φ-ing is wrong because it causes negative emotional consequences for me. By level four, φ-ing is wrong because I regard it negatively in itself. There is a transfer of emotions from egocentric consequences of transgression to transgression itself. The developmental change fits beautifully within an emotionist framework.

On the view that I have just described, people at Kohlberg's level four justify moral judgments by appeal to rules because rules are regarded as intrinsically valuable, and rules achieve this status because they are grounded in emotion, rather than reason. To test this interpretation, subjects in Kohlberg-style experiments must be pressed a bit more. When they say that φ-ing is wrong because of a rule against φ-ing, they should be asked, “Why should we do what the rules command?” I predict that most people would have difficulty articulating an answer. As in Murphy et al.'s (2000) dumbfounding research, I would expect people to become befuddled or to express emotions. They should say, “φ-ing is horrible,” or something along those lines. In referring to stage four as “conventionalist,” Kohlberg seems to have a very different interpretation in mind. The label implies that people at stage four think of morality in conventional (p.35) terms. If asked, “Why should we do what the rules command?” a person with a conventional conception of morality should respond by saying “Well, that's what members of my community do.”

In sum, the fact that most people progress to level four can be interpreted as evidence for emotionism. Kohlberg calls this the conventional stage, but I think that label is misleading. I do think moral rules are essentially conventional, but I don't think people view them that way. The fact that people justify their moral attitudes by appeal to law and order does not entail that they regard them as conventions. On the alternative interpretation that I just offered, people at this stage actually regard rules as having intrinsic value. Violating a moral rule just feels wrong. The appeal to law and order in moral reasoning is not an appeal to convention, but rather an appeal to emotionally grounded norms. My emotionist interpretation of Kohlberg's results makes two empirical predictions. It predicts that moral maturation is achieved through a process of emotion training, and it predicts that people at the so-called conventional stage do not really think moral rules hold simply in virtue of societal conventions. Putting these two predictions together, the emotionist account predicts that people come to regard moral rules as different from conventional rules by assigning emotional significance to moral rules. Evidence from developmental psychology supports all of these predictions.

First of all, there is evidence to suggest that moral education is a matter of emotional training. Children are given moral instruction via careful manipulation of emotions. Psychologists emphasize three primary methods used by caregivers to promote good conduct (Hoffman, 1983: Eisenberg, 2000). One method is power assertion. Caregivers punish or threaten to punish their children. Punishment promotes fear, and, by imitation, children who have been punished for doing something bad are likely to become angry at others who behave badly in the future. Another method is love withdrawal. When children do something bad caregivers sometimes express disappointment and refuse to signal affection. This makes children feel sad, and that may be the wellspring of regret. Caregivers also use a technique called induction. They call children's attention to the harms that their misdeeds cause. When children recognize that their actions have made someone suffer, they naturally feel sympathy and vicarious distress for the victim. Each of these methods has been associated with the development of guilt and shame, and with the development of pro-social behavior.

Second of all, there is evidence that ordinary children do not regard moral rules as merely conventional. Smetana (1981), Turiel (1983), and Nucci (2001) have pioneered research on this issue. They have demonstrated that children distinguish between moral and conventional rules. For example, children draw a distinction between rules prohibiting hitting and stealing, on the one hand, and rules prohibiting speaking without raising your hand in class and rules proscribing dress codes, on the other. To establish that this distinction is understood, Smetana, Turiel, and Nucci present children with examples of rule violations, not indicating which ones are moral and which ones are conventional. (p.36) They then ask children questions about seriousness (e.g., How wrong was behavior in the example?), authority dependence (e.g., Would the behavior be wrong if teachers permitted it?), and justification (e.g., Why is it wrong?). Children answer these questions differently for moral and conventional rules. They treat moral rule violations as more serious and less dependent on authorities. Hitting would be wrong no matter what the teacher says, but talking out in class would be fine if the teacher allowed it. Children tend to justify moral rules by appeal to harms inflicted on others, whereas they justify conventional rules by appealing to conventions.

Smetana, Turiel, and Nucci's results can be given an emotionist explanation. Why do children find moral transgressions more serious? Perhaps they have greater emotional consequences. Seriousness may be an emotional assessment. Why do children appeal to harms explaining what's wrong with moral transgressions? Perhaps harms cause sympathy and distress. Why do children consider moral rules independent of authority? Perhaps thoughts about moral transgressions stir up negative emotions, and these remain in place even when children imagine prohibitions being lifted.

