By James Q. Wilson, American Political Science Review, March 1993 v87 n1 p1(11)

Presidential address, American Political Science Association, 1992.

The central problem for social science is to explain social order. How do people manage to live together? One can discern two ways of answering that question. The first view is normative and communal: people learn from their culture customs that provide an internal compass guiding them to act in ways that minimize conflict and ensure comity. The second view is rationalistic and individualistic: order is created by explicit and implicit agreements entered into by self-seeking individuals to avert the worst consequences of their predatory instincts. In the first view, order is natural and prior to any social contract or government institution; in the second, it is contrived and dependent on agreements and sanctions. Rules are obeyed in the first case because they have moral force, in the second because they convey personal advantage. In the first view, compliance is automatic and general; in the second, it is strategic and uncertain.(1)

The normative view has been under heavy attack for several decades for at least three reasons: it seems to imply a complacent functionalism; it appears to minimize or deny the value of conflict; and it lacks the theoretical power found in the assumption that people always seek their own interests. I believe that one can grant, up to a point, all of these objections and still be left dissatisfied with the alternative, namely, that social order is contrived, based on calculation, and dependent on individual assent.

I wish to reestablish a version of the normative view. My argument is that while conflict within societies is ubiquitous and diversity among them obvious, people everywhere have a natural moral sense that is not entirely the product of utility or convention. By moral sense I mean a directly felt impression of some standards by which we ought to judge voluntary action. The standards are usually general and imprecise. Hence, when I say that people have a moral sense, I do not wish to be understood as saying that they have an intuitive knowledge of moral rules. Moral rules are often disputed and usually in conflict; but the process by which people resolve those disputes or settle those conflicts leads them back to sentiments that seem to them to have a worth that is intuitively obvious. These sentiments constitute the fundamental glue of society, a glue with adhesive power that is imperfect but sufficient to explain social order to some degree. The philosophers of the Scottish Enlightenment, in particular Francis Hutcheson writing in 1742, David Hume in 1740, and Adam Smith in 1759, explored with care and subtlety the reasons why certain sentiments commend themselves to us as worthy. I am under no illusion that I can improve on what they accomplished; but I hope to show, by drawing on the social and biological sciences, that their fundamental claims are consistent with much of what we have learned since the mid-eighteenth century.

One can infer the existence of a moral sense from behaviors that cannot easily be explained by even enlightened self-interest. There is less crime than one would expect from the probability of detection and punishment.(2) Even in the poorest neighborhoods, a complete breakdown of law and order does not lead most people to engage in looting. There are more obligations honored than one can explain knowing only that it is often useful to honor them. For example, we sometimes keep promises when it is not in our interest to do so, we often vote in elections even though we cannot affect the outcomes, we make charitable donations to organizations that confer no recognition on us, and some of us help people in distress even when no one is watching to applaud our good deed.

It will be objected that voting, donating, and helping are far from common actions and thus that little of moral significance can be inferred from their occasional performance. I do not fully accept that objection, but I recognize its force. Let me turn, then to behavior that is well nigh universal, that cannot be explained by individual calculations of utility or by negotiated social contracts, and that has obvious moral implications in any plausible meaning of the word moral: child care.


People bring children into this world and nurture them through long years of dependency. They do this with no hope of immediate gain and every expectation of sleepless nights, financial burdens, and daily vexations. David Hume, in his attempt to base morality upon sentiments, was led ineluctably to the parent-child relationship as the founding sentiment. Justice, he argued, was an artifice, a set of rules useful because they make people secure in their property and enable them to transmit it in an orderly fashion. But why do people care about the transmission of property? Because of "the natural affection, which they bear their children" (1978, 486). That natural affection implies an obligation; people everywhere praise those who care for their children and despise those who do not. Note how easily Hume inferred an "is" statement from an "ought" statement scarcely eight pages beyond the famous passage in which he suggested that this cannot be done.

Two objections may be made to this view. Scholars bent on finding self-serving explanations for behavior will argue that parents produce offspring because the latter are useful as unpaid laborers and future breadwinners. While this no doubt occurs and may explain why some parents feel a duty to their children, it cannot explain why children should feel any obligation to their parents. The youngsters are free-riders who benefit from nurturance whether they later support those who nurtured them or not. Yet children feel and act upon obligations to their parents despite the fact that such actions are unprofitable.

Scholars determined to explain all behavior as culturally determined and thus morally relativistic will argue that child care is by no means universal. Many writers have asserted that a sense of childhood is a recent invention, preceded, at least in Europe, by centuries of neglect, abuse, death, and abandonment (Aries 1962; de Mause 1974; Shorter 1975; Stone 1977). If mothers still kill their own babies and if parents have only recently stopped sending them to foundling homes, the love of children must be a recently acquired and thus socially learned disposition. In the words of Lloyd de Mause, the history of childhood is a "nightmare from which we have only recently begun to awaken" (1974, 1). The further back in time we go, the more likely the child was to be killed, abandoned, beaten, terrorized, and sexually abused. Edward Shorter gathered data on the use of wet nurses and foundling homes in the eighteenth century that, to him, bespoke a "traditional lack of maternal love" (1975, 203-4). A sense of childhood--and the love of children--is seen as a modern invention, ascribed variously to religion, capitalism, and the Enlightenment.

It is odd that this view should have had so large an impact, since we already knew from studies of contemporary primitive societies, such as the San of the Kalahari Desert, that children are not only loved but indulged without benefit of modern science, enlightened teaching, or capitalist requirements (Konner 1982, 301-4; see also LeVine 1970). It is even odder that conclusions about people's feelings about children were inferred from data about how children were treated, since the treatment of children might well have been shaped as much by circumstances as by attitudes (Anderson 1980; Macfarlane 1979). For example, wet-nursing was used by women who, because of their employment in agriculture, could not breast feed their own infants and for whom no other safe source of food was available (Sussman 1977; Wrightson 1982). In any event, wet-nursing was far less common than Shorter would have us believe. Perhaps 10% of all Parisian women, beyond that small percentage who were physiologically unable to nurse, put their children out to wet nurses (Ozment 1983, 220, n. 83; cf. also pp. 118-19). There is no evidence to support the assumption that wet nurses or foundling homes were used because most or even many parents were indifferent to their children.

