What in moral psychology is innate? Joshua Greene Department of Psychlogy; Center for the Study of Brain, Mind, and Behavior Princeton University [Note to readers: This is a very rough draft, as will be evident to anyone who reads it. Please do not cite or circulate. Feedback welcome: jdgreene@princeton.edu] If you go to www.dictionary.com, type in the word “innate,” and hit enter, this is what you’ll get: adj 1.Possessed at birth; inborn. 2.Possessed as an essential characteristic; inherent. 3.Of or produced by the mind rather than learned through experience: an innate knowledge of right and wrong.1 Of all the things in the world one might use to illustrate the concept of innateness, this dictionary offers moral knowledge. I find this amusing—the idea that someone who is not exactly sure what “innate” means would benefit from knowing that one of the most complex and least understood of human capacities could plausibly be described as “innate.” And yet this choice, I suspect, is no accident. Our capacity for moral judgment, perhaps more than anything else, strikes people as both within us and external to us, as essentially human and at the same time possessing a mysterious external authority, like the voice of God or Nature calling us at once from within and beyond. But however obvious the reality of an innate capacity for moral judgment may be to theologians, lexicographers, and the like, it is not at all obvious from a scientific point of view. What is clear is that the days of the tabula rasa are over [Pinker], and insofar as an innate capacity for moral judgment is simply the opposite of a moral blank slate, we can say that human moral judgment is innate, or has an innate component. But one might think that we humans are genetically equipped with a moral psychology that is more than merely un-blank. One might think ourselves endowed with a specialized moral sense [ref], delicately tuned to detect nature’s moral signals, or a universal moral grammar [ref], hard-wired in to enable the fast and efficient generation of well-formed moral judgments and behaviors. Or perhaps we have an innate capacity for moral reasoning. How plausible are these more ambitious hypotheses? In this chapter I will examine evidence from a variety of sources that will allow us to make some educated guesses. 1 [Note that this is not a very good definition for various reasons. This is in June, 2003. Note source of definition. The American Heritage® Dictionary of the English Language, Fourth Edition Copyright © 2000 by Houghton Mifflin Company. Published by Houghton Mifflin Company. All rights reserved.] What do we mean by “innate?” I will not dwell on this ubiquitous question, but it’s sufficiently important to warrant a few remarks. (For more in-depth discussions see [? , ?, and ?] in this volume and [Elman et al., Samuels]) First, note that the dictionary definition above is a poor one for our purposes. Traits such as secondary sexual characteristics may be innate, but they are not present at birth. Likewise, it’s not clear what counts as an “essential” characteristic. My semi-detached earlobes may well be innate, but are they essential? Essential for what? Finally, there is no meaningful dichotomy between things that are produced by the mind and things that are learned through experience, as nearly everything with which we are concerned in this chapter is a product of the mind, though only some of these mental products are learned through experience. Second, there simply is no straightforward scientific definition of “innate.” One can’t say that a trait is innate if it’s determined by genes because almost nothing of interest is “determined,” i.e determined entirely, by genes. The unfolding of even the most rigid developmental programs typically requires input from the pre- and/or post-natal environment. Rather than saying that an innate trait must be “determined” by genetic influences, one might say that it must be importantly “affected” or “shaped” or “constrained” by genetic influences. This approach has the unfortunate consequence of making nearly everything, from baseball to bouffant hairdos, innate. In light of this one might then make the rather obvious point that innateness is a matter of degree. This is certainly true, but this facile concession masks the depth of the problem, suggesting that we will merely have to settle for such claims as, “Baseball is fifteen percent innate and eighty-five percent learned.” The contributions of genes on the one hand and learning, culture, and the environment on the other are very difficult to distinguish, let alone quantify. It’s not as if baseball would exist in a form that is eighty-five percent similar to its present form (whatever that means) if humans had radically different eyes, ears, and bodyplans, nor would we have fifteen percent of baseball if the baseball-playing cultures that exist today had taken a sufficiently different course through recent history. One can meaningfully talk about the percentage of the observed variance in a trait that is accounted for by genes. For example, it might turn out that, say, fifty percent of the observed variance in professional baseball players’ batting averages are attributable to genetic factors. But a trait’s having a large proportion of its variance accounted for by genetic factors is not the same thing as that trait’s being innate. As supposed above, it may turn out that variance in batting skill is largely accounted for by genetic factors, but that does not mean that baseball-playing behavior is innate, at least not in any ordinary sense. In principle, innateness does not require variance at all. If, for example, human eyes were all exactly the same, there would be no variance in human eye structure to account for, but this wouldn’t mean that human eyes have no innate structure. Indeed, such uniformity might be thought to imply that eye structure is more innate. More generally, one might suppose that uniformity or universality implies innateness, or vice versa. These seductive ideas are problematic as well. First, starting with the “vice versa,” an innate trait need not be universal. For example, when a new adaptive trait first appears within a sub-population of a species it is not universal, and yet it may be as innate as anything else. In addition, an observable trait may be innate without being universal because it is the result of a universal disposition to exhibit that trait in some circumstances rather than others [ref]. Second, universality or uniformity does not imply innateness. Hairstyling and poetry are reported to exist in all human cultures [Brown], but are these things therefore innate? It’s not obviously true, and it’s not clear what could settle the question. Clearly humans tend toward these activities in normal environments, and clearly this state of affairs depends on genetic input, but to say that poetic composition and hairstyling are innate seems to go beyond the facts of universality and genetic dependence. Perhaps one should say, following Sober [1999], that a trait is innate if the trait reliably emerges in a sufficiently wide range of environments given the genes that code for that trait. (Here the claim is not that universality implies or is implied by innateness, but rather that universality within the population that has the relevant genes implies and is implied by innateness.) As Samuels [2002] points out, this criterion suffers from a similar problem to one described above. Nearly everyone ends up with the belief that water is wet, and yet this belief is not obviously innate and is likely to be learned. Perhaps the difference that makes the difference is natural selection: If poetry and hairstyling were not specifically selected for during the course of human evolution, then they are not innate. Unfortunately, the natural selection criterion probably excludes too much. First, it would fail to count traits caused by new mutations as innate. Second, there are plausibly innate traits that are not specifically selected for, e.g. the human susceptibility to cocaine addiction . This susceptibility certainly isn’t learned, and it doesn't appear to be culture-bound. If requiring that innate traits be specifically selected for is asking too much, perhaps we should say that a trait is innate if it follows as a direct consequence of traits that were selected for. Unfortunately, this criterion risks classifying as innate everything that depends in any significant way on genes, which is pretty much everything of interest. Of course it all depends on what one means by “directly.” One can certainly make progress along these lines by specifying the various ways that genes can affect traits [Elman et al], some of which are more naturally classified than others as “direct,” but it is unlikely that such efforts will return an account of innateness that is at once intuitively appealing and scientifically useful.2 I mention the above complexities to illustrate the simple point that the concept of “innateness” is not so simple, and may even be more trouble than it’s worth. Still, one may emerge from this barrage of logical distinctions and cautionary examples with the sense that, for the things that matter, it can’t really be all that complicated. “We all know perfectly well,” one might say, “that the visual system is basically innate and that baseball is not!” This is mostly half right. There are some things, like the human visual system, that have all the characteristics that we expect from the things we call “innate.” The problem is baseball. It’s very hard to say what exactly it is that makes baseball not innate. This is especially true once one realizes that baseball may be considered at different levels of abstraction. Baseball, like the English language, was not specifically selected for, but a disposition to engage in physically challenging 2 A noteworthy proposal along these lines is Richard Samuels' [2002] primitivist account of innateness. Samuels argues that a psychological trait is innate if it is "psychologically primitive," and a trait is psychologically primitive if correct psychological theory (as opposed to biological theory) does not explain how this trait is acquired. Thus, this view distinguishes the innate from the non-innate by appeal to the distinction between psychology and biology. In my estimation, Samuels' account elegantly captures many people's intuitive sense of what "innate" means and in many ways gets to the heart of the matter. There does seem to be an important distinction between psychology and biology, and I agree with Samuels that our intuitive sense of innateness depends on this distinction. But, I maintain, appearances are deceiving. The emerging field of cognitive neuroscience is, in a nutshell, the dissolution of the intuitively appealing but ultimately illusory distinction between psychology and biology. If Samuels' account of innateness is the best one can give—and it may well be—then what Samuels' account reveals, above all else, is that the usefulness of the concept of innateness will diminish as the science of mind (psychology) and the science of brain (neuroscience) continue to merge. Some of the results discussed in this chapter will, I hope, illustrate this merging process. and competitive leisure activities may have been selected for, just as the grammar underlying the English language may have been as well [Pinker]. But even if we come to the firm conclusion that baseball is not innate, this does not mean that evolution and baseball have nothing to do with each other. On the contrary, if we want to understand why baseball is so popular, why a curve ball is so hard to hit, why some players streak while others slump, why baseball’s rules seem reasonable, and why we tend to root for the home team, an evolutionary perspective may be absolutely essential. Moreover, the conclusion that baseball is not innate is entirely compatible with the claims that ninety percent of the variability in baseball performance is attributable to genetic factors and that a game like baseball was destined to evolve sooner or later. In other words, it’s not entirely clear what we gain or lose in accepting or rejecting the innateness of baseball. And likewise—to return us to the matter at hand—it’s not clear what matters of substance turn on the innateness of morals. At the same time, however, this indifference to the question of nativism in moral psychology is something of an achievement. It comes only with the recognition that our capacity for moral judgment, like language and memory and the rest of our cognitive capacities, is yet another natural phenomenon to be understood in terms of the complex interactions between genes and their environments. This indifference is earned only after one has given up on the idea that there is a sphere of moral thought that is comfortably insulated from crass biology. Once one accepts that the human mind is a product of natural selection and everything that follows from this, the hopes and fears that motivate the question “Is morality innate?” either disappear or are redirected toward new questions that are ultimately more meaningful: Which biological adaptations shape our moral judgment, and how do