Support for this interpretation can be found in the responses that children give to questions about moral rules. For example, in one series of studies, Nucci (2001) asked children if stealing would still be wrong if God said that stealing was permitted. The overwhelming majority of children answered affirmatively. They insisted that stealing would be wrong regardless of what God says. When Nucci asked the children to justify their answers, they tended to appeal to emotions. Here's what an eleven-year-old boy tells us: ‘if people would steal, then the world wouldn't be a happy place…it would still make everybody unhappy…Like when my sister stole my batteries, it really irritated me. If everybody's stuff kept getting stolen, everyone one would be mad’ (Nucci, 2001: 36). This response is typical. A nine-year-old girl explains that stealing is wrong because, “the one who got stealed from would get real angry” (p. 38). A ten-year-old girl echoes this justification: “You're taking another person's stuff and they would probably get upset” (p. 47). There is considerable evidence that, when people attribute emotions, they also experience them (Goldman and Sripada, 2005). It is likely, then, that the children in Nucci's study feel vicarious anger on the part of crime victims. The very idea of stealing makes them feel mad because they imagine the anger of others.

There is evidence that children begin to appreciate the emotional consequences of bad behavior considerably earlier. By the time they are two years old, children show signs of guilt and shame when they do something wrong (Barrett et al., 1993; Zahn-Waxler and Robinson, 1995; Kochanska et al., 1995). In one experimental paradigm, experimenters give toddlers a toy doll that is rigged to fall apart when they play with it. Two-year-olds show signs of self-conscious distress when such “mishaps” occur. For (p.37) example, they avoid eye-contact, they squirm, they hang their heads down, and they cover their faces. Such negative feelings may help children acquire mastery of the moral conventional distinction, which begins to appear shortly before the third birthday. Smetana and Braeges (1990) found that children at that age regard moral rules differently from conventional rules, judging that the former are more generalizable than the latter (e.g., more likely to be followed at other schools). By early childhood, the moral–conventional distinction is very well entrenched. Children come to recognize that moral wrongs are more authority-independent and more serious than conventional wrongs.

At slightly older ages, children become very sensitive to the effects that bad behavior has on others. Arsenio and Lover (1995) described a series of bad behaviors to kindergarteners, and then asked them to pick facial expressions that capture what the victims and observers of those behaviors would feel. Kindergarteners reliably choose facial expressions of negative emotions, such as fear, anger, and sadness. Interestingly, kindergarteners expect that the perpetrators of moral transgressions will feel happy. This is called the happy victimizer effect. Presumably, they think victimizers are happy because their misdeeds achieve intended goals. For example, a person who steals some candy will be happy to have obtained the candy. Kindergarteners do not judge happy victimizers to be any more reprehensible than victimizers who show remorse. They begin to regard happy victimizers as worse than remorseful victimizers by the time they are eight years old. This is an important developmental milestone. It indicates that eight-year-olds do not merely have norms about how people ought to behave; they have norms about how people ought to feel. Happy victimizers commit two wrongs: one in their conduct and the other in their attitude.

This developmental trajectory is probably the product of both nature and nurture. Children are naturally prone to empathetic distress, fear, and personal attachments. When they do bad things, they discover emotional costs. The costs are likely to be especially bad in the case of moral wrongs, since violations of social conventions tend to produce weaker reactions in others, and caregivers respond with more severe forms of discipline (Smetana, 1989; Grusec and Goodnow, 1994; Nucci and Weber, 1995). In this way, children are conditioned to associate moral wrongs with strong negative emotions. This explains why moral rules are psychologically distinct from conventional rules, and it explains why moral rules are regarded as more serious and less dependent on authority. The systematic relationship between emotions and moral comprehension in children supports the hypothesis that moral concepts are emotionally implemented. This story contradicts Kohlberg's claim that people tend to understand morality conventionally. The fact that people appeal to rules in justifying moral judgments is better explained in emotionist terms.

(p.38) 1.2.5 Moral Mary

After reviewing the literature on moral development, one might concede that emotions play a central role in acquiring moral competence, while insisting that this role is merely contingent. Children are too unsophisticated to understand complex philosophical arguments, so they must be emotionally conditioned to behave in accordance with moral rules. But adults are different. We are responsive to reason. To explore this possibility, I want to construct a thought experiment that describes an individual who learns about morality without the benefit of emotional conditioning.

