Скачать 100.05 Kb.
Fisher long ago explained why we should expect to see 50/50 sex ratios (Fisher, 1930). The failure of numerous insect lineages to conform to these expectations was addressed by Hamilton and others in the sixties. Trivers and Willard (1973) argued that even in non-insects there are certain circumstances in which organisms that are in good physical condition or are well provided for in other ways should invest heavily in male offspring, since these will inherit their parents’ advantages, making them more likely to be chosen as mates by female conspecifics. This should provide the original parents with more grandchildren. On the other hand, females can generally mate much more easily, so less robust parents, or parents endowed with fewer resources, will get more grandchildren if they favor daughters over sons. We should expect to see the Trivers-Willard effect, then, when it pays parents to invest differently in male or female offspring depending on parental condition.
It turns out that there is good evidence for the effect in several species, the most famous example being that of red deer on the Isle of Rhum. There, year after year, mothers in better condition had more sons and less flourishing mothers gave birth to a disproportionate number of daughters. Another study found that female opossums given extra rations during pregnancy produced one and a half times as many sons as did females who were fed normally. Females fed less than normal rations were twice as likely to bear female offspring (Hrdy, 1987; Allport, 1997).
Trivers and Willard suggested that one of the species where the conditions obtain — and therefore one in which we could look for the effect — was our own. In humans, as in some other primate species, reproductive success depends not just on size but on other factors which lead to greater access to resources. In humans this might include status and income differences. And indeed, there is convincing evidence for a mild effect in humans that shows up where one would expect it: at the ends of the spectrum of social class. Research has borne out the expectation that higher status parents invest more in sons and lower status parents invest more in daughters. This occurs not just among elite families, but throughout the classes. Research on the demographics of early modern Portugal found that Grandees had more sons who reached adulthood and petty nobles more daughters (Boone, 1986). In eighteenth-century Schleswig-Holstein, farming families with more land invested more heavily in sons whereas landless peasants invested more heavily in daughters (Voland, 1984).
There is little evidence that humans give birth to more sons or daughters depending on their resources. However, the concept of parental investment ranges more widely than just birth numbers. The measure of parental investment in the work of Boone and Voland is differential survival to adulthood of sons and daughters. Using that evidence to search for a Trivers-Willard effect in a modern western society is likely to be inconclusive, because improvements in nutrition and medical care mean that survival to adulthood now has more to do with wider social factors and less to do with parental investment. However, the most startling evidence for a Trivers-Willard effect among humans used a different measurement of parental investment.
Gaulin & Robbins (1991) found that in lower-income families (< $10,000 per annum) more than half of the daughters were breast-fed, and fewer than half of the sons. In higher-income families (>$60,000 per annum) the rates of nursing were reversed; some 60% of the daughters were breast-fed, but almost 90% of the sons. Furthermore, parental investment in this study also showed up when different-sex siblings, who compete for parental attention, were added to the family. Affluent parents who had a son first waited on average some 3.9 years to add to the family. If they had a daughter first they produced another child in less than 3.2 years. The figures for lower-income families were 3.5 and 4.3 years, respectively.
Gaulin and Robbins took presence or absence of a male parent in the household as another measure of resources. It turned out that women without a male to share the child-rearing breastfed girls for about 8.5 months and boys only about a third as long, whereas boys were nursed longer by mothers who had a man around. The general picture emerges from Gaulin & Robbins’ work of a conditional bias to invest differentially that reverses itself at extremes of parental condition, as Trivers and Willard expected.
Evolutionary approaches to the mind are often accused of making few interesting hypotheses and predictions. However, evidence that we are apparently able to compute Trivers-Willard and invest in offspring depending upon status supports a highly counter-intuitive prediction. Moreover, even though the hypothesis depends on a piece of adaptationist speculation, the evidence for a Trivers-Willard effect comes from the demography and child-rearing of modern humans, not allegedly pointless conjectures about the EEA.
Trivers and Willard are forward-looking adaptationists. They began with apparent conflicts between the general logic of sex ratios in mammals and some apparent counter-evidence, and made predictions about parental investment based on hypotheses about what would be adaptive in a given set of circumstances. Then they predicted that the same pattern should exist in humans. The argument is that an ability to invest differentially in offspring depending on one’s status, resources and physical condition would be a good adaptation to have, consistent with our knowledge of general evolutionary factors. To test this theory, one doesn’t need to tell a story about the distant past, because the evidence should be available among existing populations, or populations for whom we have ample historical records.