they do it? Which of these adaptations are distinctively human? How much of the variability in moral judgment is attributable to genetic differences? How much of the variability in moral judgment is attributable to cultural differences? What kinds of environmental inputs have the strongest effects on one’s moral sensibility? Is there an underlying structure to human moral psychology, “a universal moral grammar?” Is there a morality module? Are there things that we are biologically programmed to see as right or wrong? In what follows I will discuss a variety of data that bear on these questions. Social emotions and behavior in non-human primates An important source of evidence concerning the biological forces that have shaped human morality comes from studies of our nearest living relatives. I will not discuss these studies in great detail because their results and implications have already received excellent treatment by experts in the field [Flack and deWaal, Boehm, deWaal, Hauser], but instead will offer an editorial summary, largely following Flack and deWaal (2000). Non-human primates lead intensely social lives, and, moreover, their social lives and the behavioral tendencies that structure them bear a striking resemblance to those of humans. To begin, monkeys and apes appear to have a rudimentary sense of fairness with respect to exchanges of food and other goods, which itself depends on an ability to keep track of different individuals' past behaviors and to use that information for the purposes of social calculation. For example, adult female brown capuchin monkeys exchange food in a reciprocal way, repaying individuals who have behaved altruistically in the past [de Waal, 1997b]. Among chimpanzees, the exchanges are monitored, but not on an exchange-by-exchange basis. Pairs of chimpanzees tend to share with each other in a symmetrical fashion over the long term, but sharing is often asymmetrical over the course of a day [de Waal 1989b]. Aggression is more often directed against beggars for food rather than possessors of food. This pattern is noteworthy because aggression toward "have-nots" rather than "haves" appears to be aimed not so much at the defense of one's food—after all, the beggar is begging for food, not threatening to take it—but rather aimed at defending a system of norms that determine when food-sharing is or is not appropriate. (Of course, this does not require an explicit understanding of norms and a deliberate attempt to defend them.) Another noteworthy behavior is the chimpanzee's "respect for possession." [Goodall, 1971]. High-ranking individuals will often allow lower-ranking individuals to retain food that could easily be taken and in some cases will beg for food from lower-ranking individuals. Social expectations also seem to be exhibited in reciprocal patterns of conflict intervention. Chimpanzees exhibit "a revenge system," whereby an individual chimpanzee A is more likely to intervene in a conflict against chimpanzee B if B had previously intervened in a conflict against A. Chimpanzees also have elaborate mechanisms for moderating conflict. First, reconciliation behavior repairs damaged relationships after a conflict has occurred. Chimpanzees will sometimes effect reconciliation through such human-like gestures as extending a hand for kissing, and mouth-to-mouth kissing [de Waal, 1989]. Another social mechanism related to conflict is the adoption of the "control-role." In chimpanzees a dominant individual will break up conflicts, impartially punishing any combatants who continue to fight [de Waal, 1982, Boehm, 2000], and chimpanzees are known to form large coalitions to suppress unwanted individual behavior [Boehm, 2000]. Yet another such mechanism is the phenomenon of third party mediation, whereby two chimpanzees who have recently fought with one another are brought together for reconciliation by the mediating actions of an individual who was not involved in the original conflict [de Waal and van Roosmalen, 1979]. Indeed, chimpanzees appear to take a keen interest in the resolution of conflicts, as demonstrated by repeated observations of group-wide celebration following the resolution of dramatic conflicts [de Waal, 1996]. In addition to social behaviors related to exchange and conflict, chimpanzees exhibit a softer, emotionally-based concern for other individuals that is naturally described as sympathy or empathy. De Waal [1982] has observed, for example, a juvenile chimpanzee embrace an adult male who has just lost a confrontation with this rival. This appears to be part of a broader pattern of active consolation in chimpanzees, a pattern that is not observed outside of the great apes [de Waal and Aureli, 1996]. In short, chimpanzees exhibit a wide range of social behaviors that, in the words of Flack and de Waal [2000] form the "building blocks" of much of human morality. They exhibit social sensibilities that are strikingly similar to human moral sensibilities, including such phenomena as moral indignation/anger, compassion, a sense of fairness, and community concern. Of course human morality is far more developed and complex than the moralities or proto-moralities of our nearest relatives, but these similarities are significant nonetheless and in need of explanation. More specifically, one might wonder whether these similarities between humans and non-human primates stem from a common genetic inheritance [de Waal; Darwin, Descent], or whether our perception of similarity stems from a tendency on our part to attribute more to these animals than is actually there. With respect to this issue, Flack and de Waal appeal to a principle of 'evolutionary parsimony' according to which apparently similar behaviors observed in closely related species are reasonably attributed to underlying genetic similarities. Nativism and modularity Imagine the following scenario. A woman is brought to the emergency room after sustaining a severe blow to the head. At first, and much to her doctors' surprise, her neurological function appears to be completely normal. And for the most part it is, but it soon becomes clear that she has acquired a bizarre disability. As a result of her accident, this woman can no longer play basketball.3 Her tennis game is still top notch, as is her golf swing, and so on. Only her basketball game has been compromised. Could such an accident really happen? Almost certainly not. The way the brain is organized, it is virtually impossible that something like a blow to the head could selectively destroy one's ability to play basketball and nothing else. This is because the neural machinery required to play basketball isn't sitting in one place, like a car's battery. Instead, this machinery is spread all over the brain, and its various components are used in the performance of any number of other tasks. But what if someone were to have an accident like this? If, somehow, this were to happen, we would be forced to conclude that the brain has structures that are, in some cases at least, specifically dedicated to playing basketball and that these structures are sufficiently independent from the rest of the brain that rendering them inoperable need not affect the rest of the brain's function. In other words, we would have strong evidence for the existence of a "basketball module."4 Talk of modules and modularity is often associated with nativist hypotheses and evolutionary psychology [Cosmides and Tooby, Pinker]. Such nativists argue that the human mind is essentially a collection of modules, a cognitive Swiss Army knife if you will, with each of its several components designed by natural selection to perform a specific function in a relatively independent fashion. Nativists like to see cases of brain damage leading to selective cognitive deficits, not because nativists are cruel, but because such cases demonstrate that the mind's structure is modular, and that, they argue, is evidence that our minds have been specifically adapted to perform the tasks that are performed more by those mental modules. This argument, however, is controversial. Many theorists, especially those in the "connectionist " camp, have argued that modular architecture can develop without the help of any specific adaptations to that end and that therefore modularity is no evidence for nativism [Elman et al]. Certainly, modularity and nativism are independent in principle, and there do appear to be in instances in which modules (of certain sort, at least) are acquired through learning [Schiffrin and Schneider, O'Reilly and Munakata].5 But, say nativists, there are modules and there are modules. 3 The basketball example is adapted from Casebeer and Churchland [2003]. A module is a "cognitive organ," a computational structure that computes outputs from inputs by way of intervening processes are computationally isolated from the rest of the system in which it is in embedded, i.e. “encapsulated” [Fodor]. A modular structure gains efficiency at the expensive of flexibility, much as a hand-held calculator does a splendid job of quickly and accurately performing arithmetic calculations while being utterly useless for most other tasks, such as making dinner. 5 Of course, it depend s on what one means by "module." According to some definitions, a module must be innate. Needless to say, there is no such thing as a learned module according to this definition. But the central idea behind 4 Some kinds of moduralish learning may take place, but there's only so much modular structure that one can acquire without the help of natural selection. I mention this controversy not because I intend to help resolve it, but because it bears directly on the data I will be discussing in the next three sections. Studies of psychopaths and patients with brain lesions along with neuroimaging studies of normal individuals reveal a number of noteworthy cognitive dissociations. To dissociate two cognitive processes is simply to show that those processes are independent to some extent. 6 For example, in the case of the imaginary brain damage described above, the neural bases of basketball-playing are dramatically dissociated from all other cognitive processes. To say that two processes are dissociable from each other is not too far from saying that they make use of different modules. Here's the argument: If two processes are dissociable, they must make use of different cognitive structures to some extent. But why would that be? Presumably it is because these different cognitive structures perform different functions. But for different parts to perform different functions, they must, to some extent, stay out of each others' business, else there would be no functionally distinct parts of which to speak. And a functionally distinct cognitive structure that, more or less, takes care of its own business is, more or less, a module. Let us suppose, then, that a dissociation between two processes implies that those processes draw on different modules and, a fortiori, shows that modules relevant to these processes exist. If you're the kind of nativist who believes that modular structures are, as a general rule, innate structures, you will then conclude that there is probably a significant "innate component" to one or both of the dissociated processes. If, in contrast, you think that modules develop without any special (i.e. domain-specific) help from natural selection, you'll likely resist this conclusion. For now, I will simply leave the reader to his/her prejudices, although I will make some partisan comments later. My aim at this point is simply to explain how the data discussed below may bear on the issue of innateness in moral psychology. But, as suggested above, the issue of innateness may prove to be a red herring. Best, at this point, to let the data speak for themselves. Lesion data As noted above, the idea that a well-placed blow to the head could selectively rob one of one's ability to play basketball is pretty far-fetched. And yet there have been cases in which brain damage has appeared to rob individuals of their moral sensibilities in a strikingly selective way. By far, the most celebrated of such cases is that of Phineas Gage, a Nineteenth Century railroad foreman working in Vermont. One fateful day, an accidental explosion sent a tamping iron through Gage's eye socket and out the top of his head, destroying much of his medial prefrontal cortex. Gage not only survived the accident; at the time he appeared to have emerged with all of his mental capacities intact. After a two-month recuperation period Gage was pronounced cured, but it was soon apparent that Gage was damaged. Before the accident he was admired by his colleagues for his industriousness and good character. After the accident, he became lawless. He wandered around, making trouble wherever he went, unable to hold down a steady job due to modularity is informational encapsulation, and the point illustrated by cases of learned modularity is that a system can spontaneously develop encapsulated processing structures in the course of learning how to solve a problem that it was not specifically adapted to solve. 