Such direct evidence as we have about the feelings of European parents and children toward one another is inconsistent with the view that a caring family is a recent invention. The painstaking research of Linda Pollock (1983) in over four hundred diaries and autobiographies, including many written by children, suggests quite strongly that British and American parents and children from the sixteenth to the nineteenth centuries felt toward each other much what we feel today: mutual attachment and great affection.(3)

But surely the existence of infanticide throughout history confirms the purely conventional nature of family attachments. Few could a natural sentiment--affection for a child--ever coexist with the deliberate killing of that child? It is a profoundly important question and one that cannot be entirely resolved on the basis of the available historical and anthropological evidence. In his brief but chilling history of infanticide, William L. Langer noted that it has existed everywhere since time immemorial as an accepted procedure for disposing of deformed infants and limiting the size of the population during periods of extreme privation. Jews had always condemned the practice (Noonan 1965, 86); but only with the advent of' Christianity did there begin, in Europe at least, the widespread condemnation of the practice on moral grounds, and only with the spread of Christianity did the secular authorities make it a crime (Langer 1974; see also Herlihy 1985, 23-27 and Noonan 1965, 85-87). In hopes of providing an incentive that would reinforce the sanctions of the criminal law, many states created foundling homes in which mothers could leave their unwanted infants. These hopes were realized beyond the capacity of the system to accommodate them. In 1833, 164,319 babies were left in French foundling homes. At about the same time, one such home in Saint Petersburg had 25,000 infants on its rolls, with 5,000 being admitted yearly (Langer 1974, 358-359). The crowding was so great, and regulation so lax, that conveying a baby to a foundling home was often tantamount to sentencing it to death from neglect.

While this grisly history confirms how often infants were killed or abandoned, it is not very clear about the sentiments and motives of those who did it. Langer suggests that the motive for infanticide reflected extreme circumstances, typically a child who was either deformed or beyond the capacity of its poor parents to feed. To this must be added the threat of social stigma, moral obloquy, or penal sanction faced by unwed mothers. But these are only suggestions, not conclusions based on a close study of parental feelings. For the history of infanticide to shed light on the existence of a moral sense, it is essential to know how the parents, and especially the mothers, felt about what they did. What proportion disposed of the baby without remorse as a matter of convenience, and what proportion did so in anguish and out of necessity? Langer attributed the decline in infanticide in the late nineteenth and early twentieth centuries to the advent of modern contraception coupled with more stringent state regulation. This is a troubling hypothesis, for it implies that convenience dictated whether the baby would be killed or not. It neglects entirely what may have been the more important causes of the decline: a rise in the standard of living sufficient to enable poor parents to support several children, a change in the attitude toward unwed mothers great enough to make it thinkable to keep an illegitimate child, and an improvement in medical care adequate to ensure the ultimate good health of sickly infants.

The only way to assess the moral significance of infanticide is either to examine the feelings of the parents directly or to consider what happens when the conditions giving rise to it change. As with child neglect more generally, it is a mistake to infer sentiments from actions. In the modern world, infanticide still occurs; but there are no reliable data on how often (Dickemann 1984, 433; Scrimshaw 1984, 449-52). The closest thing we have to systematic data is a survey of 112 preindustrial societies from which the authors concluded that infanticide was "common" in about a third of them (Divale and Harris 1976; see also Scrimshaw 1984). The word common was not defined, nor does anyone have any idea how many times it actually occurs in any society.

So far as one can tell, infanticide occurs today under essentially the same conditions as in the past but less frequently: there is so little food that the child cannot be fed (especially a problem with twins), or the child is born so deformed or sickly that it chances of survival are slight (Daly and Wilson 1984, 488-95; idem 1988, chap. 3; Scrimshaw 1984, 444-60). Infanticide may also occur when the child's paternity (hence legitimacy) is in dispute. Of the 112 instances in which a cultural justification for infanticide could be found in the anthropological literature, all but 15 involved food shortages, deformity, or uncertain paternity (Daly and Wilson 1987, 207; see also Konner 1990, 173-76). Less common are instances of female infanticide in order to minimize the number of girls for whom dowries must be provided (Dickemann 1979, 456; Konner 1990, 173-76). No sentiment is sovereign; each must compete with others. A mother's affection for one infant must compete with her affection for another and with her own desire to survive. In a poor area of Brazil, mothers cope with this competition by not naming the baby until its survival seems assured (Scheper-Hughes 1987, 203-4).

That a mother's affection for her infant is not sovereign, however, does not mean that it is natural. Suppose for a moment that mother-child attachments were purely a matter of convention such that infanticide was governed by personal advantage or cultural practice. Under those circumstances, one would expect to find some societies--perhaps many--in which babies were killed even though food was plentiful, paternity certain, and the child healthy. After all, even healthy, easy-to-support children can be, for many years, a great nuisance. We would also expect to find some societies in which children were killed in the second or third year of life, rather than immediately after birth, especially since (as every parent knows) a two-year-old child is often a greater burden than a newborn infant.

But in fact, all of the predictions that follow from a purely relativist view of human nature are, so far as one can tell, false. When economic stresses end, infanticide becomes far less common and is almost always made a criminal act. Healthy babies of certain paternity are rarely destroyed. Infanticide almost never occurs after the first year of life; indeed, it rarely occurs except during the first few hours of life (Daly and Wilson 1987, 208-9; Idem 1988, 75-76; Eibl-Eibesfeldt 1989, 194; Scrimshaw 1984, 440, 448-49; Trevathan 1987, 231-2). That is because infanticide must be committed before bonding takes place. If the baby does not die almost immediately, the mother's distress is very great, at least in the few instances in which scholars have been on the scene to record the events (Eibl-Eibesfeldt 1989, 193-94)^.


The view that there is not a natural moral sense sufficient to account, to any significant degree, for social order began, I believe, when philosophers argued that the human mind was a tabula rasa. If everything, including morality, had to be learned, then anything could be learned. Cultural relativism was the inevitable result of viewing human nature as wholly passive and completely malleable.

