In the village McCabe studied, “first cousins grew up in an association as close as that of siblings” (1983:58). She found that the relationship between a boy and his father’s brother’s daughter was essentially the same as between a boy and his sister: it rested on a constant and intimate interaction from birth (including sexual exploration when very young), and was characterized by “informality, candor, teasing, tattling, quarreling, laughing, joking” and the exchange of confidences (1983:59).
But marriages between patrilateral parallel cousins produced 23 percent fewer children during the first 25 years of marriage and were four times more likely to end in divorce than all other marriages. McCabe (1983:61) cites other scholars who, from early in this century, had noted signs of greater “sexual apathy” or “coolness” in patrilateral cousin marriages. As in the Chinese case, McCabe argues, it is parents or others, not the ones who actually marry, who prefer patrilateral parallel cousin marriages.
If the Westermarck effect is real, an important issue is the age limits within which it is created. Wolf and Huang (1980:185) offered some insight into the matter by noting that minor marriages in which the children were brought together before age 4 were two times more likely to end in divorce than minor marriages in which the children became acquainted at age 8 or later. Joseph Shepher (1983) has looked at the matter more closely. Born and raised in an Israeli commune himself, Shepher conducted the most thorough study of marriage in Israeli communes, getting data on 2769 married couples in 211 kibbutzim. Among them he found only 20 marriages between members of the same commune and only 14 that allegedly took place between persons who had been in the same peer group. But on contacting these 14 couples he found that all cases dissolved: there was not a single case of marriage between a boy and girl who had spent the first 6 years of their lives in the same peer group. In the one commune (his own) in which he could get reliable data on premarital sex he also found that none had occurred between persons raised from infancy in the same peer group. Boys and girls brought into the group at later ages sometimes did have an intense attraction to one of their group mates.
There was no attempt in the communes to stop the sexual experimentation of young children. There was no attempt to keep adolescents and young adults from dating or marrying their commune mates, though they were supposed to refrain from sex in general during high school. There was in fact some encouragement of intracommunal marriage.
Examining the pattern of entry and exit from peer groups, and the resulting pattern of attraction or sexual interest or uninterest among the relevant parties, Shepher concludes that a form of imprinting (or negative imprinting) occurred, that it was complete by the age of 6, and that it took about 4 years. He argues that this imprinting is a phylogenetic adaptation to reduce the harmful effects of inbreeding.
Certain lines of research conducted largely outside of anthropology also have a close bearing on the Westermarck hypothesis. They include studies of the physical or medical consequences of inbreeding among humans (as well as other animals), studies of evolved inbreeding avoidance mechanisms in nonhuman species, and studies of the social consequences of human incest.
Reviewing the scanty literature on the empirical consequences of inbreeding among humans, Shepher (1983) finds that full-sibling or parent-child incest results in about 17 percent child mortality and 25 percent child disability, for a combined result of about 42 percent nonviable offspring. The negative consequences decline rapidly for more distant inbreeding. If the figures Shepher cites are even approximately correct, mechanisms to avoid the costs of incest between close kin are quite expectable.3
A study of 38 captive mammalian species found a cross-species average of around 33 percent offspring mortality resulting from closely incestuous matings (the range of nonviability—measured rather conservatively in terms of “juvenile survival”—was all the way from 0 to nearly 100 percent) (Ralls, Ballou, and Templeton 1988). As the apparent consequence of this widespread phenomenon, equally widespread mechanisms have evolved that enable animals to avoid incest. These mechanisms operate in three distinguishable ways: by prohibition (only among humans, of course), by prevention, and by inhibition (Shepher 1983). In the case of prevention, incest simply cannot occur (because, for example, parents die before their offspring become fertile, or siblings are so widely dispersed that there is minuscule likelihood of their mating).
Inhibition, apparently brought about by imprinting or related processes, occurs when closely related and fertile individuals are in proximity but avoid mating with each other. Wolf, McCabe, and Shepher all provide evidence for some kind of negative imprinting that would, in the normal course of events, inhibit brother-sister incest among humans. Although the evidence he presents is minimal, Shepher (1983:108–110) argues that mother-son incest is also inhibited by imprinting (see also Fox 1980, Arens 1986). The opportunity for their imprinting is of course excellent, due to the prolonged and intimate contact of mother and child. In a great many societies the opportunities for developing aversion between father and child are the least.
Incest avoidance, via mechanisms of prevention or inhibition, is widely reported among many animal species (Bischof 1972), so that parent-offspring or sibling incest among animals in the wild is “apparently rather rare” (Lewin 1989:482). Consequently, the assumption that human incest avoidance is fundamentally a cultural phenomenon now rests on the inelegant assumption of a double discontinuity with the animal kingdom: unlike other species we lack innate avoidance mechanisms; unlike other species we therefore avoid incest via cultural prohibition (Arens 1986:94).