Imagine a woman named Mary who was never exposed to any moral education while she was growing up, but her other cognitive capacities developed normally. She is now an intelligent adult. Imagine that Mary has no intact innate moral attitudes. She doesn't feel guilty or indignant about anything. But she decides that she wants to learn what morality is all about, so she coops herself up in a room with masterworks by Kant, Mill, and other normative ethicists. She learns their theories, and she becomes very adept at identifying the kinds of considerations that they bring to bear. For any action that she considers, Mary is able to determine (a) whether it would maximize utility and (b) whether it would lead to any practical contractions if it were pursued by all agents. Indeed, she can discern any of the facts emphasized by leading normative theories. Now here's the crucial question. Suppose Mary discovers that doing X will in fact maximize utility. Is that sufficient for her knowing that doing X is morally right? Can she wonder whether X is morally required even though she knows that it maximizes utility? The answer is obvious. Mary can wonder. She may be totally unsure about whether X is an action that morality demands. Suppose Mary also contemplates another course of action Y. She knows that doing Y would lead to a practical contradiction if everyone did it; perhaps it requires using another person as a means rather than as an end. Kant would say Y is morally wrong, but Mary can wonder. She knows that Y is practically irrational, but she doesn't know whether it is immoral. Suppose the Kantian and Millian recommendations for action come into conflict. Can Mary decide which option is morally superior? Certainly not. Mary began her training wondering what people are talking about when they moralize, and she is still in the dark; reading normative ethics books proved fruitless. Intuitively, Mary can be a perfect detector of the features that normative ethicists identify as the basis of morality, and she can have no idea whether those features have any moral significance.

Indeed, moral Mary might not even be able to pose the relevant questions. She could not ask herself whether maximizing utility is morally required, because the concept of moral requirement would elude her. She would have no understanding, given her deviant development, of what people mean when they refer to something as “right” or “wrong.” At best, she could ask whether these (p.39) words, used by members of her community, designate the properties discussed by Kant and Mill. She can mouth the words “right” and “wrong,” but she cannot understand them. This strongly suggests that the concepts of right and wrong (or good and bad) are not explicable in terms of the concepts introduced by Kant, Mill, and other normative ethicists.

This thought experiment is just a version of G. E. Moore's “open-question argument.” Moore (1903) tried to refute the thesis that morality can be reductively analyzed, by showing that, for any “natural” property P, it is an open question whether possessing P is good (though what Moore meant by “natural” isn't exactly clear, see Dreier, 2006). In a similar spirit, I am using Moral Mary to argue that it is unlikely that moral concepts can be analyzed in terms of the constructs used in familiar normative ethical theories. Despite this resemblance, I want to distinguish my argument from Moore's because his argument trades on a mistake. Moore was arguing for a metaphysical conclusion. He took the fact that it was an open question whether any property P was good as support for the conclusion that the property of being good is not constituted by any such property. This is certainly a mistake. Since Moore's time, philosophers have come to recognize the existence of a posteriori identities—metaphysical facts that cannot be discovered through conceptual analysis. Before modern chemistry, no one knew that alcohol is a hydroxyl compound, with a molecular structure CnH2n+1OH. The claim that alcohol is a hydroxyl compound is certainly not an obvious truth, because one can possess the concept of alcohol without knowing any chemistry. But it would be a mistake to challenge this identity by pointing out that one can intelligibly wonder whether it is true. Likewise, it was a mistake for Moore to infer that moral properties are non-natural, and it would be a mistake to infer from the thought experiment about Moral Mary that moral properties do not refer to the kinds of properties discussed in the normative ethical theories of Kant and Mill.

Open-question arguments cannot establish metaphysical conclusions, but they can establish conceptual conclusions. If it is an open question whether some natural property N is good, then the concept of N cannot be part of the concept of goodness. Were there a conceptual link between the two, the question would be closed. For example, it might be part of the ordinary concept of alcohol (i.e., the concept used by most adults when thinking about alcohol on most occasions) that drinking alcohol can result in intoxication. If that is part of the ordinary concept, then no one who grasps that concept could intelligibly wonder whether drinking can result in intoxication. Where there is a conceptual link there is no open question. Likewise in the case of Moral Mary. If Moral Mary cannot acquire the ordinary concepts of good and bad or right and wrong by reading about the properties described in normative ethical theories, then it cannot be the case that those properties are constitutive of our ordinary moral concepts. Moral concepts may, for all I have argued so far, refer to Kantian or Millian properties, but it is not a conceptual truth that the right is (p.40) that which maximizes utility or that it is wrong to do that which cannot be universalized.

The Moral Mary thought experiment can also be compared and contrasted with Frank Jackson's celebrated argument against materialism. Jackson has us imagine a brilliant neuroscientist named Mary, who is trapped in a black and white room. Mental Mary knows everything about what happens in the brain when people see red, but this does not help her understand what it would be like to have a red experience. Like Moore, Jackson tries to draw a metaphysical conclusion. He says that the experience of seeing red cannot be identical to a brain state. That conclusion is very controversial. Many commentators think that it is fallacious to infer a metaphysical conclusion about the basis of red experience from epistemological or semantic premises about whether we can infer what red is like from knowledge of the brain. But just about everyone is willing to grant that Jackson's argument establishes an epistemological conclusion. The concepts involved in knowing what experiences are like differ from the concepts involved in knowing about brain states as such: one cannot infer one class of concepts from the other.