Modern science has destroyed that view. It is now clear that nature has prepared the child to be an active participant in his social development and disposed him to see and judge the world in moral terms. In the words of certain anthropologists, the child is an "intuitive moralist" (Shweder, Turiel, and Much 1981).

Newborn infants engage in social activity before they are taught it. They root, suck, and express distress at the sound of other babies crying. They prefer human sounds to other sounds, female sounds to male ones, and maternal sounds to other female sounds (Eibl-Eibesfeldt 1989, 200; Field 1990, 27-39; Hay and Rheingold 1983). This prosocial behavior is not learned. Infants born blind will smile though they have never seen a smile; infants born both deaf and blind will laugh during play, though they have never heard laughter, and frown when angry, though they have never seen a frown (Eibl-Eibesfeldt 1989, 30-31). The newborn infant can tell its mother's voice from that of another. It will imitate several facial and hand gestures within two weeks of birth and some gestures within 32 hours of birth (Meltzoff 1988; Meltzoff and Moore 1977). Most, if not all, of the universal human facial expressions--those expressing happiness, sadness, surprise, interest, disgust, anger, and fear--can be observed in the newborn child (Field 1990, 61). Within two weeks infants will reach for a presented object (Eibl-Eibesfeldt 1989, 53) and will cry at the sound of another baby crying (Radke-Yarrow, Zahn-Waxler, and Chapman 1983, 479) but not at the recorded sound of their own crying (Field 1990, 31; Sagi and Hoffman 1976). Within six months babies can tell the difference between the face of a friendly and an unfriendly adult (Radke-Yarrow, Zahn-Waxler, and Chapman 1983, 480). Within two years, children will share toys, and offer help, and console others who are in distress.(4)

The mother responds to these prosocial behaviors with nurturance, affection, and communication. She smiles at the child's smile and laughs at its laughter, picks it up when it cries, feeds it when it is hungry, and plays it when it is bored. Some people believe that if you reward behavior it will be repeated. One might infer from this general truth the particular claim that if you pick up a crying baby, it will always cry and that if you play with a fussy baby, it will always fuss. Not so. The natural sociability of the child inclines it to acquire greater autonomy and confidence, not greater dependence and manipulativeness, when its desire for attachment is met with an equivalent response from its parent.

This is the great paradox of attachment. Bonded children will grow up to be, not dependent, but independent, at least within such latitude as the culture allows. Human infants become attached to humans who make eye contact, whether or not they supply food. Bonding, once it has occurred, will so persist despite punishment that abused children will remain attached to abusive parents. It is clear from these facts that bonding is driven by powerful biological forces and is not simply the result of a utility-maximizing organism engaging in whatever behaviors bring it immediate rewards.

The child has within it, so to speak, a template that makes some kinds of learning quite easy and others very difficult. For example, a child can be conditioned to fear rats and spiders but not to fear opera glasses (Rachman 1990, 156-58). He is preprogrammed, if you will, to discriminate between things that are relevant to his life (because they can hurt) and things that are not,(5) much as he is preprogrammed to learn language at a certain time and in a certain way (Lenneberg 1972).

This natural sociability shapes the child's relationships with its parents, siblings, and peers. In all of these encounters, the child is not a passive organism repeating whatever acts are rewarded; nor is it a blank slate on which the world can write any message. The child is an active partner not only in shaping, but in judging, its experiences, as is evident when we consider the emergence of two of the moral sentiments, sympathy and fairness.


Children are by nature sociable; in the family they learn to extend sociability into generosity. This extension requires the instruction and example of parents, other kin, and older playmates; but the original impulse requires no instruction. The innate sociability of children makes them sensitive to the moods and actions of others. At first they try to control those moods and actions simply for their own pleasure; later they grasp that what pleases them may not please others, and so they act on the basis of some knowledge of the feelings of others. Children learn without much instruction that their own happiness is in some ways affected by the happiness of other; with some instruction, they learn that the happiness of others can be improved by modest sacrifices in their own well-being. Their own experiences and the teachings of others produce habits of action that routinely take into account the feelings of others. All this occurs early in life, before the children have understood sermons, mastered moral precepts, or read cautionary tales.

No infant needs to learn to assert its own needs; it cries when it is hungry or in distress. Until recently, however, many psychologists assumed that it had to learn everything else. Its capacity for sympathy was, in this view, an acquired characteristic. No doubt how a child is raised will affect the extent to which it is empathic or altruistic, but we now know that the infant brings to its own rearing a keen sensitivity to the distress of others. As early as 10 months of age, toddlers react visibly to signs of distress in others and often becoming agitated; when they are one-and-a-half years old, they seek to do something to alleviate the other's distress; and by the time they are two years old they verbally sympathize, offer toys, make suggestions, and look for help (Dunn 1987; Dunn and Munn 1986; Zahn-Waxler and Radke-Yarrow 1982). Though these youngsters are no doubt expressing some learned reactions to distress, they seem prepared to learn these things. It is obvious that infants are biologically inclined to seek help and attention; it may also be that they are biologically inclined to offer help and give attention (Rheingold and Hay 1978, 119).

This innate sensitivity to the feelings of others--a sensitivity that, to be sure, varies among individuals--is so powerful that it makes us grasp not only the feelings of friends and family members but also those of some strangers, many fictional characters, and even animals. We wince when the motion picture hero is threatened and exult when he is triumphant; we are disturbed by the sight of a wounded dog and pleased by the sight of someone else's baby.

It is sometimes argued that we display these feelings because it is expected of us or because we hope to curry favor with others or make ourselves seem worthy of reciprocal benevolence. That is certainly part of the story, but it is not the whole story. In a remarkable series of experiments, Bibb Latane and J. M. Darley (1970) showed that these explanations were probably wrong. They staged a number of "emergencies" in stores, offices, and laundromats, ranging from medical problems and fire alarms to thefts and disorderly conduct. In every case, a lone bystander was more likely to help the "victim" then was a group of bystanders.