A third line of research, conducted mostly by psychologists and sociologists, and mostly in recent decades, concerns actual cases of human incest—a topic curiously neglected during most of the period in which the incest taboo has exercised the anthropological imagination. One of the most important consequences of these studies is their dismissal of the sociological or functionalist explanation of the incest taboo. In a line of thought that Arens (1986:29) traces back as far as Jeremy Bentham—but in more recent times through many distinguished anthropologists—it has often been argued that incestuous relations would confound the organization of the family, rendering it inefficient and thereby rendering society inefficient. As persuasive as this line of reasoning has been—in the absence of empirical tests—it now appears to be incorrect.
Bagley (1969; summarized in Arens 1986) analyzed 425 published cases of incest, finding 93 instances in which incest was the means that allowed the family to maintain its functional integrity. Typically, a father-daughter relationship replaced the father-mother relationship when the mother was either unable or unwilling to fulfill her role. Bagley (1969) describes this as “functional incest.” Whatever the psychological costs may be to individuals, the study of actual cases of incest gives no obvious support to the assumption that society, or even the family, is necessarily threatened by incest (Arens 1986; see also Willner 1983 and La Fontaine 1988).
A recent study (Parker and Parker 1986) of incestuous relationships has a more direct bearing on the Westermarck hypothesis. Although the actual frequencies of the various forms of nuclear family incest—brother-sister, mother-son, and father-daughter—is a matter of uncertainty, there is substantial agreement that father-daughter incest is much commoner than mother-son incest. Furthermore, the variant of stepfather-stepdaughter incest seems to be disproportionately common. There are a number of explanations for this, not all of them mutually exclusive. One of them has to do with imprinting: if some form of imprinting results in the inhibition of incestuous desires, on the average it would, as noted earlier, probably work best between mother and son, not so well between father and daughter, and even less well between stepfather and stepdaughter.
Parker and Parker (1986) tested this line of thought by comparing sexually abusive and nonabusive fathers with comparable backgrounds. Comparing fathers who had been present in the household during the first three years of their daughters’ lives, the P
arkers found that abusers had been “much less frequently involved in caring and nurturing activities” (1986:540). They also found that in general stepfathers or adoptive fathers were more likely to be abusive, apparently because such fathers were less likely to have an effective bonding (imprinting) experience. When biological fathers were compared to step- or adoptive fathers with similar degrees of early childhood contacts with their daughters, no significant differences in abuse were found (1986:541). These findings support the Westermarck hypothesis and extend it beyond the brother-sister relationship that has been the principal focus of recent anthropological studies.
But in spite of the mounting evidence that supports the Westermarck hypothesis, and fails to support its rivals, such as the functionalist hypothesis, the dust has not settled on all the issues involved. Ancient Egyptian materials, for example, pose a problem precisely where the evidence for the Westermarck effect seems strongest: inhibition of brother-sister incest. Keith Hopkins (1980) provides evidence that brother-sister marriages were actually common for a period in Egypt and, hence, that incest avoidance in general, not merely the taboo, may not in fact be universal.
About 44 years after Alexander the Great conquered Egypt in 332 B.C., a Greek king of Egypt divorced his wife and married his full sister (who was about 10 years older than he). While there may have been some Greek precedent for his action—half-sibling marriages were alleged to be possible in certain ancient Greek communities—he was also following an ancient Egyptian custom. Whatever the case, 7 of the next 11 Greek kings in Egypt married their sisters. There is some vague evidence that the custom was penetrating other parts of the populace. Egypt subsequently passed to Roman rule.
Beginning in A.D. 19–20 and lasting until 257–258, the Roman administrators of Egypt conducted periodic censuses of the Egyptian population. Some 270 actual household returns survive; 172 returns, listing 880 persons, are in good enough condition to be used. While not in any sense a random sample, they report households widely spread in time, space, and social class. Seventeen of the 113 marriages ongoing at the time of the censuses were definitely between brother and sister, another 6 may have been. Thus some 15 to 21 percent of the ongoing marriages reported in these returns were brother-sister marriages. Eleven or 12 marriages were between full siblings, 8 between half siblings; in 3 the kind of sibling relationship is uncertain. Given the probable demographic structure of the family under the conditions of the time, there was only about a 40 percent likelihood of any family having a brother and sister of marriageable age. Thus a third or more of those who could marry their sisters did so. This is a very high proportion and, if correct, it provides the only known case in which brother-sister marriages were common throughout a populace.
Other forms of documentation—such as wedding invitations, letters, and marriage contracts—routinely mention brother-sister marriage, which indicates not only that it occurred but that it was considered normal. Some letters indicate real affection between the sibling couples, although this line of evidence is weakened by the Egyptian use of the term “sister” as a euphemism or term of endearment for women who were not actually one’s sister (Arens 1986:111–112).
The marriages were fertile, and no source indicates an awareness of harmful genetic consequences. But Hopkins does not indicate how fertile they were, and perhaps it should be asked whether the high rates of infant mortality in preindustrial societies might not tend to mask any mortality brought about by inbreeding (recall also that the Chinese seemed unaware of the lesser fertility of their minor marriages).