Likewise, the Moral Mary argument firmly establishes an epistemological conclusion. The concepts invoked in normative ethical theories differ from the concepts of right and wrong. One cannot infer that an action is morally right or wrong from a Kantian or Millian description of that action. This is enough to establish that Kant, Mill, and other normative ethicists fail to explain moral concepts. That does not prove that metaphysical emotionism is right. We cannot infer that Kant and Mill are wrong about moral properties from the fact that they are wrong about moral concepts. It is possible (though, for reasons given in chapters 3 and 4, implausible) that moral concepts designate Kantian or Millian properties. But conceiving these properties is not sufficient for grasping moral concepts. Mary understands what it is to maximize utility and what it is to universalize a behavioral maxim, but she does not understand what people ordinarily grasp when they use the words “right” and “wrong.”

Unlike Jackson and Moore, I am not trying to establish a metaphysical conclusion; that must be argued for independently. So far, I have tried to establish only a negative thesis about ordinary moral concepts: those concepts are not bound to the properties emphasized by normative ethicists. If moral concepts refer to such properties that would be an a posteriori discovery. I now want to extend the Moral Mary thought experiment to establish a positive thesis. Moore thought that no analysis of moral concepts would lead to a closed question, and he concluded that moral concepts are conceptually primitive. Here too I think Moore was mistaken. Suppose Mary develops the capacity for human empathy. Suppose she starts to feel spontaneously happy when others are happy, and sad when others are sad. Suppose, now, she takes special delight in maximizing the good because she knows it will bring happiness to others. Suppose she feels terribly guilty when she doesn't maximize the good because she knows that (p.41) she failed to bring pleasure to someone who could have been made happier. Suppose she experiences rage at other people when they fail to maximize utility. At this point, it seems perfectly okay to say that Mary believes it is wrong not to maximize utility and right to maximize utility. We can attribute the moral belief to her, once she starts to have an emotional attitude toward utility maximization. Had she instead developed emotional attitudes toward the actions that Kantian ethics deem morally compulsory, we would attribute moral attitudes toward those actions. This is an important positive thesis. We attribute moral attitudes to Mary once she has emotional attitudes. Once certain actions lead her to feel guilty and outraged, we say she has a moral point of view. The intuitions about the moral Mary case suggest that there is a conceptual link between moral judgments and emotional responses. We attribute moral attitudes if and only if a person has certain emotional responses.

This positive thesis can be expressed by saying that, once Mary acquires emotional attitudes, certain questions become closed. But one needs to be careful in stating what the closed question is. Suppose Mary would feel guilty if she were to steal and outraged if someone else were to steal. There seems to be no question about whether Mary has a moral attitude toward stealing. Clearly she regards stealing as wrong. The question of Mary's attitude is closed. If I am right, Moore was mistaken when he claimed that moral concepts admit of no analysis. Ordinary moral concepts are conceptually linked to emotions, and this allows us to say that someone who harbors guilt and anger is taking a moral stance.

In response, one might object that the link between moral concepts and emotions leaves us with open questions. Suppose Mary feels outrage at those who steal. She can still wonder whether stealing is really wrong; she can wonder whether the attitude is justified. Doesn't this open question undermine my claim that there is a conceptual link between wrongness and certain emotions? Absolutely not. The reason why Mary can wonder whether stealing is really wrong has to do with a gap between sense and reference. Many of our concepts are grasped by means of features that are neither necessary nor sufficient for category membership (see e.g., Prinz, 2002). We think of birds as creatures with wings and feathers, but there could be a wingless featherless bird. As a result, we can usually wonder whether a concept that we are applying, on any given occasion, really does apply. Mary knows that she regards stealing as wrong, but she can wonder whether stealing really falls under her concept. Because sense and can come apart, there are often open questions about what a given concept refers to, even if we know that the sense by which we grasp the concept has been satisfied. The open question in the present case is a question about what moral concepts refer to, not about their sense. The fact that we can say Mary is moralizing simply in virtue of her emotional attitudes suggests that there is a conceptual link between moral concepts and emotions.

At this stage, I don't want to rule out the possibility that one can think about morality without possessing emotions. If moral concepts refer to Kantian (p.42) properties or Millian properties, then surely one can. But I think the argument establishes that our ordinary way of thinking about morality depends on the emotions. Above, I introduced the term “standard” concept to refer to an ordinary way of conceptualizing something. The standard concept for a property is the concept that is most commonly used within a community; it is usually the concept that we learned first, and it is the concept we would try to convey if we were teaching someone a corresponding term. We standardly construe birds as feathered, we standardly construe alcohol as intoxicating, and we standardly conceive of colors by imagining the phenomenal qualities that they cause. The intuitions behind the thought experiment here suggest that Mary does not have standard moral concepts until she develops moral emotions. Without moral emotions, she cannot form moral judgments in the ordinary sense. She cannot take the moral stance.