This finding casts great doubt on the notion that altruism among strangers is merely a form of reciprocity by which helpers get credit for good deeds that can later be cashed in for other rewards, such as status. If altruism were really a self-interested investment in the future, then people should more frequently help victims when others can witness the good deed. But they do so less frequently. There is a social inhibition against helping that probably derives from the fact that in a group, the sense of personal responsibility is diffused. It is as if each person in a groups says to him- or herself, "Maybe somebody else will do it." When we are alone, we feel more keenly a sense of responsibility; we must answer not to the public but to the voice within.

Benevolence is often motivated by a desire for fame, status, or favors; but if that were all there was to it, our language would not be rich in words designed to distinguish selfish from selfless actions, kind from unkind persons, and heroism from bravado. If we really believed that altruism was merely reciprocity, we would purge our language of all such distinctions, and then the only difference between Tiny Tim and Scrooge would be age.


Perhaps the first moral judgment uttered by the child is, "That's not fair!" At first this claim may be largely self-interested, a way of making persuasive the real claim, "I want!" But at a very early age, the claim of fairness begins to take on the quality of a disinterested standard. It does so because fair play--taking turns, sharing toys, following rules--is a necessary condition for the child to satisfy its natural sociability. Judy Dunn (1988), who closely observed children between the ages of 18 and 36 months, found that about half spontaneously offered to share things and noted instances, familiar to every parent, of even younger babies offering a toy, pacifier, or piece of food to another person. These offerings reflect a desire on the part of the toddler to win approval, initiate play, or maintain contact. Around the world, children offer food as a way of establishing friendly relations even before they are able to talk (Eibl-Eibesfeldt 1989, 340-41). This tendency to share increases with age and is accompanied by a rapid growth in the sense of what rules ought to govern play and contact.

Jean Piaget formulated his theory of the development of moral judgment by watching children play marbles. What struck him most forcibly is that the complex and subtle rules of that game are not taught by adults to children, they are taught by children to each other (Piaget 1965, 14). Out of ordinary play and interaction there emerges a fairly clear sense of rules and justifications: "principles of possession, positive justice, !and^ excuses on grounds of incapacity or lack of intention" (Damon 1988, 172). These rules are not specific to particular situations but are understood generally, so that they can be applied differently in different contexts without sacrificing the principle underlying the rule.

Children, from infancy on, court other people. They differ in the skill and enthusiasm with which they do this and the clarity and consistency of the rules that they infer from this courtship, but the process is not driven by self-interest narrowly conceived. A child, especially a two-year-old, is learning that it has a self that is different from the self of others; but it is also beginning to learn that its self requires the presence of others to achieve happiness. In the language of economists, children learn that utilities are interdependent--that one's happiness depends to some degree on the happiness of others--long before they can say "interdependent." Children learn that they ought to obey certain rules because it pleases others at the same time that they learn that breaking rules can be fun (up to a point) because it teases others.

By the time that they are in elementary school, the idea of fairness has acquired a fairly definite meaning: people should have equal shares. But once the equality principle is grasped, exceptions to it become apparent. For example, it does not seem right that a lazy boy should be paid as much as an energetic one when working on the same task. By the time they have left elementary school, children will go to great lengths to discuss and weigh competing principles (merit, age, need, etc.) for allocating things in a fair way. Most striking about this process, notes Ann Cale Kruger, is that they discuss these matters almost entirely without reference to adult authority figures or adult rules, regardless whether they are being interviewed by adults or secretly overheard (cited in Konner 1991, 299-301). Far from expressing an internalized set of adult rules or looking furtively over their shoulders for any sign of adult power, older children discuss, in sophisticated detail, principles of justice that have evolved out of their own interactions. Their affiliation with others in natural social groupings is the continuing source of their moral judgments.

A vast body of research on adult behavior provides compelling evidence for the importance of fairness as a guide to how we behave. In these studies (and here as well), fairness is defined much as Aristotle defined distributive justice: "What is just ... is what is proportionate" (Nichomachean Ethics 1131b17)--that is, things should be divided among people in proportion to their worth or merit. In modern equity theory, a division of something between two people is fair if the ratio between the first person's worth (effort, skill, or deeds) and gains (earnings, benefits, or rewards) is the same as the ratio between the second person's worth and gains.

In a famous set of studies in the 1960s, various experimenters hired men to conduct interviews, paying them on a piece-rate basis. During the hiring, the "employer" (an experimenter) made clear to some men that he thought them unqualified (implying that they would be overpaid for the work to be done) while saying to others that they were fully qualified (implying that they would be equitably paid). The men were then sent out to work. Those who were made to feel unqualified (and hence overpaid) produced fewer but better interviews than did the men who were given to believe that they were being fairly paid. When some employees were made to feel that they were underpaid (i.e., that their skills were worth more than they would earn), they produced far more interviews (but of lower quality) than did employees who believed they were fairly paid (Adams 1963; Adams and Jacobson 1964; Adams and Rosenbaum 1962; Lawler and O'Gara 1967; Walster, Walster, and Berschied 1978).

This is not what one would expect if people were only interested in maximizing their income. Both the "overpaid" and the "equitably" paid workers earned the same amount per interview completed. If getting the most money was all that mattered, both groups would try to complete as many interviews as possible and the earnings of each group (the employees were randomly assigned) would be identical. What their employer thought of them would be irrelevant. The fact that the "overpaid" workers did less work (thereby sacrificing earnings) but did work of higher quality (thus sacrificing effort) can be explained in terms of their concern for equity.

There are many circumstances in which the self-interest of a person is not at all engaged, yet that person experiences--and often acts upon--strong feelings of fair play. Most of us are outraged at members of Congress who bounce checks even though the cost to us is close to zero. We are upset if, while waiting in a line to buy tickets to attend the theater, someone cuts in line ahead of us even though the addition of one person to the line almost certainly will not affect our chances of entering the theater and only trivially affect our choice of seats. But we believe that we have been treated unfairly because, by arriving earlier than the intruder, we have established a stronger claim (however tiny the difference) and thus are entitled to a greater reward (however small the increment). On the other hand, if the intruder can show that in fact he was there all night waiting for a seat, and only stepped away momentarily to get a cup of coffee, we will probably acknowledge his right to reclaim his place and will even (though less cheerfully) acknowledge the rights of his five companions, all being in the same position and all with the same prior claims.