6 What I am calling a dissociation here is, strictly speaking, a double dissociation. his anti-social behavior. For a long time no one understood why Gage’s lesion had the profound but remarkably selective effect that it had. More recent cases of patients with similar lesions have shed light on Gage’s injury. Damasio and colleagues [refs] report on a patient named “Elliot” who suffered a brain tumor in roughly the same region that was destroyed in Gage. Like Gage, Elliot maintained his ability to speak and reason about topics such as politics and economics. He scored above average on standard intelligence tests, including some designed to detect frontal lobe damage, and responded normally to standard tests of personality. However, his behavior, like Gage’s, was not unaffected by his condition. While Elliot did not develop anti-social tendencies to the extent that Gage did, he, too, exhibits certain peculiar deficits, particularly in the social domain. A simple laboratory probe has helped reveal the subtle but dramatic nature of Elliot’s deficits. When shown pictures of gory accidents or people about to drown in floods, Elliot reported having no emotional response but commented that he knew that he used to have strong emotional responses to such things. Intrigued by these reports, Damasio and colleagues employed a series of tests designed to assess the effects of Elliot’s damage on his decision-making skills. They asked him, for example, whether or not he would steal if he needed money and to explain why or why not. His answers were like those of other people, citing the usual reasons for why one shouldn’t commit such crimes. Saver and Damasio followed up this test with a series of five tests of moral/social judgment [Saver and Damasio, 1991]. As before, Elliot performed normally or above average in each case. It became clear that Elliot’s explicit knowledge of social and moral conventions was as good or better than most people’s, and yet his personal life, like Gage’s, has deteriorated rapidly as a result of his condition (although he does not seem to mind). Damasio attributes Elliot's real-life failures not to his inability to reason (in the sense above) but to his inability to integrate emotional responses into his practical judgments. “To know, but not to feel,” says Damasio, is the essence of his predicament. In a study of Elliot and four other patients with similar brain damage and behavioral deficits, Damasio and his colleagues observed a consistent failure to exhibit typical electrodermal responses (a standard indication of emotional arousal) when these patients were presented with socially significant stimuli, though they responded normally to non-social, emotionally arousing stimuli [Damasio, Tranel, and Damasio, 1990]. A more recent study of patients like Elliot used the "Iowa gambling task" to study their decision-making skills [Bechara et al, 1996]. In this task, patients draw cards from a set of four decks and receive points depending on which cards they choose. The decks are stacked so that two of them yield great gains but even greater losses while the other two decks yield moderate gains but even smaller losses. Thus, the best long term strategy is to pick from the "safe" decks rather than the "risky" decks. Normal individuals sample all four decks initially but eventually figure out that the safe decks are a better choice. Moreover, during the learning phase, normal subjects exhibit anticipitatory electrodermal responses immediately prior to choosing from the risky decks, indicating a negative emotional response to the intention to make a disadvantageous choice. (Amazingly, these electrodermal responses are typically observed before the subject is consciously aware that the disadvantageous decks are disadvantageous [Bechara et al, 1994].) Patients like Elliot, however, typically fail to realize that the risky decks make bad choices and, what's more, they fail to have the anticipatory electrodermal responses to the risky decks exhibited by normal individuals, suggesting, as predicted, that their failure to perform well in the gambling task is related to their emotional deficits. They can't feel their way through the problem. While the subjects in the above studies exhibit “sociopathic behavior” as a result of their injuries, they are not “psychopaths." Most often they themselves, rather than other people, are the victims of their poor decision-making. However, a more recent study (Anderson et al., 1999) of two subjects whose ventral, medial, and polar prefrontal cortices were damaged at an early age (three months and fifteen months) reveals a pattern of behavior that is characteristically psychopathic: lying, stealing, violence, and lack of remorse after committing such violations. These developmental patients, unlike Elliot et al., exhibit more flagrantly anti-social behavior presumably because they did not have the advantage of a lifetime of normal social experience involving normal emotional responses. Both patients perform fairly well on IQ tests and other standard cognitive measures and perform poorly on the Iowa gambling task, but unlike adultonset patients their knowledge of social/moral norms is deficient. Their moral reasoning appears to be, in the terminology of Kohlberg, “preconventional,” conducted from an egocentric perspective in which the purpose is to avoid punishment. Other tests show that they have a limited understanding of the social and emotional implications of decisions and fail to identify primary issues and generate appropriate responses to hypothetical social situations. Grattan and Eslinger (1992) report similar results concerning a different developmental-frontal patient. Thus, it appears that the brain regions compromised in these patients include structures crucial not only for online decision-making but also for the acquisition of social knowledge and dispositions toward normal social behavior. What can we learn from these damaged individuals? In Gage—the legend if not the actual patient—we see a striking dissociation between "cognitive"7 abilities and moral sensibilities. Gage, once an esteemed man of character, is transformed by his accident into a scoundrel, with little to no observable damage to his "intellectual" faculties. A similar story emerges from Elliot's normal performance on questionnaire-type assays of his social/moral decision-making. Intellectually or "cognitively," Elliot knows the right answers, but his real life social/moral decision-making is lacking. From this pattern of results, one might conclude that Gage, Elliot, and the like have suffered selective blows to their "morality centers." Other results, however, complicate this neat picture. Elliot and similar patients appear to have emotional deficits that are somewhat more general and that adversely affect their decision-making in nonsocial contexts as well as social ones (e.g. on the gambling task). And to further complicate matters, the developmental patients studied by Anderson and colleagues appear to have some "cognitive" deficits, although these deficits are closely related to social decision-making. Thus, what we observe in these patients is something less than selective damage to these individuals' moral judgment abilities, but something more than a general deficit in "reasoning" or "intelligence" or "judgment." In other words, these data suggest that there are dissociable cognitive systems that contribute asymmetrically to moral judgment but give us little reason to believe that there is a discrete faculty for moral judgment or a "morality module."8 What these 7 The term "cognitive" has two uses. In some contexts, "cognitive" refers to information processing in a general way as in "cognitive science." In other contexts, "cognitive" refers to a subset of cognitive (first meaning) processes that are to be contrasted with affective or emotional processes. Unfortunately, there is no good, unambiguous word for these non-emotional processes, largely because, at the present time at least, they lack theoretical unity. In spite of these difficulties I will use the term "cognitive" with scare quotes to indicate this second, more specific meaning. 8 There is a sizable literature reporting on patients with morally aberrant behavior resulting from frontal damage, and the cases discussed above are not necessarily representative. I have chosen to focus on these cases because they, of all the cases reported in the lesion literature, come the closest to reporting a dissociation between the capacity for moral judgment and other cognitive abilities. The thought is that if there were a case in which moral judgment were selectively knocked out, that single case would tell us much more about the architecture of the moral mind than a data do suggest is that there is an important dissociation between affective and "cognitive" contributions to social/moral decision-making and that the importance of the affective contributions have been underestimated by those who think of moral judgment primarily as a reasoning process.9 Indeed, Jonathan Haidt has amassed an impressive body of evidence in support of the conclusion that moral judgment is driven almost entirely by emotion [Haidt, 2001]. Anti-social behavior The studies described above are of patients whose social behavior has been compromised by observable and relatively discrete brain lesions. There are, however, many cases of individuals who lack macroscopic brain damage and who exhibit pathological social behavior. These people fall into two categories: people with anti-social personality disorder (APD) and the subset of these individuals known as psychopaths. Anti-social personality disorder is just a catch-all label for whatever it is that causes some people to habitually violate our more serious social norms, typically those that are codified in our legal system [DSM IV].10 If you habitually lie, steal, or beat people up, you have APD, but you are not necessarily a psychopath. This is because the leading model of psychopathy [Hare, ???] has two factors. In addition to exhibiting pathologically anti-social behavior, psychopaths are additionally characterized by a pathological degree of callousness, lack of empathy or emotional depth, and lack of genuine remorse for their anti-social actions. In more intuitive terms, the difference between plain old APD and psychopathy is the difference between a seriously flawed human being (e.g. the hot-headed barroom brawler) and someone who is just inhuman (the cold-blooded killer who murders his parents for their money). Psychopaths appear to be special in a number of ways [Blair, 2001]. First, while the behavioral traits that are used to diagnose APD correlate with IQ and socio-economic status, the traits that are distinctive of psychopaths do not [Hare et al, 1991]. Moreover, the behaviors associated with APD tend to decline with age, while the psychopath's distinctive socialemotional dysfunction holds steady [Harpur et al, 1994]. The roots of violence appear to be different in psychopaths as compared to similarly violent non-psychopaths. The anti-social behavior of non-psychopaths appears to be more contingent in two ways. First, positive parenting strategies appear to influence the behavior of non-psychopaths, whereas psychopaths appear to be impervious in this regard. Second, an probably not incidentally, the violence of psychopaths is more often instrumental rather than impulsive [Blair, 2001]. Experiments using psychopaths as subjects reveal further, more subtle differences between psychopaths and other individuals with APD. Psychopaths exhibit a lower level of tonic electrodermal activity and show weaker electrodermal responses to emotionally significant hundred cases in which brain damage results in a hodgepodge of behavioral deficiencies, including deficiencies in moral judgment. For studies of some more "hodgepodgy" cases (or cases that have not been shown to be not "hodgepod y") see Blair and Cipolotti (2000), Grafman et al. [?], Pincus (2001), Cohen et al. (1999), and Davidson et al. (2000). For a striking case of acquired pedophilia due to a frontal lesion see Burns and Swerdlow (2003). 9 By "reasoning" I refer to processes that are relatively slow and effortful with intermediate steps that are consciously accessible. Thse stand in contrast to intuitive processes that are quick, effortless, and with intermediate steps that are not consciously accessible. 