1.2.6 Psychopaths

The Moral Mary thought experiment suggests that emotions are necessary for making moral judgments in a standard way. This is a conceptual argument for epistemic emotionism. Parallel conclusions can be drawn from empirical research. Mary is someone who fails to grasp moral concepts, despite her high level of intelligence. There are people who have this profile in the real world. There are people who can learn everything there is to know about the consequences of an action without understanding that it is immoral. The people I have in mind are psychopaths.

Psychopaths present an important test case for ethical theory. Epistemic emotionists are committed to motivational internalism: the view that moral judgments are intrinsically motivating. Internalism faces a serious challenge. Critics believe that a person could form a moral judgment without being motivated to act. They call such hypothetical individuals “amoralists” (Brink, 1989). When this challenge is offered, internalists and externalists tend to get embroiled in a battle of intuition-mongering. Internalists insist that a person who was unmoved by moral judgments would not really comprehend morality. Externalists disagree. If we use emotional or motivational criteria as the litmus test for moral competence, that would beg the question against externalists. If we reject these criteria, internalists will cry foul. I think psychopaths offer a very promising way out of this quagmire.

Psychopaths seem to be the closest thing we have to real-world amoralists. They are perfectly intelligent and articulate. They seem to comprehend moral values, but they are utterly indifferent to them. They engage in chronic antisocial behavior, from lying and stealing to torturing and killing, and they commit these crimes without emotional cost. When psychopathic killers hear words pertaining to violence, they do not have a normal emotional response (Gray et al. 2003). They also show little empathy, guilt, shame, or remorse (Hare, 1998). Hare (p.43) (1993) and Cleckley (1941) catalogue many examples of psychopaths showing an astonishing indifference to the harm that they cause. For example, one psychopathic murderer insisted that he was blameless because he made a clean wound when he killed his victim.

On the face of it, psychopathy seems to demonstrate that amoralism is not only possible, but actual. Closer examination, however, reveals a very different story. Psychopaths seem to comprehend morality, but they really don't. They use moral terms in a way that deviates strikingly from the way non-psychopaths use those terms. These deviations suggest that they do not possess moral concepts; or at least that their moral concepts are fundamentally different from ours. Here's a passage from Cleckley:

The [psychopath] is unfamiliar with the primary facts or data of what might be called personal values…. Beauty and ugliness, except in a very superficial sense, goodness, evil, love, horror, and humor have no actual meaning, no power to move him…. It is as though he were colorblind, despite his sharp intelligence, to this aspect of human existence. It cannot be explained to him because there is nothing in his orbit of awareness that can bridge the gap with comparison. He can repeat the words and say glibly that he understands, and there is no way for him to realize that he does not understand.

(Cleckley 1941: 40)

Externalists might resist Cleckley's assessment because he seems to presuppose that emotional states are required for comprehension. Cleckley says that values have “no actual meaning, no power to move” the psychopath. If Cleckley is using observations about moral motivation to support the thesis that psychopaths cannot understand moral concepts, then he is begging the question against externalists. Fortunately for the internalist, the evidence for moral retardation in psychopaths is not restricted to evidence pertaining to their motivational states. Crucially, psychopaths' incomprehension of moral concepts can be established using criteria that both externalists and internalists should agree to. Let me explain.

Both externalists and internalists should agree that moral rules differ from other kinds of rules. “Morally right” and “morally wrong” mean something different from “conventionally right” and “conventionally wrong” (for a discussion of how this can be reconciled with relativism, see chapter 5). Anyone who grasps moral concepts should be able to draw this distinction. As we saw already, healthy children have begun to master the moral/conventional distinction by the time they are three years old. If psychopaths fail to grasp the distinction, then they cannot be said to understand what healthy people mean by morality. Concepts are individuated by sense and reference. If psychopaths place moral wrongs and conventional wrongs in the same category, then their concept of “moral wrongs” has different reference and, presumably, different sense from the concept we ordinarily express when we talk about moral wrongs.

Blair (1995) demonstrates that psychopaths fail to grasp the moral/conventional distinction. Using the methods of Turiel, Smetana, and Nucci, he asked criminals (p.44) who had been diagnosed as psychopaths to consider various scenarios in which rules had been violated. Some of the rules were moral and some were conventional, but the psychopaths were not alerted to this fact. They were simply asked to rate the wrongness and seriousness of the violations, and to justify their answers. They were also asked whether the described behavior would have been wrong if an authority had allowed it. The results were striking. Psychopaths did not treat moral and conventional wrongs significantly differently. Unlike a control group of non-psychopathic criminals, they tended to ignore victim's welfare when justifying their answers about moral wrongs.