The natural sociability of mankind gives rise to sentiments of sympathy, fairness, and reciprocity in every culture that we can imagine. Though custom will shape the reach of these sentiments by determining who is worthy of sympathy and what constitutes equality of worth or effort, the sentiments themselves emerge spontaneously. They do so because they are essential to human reproduction, family life, and small-group cohesion. These sentiments, in short, confer reproductive fitness. Ordinarily, parents who are innately disposed to care for their children produce more surviving children than do people lacking that disposition and so come to constitute an ever larger fraction of the gene pool.

Even John Stuart Mill readily admits this natural sociability but then, unaccountably (and quite unlike Hume and Smith), argues that "the moral feelings are not innate but acquired," implying that the "social feelings of mankind" and the "deeply rooted" sense that "there should be a natural harmony between his feelings and aims and those of his fellow creatures" have no moral content (1979, 30-31, 33). This was, as Charles Darwin was to remark eight years later, "extremely improbable": if the moral sentiments had no innate basis (i.e., had not been selected for by evolution), it is quite unlikely that so many of us would acquire them (1981; 71, n. 5; see also chap. 3).

The moral sentiments can lead to altruistic behavior, as when one individual risks his life that another might survive. Biologists have devised two theories to explain why altruism might spread in a population, rather than be extinguished by the greater rate of survival of wholly self-interested individuals. The first is the notion of inclusive fitness (Hamilton 1964), which holds that evolution will select for creatures that run risks for the benefit of others with whom they share genes and in proportion to the degree of that sharing. Flippantly but not inaccurately, J.B.S. Haldane (1955) put it this way: I will risk death to save my child from a raging river if the odds are at least two to one that I will succeed (because she shares roughly half my genes); but I will jump in the river to save my cousin only if the odds favoring success are seven in eight (because she has only one-eighth of my genes). Trying to save my grandmother makes no sense at all because, being past child-bearing age, she can pass on none of my genes to the next generation.

This calculus may explain why, on the average, we strive harder to save our own children than somebody else's, and it is certainly consistent with the fact that children are more likely to be abused by their step-parents than by their natural parents (Wilson and Daly 1987); but it cannot explain why we should ever run any risk at all of saving our grandmother, our adopted child, or our dog. Yet many people will jump into the river for grandmothers, adopted children, and even dogs.

Studies of adoptive families provide no evidence that parents of adopted children are any less loving, solicitous, or protective than are the parents of biological children. Mothers who have both an adopted and a biological child report no difference in their feelings toward them (Smith and Sherwen 1983, 95). If anything, adoptive parents are more protective and less controlling than biological ones (Hoopes 1982, 97-98), a puzzling finding if one believes that investment in child care is driven by a desire to reproduce one's genes. Adopted children report that they were loved as if they were natural children (Triseliotis and Hill 1990).(7)

People in primitive, as well as advanced, societies form strong attachments to animals (Serpell 1986, chap. 4). There are, of course, great variations in which animals are cherished; but beneath these variations there is a deeper constancy: in virtually every society and in virtually every historical period, people have been attracted to certain kinds of animals in ways that are hard to distinguish from the way in which we treat infants--difficult to explain in terms of economic necessity (the desire for food or help) and impossible to explain in terms of reproductive fitness.(8)

One can attempt to solve these puzzles of affectional behavior directed toward non-kin and nonhumans while remaining within a narrow interpretation of the evolutionary perspective by advancing the notion of reciprocal altruism: we engage in altruistic acts, such as helping nonrelatives, caring for adopted children, or being affectionate toward pets, in order to impress others with our dependability and hence to increase our opportunities to have profitable exchanges with these others (Alexander 1987; Trivers 1971). There is a great deal of truth in this; having a reputation for doing one's duty, living up to promises, and helping others will enhance one's own opportunities. Moral behavior is far more likely when utility conspires with duty, and the strongest moral codes are invariably those that are supported by considerations of both advantage and obligation (see Sidgwick 1981, 386-87).

But sometimes, sentiment alone, unsupported by utility, motivates our actions, as when someone makes an anonymous benefaction or a lone bystander helps an endangered person. While anonymous giving may be relatively rare, it is generally the case that a lone bystander is more likely to go to the aid of a threatened person than a bystander who is part of a group--the opposite of what one would predict if reputation enhancement were the motive for altruistic actions (Latane and Darley 1970).

Evolutionary biology provides a powerful insight into human behavior at the level of the species, but it fares less well at the level of daily conduct. This deficiency arises in part because evolutionary biologists ordinarily do not specify the psychological mechanism by which a trait that has been selected for governs behavior in particular cases (Cosmides and Tooby 1987). The strict and exclusive altruism of social insects is different from the more inclusive altruism of humans.

Sympathy for persons who are not offspring and creatures that are not human is a characteristic of almost all humans. Indeed, we regard as inhuman anyone who acts as if they had no feeling for others and we criticize as insincere people who merely feign such fellow feeling. If sympathy is widespread, it must have been adaptive; but what was selected for was not a simple desire for reproductive success: it was a generalized trait that both encouraged reproductive fitness and stimulated sympathetic behavior.

That trait, I suggest, is affiliative behavior. Evolution has selected for the attachment response: if infants and parents were not predisposed to develop strong attachments for one another, it would be impossible to provide for the postpartum development of the human central nervous system. A predisposition to attachment is necessary if a child is to be sustained for that long period of dependency and nurturance during which the human brain becomes fully developed.

But it is a mistake to suppose that the psychological predisposition for which evolution has selected will be as precise in its effect on human behavior as the social instinct is among ants. Our large brain almost guarantees that the effect of attachment will be complex and diffuse if for no other reason than that the human brain not only makes possible complex actions but also makes possible our imagining such actions. The predisposition to attachment is a pervasive but somewhat general drive that imperfectly discriminates between parents and parent substitutes; is evoked by adoptive, as well as natural, infants; extends to creatures that have just a few of the characteristics of the human infant; and embraces not only family but also kin and many non-kin.