Hopkins is unable to find any reason peculiar to the Egyptian condition that may have induced parents to foist this kind of marriage on their children (though the late average age of first marriages—in their mid-twenties—does suggest parental involvement). Hopkins cites marriage contracts between brothers and sisters that specify dowry and/or separate property and hence suggest that sibling marriage was not a device to avoid marriage expenses or the division of family property.
Addressing the problem of how else to explain brother-sister marriage, Hopkins presents what can only be called a classically cultural explanation. He draws attention to the importance in Egyptian religion of Isis and Osiris, who were brother and sister, husband and wife4; a romantic tradition of idealizing brother-sister love in story and poetry; and the evidence that the status of women was high and that they therefore exercised some autonomy in marriage and divorce. That love was a basis for marriage, and its cessation a basis for divorce, is well attested. Hence, Hopkins is left with the possibility that brothers and sisters married because they wanted to.
In A.D. 212–213 the Egyptians were made Roman citizens, for whom marriage with near-kin was prohibited. Sibling marriage disappeared.
Given the spottiness of the Egyptian data it is difficult to decide how much credence to give them. But a few points should be noted. Hopkins gives the ages of five sibling couples; they were separated in age by 7, 8, 4, 8, and 20 years. With one exception, then, these are not necessarily couples who were raised together as children or, at any rate, who were raised together in the manner that produces the Westermarck effect. It would be of interest to know more about child-rearing practices among Roman Egyptians.
Shepher’s (1983) response to the Egyptian case was to dismiss it on the grounds that the data were few and that a single exception can carry little weight (he thereby reversed, by the way, the de facto opinion of many anthropologists that a single exception is all it takes to dismiss claims of universality). In this context, Shepher argued that unrestricted universals were not very likely to occur anyway—since nature operates by probability—so that a near-universal was the most to be expected.
Spiro (1982) summarizes other criticisms of the Westermarck hypothesis and adds his own. He notes, for example, an alternative interpretation of the kibbutz case. Spiro says it is not the child rearing but rather the adolescent repression of sexuality that produces the strong tendency for boys and girls to go outside their peer group and kibbutz to find mates. In adolescence, children were still living together, but their childhood exploration of sexuality was to stop. They were strongly urged to forgo sex until education was complete. In Spiro’s view, this adolescent frustration resulted in peer group members’ lack of interest in one another—they responded, in effect, to a consciously stated taboo.
To support his argument, Spiro cites Kaffman (1977), a psychiatrist employed by the kibbutz movement, who says that liaisons between children raised in the same peer group do in fact now occur. Since infant socialization has not changed, but adolescent controls have been relaxed, it is adolescent conditions that are critical. Unfortunately, Kaffman gives no data. Shepher (1983) dismisses Kaffman’s argument and notes that marriages between those who had been adolescent (but not childhood) peers did occur before; hence, adolescent repression of sexuality could not have been the crucial factor. (But note that such marriages weren’t at all common. A defect in Shepher’s contribution is that by narrowing imprinting to a 4-year period that must occur in the first 6 years of life he has made this a small part of what must be various controls on incest, since even individuals who were not reared in the same peer group but who were resident in the same commune seem to marry rather infrequently. The low rate of intra-kibbutz marriage in general must find some of its explanation in some other factors.)
Spiro also draws attention to two further considerations. One is the smallness of the peer groups, which makes finding a mate outside them statistically expectable. The other is that the boys and girls in the peer groups were the same age; since young girls tend to be interested in older boys, and older boys in younger girls, they therefore tend to seek mates outside the peer group.
What lessons, in conclusion, may be derived from the recent efforts to understand the incest taboo/avoidance? One is the sobering reflection that an alleged universal that has exercised the anthropological imagination for over 100 years is still not explained to everyone’s satisfaction. It i
s not even certain that the phenomenon is a universal. The incest taboo clearly is not universal, though it surely is a statistical universal and might be a near-universal. On the other hand, incest avoidance may be universal.
Even more sobering has been the impact of biological considerations that for decades were all but banned from mainstream anthropological thought. The ethological discovery that humans are far from unique in avoiding incest has entirely reoriented the problem. The resuscitation of the Westermarck hypothesis has provided a successfully tested explanation for part of the phenomenon. In eliminating possible hypotheses, and in accumulating relevant data, then, there has been progress. This experience suggests that anthropologists might do well to look into other lines of thought that may have been neglected for no good reason (a lesson no less applicable to sociology; see Scheff 1985).
Also important to notice in the incest-avoidance example is the clear attempt to explain the phenomenon by clarifying the ultimate (evolutionary) conditions that generate the mechanisms and by specifying the proximate mechanisms that generate the universal—infant (negative) imprinting, resulting in specified psychological states in the individual. Equally important has been the exploitation of natural experiments and the role that quantitative testing or analysis has played.
Human Universals Page 19