There was an unexpected finding in Blair's study. Psychopaths tended to treat both moral and conventional wrongs as if they were authority-independent. It's wrong to speak out in class no matter what the teacher says. On the face of it, this seems to suggest that psychopaths are interpreting conventional as moral, rather than the other way around. But Blair offers another, more plausible explanation. He speculates that his subjects were trying to sound as if they were sensitive to moral rules. Inmates have great motivation to do that, because moral sensitivity can hasten release. Psychopaths tended to treat all rules as inviolable in an effort to convince the experimenter that they were mentally healthy. The plan backfired. Non-psychopathic criminals in the control group who were equally motivated to impress the experimenter answered in line with normal subjects. They treated conventional transgressions as authority-dependent. If psychopaths understood the difference between moral and conventional wrongs, they would have treated the transgression-types differently, in order to convey their moral health. The compensatory strategy of feigning moral rectitude actually revealed the profundity of their deficit. This interpretation finds confirmation in another study. Instead of looking at incarcerated adults, who have strong motivation to lie, Blair (1997) administered the moral/conventional test on children with psychopathic tendencies, and they found that these children tend to treat moral transgressions as if they were conventional. Blair concludes that psychopaths fail to grasp the difference between moral and conventional wrongs, and they tend to regard all wrongs as merely conventional. For them, morality is like etiquette, like conventions about which side of the road to drive on, or like chess: a group of more or less arbitrary conventions that place demands on us only because they have been adopted by a social group. Psychopaths can give lip service to morality, but their comprehension is superficial at best.

Psychopaths are not real-world amoralists, so they cannot be used as evidence against internalism. In fact, they seem to furnish internalists with a useful piece of supporting evidence. In psychopathy, a deficit in moral motivation co-occurs with a deficit in moral competence. This suggests that the two are linked. In fact, leading explanations of psychopathy maintain that the deficit in moral comprehension is a direct result of the emotional deficit.

First, consider the explanation offered by Blair (1995). Blair speculates that psychopaths suffer from an abnormality in a psychological system designed to (p.45) inhibit violence. The idea of a violence-inhibition mechanism (VIM) is inspired by work in ethology. Dogs and other animals withdraw during acts of aggression when a victim displays a submission cue, such as bearing the throat. This response could be mediated by a VIM. An homologous mechanism might be responsible for the well-established fact that healthy humans are emotionally disturbed when they see others in distress. The vicarious distress response is even present in infants, but it is seriously impaired in psychopaths (House and Milligan, 1976; Blair et al. 1997). Blair takes this as evidence for a VIM dysfunction. He thinks that psychopaths never master the moral emotions, because they lack the mechanism that makes them sensitive to others in distress. Antisocial behavior is a consequence of impaired violence inhibition and also a secondary by-product of a more general lack of emotional concern for others, brought on by that impairment.

I think Blair's explanation is partially right and partially wrong. Psychopathy involves impaired inhibition, but it is not restricted to violence inhibition. The VIM story explains why psychopaths may lack moral emotions, but it leaves other symptoms unexplained. First, the focus on violence inhibition seems a bit narrow, given that psychopaths engage in all manner of crime, not just violence. This is confirmed by the two diagnostic tools that are most frequently used to identify psychopaths: Hare's (1991) Psychopathy Checklist includes an item labeled criminal versatility, and the DSM-IV criteria for antisocial personality disorder includes items for deceitfulness and failure to meet financial obligations (American Psychiatric Association, 1994). More seriously, Blair's account does not explain a general pattern of general cognitive abnormalities found in psychopaths. For example, psychopaths make numerous errors when completing mazes of increasing difficulty (Schalling and Rosen, 1968). They are tempted to go down paths that will not succeed. And finally, the VIM story does not explain the full range of emotional deficits in psychopathy. Another diagnostic criterion is flattened affect (Hare, 1991). Psychopaths lack moral emotions, but they are also alarmingly deficient in non-moral emotions. Psychopaths show deficits in fear, as evidence by a diminished capacity for electric-shock conditioning and distress-induced increases in startle response (Davies and Maliphant, 1971; Patrick et al., 1993). They also show a deficit in sadness. Cleckley (1941) discusses a teenage psychopath who showed only superficial concern when her pet dog was run over by a car (p. 71), and he also describes a man who shows no sorrow after murdering his mother and numerous others without provocation (p. 266). It has also been shown that psychopaths have difficulty recognizing facial and vocal expressions of sadness (Blair et al., 2001; Stevens et al., 2001).