One of the cues (in evolutionary jargon, "releasers") that stimulates this affectional response in adults is "cuteness"--by which I mean that set of traits by which we judge an organism to be delightfully attractive. We respond to certain features of people and animals in ways that suggest that we share a roughly common definition of cuteness: eyes large relative to the skull, chubby cheeks and a rounded chin, awkward movements, a cuddly epidermis, small size, and a distinctive smell (Eibl-Eibesfeldt 1989; Lorenz 1943; McKenna 1987, 161; Sternglanz, Gray, and Murakami 1977; Super 1981). Nonparents, as well as parents, respond to these cues; and the response extends beyond the human infant to other creatures with these infantile traits. I suggest that social scientists and moral philosophers have paid too little attention to the concept of "cute." People use the word all the time; philosophers and scientists almost never use it. Its frequent use suggests that it may refer to an important mechanism by which our moral sentiments are extended beyond ourselves and our immediate families.

There are other concepts, just as important and just as poorly understood, such as being a "fan," a "loyalist," or a "sentimentalist." Each suggests that our affiliative impulse is so strong that it can be evoked almost by remote control. We identify with people whom we do not know and who do not know us and with people who are entirely fictional. When we are in the audience of a play or motion picture, we are moved by the plight of imaginary people; when we watch an athletic spectacle or a military parade, we are moved by the exploits--and sometimes the mere sight--of people who are unaware of our existence. There is little in behavioral psychology or evolutionary biology that explains these emotions and their tendency to evoke in us moral sentiments.


Our moral sense, however weak or imperfect, helps explain social order because that sense grows out of, and reflects, the fact that we are social beings, dependent upon one another and because we are able to avail ourselves of the essential help of others, at least in the intimate precincts of life, only on the basis of understandings that arise spontaneously out of, and necessarily govern, human relationships: the need to show some concern for the well-being of others, treat others with minimal fairness, and honor obligations. This natural sociability and the patterns of sympathy and reciprocity on which it rests are the basis, in my view, of Aristotle's argument for natural law: man is by nature social, and social groupings aim at some good.

It can be put in the form of a thought experiment. Imagine people stripped of every shred of their social experiences and set loose in some Arcadian paradise, free to invent "culture." What would emerge? If they are young boys, the answer may be something akin to William Golding's Lord of the Flies (1962), though my guess is that a close study of abandoned children in war-torn nations would disprove even this hypothesis. But if they are men and women, what emerges would almost surely be something with strange customs, odd dress, and unfamiliar gods but invariably with familiar systems of infant care, familial obligation, kinship distinctions, and tribal loyalties.

The results of such a thought experiment taken together with the findings of modern science cast doubt, in my mind, on the philosophical value of imagining a man who is presocial, driven by a single motive, or unware of the main and necessary features of social life. John Rawls (1971) may ask us to imagine ourselves in an "original position" behind a "veil of ignorance"; but no human being is ever in such a position and, to the extent he is human, cannot possibly be ignorant. Locke (1690/1979) may ask us to believe that experience is the sole source of ideas; but if we accept that, we will have difficulty explaining why all children learn a language at roughly the same time and without having the rules of that language explained to them. Hobbes (1957) may ask us to believe that man is driven by the fear of violent death; but were that our overriding concern, we would not give birth to children or lavish so much care on them. Why expend so much effort on something so perishable, whose birth threatens the mother's life, and whose protection increases our vulnerability to the predation of others? Rousseau may imagine an equally implausible alternative, man born with no inclination to civil society; but no such man can exist and, if he did exist, could not learn goodness by reading Robinson Crusoe (1979, 184; see also Bloom 1978).

A proper understanding of our natural disposition to sociability not only helps explain social order, it provides the grounding for our judgments about that order. We cannot imagine praising a man who laughs while torturing an innocent baby (Thomson 1989); we cannot defend a principle that says that every man is entitled to be the judge in his own case. We are not limited to condemning Auschwitz contingently and ironically; we can condemn it absolutely and confidently (see Rorty 1989, 189).

Moral philosophy, like social science, must begin with a statement about human nature. We may disagree about what is natural; but we cannot escape the fact that we have a nature, that is, a set of traits and predispositions that limits what we may do and suggests guides to what we must do. That nature is mixed: we fear violent death but sometimes deliberately risk it; we want to improve our own happiness but sometimes work for the happiness of others; we value our individuality but are tormented by the prospect of being alone. It is a nature that cannot be described by any single disposition, be it maximizing our utility or enhancing our reproductive fitness. Efforts to found a moral philosophy on some single trait (the desire for happiness or the fear of punishment) or political philosophy on some single good (avoiding death or securing property) will inevitably produce judgments about what is right that at some critical juncture are at odds with the sober second thoughts of people who deliberate about what constitutes praiseworthy conduct and who decide, out of that deliberation, to honor the hero who risked violent death or to sympathize with the mother who sacrificed one child to save another.

As Ian Shapiro (1990, 296) has noted, much contemporary political theory is "locked into a series of antinaturalist assumptions about human nature"; for example, original positions, veils of ignorance, the priority of rights. Any reasonable theory must have a "view of human nature and human interests and an argument about the injunction for action this entails given a plausibly defended account of the pertinent causal structure of the social world" (ibid.).

Aristotle gave such an account, but his views became unfashionable among those who sought to base moral or political philosophy on a single principle (e.g., utility, liberty, or self-preservation), who worried about Aristotle's teleology, or who believed in the priority of the right over the good. But if one acknowledges that there is no single moral principle but several partially consistent ones and that neither happiness nor virtue can be prescribed by rule, one is better prepared for a more complete understanding of man's moral capacities, an understanding stated by Aristotle in phrases that in most respects precisely anticipate the findings of modern science. Though Aristotle's account is often dismissed as teleological (much as those of later scientists were dismissed as functionalist), his view does not involve any "mysterious non-empirical entities" (Nussbaum 1978, 60) or any suspiciously conservative functionalism.