10 Pincus (2001) compares the APD designation to other such laughably empty medical terms such as ??? ani ("itchy anus"). stimuli than normal individuals (Hare and Quinn, 1971). A more recent study (Blair et al., 1997) compares the electrodermal responses of psychopaths to a control group of criminals who, like the psychopaths, were serving life sentences for murder or manslaughter. While the psychopaths were like the other criminals in their responses to threatening stimuli (e.g. an image of a shark’s open mouth) and neutral stimuli (e.g. an image of a book), they showed significantly reduced electrodermal responses to distress cues (e.g. an image of a crying child’s face) relative to the control criminals, a fact consistent with the observation that psychopathic individuals appear to have a diminished capacity for emotional empathy. Blair (1995) hypothesized that psychopaths’ diminished capacity for emotional empathy should prevent them from drawing a distinction between what are sometimes referred to as “moral” and “conventional” rules and that psychopaths, as compared to other criminals, should make fewer references to the pain or discomfort of victims in explaining why certain harmful actions are unacceptable. Both of these predictions were confirmed. “Moral” transgressions were defined as those having negative consequences for the “rights and welfare of others” and included instances of one child hitting another and a child smashing a piano. “Conventional” transgressions were defined as “violations of the behavioral uniformities that structure social interactions within social systems” and included instances of a boy wearing a skirt and a child who leaves the classroom without permission. While the “normal” subjects (non-psychopathic incarcerated criminals) drew a general distinction between moral and conventional transgressions, the psychopaths did not. Normal subjects found a greater difference in permissibility and seriousness between moral and conventional transgressions than did the psychopaths. The most striking finding, however, concerned the psychopath’s judgments concerning “modifiability.” Each of the transgression stories were set in school, and in each case the subjects were asked whether or not it would be permissible for the child to perform the transgressive action if the teacher had said earlier that such actions were permitted. Nonpsychopathic criminals tended to say that the conventional transgressions would become permissible if the teacher were to explicitly allow their performance but that the moral transgressions would not be permissible in either case. Psychopaths, however, treated all transgressions as impermissible regardless of what the teacher says.11 In addition, the psychopaths were, as predicted, less likely to appeal to the pain and discomfort of victims and more likely to appeal to the violation of rules in explaining why various transgressions are impermissible. A different cognitive test shows that psychopathic murderers fail to have normal unpleasant associations with violence. The study conducted by Gray et al. (2003) uses an adapted version of the Implicit Associations Test (IAT) (Greenwald et al., 1998). Subjects classify uppercase words as "pleasant" or "unpleasant" and lowercase words as "violent" or "peaceful." Normal individuals respond more quickly when the same button is used to indicate both "pleasant" and "peaceful" and more slowly when different buttons are used, but psychopathic murders show no such effect. Surprisingly these results hold only for psychopathic murderers and not for psychopathic non-murderers, non-psychopathic murders. 11 While Blair predicted that the psychopaths would fail to draw the moral/conventional distinction, he predicted more specifically that the psychopaths would treat both types of transgression as normal subjects treat conventional ones. Nevertheless, he found that the psychopaths treated both types of transgressions as normal subjects treat moral ones. He attributes this result to the fact that his psychopathic subjects (all incarcerated) have an interest in demonstrating that they have “learned the rules.” According to Blair (Blair et al., 1997), “The clinical and empirical picture of a psychopathic individual is of someone who has some form of emotional deficit." This conclusion is bolstered by the results of a recent neuroimaging study in which psychopaths and control criminals processed emotionally salient words. In the psychopaths, part of the posterior cingulate gyrus, a region that exhibits increased activity during a variety of emotion-related tasks (Maddock, 1999), was less active than in the control subjects. At the same time, other regions were more active in psychopaths during this task, leading Khiel et al. to conclude that the psychopaths were using an alternative cognitive strategy to perform this task, much as Blair argues that psychopaths answer questions about the "modifiability" of moral transgressions through non-emotional means. Thus, so far, a host of signs point to the importance of emotions in moral judgment [Haidt, 2001]. In light of this, one might come to the conclusion that a psychopath, with his dearth of morally relevant emotion, is exactly what we're looking for—a human being "with everything— hold the morality." Indeed, Schmitt et al. (1999) found that psychopaths performed normally on the Iowa gambling task, suggesting that their emotion-based decision-making deficits are not general, but rather related specifically to the social domain. As before, however, the empirical picture is not quite so simple, as psychopaths appear to have other things "held" as well. To begin, two studies, one of adult psychopaths [Mitchell et al, 2002] and one of children with psychopathic tendencies [Blair et al, 2001], found that psychopathic individuals do perform poorly on the Iowa gambling task. (These authors attribute the conflicting results to Schmitt et al.'s failure to use the original task directions, which emphasize the strategic nature of the task.) Moreover, there are several indications that psychopaths have deficits that extend well beyond their apparently stunted social-emotional responses. Lapierre et al. (1995) find that psychopaths have a hard time inhibiting a learned response in a Go/No-Go task12, and Kiehl et al. (2000) observed abnormal electrophysiological signals (brain waves) in psychopaths performing this task. Newman et al. (1997) found similar result using a "response reversal" task in which subjects must adjust their strategy due to changing pay-off contingencies. Along similar lines, Bernstein et al. (2000) found that psychopaths had a hard time recalling incidental information about the spatial location of words they were told to memorize, but only for words presented in the right spatial field (which are processed by the left hemisphere). They argue that psychopaths have a hard time attending to secondary cues or peripheral information once their left hemisphere-based motivational system is engaged. Kiehl et al. (1999a) found that psychopaths produced abnormal electrophysiological signals during the performance of a "visual oddball" task" in which one must detect target stimuli (large squares) within a series of distractor stimuli (small squares). Psychopaths exhibited decreased P300 responses (electrically positive responses at a latency of approximately 300 milliseconds) to target stimuli, and these diminished responses were less lateralized than in normal individuals. Finally, Kiehl et al. (1999b) have also found abnormal electrophysiological activity in psychopaths during linguistic tasks. Thus, it's pretty clear that psychopathy involves a number of suite of superficially varied cognitive deficits and abnormalities. The psychopathy literature sends mixed signals regarding the "impulsivity" of psychopaths. On the one hand, psychopaths are to be distinguished from other violent individuals by their "cold-bloodedness," their lack of empathy, etc. Once again, their violence is supposed to be "instrumental" rather than "reactive" [Blair 2001] At the same time, however, 12 This is a simple cognitive task in which the subject must, depending on the stimulus cue, either perform an action or do nothing. [ref] some of the evidence described above suggests that psychopaths have a hard time inhibiting disadvantageous behavior, even during the performance of "dry" cognitive tasks. Most likely, these apparently contradictory results reflect two sides of the same cognitive coin, although it is not clear how to reconcile them. Compared to some anti-social individuals, psychopaths are "cool and collected," but a closer examination reveals that psychopaths have a kind of impulsivity or one-track-mindedness that distinguishes them from normal individuals in a subtle, but fairly general way. The results of a neuroimaging study of "predatory" vs. "affective" murderers by Raine et al. (1998) gestures toward a synthesis. They argue that excessive subcortical activity in the right hemisphere leads to violent impulses, but that "predatory" murderers, who unlike "affective" murderers exhibit normal levels of prefrontal activity, are better able to control these impulses. (In a more recent study, Raine et al. (2000) found that a sample of individuals diagnosed with APD (some of whom, however, may have been psychopaths [right?]) tended on average to have decreased prefrontal gray matter.) Thus, according to these results, the difference between a violent "hot head" and a violent "predator" is a lack of cognitive control, while the difference between a violent "predator" and normal person is the presence of anti-social impulses. These conclusions, however, are based on fairly crude neuroimaging data that are difficult to interpret.13 Additionally, it's not clear how to reconcile the claim that "predatory" and "affective" murderers act on the same underlying impulses with the claim that psychopathic violence is "instrumental" rather than "impulsive." To make a long story short, psychopaths are not nature's controlled experiment with amorality. Psychopathy is a complicated syndrome that has subtle and not-so-subtle effects on a wide range of behaviors, including many behaviors that, superficially at least, have nothing to do with moral judgment and behavior. At the same time, however, psychopathy appears to be a fairly specific syndrome. Psychopaths are not just people who are very violent or very bad. Using the proper methods, psychopaths are clearly distinguishable from others whose behavior is comparably anti-social, suggesting that the immoral behavior associated with psychopathy stems from the malformation of specific cognitive structures that make important contributions to moral judgment. Moreover, these structures seem to be rather "deep" in the sense that they are not well-defined by the concepts of ordinary experience and, more to the point, ordinary learning. Psychopaths do not appear to be people who have, through some unusual set of experiences, acquired some unusual moral beliefs or values. (Recall, once again, that psychopathic tendencies, unlike general anti-social tendencies, do not appear to correlate with things that ordinarily affect learning: IQ, socio-economic status, and age.) Rather, they appear to have an abnormal but stereotyped cognitive structure that affects a wide range of behaviors, from their willingness to kill to their inability to recall the corner of the screen on which a given word has appeared.14 Neuroimaging studies of moral judgment and decison-making 13 It should also be noted that Raine et al.'s criteria for "predatoryness" is different from Blair's criteria for psychopathy. Blair uses Hare's PCL-R, the two-factor model described above, while Raine et al. used the judgments of raters who considered a wide variety of information including clinical assessments, legal documents, interview transcripts, and media reports. 14 Blair [2001?] argues that this suite of cognitive features results from developmental dysfunction in the amygdala. Consider the following moral dilemma (the trolley dilemma):15 A runaway trolley is headed for five people who will be killed if it proceeds on its present course. The only way to save these people is to hit a switch that will turn the trolley onto an alternate set of tracks where it will run over and kill one person instead of five. Is it okay to turn the trolley in order to save five people at the expense of one? Most people say that it is, and they tend to do so in a matter of seconds (Greene et al., 2001).16 Now consider a slightly different dilemma (the footbridge dilemma): A runaway trolley threatens to kill five people as before, but this time you are standing next to a large stranger on a footbridge spanning the tracks, in between the oncoming trolley and the five people. The only way to save the five people is to push this stranger off the bridge and onto the tracks below. He will die as a result, but his body will stop the trolley from reaching the others. Is it okay to save the five people by pushing this stranger to his death? Most people say that it's not and, once again, they do so rather quickly. These dilemmas were devised as part of a puzzle for moral philosophers [Thomson] by which the aim is to explain why it's okay to sacrifice one life to save five in the first case but not in the second case. Solving this puzzle has proven very difficult. While many attempts to provide a consistent, principled justification for these two intuitions have been made, the justifications offered are not at all obvious and are generally problematic. The fact that these intuitions are not easily justified gives rise to second puzzle, this time for moral psychologists: How do people know (or "know") to say "yes" to the trolley dilemma and "no" to the footbridge dilemma if there is no obvious, principled justification for doing so? If these conclusions aren't reached on the basis of some readily accessible moral principle, they must be made on the basis of some kind of intuition. But where do these intuitions come from? To try to answer this question, my