The pattern of cognitive and emotional deficits in psychopathy suggests that the dysfunction cannot be restricted to violence inhibition. Instead, there seems to be a dysfunction in a more global and rudimentary affective system that underlies numerous emotions and inhibitory processes in cognition. Gray (1987) (p.46) has postulated a Behavioral Inhibition System (BIS) that fits this profile perfectly. It is a core component of negative emotions such as fear and sadness, as well as moral emotions, and it also plays a role in cognitive tasks that require a change or cessation in behavior. Some authors have suggested that BIS deficiencies could explain the core symptoms of psychopathy (Fowles, 1980).

The main difference between Blair's VIM theory and the BIS theory has to do with the extent of the initial defect. Blair attributes psychopathy to a specific kind of inhibition impairment, rather than a general inhibition impairment. Both theories agree on the point of central interest here. They entail that moral retardation results from emotional retardation. Poor performance on the moral/conventional distinction derives from a deficit in moral emotions caused by a deficiency in inhibition. The key point is that both accounts see psychopathy as an emotional disorder. The moral blindness of psychopaths issues from an emotional blindness. If this is right, psychopathy provides positive evidence for internalism. Internalism, I have said, is entailed by epistemic emotionism. If moral judgments are intrinsically motivating, it may be due to the fact that standard moral concepts are essentially emotion-laden. That is precisely what research on psychopathy seems to confirm.

This account of psychopathy is primarily intended as an account of moral concepts, but it may have a metaphysical upshot. If moral properties were not essentially emotion-involving, then there should be a way of drawing the moral/conventional distinction without appeal to emotions. Psychopaths should be able to learn the difference. After all, psychopaths generally have intelligence quotients within the normal range. Some are extremely bright. They are often articulate and cunning. They have a great interest in learning to distinguish right from wrong. Most psychopaths have long histories of misconduct (that is actually a diagnostic criterion of the DSM-IV). These histories put them into contact with people who take special care in helping them grasp morality. They often encounter concerned parents, teachers, lawyers, and law enforcers. Given this combination of exposure and motivation to learn, psychopaths should be more likely than others to develop ways of reliably identifying moral properties. The fact that they fail to master the moral/conventional distinction suggests that there may be no way to draw that distinction without adverting to or experiencing emotional responses. The distinction may be emotional to the core. Rightness and wrongness may be constituted by emotional reactions in us. Subtract these reactions and the distinction becomes as invisible as the color spectrum is to the blind. Psychopaths can carefully monitor the moral judgments of healthy individuals. They can sort familiar examples of good conduct into one bin and bad conduct into another, but they lack insight into the very essence of this division. They would have difficulty reasoning about morality in modal contexts (“would it remain wrong if…?”), and they would have difficulty extending the category to unusual cases. Epistemic emotionism predicts that emotionally deficient individuals will not develop normal moral concepts. Metaphysical (p.47) emotionism predicts that it will be difficult to compensate for that deficiency by acquiring concepts that have the same referents but different senses. Like color concepts, moral concepts may be a case where sense and reference are intimately linked.

1.2.7 The Disunity of Morality

I will conclude with a further argument for metaphysical emotionism, since most of my arguments pertain to the epistemic thesis. Epistemic emotionism does not entail metaphysical emotionism. Moral judgments could be essentially tied to emotional responses even if moral properties were not. The good might have an essence that is independent of our emotions. Compare Litmus paper. It tests for acids by changing color. Litmus paper detects acid by the effect that acid has on it. But acid is not defined by its effect on Litmus paper. Acid has an essence that can be independently characterized. By parity of reasoning, we might detect good and evil by measuring our emotional responses even if good and evil have some independent essence.

Epistemic emotionism would support metaphysical emotionism if there were no mind-independent way to characterize the essence of good and evil. Suppose that good things don't have anything in common other than the fact that they cause positive emotions in us. Suppose evil things share nothing but their capacity to excite our negative emotions. If that is how things are, then metaphysical emotionism is a plausible thesis. Good and evil are essentially response-dependent; they are defined by their effects on us. Compare poisons. There is no intrinsic unity in the class of things that are poisonous. They have wildly diverse chemical constitutions, and their actual effects on the body are varied. Poisons are defined by the fact that they cause illness and death in us. Moral properties could be like that.

How do we decide between these two options? The evidence from psychopathy is suggestive, because psychopaths do not succeed in finding an emotion-independent way to identify moral rules, even though it is in their interest to do so. But the case for metaphysical emotionism should not rest with psychopaths. Their failure to find a mind-independent essence of good and evil might derive from the fact that they don't care enough about values or they don't have enough philosophical training. To prove that morality lacks a mind-independent essence, one would really need to examine philosophical theories. One would need to show that philosophers have been unable to identify a mind-independent essence for morality. I examine some of those attempts in chapters 3 and 4. Here I want to advance the case for metaphysical emotionism by doing some burden-shifting. I want to argue that, on the face of it, morality is more like the poison case than the acid case. The class of moralized behaviors seems to be disunified. If good and evil deeds appear disunified, then opponents of metaphysical emotionism have the burden of proof; they must show that appearances deceive. They must (p.48) establish that good and evil have unifying essences that are independent of human emotional responses.