There is certainly nothing mysterious or nonempirical about Aristotle's assertion that men and women unite out of a "natural striving to leave behind another that is like oneself" (Politics 1252a30) because a "parent would seem to have a natural friendship for a child, and a child for a parent" (Nichomachean Ethics 1155a17) or that "the household is the partnership constituted by nature for !the needs of^ daily life" (Politics 1252b11). These are as close to self-evident propositions as one could utter. Only slightly less obvious, but still scarcely mysterious, are the arguments that "in the household first we have the sources and springs of friendship, of political organization, and of justice" (Eudemian Ethics 1242b1) and that "there is in everyone by nature an impulse toward this sort of partnership !that is, toward the city^" (Politics 1253a29).

These natural moral sentiments are an incomplete and partial guide to action. They are incomplete in that they cannot resolve a choice we must make between two loved persons or between the desire to favor a loved one and the obligation to honor a commitment. They are partial in that these sentiments extend chiefly to family and kin, leaving non-kin at risk for being thought nonhuman. Resolving conflicts and extending our sentiments across the high but necessary walls of tribe, village, and racial grouping--an extension made more desirable by the interdependence of cosmopolitan living--requires moral reasoning to take up the incomplete task of moral development.

These deficiencies can lead the unwary philosopher to suppose that if a sentiment does not settle everything, it cannot settle anything or to infer that if people make different choices, they must do so on the basis of different sentiments. The first error leads to logical positivism; the second, to cultural relativism; and the two together, to modern nihilism or, at best, to "liberal irony." A proper understanding of human nature can rarely provide us with rules for action, but it can supply what Aristotle intended: a grasp of what is good in human life and a rough ranking of those goods (Arnhart 1990; Salkever 1990).

Antinaturalist assumptions have impeded the search for explanations for social order, as well as efforts to justify different systems of order. Normative theories have stressed that order is the product of cultural learning without pausing to ask what it is we are naturally disposed to learn. Utilitarian theories have confidently responded by saying that we are disposed to learn whatever advances our interests without pausing to ask what constitutes our interests. And despite their differences in approach, they have both supported an environmental determinism and cultural relativism that has certain dangers. If man is infinitely malleable, he is as much at risk from the various despotisms of this world as he would be if man were entirely shaped by some biochemical process. Anthropologist Robin Fox has put the matter well: "If, indeed, everything is learned, then surely men can be taught to live in any kind of society. Man is at the mercy of all the tyrants who think they know what is best for him. And how can he plead that they are being inhuman if he doesn't know what being human is in the first place?" (1973, 13). Despots are quite prepared to use whatever technology will enable them to dominate mankind; if science tells them that biology is nothing and environment everything, then they will put aside their eugenic surgery and selective breeding programs and take up instead the weapons of propaganda, mass advertising, and educational indoctrination. The Nazis left nothing to chance; they used all methods.

Recent Russian history should have put to rest the view that everything is learned and man is infinitely malleable. After 75 years of cruel tyranny during which every effort was made to destroy civil society to create the New Soviet Man, we learn that people kept civil society alive, if not well. The elemental building blocks of that society were not isolated individuals easily trained to embrace any doctrine or adopt any habits; they were families, friends, and intimate groupings in which sentiments of sympathy, reciprocity, and fairness survived and struggled to shape behavior.

Mankind's moral sense is not a strong beacon light, radiating outward to illuminate in sharp outline all that it touches. It is, rather, a small candle flame, casting vague and multiple shadows, flickering and sputtering in the strong winds of power and passion, greed and ideology. But brought close to the heart and cupped in one's hands, it dispels the darkness and warms the soul.


An expanded version of this essay will appear in 1993 under the title The Moral Sense from the Free Press of New York City.

Emile Durkheim (1951, 1960, 1961) is an important exponent of the first view, Thomas Hobbes (1957) of the latter. For an analysis, see Elster (1989).
In 1986, three in one hundred thefts reported to the police resulted in prison sentences. Since at the most only one-fourth of all thefts are reported to the police, the true probability of imprisonment, given a theft is much less than one chance in a hundred (Langan 1991).
Pollock's conclusion is worth quoting, resting, as it does, on the most systematic review of the evidence that we have: "Despite the individual differences in child-rearing techniques, there are limits of variation. These limits are the dependency of the child and the acceptance of responsibility for the protection and socialisation of that child by the parents. From the material gathered here, it is clear that the vast majority of parents from earlier centuries were operating within these constraints" (1983, 271). A similar judgment about child care was reached by David Herlihy (1985) for medieval Europe, Steven Ozment (1983) for sixteenth-century Europe, and Keith Wrightson (1982) and Michael MacDonald (1982) for seventeenth-century England.
Mothers also behave in distinctive ways toward children whatever their cultural background. For example, whether they are right-handed or left-handed, the vast majority of mothers carry their infants on the left side; it is almost inconceivable that this is a learned behavior (Trevathan 1987).
There is a good deal of research on the limits of conditioning even among animals. Rats can be trained to avoid a taste that is followed by induced nausea or to avoid a light that is followed by an electrical shock; but they cannot be conditioned to avoid a certain taste that is followed by a shock or avoid a light that is followed by nausea. They are preprogrammed to associate nausea with eating and light with shocks but not the reverse (Garcia and Koelling 1966; Seligman and Hager 1972).
Moreover, the much greater abusiveness of step-parents cannot be explained by economic differences, since natural parent and step-parent families do not, on average, differ in income (Wilson and Daly 1987, 225).
The fact that adopted children are at greater risk than natural children for psychological problems and conduct disorders does not invalidate the argument that the former are cherished equally with the latter. Adopted children have more personality problems because their biological parents had these problems, which have a large genetic component (Bohmann and Sigvardsson 1990; Cadoret 1990).
The animals that are most likely to become pets are those that have some of the characteristics of the human infant, such as eyes that are large relative to the face, a soft epidermis, and a prosocial disposition. The most common pets supply an object of affection that returns the affection through obedience, loyalty, purring, or posing (Serpell 1986, 66-68, 114-15).

Adams, J. Stacy. 1963. "Toward an Understanding of Inequity." Journal of Abnormal and Social Psychology 67:422-36.

Adams, J. Stacy, and P. R. Jacobsen. 1964. "Effects of Wage Inequities on Work Quality." Journal of Abnormal and Social Psychology 69:19-25.