colleagues and I conducted an experiment in which subjects responded to these and other moral dilemmas while having their brains scanned (Greene et al, 2001). Our hypothesis was that the thought of pushing someone to his death with one's bare hands is more emotionally salient than the thought of bringing about similar consequences by hitting a switch. More generally, we hypothesized that moral violations of an "up close and personal" nature, as in the footbridge case, are more emotionally salient than moral violations that are more impersonal, as in the trolley case, and that this difference in emotional response explains why people respond so differently to these two cases. The rationale for this hypothesis is evolutionary. As noted above, it is very likely that we humans have inherited many of our social instincts from our primate ancestors, among them instincts that reign in the tendencies of individuals to harm one another. These instincts are emotional instincts, triggered by behaviors and other elicitors that were present in our ancestral environment. This environment did not include opportunities to harm other individuals using complicated machinery, but it did include opportunities to harm other individuals by pushing them into harms way (e.g. off a cliff or into a river). Thus, one might suppose that the sorts of basic, interpersonal violence that threatened our ancestors back then will "push our buttons" today in a way that peculiarly modern harms do not. With all of this in mind, we operationalized the "personal"/"impersonal" distinction as follows: A moral violation is personal if it is (a) likely to cause serious bodily harm (b) to a particular person (c) in such a way that the harm does not result from the deflection of an 15 This and the following dilemma are adapted from Thomson [ref]. Note that here and elsewhere, "most people" may not hold cross-culturally, as the people who have been tested using these dilemmas are primarily American college students. 16 existing threat onto a different party.17 A moral violation is impersonal if it fails to meet these criteria. One can think of these criteria for personal harm in terms of ME HURT YOU and as delineating roughly those violations that a chimpanzee can appreciate.18 Condition (a) (HURT) picks out roughly those harms that a chimp can understand (e.g., assault vs. tax evasion). Condition (b) (YOU) requires that the victim be vivid as an individual. Finally, condition (c) (ME) captures the notion of “agency,” the idea that the action must spring in a vivid way from the agent’s will, must be “authored” rather than merely “edited” by the agent. Pushing someone in front of a trolley meets all three criteria and is therefore "personal," while diverting a trolley involves merely deflecting an existing threat, removing a crucial sense of “agency” and therefore making this violation "impersonal." Other moral dilemmas (about forty total) were categorized using these criteria as well. Before turning to the data, the evolutionary rational e for the "personal"/"impersonal" distinction requires a bit more fleshing out. Emotional responses may explain why people say "no" to the footbridge dilemma, but why do they say "yes" to the trolley dilemma? Here we must consider what's happened since we and our closest living relatives parted ways. Of course, a great deal has happened since then, but among the most important developments is that of our capacity for general-purpose abstract reasoning, a capacity that can be used to think about anything one can name, including moral matters. Thus, one might suppose that when the heavyduty, social-emotional instincts of our primate ancestors lay dormant, abstract reasoning has an opportunity for greater influence. And, more specifically, one might suppose that in response to the trolley case, with its peculiarly modern method of violence, the powerful emotions that might otherwise say "No!" remain quiet, and a faint little rational voice can be heard: "Isn't it better to save five lives instead of one?" That's a hypothesis. Is it true? And how can we tell? This hypothesis makes some strong predictions regarding what we should see in the brain scanner while people are responding to personal and impersonal moral dilemmas. The contemplation of personal moral dilemmas like the footbridge case should produce relative increases in neural activity in brain regions associated with emotional response and social cognition, while the contemplation of impersonal moral dilemmas should produce relatively greater activity in regions associated with "higher cognition." This is exactly what was observed (Greene et al., 2001). Contemplation of personal moral dilemmas produced relatively greater activity in two emotion-related areas, the posterior cingulate cortex and the medial prefrontal cortex (one of the areas damaged in both Gage and Elliot), as well as in the superior temporal sulcus, a region associated with various kinds of social cognition [Allison et al]. At the same time, contemplation of impersonal moral dilemmas produced relatively greater neural activity in two classically "cognitive" brain areas associated with working memory function in the inferior parietal lobe and the dorsolateral prefrontal cortex. This hypothesis also makes a prediction regarding people's reaction times. According to the view I've sketched, people tend to have emotional responses to personal moral violations that incline them to judge against performing those actions. That means that someone who judges a moral violation to be appropriate (e.g. someone who says it's okay to push the man off the bridge in the footbridge case) will most likely have to override an emotional response in order to do it. That overriding process will take time, and thus we would expect that "yes" answers will take 17 This third criterion mirrors Thomson's [ref] "no new threat" principle. Her principle, however, is a normative principle, whereas this criterion is part of a set of criteria that are simply descriptive. 18 We didn't evolve from chimpanzees, of course. Chimpanzees are simply a stand-in for the ancient primate species that gave rise to both chimpanzees and us. longer than "no" answers in response to personal moral dilemmas like the footbridge case. At the same time, we have no reason to predict a difference in reaction times between "yes" and "no" answers in response to impersonal moral dilemmas like the trolley case because there is, according to this model, no emotional response to override in such cases. Here, too, the prediction holds. Trials in which the subject judged in favor of a moral personal violation took significantly longer than trials in which the subject judged against them, but there was no comparable reaction time effect observed in response to impersonal moral violations (Greene et al., 2001). [Note: the following data from my own work presented here are either unpublished or published only in conference abstracts. Presumably they will have been published by the time this book comes out, but if that is not the case then I may not be able to include them in this discussion. Also, some results may change as more data are collected , although the main results are fairly robust at this point.] Further results support this model as well. Above we contrasted the neural effects of contemplating "personal" vs. "impersonal" moral dilemmas. But what should we expect to see if we subdivide the "personal" moral dilemmas into two categories based on difficulty (i.e. based on reaction time)? Consider the following moral dilemma (the crying baby dilemma): It's wartime, and you and some of your fellow villagers are hiding from enemy soldiers in a basement. Your baby starts to cry, and you cover his mouth to block the sound. If you remove your hand your baby will cry, the soldiers will hear, and they will find you and the others and kill all of you, including you and your baby. If you do not remove your hand, your baby will smother to death. Is it okay to smother your baby to death in order to save yourself and the other villagers? This is a very difficult question. Different people give different answers and nearly everyone takes a relatively long time to answer. Here's a similar dilemma (the infanticide dilemma): You are a teenage girl who has become pregnant. By wearing baggy clothes and putting on weight you have managed to hide your pregnancy. One day during school, you start to go into labor. You rush to the locker room and give birth to the baby alone. You do not feel that you are ready to care for this child. Part of you wants to throw the baby in the garbage and pretend it never existed so that you can move on with your life. Is it okay to throw away your baby in order to move on with your life? For the people in our test sample, at least, this is a very easy question. All of them say fairly quickly that it would be wrong to throw the baby away. What's going on in these two cases? My colleagues and I hypothesized as follows. In both cases there is a prepotent, negative emotional response to the personal violation in question, killing one's own baby. In the crying baby case, however, there are powerful, countervailing rational19 considerations that push one toward smothering the baby. After all, the baby is going to die no matter what, and so you have nothing to lose20 and much to gain by smothering it, awful as it is. In some people the emotional response ("Aaaahhhh!!! Don't do it!!!") dominates, and those people say "no." In other people, a "cognitive," cost-benefit analysis ("But you have nothing to gain, and so much to lose...") wins out, and those people say "yes." 19 By calling these considerations "rational" my intention is not to endorse them, but rather to describe the type of psychological process that underlies them as one that is "cool" rather than "hot," "cogntive," deliberate, etc. 20 At least in terms of lives lost or saved. What does this model predict that we'll see in the brain scanner when we compare cases like crying baby to cases like infanticide? First, this model supposes that cases like crying baby involve an increased level of "response conflict," i.e. conflict between competing representations for behavioral response. Thus, we should expect that difficult moral dilemmas like crying baby will produce increased activity in a brain region that is associated with response conflict [Botvinick et al], the anterior cingulate cortex. Second, according to our model, the crucial difference between cases like crying baby and cases like infanticide is that dilemmas like crying baby involve "cognitive" considerations that compete with the prepotent, negative emotional response. Thus, we should expect to see increased activity in classically "cognitive" brain areas when we compare cases like crying baby to cases like infanticide, even though dilemmas like crying baby are personal moral dilemmas. Finally, according to our model, cases like crying baby involve a competition between "cogntive" and emotional forces, and so we should expect to see some evidence of the emotional forces making a stand. At the same time, however, the emotional response is supposed to be common to both kinds of dilemma in this comparison, and so some of the emotion-related activity associated with personal moral dilemmas should subtract out. All of these predictions held (Greene et al, 2002). Comparing high reaction time personal moral dilemmas like crying baby to low reaction time personal moral dilemmas like infanticide revealed increased activity in the anterior cingulate (conflict), the anterior dorsolateral prefrontal cortex ("cognitive"), the inferior parietal lobes ("cognitive"), the posterior cingulate (emotional), and the precuneus (a region near the posterior cingulate that has been associated with visual imagery and that appears to work in concert with the posterior cingulate). The activity in the medial prefrontal cortex or the superior temporal sulcus subtracted out. So far we have talked about neural activity correlated with the type of dilemma under consideration, but what about activity correlated with subjects' behavioral response? Does a brain look different when it's saying "yes" as compared to when it's saying "no" to questions like these? To answer this question we subdivided our dilemma set further by comparing the trials in which the subject says "yes" to difficult personal moral dilemmas like crying baby to trials in which the subject says "no" in response to such cases. Once again, we turn to the model for a prediction. If cases in which people say "no" are cases in which emotion wins, then we would expect to see more activity in the posterior cingulate and possibly the precuneus in those cases. Likewise, if the cases in which people say "yes" are cases in which "cognition" wins, then we would expect to see more activity in the dorsolateral prefrontal cortex and/or parietal lobes in those cases. The first of these predictions held. That is, the posterior cingulate and precuneus showed relatively greater activity at the end of trials in which the subject said "no." There was no significant effect in keeping with the second prediction for this comparison, although there was a trend