I will focus on the class of evils. Do immoral acts have anything in common? Ostensibly, the answer is no. Immoral acts comprise a hodgepodge: lying, stealing, hoarding, hurting, killing, neglecting, harassing, polluting, insulting, molesting, vandalizing, disrespecting, and so forth. What do these things have in common other than the fact that we frown on all of them? It is tempting to reply that they are all harmful. Perhaps harm is the essence of iniquity. There are three fatal objections to this proposal. First, the notion of harm is itself response-dependent. There are many things that are difficult or painful, but not regarded as harms: writing term papers, exercizing, spending eight hours at work, venturing out in the cold, saying “goodbye” to a loved one before a business trip, eating spicy food, going to the dentist, and so on. Harms seem to be those psychological and physical ordeals that we regard as candidates for moral concern. Second, some harms are not considered morally bad. Many cultures have harmful initiation rights, and most consider it morally acceptable to harm people in self-defense or retribution. Third, and most importantly, some of the things that we morally condemn do not cause harm; some wrongs lack victims. Consensual incest strikes many people as immoral even though no one seems to be harmed in any obvious way. Consider Murphy et al.'s (2000) scenario in which a brother and sister have consensual sex using contraception. They do it only once, they enjoy it, and they keep it a secret. People tend to regard this as morally wrong, even though no one is harmed.

Victimless wrongs present a serious problem for other attempts to find unity in the class of iniquities. When there is no victim, there is no violation of rights, and no reduction in net happiness. Victimless crimes are also a problem for Kant, who tried to show that all wrongs are unified by a particular kind of irrationality. In particular, Kant suggested that wrongs are those actions that cannot be coherently universalized. The difficulty is that there is no contradiction in wanting consensual incest to be a universal law. No contradiction arises in a world where all siblings have protected sex with each other when they want to. In such a world, sibling sex is like play fighting between siblings. Play fighting is morally acceptable because it is good practice and no one gets seriously hurt. Sex is a central human activity, and a bit of practice does some good. Still, consensual incest strikes us as very wrong. This is not a refutation of Kant. Kant could argue that there is a hidden irrationality in consensual incest, or he could say that it is not really wrong. But, on the face of it, Kant's proposal and other normative ethical theories don't seem to get the pretheoretical taxonomy right. There are things that ordinary people consider wrong, but are not wrong on leading philosophical accounts. Those accounts tend to focus in on categories that are narrower that the class of things that are regarded as wrong pretheoretically. When we consider the full class of things that we ordinarily regard as wrong, prospects for finding a unified theory begin to look grim.

(p.49) I will briefly discuss Kant again in chapter 3, and in chapters 4 and 5, I will dwell further on the theme of disunity. My present goal is to show that there is no obvious intrinsic property that unifies the class of wrongs, and that suggests that the wrongness of an act might have more to do with our reaction than with the act itself. Consensual incest and stealing might belong to the same category in virtue of offending our sensibilities. Sentiment could be the thread that binds together all vices and virtues. Those who believe in the intrinsic unity of wrongs have the burden of proof. On the face of it, we moralize a heterogeneous class of things. The only thing that items in this class seem to have in common is our moral attitudes toward them.

1.2.8 The Emotionist Upshot

I have offered a battery of arguments in favor of emotionism. I intermingled philosophical arguments with evidence from experimental psychology, neuroscience, child development, and psychopathology. Emotionism enjoys convergent support from armchair intuitions and empirical work. Moral concepts seem to be bound up with emotional responses. Those who lack moral emotions lack moral concepts as well. This supports epistemic emotionism. Furthermore, there is little evidence that anyone can successfully pick out moral properties without the help of emotions, and moral properties seem to lack any unifying characteristics independent of our reactions to them. This supports metaphysical emotionism. There may be other ways to make sense of all the evidence that I have been discussing, but emotionism emerges as the most obvious explanation.

The case that I have been presenting is not intended to be a demonstrative proof. A knock-down argument for emotionism may be too much to hope for. For now, I will settle for prima facie support. If emotionism has prima facie support, then it is worth exploring. In the following chapters I will present an emotionist theory in more detail, and I will argue that it can withstand objections. I will also present objections to non-emotionist ethical theories, such as Kant's, and weak emotionist theories, such as utilitarianism and expressivism. I will conclude that the case for strong emotionism is compelling. Emotions are implicated essentially in both the ontology and epistemology of morals.