Adams, J. Stacy, and W. B. Rosenbaum. 1962. "The Relationship of Worker Productivity to Cognitive Dissonance about Wage Inequities." Journal of Applied Psychology 46:161-64.

Alexander, Richard D. 1987. The Biology of Moral System. New York: De Gruyter.

Anderson, Michael. 1980. Approaches to the History of the Western Family, 1500-1914. London: Macmillan.

Aries, Phillipe. 1962. Centuries of Childhood. New York: Vintage.

Arnhart, Larry. 1990. "A Sociobiological Defense of Aristotle's Sexual Politics." Presented to the annual meeting of the American Political Science Association, San Francisco.

Bloom, Allan. 1978. "The Education of Democratic Man: Emile." Daedalus 107:135-53.

Bohmann, Michael, and Soren Sigvardsson. 1990. "Outcome in Adoption: Lessons from Longitudinal Studies." In The Psychology of Adoption, ed. David M. Brodzinsky and Marshall D. Schecter. New York: Oxford University Press.

Cadoret, Remi. 1990. "Biologic Perspectives of Adoptee Adjustment." In The Psychology of Adoption, ed. David M. Brodzinsky and Marshall D. Schecter. New York: Oxford University Press.

Cosmides, Leda, and John Tooby. 1987. "From Evolution to Behavior: Evolutionary Psychology as the Missing Link." In The Latest on the Best, ed. John Dupre. Cambridge: Massachusetts Institute of Technology Press.

Daly, Martin, and Margo Wilson. 1984. "A Sociobiological Analysis of Human Infanticide." In Infanticide, ed. Glenn Hausfater and Sarah Blaffer Hrdy. New York: De Gruyter.

Daly, Martin, and Margo Wilson. 1987. "Children as Homicide Victims." In Child Abuse and Neglect, ed. Richard J. Gelles and Jane B. Lancaster. New York: De Gruyter.

Daly, Martin, and Margo Wilson. 1988. Homicide. New York: De Gruyter.

Damon, William. 1988. The Moral Child. New York: Free Press.

Darwin, Charles. 1981. The Descent of Man, and Selection in Relation to Sex. Princeton: Princeton University Press.

de Mause, Lloyd, ed. 1974. The History of Childhood. New York: Psychohistory.

Dickemann, Mildred. 1984. "Concepts and Classification in the Study of Human Infanticide." In Infanticide, ed. Glenn Hausfater and Sarah Blaffer Hrdy. New York: De Gruyter.

Divale, W. T., and M. Harris. 1976. "Population, Warfare, and the Male Supremacist Complex." American Anthropologist 78:521-38.

Dunn, Judy. 1988. The Beginnings of Social Understanding. Oxford: Basil Blackwell.

Dunn, Judy, and Penny Munn. 1986. "Siblings and the Development of Prosocial Behavior." International Journal of Behavioral Development 9:265-84.

Durkheim, Emile. 1951. Suicide. Trans. J. A. Spaulding and George Simpson. Glencoe, IL: Free Press.

Durkheim, Emile. 1960. The Division of Labor. Trans. George Simpson. Glencoe, IL: Free Press.

Durkheim, Emile. 1961. The Elementary Forms of the Religious Life. Trans. J. W. Swain. New York: Collier.

Eibl-Eibesfeldt, Irenaus. 1989. Human Ethology. New York: De Gruyter.

Elster, Jon. 1989. The Cement of Society. Cambridge: Cambridge University Press.

Field, Tiffany. 1990. Infancy. Cambridge: Harvard University Press.

Fox, Robin. 1973. Encounter with Anthropology. New York: Harcourt Brace Jovanovich.

Garcia, John, and Robert A. Koelling. 1966. "Relation of Cue to Consequence in Avoidance Learning." Psychonomic Science 4:123-24.

Golding, William. 1962. Lord of the Flies. New York: Coward-McCann.

Haldane, J. B. S. 1955. "Population Genetics." New Biology 18:34-51.

Hamilton, William D. 1964. "The Evolution of Social Behavior." Journal of Theoretical Biology 7:1-52.

Hay, Dale F., and Harriet L. Rheingold. 1983. "The Early Appearance of Some Valued Social Behaviors." In The Nature of Prosocial Development, ed. Diane L. Bridgeman. New York: Academic.

Herlihy, David. 1985. Medieval Households. Cambridge: Harvard University Press.

Hobbes, Thomas. 1957. Leviathan. Ed. Michael Oakeshott. Oxford: Basil Blackwell.

Hoopes, Janet L. 1982. Prediction in Child Development: A Longitudinal Study of Adoptive and Nonadoptive Families. New York: Child Welfare League of America.

Hume, David. 1978. A Treatise of Human Nature. 2d ed. Ed. L. A. Selby-Bigge and P. H. Nidditch. Oxford. Oxford University Press.

Hutcheson, Francis. 1742 !1969^. An Essay on the Nature and Conduct of the Passions and Affections with Illustrations on the Moral Sense. 3d ed. Reprint. Gainesville, FL: Scholars' Facsimilies and Reprints.

Konner, Melvin. 1982. The Tangled Wing: Biological Constraints on the Human Spirit. New York: Holt, Rinehart, and Winston.

Konner, Melvin. 1990. Why the Reckless Survive. New York: Viking.

Konner, Melvin. 1991. Childhood. Boston: Little, Brown.

Langan, Patrick. 1991. "America's Soaring Prison Population." Science, p. 1572.

Langer, William L. 1974. "Infanticide: A Historical Survey." History of Childhood Quarterly 1:353-66.

Latane, Bibb, and J. M. Darley. 1970. The Unresponsive Bystander: Why Doesn't He Help? New York: Appleton-Crofts.

Lawler, E. E., and P. W. O'Gara. 1967. "The Effects of Inequity Produced by Underpayment on Work Output, Work Quality, and Attitudes toward Work." Journal of Applied Psychology 351:403-10.

Lenneberg, Eric H. 1972. "On Explaining Language." In Biological Boundaries of Learning, ed. Martin E. P. Seligman and Joanne L. Hager. New York: Appleton-Century-Crofts.