in the right direction in the right anterior dorsolateral prefrontal cortex (but see below for a significant effect in this area). Additionally, the insula, a region associated with disgust, anger, and autonomic arousal [refs] showed increased activity for "yes" answers, a surprise given our model (according to which emotional responses are associated with "no" answers for these questions). However, an examination of the time course of the activity in this area revealed that the differences in activity in this region probably occurred after the decisions were made. Thus, we interpret this activity as a reactive effect, an emotional backlash to the subject's approving judgment of a personal moral violation. (After all, how you would you feel if you just decided that it would be okay to smother your own baby?) The above analysis was performed at the level of individual trials, but we can also perform a similar analysis at the level of individual subjects. Individual subjects can be characterized along a continuous "utilitarian" -"Kantian" dimension depending on how often and how quickly they are willing or unwilling to sacrifice an individual's welfare in the name of the greater good. It turns out that, at the a well-chosen time point, the level of neural activity in the precuneus correlates positively (r = .38) with Kantian behavioral tendencies. In contrast, the level of neural activity in the right anterior dorsolateral prefrontal cortex correlates positively with utilitarian behavioral tendencies (r = .44). These effects are even stronger if the analyses are restricted to the most Kantian subjects for the precuneus (r = .73) and the most utilitarian subjects for the right anterior dorsolateral prefrontal cortex (r = .94!). In other words, if you're a Kantian (i.e. someone who tends to reach Kantian conclusions, not necessarily someone who reasons as Kantians recommend) then your precuneus activity reflects just how Kantian you are. Likewise, for the right anterior dorsolateral prefrontal cortex if you're a utilitarian. Together, the levels of activity in these two brain areas account for about a third of the variance in people's overall Kantian-utilitarian tendencies (R2 = .33). The above results, taken together, provide strong support for the model sketched above according to which moral decisions are produced through an interaction between emotional and "cognitive" processes subserved by anatomically dissociable brain systems. Another recent brain brain imaging experiment adds further support to this model of moral judgment. Alan Sanfey, Jim Rilling, and colleagues (2003) conducted a brain imaging study of the Ultimatum Game in order to study the neural bases of people's sense of fairness. The Ultimatum Game works as follows: There is a sum of money, say $10, and the first player (the proposer) makes a proposal as to how to divide it up between herself and the other player. The second player, the responder, can either accept the offer, in which case the money is divided as proposed, or reject the offer, in which case no one gets anything. When both players are perfectly rational, purely motivated by financial self-interest, and these facts were known to the proposer, the outcome of the game is guaranteed. Because something is better than nothing, a rationally and financially self-interested reponder will accept any offer. A rational and financialy self-interested proposer who knows this will therefore offer the responder as small a share of the total as possible, and the thus the proposer will get nearly all and the responder will get nearly none. This, however, is not what usually happens when people play the game, even when both players know that the game will only be played once. Proposers usually make offers that are fair (i.e. fifty-fifty split) or close to fair, and responders tend to reject offers that are more than a little unfair. Why does this happen? The answer, once again, implicates emotion. The results of Sanfey and Rilling's study show that unfair offers, as compared to fair offers, produce increased activity in the anterior insula, the region mentioned above that is associated with anger, disgust, and autonomic arousal. Moreover, individuals' average levels of insula activity correlated positively with the percentage of offers they rejected and was weaker for trials in which the subject believed that the unfair offer was made by a computer program. But the insula is only part of the story. The anterior cingulate , the region mentioned above that is associated with response conflict, and the dorsolateral prefrontal cortex, one of the regions mentioned above that is associated with "higher cognition" were also more active in response to unfair offers. Moreover, for trials in which the unfair offer was rejected, the level of activity in the insula tended to be higher than the level of activity in the dorsolateral prefrontal cortex, while the reverse was true of trials in which unfair offers were rejected. These results dovetail nicely with the imaging experiments of hypothetical moral dilemmas described above. In both studies, the insula subserves an emotional response to an action that is naturally seen as unfair. (In my study the insula responds to the subject's own decision to approve of a personal moral violation, whereas here the subject and her insula are reacting to another person's decision to allocate resources unfairly.) In both studies there is a "cognitive" rationale for not acting on the basis of one's emotions. In response to crying baby, for example, one recognizes that putting one's parental instincts aside and smothering the baby will produce the best overall consequences, whereas in the Ultimatum Game one recognizes that putting one's righteous indignation aside and accepting an unfair offer will make oneself more money. These instances of "cognitive overriding" appear to be subserved by regions in the dorsolateral prefrontal cortex. Finally, according to this model, both unfair offers and difficult moral dilemmas elicit competing emotional and "cognitive" representations, as indicated by the activation of the anterior cingulate cortex in both cases. Other neuroimaging results have shed light on the neural bases of moral judgment. Jorge Moll and colleagues have conduced two experiments using simple, morally significant sentences (e.g. "They hung and innocent") [Moll et al., 2001, 2002b] and an experiment using morally significant pictures (e.g. pictures of poor abandoned children) [Moll et al., 2002a]. These studies along with the ones described above implicate a wide range of brain areas in the processing of morally significant stimuli, with a fair amount of agreement (given the variety of tasks employed in these studies) concerning which brain areas are the most important.21 In addition, many of the brain regions implicated by this handful of neuroimaging studies of moral cognition overlap with those implicated in neuroimaging studies of "theory of mind," the ability to represent others' mental states [Frith, 2001]. (For a more detailed account of the neuroanatomy of moral judgment and its relation to related processes see Greene and Haidt (2002).) While many big questions remain unanswered, it is clear from these studies that there is no "moral center" in the brain, no "morality module." Moreover, moral judgment does not appear to be a function of "higher cognition," with a few emotional perturbations thrown in [Haidt, 2001; Damasio, 1994] Nor do moral judgments appear to be driven entirely (or even more or less entirely) my emotional responses [Haidt, 2001]. Rather, moral judgments appear to be produced by a complex network of brain areas subserving both emotional and "cognitive" processes [Greene et al., 2001; Greene and Haidt, 2002; Sanfey et al., 2003]. Other data: development, anthropology/cultural psychology, evolutionary psychology, and behavioral genetics There are many sources of valuable information that bear on the issue of innateness in moral psychology. In addition to the ones discussed above, I will briefly mention five more. The most extensive literature in moral psychology is probably the literature on moral development, including the well-known works of Piaget [ref], Kohlberg [ref, ref], Turiel [ref], and Giligan [ref], among others. In the best cases, the study of development can provide dramatic evidence for the existence of innate behaviors. Indeed, if newborn human babies, like newborn gazelles taking their first shaky steps, were capable of making rudimentary moral judgments straight out of the womb, it would be clear that our capacity for moral judgment is as 21 An exception here are the "cognitive" areas, which are only seen using paradigms that involve "impersonal" moral judgments and/or difficult moral decisions [Greene et al., 2001; Sanfey et al., 2003]. innate as anything else. Needless to say, things don't work that way, and as a result the scientific record of human moral development is difficult to interpret. The are two fundamental problems that make drawing conclusions about nativism in moral psychology from the developmental literature difficult. The first problem is methodological. Much of the most influential work in moral psychology has come from analyses of children's verbal accounts of their own moral sensibilities [Kohlberg]. As demonstrated throughout this chapter, there is a large and growing body of evidence suggesting that moral judgment is largely an intuitive, emotional affair [Haidt, 2001]. If that's correct, then it's likely that much of the work that has been done chronicling the increasingly sophisticated modes of moral thought employed by children, teenagers, and adults has inadvertently been directed toward the development of people's abilities to verbalize and rationalize their moral sensibilities rather than the development of those sensibilities themselves [Haidt, 2001] The second, and more significant, problem was alluded to above. Moral development, like so much else, is a robustly interactive affair, and, in a species that depends so much on cultrual learning, it is difficult to tease apart the genetic and environmental components of any complex social behavior simply by observing it. There are, however, some suggestive findings. For example, it is sometimes argued that behaviors that are learned during a critical period are behaviors that we are specifically and innately prepared to learn, but these claims are controversial. (The claims are most often made [Pinker] and disputed [McClelland, Elman et al] in discussions of the development of language.) As noted above, patients who sustain ventromedial damage at a young age appear more likely to exhibit anti-social behavior than those who sustain such damage later in life (Anderson et al., 1999). Rozin, Fallon, and AugustoniZiskind (1985) point out that children in cultures that do not emphasize matters of purity and pollution in moral contexts [Schweder] often develop such intuitions spontaneously around the ages of seven or eight. (Think of children’s frequent obsessions with “cooties.”) These intuitions and concerns, however, tend to wither without cultural support. Something similar appears to be the case with respect to the ethics of autonomy and community. Around the age of four children tend to go from being relatively uninterested in matters of fairness to being obsessed with them, often overgeneralizing norms of fairness to inappropriate situations (Fiske, 1991). Minoura (1992) found striking differences in the socialization processes undergone by Japanese children of different ages living in America where their fathers were temporarily transferred for work. Of these children, those who spent a few years in America during the ages of nine through fifteen tended to develop American ways of interacting with friends and of reacting to and addressing interpersonal problems. The ones who spent time in America before the age of nine showed no such lasting effects, and the ones who arrived in America after the age of fifteen did not adjust as well to American life. Such late arrivals typically felt awkward behaving in American ways even while having excellent explicit knowledge of American behavioral norms. While human newborns have little in the way of a moral sensibility, human children can be morally precocious in a way that suggest that they have received some specific help from their genes, a claim that is bolstered by observations of what appear to be analogous behaviors in chimpanzees (see above). Children exhibit empathy and other pro-social behaviors from a very young age [Hoffman, 1982, Kagan, ?]. Likewise, young children (including autistic children [Blair]) are able to draw the moral/conventional distinction (see above) [Turiel]. Regardless of what one makes of the evidence for nativist views of moral development, it's clear that culture plays an important role in shaping people's moral sensibilities, as demonstrated by the fact that moral beliefs and values vary widely from culture to culture, and from sub-culture to sub-culture. Anthropologists and cultural psychologists have documented a wide range of such differences [Haidt et al, 1993; Fiske, 1990; Nisan, 1987; Schweder; Rozin, 1997a, 1997b; Lakoff, 1996; Naravez, 1999; Edwards, 1987 what's a good summary for this?], but the general point that cultural differences involve differences in moral outlook can be ascertained simply by reading the newspaper. The more interesting question, then, is whether there are similarities in moral outlook among the various cultures of the world that point toward an innate component. Some have observed that the cultural variations on human morality appear to variations on certain central themes. According to Shweder and his colleagues (1997), moral norms and intuitions cluster around what he calls the “big three” domains of human moral phenomena: the “ethics of autonomy” which concerns rights, freedom, and individual welfare; the “ethics of community” which concerns the obligations of the individual to the larger community in the form of loyalty, respectfulness, modesty, self-control, etc.; and the “ethics of divinity” which is concerned with the maintenance of moral purity in the face of moral pollution. Rozin and colleagues (1999) argue that these three domains correspond to three basic moral emotions: anger for autonomy, contempt for community, and disgust for divinity. While all cultures appear to have practices associated with each of the "big three" moral domains to some extent, their emphases can be very different. Westerners, for example, are said to have relatively little experience with the ethics of divinity, although there are exceptions as seen, for example, in the moralization of meat-eating among some Western vegetarians [Rozin]. Others have observed that certain moral principles seem to be universal such as injunctions against rape and murder and provisions for redressing such wrongs [Brown], and some claim that social life is structured by a discrete set of principles for fair exchange [Fiske], with strong moral injunctions against exchanges that violate these principles [Tetlock et al.]. One problem in interpreting these regularities is that the presence of a trait across a wide variety of cultures, even all cultures, does not imply the trait in question is innate. As noted above, poetry and hair styling are observed in all cultures [Brown], but there seems to be a further question, and much room for skepticism, as to whether these things are innate. At the same time, however, there do seem to be some commonalities in moral outlook that smack of innateness. Perhaps the best example is the incest taboo, which is maintained in some form in all cultures [Brown]. There are three complementary reasons why this aspect of morality is such a good candidate for innateness. First, it has a strong biological rationale and is observed in other species [ref]. (Matings between first-degree relatives are more likely to produce defective offspring.) Second, the taboo is endorsed even by peoples who have no knowledge of the deleterious biological consequences of incest. Third, people who are aware of these consequences maintain that incest is wrong even when those consequences can be avoided (e.g. through birth control) but have an amusingly hard time justifying their convictions [Haidt]. The incest taboo appears to be grounded in a strongly felt moral intuition that outstrips any rationale we can provide for it, which suggesting that the biological rationale is the operative one, implemented in the form of an evolutionary adaptation. The case for the incest taboo as a biological adaptation is pretty strong, but is it an exception? Many evolutionary theorists argue that the incest taboo is just the tip of the iceberg when it comes innate moral dispositions in humans. These arguments fall into two main categories which we might call "possibility arguments" and "actuality arguments." For a long time, the existence of pro-social or altruistic behavior in humans and other animals was viewed as a problem for those who would explain the most salient aspects of human nature in terms of natural selection: If nature selects for traits that help individuals outcompete other individuals in the struggle for existence, why would any individual do anything to help others' at one's own expense? Many people, including one of evolutionary theory's earliest and most vocal proponents (Huxley, 1894), concluded that morality must be a cultural addition to biological human nature, and perhaps even an antidote of sorts. In recent decades, however, an impressive body of work, beginning with the seminal works of Hamilton [1964], Maynard Smith [1964],Williams [1966], and Trivers [1971] has explained how the amoral forces of natural selection can produce moral creatures with genuinely altruistic motives. These theories have been refined in recent years and have given rise to the field of evolutionary game theory [Axelrod; Skyrms; Boyd and Richerson; Sober and Wilson]. I will not attempt to summarize these developments, but will simply report that the existence of genuine altruism in humans and other animals is no longer regarded as an obstacle to evolutionary explanations of human nature. These days, most of the controversy in evolutionary psychology centers around specific theories according to which various aspects of human moral nature are biological adaptations. Many of these arguments have focused on the perennially alluring topics of sex and violence. For example, people have used evolutionary theory to explain patterns in gender roles and sexual preference [Symons, 1979; Perrett face study, 1999; Wright, 1994] as well as patterns in human violence [Daly and Wilson; Wright, 1994; Demonic Males; Thornhill and Palmer, 2000, Buss and Duntley, this volume.] Some have sought positive evidence for the cognitive machinery postulated by earlier theorists' possibility arguments. For example, the adaptation for reciprocal altruism described by Trivers requires that individuals be reasonably good at detecting those who would reap the benefits of social exchange without paying the costs. This led Leda Cosmides [1989] to posit the existence of a specialized "cheater detection" mechanism and to catch it in action using her well-known variation on the Wason selection task, a test of logical reasoning. When people are given a standard version of this task in which the subject must identify the conditions required to satisfy an abstract logical rule, they do poorly, but when the test is recast in terms of a social exchange whereby the subject must identify the behavioral conditions that must be met in order for a social norm to have been respected, people do much better. Cosmides concludes that this specialized ability to reason about social exchange is a biological adaptation that allows people to successfully reap the benefits of social life. Most of the arguments regarding evolutionary adaptations that bear on moral judgment and behavior are speculative, theoretical arguments of the form "Wouldn't it make sense that...?" I personally am sympathetic to evolutionary psychology with respect to understanding both morality and human nature more broadly, but I am the first to admit that the field is heavy on theory and light on hard data. The fundamental difficulty in evolutionary approaches to human behavior is that they seek to explain readily observable phenomena in terms of processes that took place long ago and that leave very little in the way of physical evidence. The fact that the phenomena to be explained by evolutionary theories are readily observable is, perhaps surprisingly, a disadvantage , because this makes it hard for theorists to make genuine predictions, i.e. to describe in advance phenomena that have yet to be observed, either formally or informally. Theorists whose evidence consists of successful "postdictions" can always be accused of retrofitting theory to data, and, regardless of one's views regarding the normative epistemology of science, novel predictions tend to be more compelling. There are, however, a handful of cases in which evolutionary theories of human behavior have made novel predictions22 that have turned out to be true. Cosmides' studies using the modified Wason Selection Task are a good example. Her evolutionary interpretation of these results has been criticized, but in this case it's the skeptics whose alternative theories are in danger of having been retrofitted to the data. In the years to come, the emerging field of behavioral genetics will contribute greatly to our emerging understanding of human moral nature. It's important, however, to understand what exactly studies correlating genes and behavior can and cannot tell us. There are, unbeknownst to many people, two broad "nature-nurture" questions. The first question is concerns the extent to which genetic influences account for the aspects of human nature that we all share. This is the question with which most of this chapter has been concerned. More specifically, we have been examining the features of moral psychology that all normal humans appear to have as well as features that we share with other species. So far as this first question is concerned, genetic evidence is difficult to interpret because it is hard, both empirically and conceptually, to partial out the effects of genes when both the genes and the effects are the uniform. The second question concerns the factors that make us different from one another, and here genetic evidence has proven incredibly powerful. Behavioral geneticists track correlations between genes and behavioral traits in two main ways. The first way is to compare the behavior of people with known degrees of genetic relatedness. Pairs of identical twins reared apart can be compared to unrelated people reared apart. Pairs of identical twins reared together can be compared to pairs of fraternal twins reared together, both of whom can be compared to pairs of ordinary siblings reared together, all of whom can be compared to pairs of genetically unrelated individuals reared together. In a similar way, the genetic heritability of traits can be deduced through the examination of family trees. Second, in contrast to the whole-genome approach employed in twin and family studies, researchers can also identify correlations between observable traits and particular genes or ensembles of genes. Four decades of research have produced some striking results. Testing confirms that identical twins, whether separated at birth or not, are eerily alike (though far from identical) in just about any trait one can measure. They are similar in verbal, mathematical, and general intelligence, in their degree of life satisfaction, and in personality traits such as introversion, agreeableness, neuroticism, conscientiousness, and openness to experience. They have similar attitudes toward controversial issues such as the death penalty, religion, and modern music. They resemble each other not just in paper-and-pencil tests, but in consequential behavior such as gambling, divorcing, committing crimes, getting into accidents, and watching television. And they boast dozens of shared idiosyncrasies such as giggling incessantly, giving interminable answers to simple questions, [and] dipping buttered toast in coffee... The crags and valleys of their electroencephalograms ... are as alike as those of a single person recorded on two occasions, and the wrinkles of their brains and distribution of gray matter across cortical areas are also similar. (Pinker, 2002, pg. 47) 22 By "novel predictions" I mean cases in which folk wisdom fails to make the same prediction. For example, the "prediction" that men are, on average, more willing to have sex with strangers than women doesn't count as novel, even before this phenomenon has been scientifically documented. If you have a longer than average version of the D4DR dopamine receptor gene you are more likely to be a thrill seeker, the kind of person who jumps out of airplanes, clambers up frozen waterfalls, or has sex with strangers. If you have a shorter version of a stretch of DNA that inhabits the seratonin transporter gene on chromosome 17, you are more likely to be neurotic and anxious... (pg. 48) ...A conventional summary is that about half of the variation in intelligence, personality, and life outcomes is heritable... (pg. 374) So far as I am aware, there haven't been any genetic studies of moral psychology per se, but the results described above concerning such traits as criminality, sexual promiscuity, and views on the death penalty come pretty close, and given the general trend described above it seems unlikely that genes will account for less of the variation in moral behavior than they do for most other kinds of behavior. That said, two very important caveats to these bold claims deserve attention. First, the heritability of a trait is relative to a population, and the more homogeneous the population the higher the heritability values are likely to be. The results described above are typically from studies of relatively homogeneous populations, and therefore they might not tell us much about the extent to which genes account for differences between cultures (Pinker, 2002, pg. 380). Nevertheless, if you're an American who wants to know why you and George W. Bush can't seem to agree on anything, genes may be a large part of the answer. The second important caveat is that individual differences in moral values and character may not be the differences that matter most. Decades of research in social psychology, including a high proportion of its most celebrated studies [Milgram, Darley and Latane], have shown that situational variables explain much more of people's behavior than is generally acknowledged23 and that personality traits explain much less than is ordinarily thought [Nisbett and Ross]. (For an excellent philosophical account of moral philosophy, moral psychology, and situationsim see Doris [?]) Thus, even if genes explain a lot about why I behave differently from you when you and I are in the same situation, the differences between our situations may be the differences that matter most. Still, this point should not obscure the fact that genes account for a great deal of behavioral variation across situations, as demonstrated by the striking similarities between twins reared apart.24 What in moral psychology is innate? In extracting from the above discussion an answer to this question, it will be useful to distinguish between the form and content of moral thought. The form of moral thought concerns the nature 23 There is some evidence to suggest that Asians do a better job of acknowledging this than Westerners. [Nisbett] There is certainly an interesting—and, to my knowledge, overlooked—tension between situationism and the apparent power of genetic factors to explain behavior. One possible resolution of this tension lies in the fact that situational factors seem to account for behavior in particular situations (e.g. a particular choice about whether or not to help a distressed individual) whereas genes seem to account for broad patterns of behavior (e.g. criminal tendencies, television watching). Perhaps genes account for people's "long run averages" while situational factors provide a better account of people's day-to-day choices. It's not clear what consequences this resolution would have for those who have found in the situationist doctrine grounds for skepticism about the existence of morally significant character traits [Doris; Harman]. 24 of the cognitive processes that subserve moral thinking, which will most likely be a function of the cognitive structures that are in place to carry out those processes. The content of moral thought concerns the nature of people's moral beliefs and attitudes, what they think of as right or wrong, good or bad, etc.. Thus, it could turn out that all humans have an innate tendency to think about right and wrong in a certain way in the absence of any genetic predisposition to come to any particular conclusions about which things are right or wrong. With this in mind, let us review the data presented above. A wide range of studies over several decades reveals striking similarities between the social and emotional dispositions that structure the social lives of non-human primates and those of humans. With respect to the form of moral thinking, the primary similarities appear to lie in the affective domain. The basic emotions that pull us human that govern human social exchange (anger, empathy, anxiety, joy) appear to be at work in non-human primates, and in similar ways. Thus, so far as the form of our moral thought is concerned, we humans apparently make use of many of the same cognitive tools as other primates. No one believes that the chimpanzee 's social emotions are cultural inventions, and the principle of "evolutionary parsimony" suggests that our strikingly similar social emotions spring from a common genetic source, courtesy of natural selection. There are, of course, aspects of human moral thinking that are not observed among our nearest relatives. As Otto from A Fish Called Wanda points out "Gorillas don't read Nietzsche!"25 Humans certainly have a complex, language-dependent capacity for abstract moral thought that outstrips anything of which other animals are capable. Jonathan Haidt (2001) has argued, however, that our that human moral reasoning is largely a post-hoc affair and plays little direct role in producing moral judgments, which are made on the basis of emotional intuitions. Thus, if the strong emotivist position developed by Haidt is correct, the form of human moral thought may, underneath all the rhetoric and mediating cognitive capacities, be very similar to that of chimpanzees. Studies of non-human primates also speak to the content of human morality. Their social lives are not only governed by social emotions, these social-emotional responses appear to be directed toward familiar contents. Chimpanzees do not, for example, fly into a rage when a member of the troupe peels his banana from the bottom instead of the top. Instead, they seem to care about the same sorts of things that we do: avoiding personal violence, equitable distribution of material resources, restrictions on sexual access, and so on. Thus it seems likely that many of our most basic moral values find their roots in the pro-social dispositions of our primate ancestors. A number of themes emerge from studies of patients with social behavioral problems stemming from brain injury, psychopaths, and the neural bases of moral judgment in normal individuals. Popular conceptions of moral psychology, bolstered by the legend of Phineas Gage and popular portrayals of psychopaths, encourage the belief that there must be a "moral center" in the brain. This does not appear to be the case. The lesion patients, both developmental and adult-onset, all have deficits that extend beyond the moral domain, as do the psychopaths that have been studied. Moreover, the results of brain imaging studies of moral judgment reveal that moral decision-making involves a diverse network of neural structures that are implicated in a wide range of other phenomena. Nevertheless, the dissociations observed in pathological cases and in the moral thinking of normal individuals are telling. Most importantly, multiple sources of evidence point toward the existence of at least two relatively independent systems that contribute to moral judgment: (1) an affective system that (a) has its roots in primate social 25 Or, perhaps, as Wanda maintains, they do read Nietzsche, but fail to understand it. emotion and behavior; (b) is selectively damaged in certain patients with frontal brain lesions; and (c) is selectively triggered by personal moral violations, perceived unfairness, and, more generally, socially significant behaviors that existed in our ancestral environment. (2) a "cognitive" system that (a) is far more developed in humans than in other animals; (b) is selectively preserved in the aforementioned lesion patients and psychopaths; and (c) is not triggered in a stereotyped way by social stimuli. I have called these two different "systems," but they themselves are almost certainly composed of more specific subsystems. In the case of the affective system, it's subsystems are probably rather domain-specific, while the system that is responsible for linguistically-based higher cognition, though composed of subsystems with specific cognitive functions, is more flexible and more domain-general than the affective system and its subcomponents. Mixed in, perhaps mistakenly, with what I've called the affective system are likely to be cognitive structures specifically dedicated to representing the mental states of others ("theory of mind") (Greene and Haidt, 2002). What does this mean for the innateness of moral thought? It seems that the form of moral thought is highly dependent on the large-scale structure of the human mind. Cognitive neuroscience is making it increasingly clear that the mind/brain is composed of a set of interconnected modules. Modularity is generally associated with nativism, but some maintain that learning can give rise to modular structure, and in some cases this is certainly true. My opinion, however, is that large scale modular structure is unlikely to be produced without a great deal of rather specific biological adaptation. Insofar as that is correct, the form of human moral thought is to a very great extent shaped by the contingent structure of the human mind, which itself is a product of natural selection. In other words, nature does not provide us our moral thinking is not the result of applying a general-purpose learning device to the problems of social life. As the stark contrast between the trolley and footbridge problem suggests, our moral judgment is greatly affected by the quirks in our cognitive design. As for the content of human morality, there are good reasons to think that genes play an important role here as well. Psychopaths, with their highly stereotyped suite of deficits, don't just get things wrong. They get things wrong in very specific ways. More specifically, they appear to be lacking in empathy, an emotional capacity that counteracts the more selfish motives that are so dramatically displayed inpsychopathic behavior. Even if there is no "empathy module" the fact that empathic emotions can be removed without damaging too much else suggests that ordinary humans, like chimpanzees, are equipped with domain-specific socialemotional tendencies that are products of natural selection and that lie at the core of human morality. So far I've argued that the form of human moral thought is importantly shaped by the innate structure of the human mind and that some basic pro-social tendencies probably provide human morality with innate content. What about more ambitious versions of moral nativism? Some have suggested, for example, that there may be a moral equivalent of "universal grammar," [Rawls, Stich, Mikhail?, others] a deep structure to our moral thought that is hidden from ordinary moral experience, but nonetheless responsible for shaping it. I have my doubts. Of course, it depends on what one means by "grammar." As I've said, I do think that the human mind has a very particular structure that is universal and that profoundly shapes both the form and content of moral judgment. You can call that "universal moral grammar" if you like, but I think this is misleading. It seems to me that the proponents of moral grammar have a normative agenda. They want not only deep form, but highly specific deep content. In other words, what they really want is more analogous to a moral language than a moral grammar. They look at people's responses to things like the trolley and footbridge cases and observe that people seem to have a detailed and highly sensitive knowledge of right and wrong that defies people's ability to explicate that knowledge, much as people's implicit understanding of grammar far outstrips their explicit understanding of how language works. Such nuggets of implanted moral wisdom encourage the thought that somewhere, deep in our cognitive architecture, we're going to find the mother lode. I am very skeptical of the idea that there are detailed moral truths written into our psyche. Rather, I think that the suggestive cases like trolley and footbridge work so well because they exploit a large-scale dissocation in our cognitive architecture. We recoil at the thought of pushing the man off the footbridge, but not at that of hitting the switch, because we're adapted to respond emotionally to good-old-fashioned interpersonal violence, and not because we have a detailed set of moral rules written into our brains, one of which tells when exactly it is okay sacrifice one life to save to save five. And besides, even if there were a detailed set of moral rules written into our brains, would it matter? It would certainly be nifty if it were true, but I don't think it would get us what we want out of normative ethics. That is, I don't think it would do anything to settle the important moral questions over which people disagree. If my deepest moral convictions were at odds with the dictates of innate moral grammar, I would simply say "So much the worse for innate moral grammar!", as could you. I believe that the question of nativism in moral psychology grips many people because our moral thought is at once both highly familiar and thoroughly alien. Our moral convictions are central to our humanity and integral to our lives, and yet their origins are obscure, leading people to attribute them to supernatural forces, or their more naturalistic equivalents. For some, it seems, the idea of innate morality holds the promise of validation. Our moral convictions, far from being the internalization of rules that we invented and taught one another, would be a gift form a universe wiser than ourselves. I believe that there is much wisdom in our moral instincts, but they, like everything else nature hands us, have their quirks and their flaws. Those who seek redemption in the study of moral psychology are bound to be disappointed, but there are, I think, enough rewards on the horizon to make it worth the trouble. they believe that in some sense, our moral instincts will be validated if they turn out to be something that we've taken from forces wiser than ours our ability to explicate that knowledge, much as people human moral behavior and moral grammar unlikely, and unlikely to give people what they want, which is really an innate moral "language" i.e. something with content.