The same applies to other forward-looking hypotheses. They may be based on general points about what could have been adaptive in past environments, or they may be based on much more general adaptationist ideas considered in the light of background knowledge about humans.2 The latter, one suspects, is often what narrow-school evolutionary psychologists are talking about when they mention the EEA, and their position might in fact be strengthened if they dropped the pretense that they are discussing a real place. Theorists who are less invested in the marriage of sociobiology and MIT cognitive science often make similar arguments, as when Sober and Wilson appeal to the anthropological record to support their claims about the circumstances in which altruism should be adaptive for human groups (Sober and Wilson, 1999).
1.3.History, Explanation, Confirmation
So far, we have seen that forward-looking adaptationist claims can be tested in the contemporary world. I now turn to the question of exactly what is confirmed when we find a psychological capacity as predicted. Obviously the claimed capacity is there, but what else have we discovered?
We can distinguish a strong and a weak claim about what a confirmed forward-looking prediction shows. The weak claim is just that in these cases forward-looking adaptationism proves its worth as a heuristic that generates interesting claims about contemporary minds or cultures. These are just predictions about the modern mind and are open to straightforward testing. The strong claim is that confirming a forward-looking adaptationist hypotheses uncovers some of the causal-historical structure of the world. Past conditions caused the psychology of humans alive today to have one form rather than another. By investigating historical conditions, then, we make predictions about the present state of the world, confirm those predictions and conclude that we have uncovered a causal relation between events in the past and what we now see around us. This idea stresses not just the heuristic value of forward-looking adaptationist thinking, but its status as a form of causal explanation.
The commonest objection to the strong claim is that there are just too many other possible explanations for a present-day state of affairs that the alleged explanation appeals to (Fodor, 1998). Some of these may be non-selectionist or otherwise ahistorical explanations. It is important to see that these may not be incompatible with evolutionary explanations; they may, for example, be about proximate psychological mechanisms rather than ultimate evolutionary causes (Mayr, 1976; Kitcher, 1990). However, some alternative ahistorical explanations may be incompatible with evolutionary explanations. I will address this problem in a moment. Whatever one’s view of ahistorical explanatory alternatives, however, one may still think that too many alternative historical explanations can be cooked up. Current evidence is compatible with indefinitely many alternative historical scenarios, and once we have the evidence we can invent some alternative histories to our hearts’ content. Again, without a time-machine we cannot discriminate among alternative histories. Perhaps we can demonstrate the heuristic value of forward-looking adaptationism, but we cannot accept the strong claim.
I defend the possibility of the strong claim, although one will not always be in a position to make it at the conclusion of a piece of research. To defend it in principle, though, I will try to deflect the objection just outlined. I will discuss the seemingly reasonable idea that there are too many other possible explanations for a trait’s function. Then I will look at the specific claim that the historical causal structure of the world is simply inaccessible.
In a particular case the force of the objection comes not from general considerations but from the availability of incompatible alternatives. It is not enough just to say that there must be some alternative explanations. They should be stated, they should have testable consequences of their own, and preferably they should make some additional predictions.
With respect to some forward-looking claims it may be easy to envisage decent alternative explanations. In other cases it is not. No other line of inquiry has predicted or explained the Trivers-Willard effect in the full range of species it applies to, let alone in humans. If we cannot imagine any alternative explanation of the evidence we have good reason to believe the explanation that Trivers and Willard provided. At the very least, our confidence that we have found the cause of the Trivers-Willard effect should be raised.
The more general problem with the objection from the excess of possible histories is that it applies to other sciences with wholly unacceptable results. Cosmologists, for one distinguished instance, have confirmed predictions about the big-bang by using our current particle physics and by studying the available contemporary astronomical evidence. They have not traveled back in time to see what happened in the first seconds of the universe.
Cosmologists rely on the assumption that the effects of the big-bang are still with us. They argue that by using contemporary evidence such as the existence of background radiation, or the distribution of cobalt atoms, we are able to suggest and confirm hypotheses about the effects of the big-bang, and hence uncover some of the history and structure of the universe. This pattern of reasoning is the same as the one I attributed to makers of the strong claim about forward-looking adaptationism. The reasoning is common throughout the historical sciences, although it has attracted little philosophical attention.
Now, I daresay that we could come up with alternative histories that explained the distribution of cobalt atoms differently. Perhaps there was no big-bang, but there is a hitherto undiscovered source of cobalt somewhere in the universe. This objection would have little force if delivered at a meeting of astrophysicists. Why should evolutionary psychologists be worried by its analogue? All manner of historical sciences rely on the idea that by hypothesizing about past conditions we can develop stories about how that past caused the conditions we now see. A well-confirmed evolutionary claim in psychology is, in principle, as credible as any other historical claim, and there are effective ways of using historical evidence to discriminate between adaptive histories (Griffiths, 1996, 1997).
The real force of forward-looking explanations comes if we can integrate them with backward-looking explanations in a mutually supporting structure. We might start with some known facts and develop an explanation about how those facts might have originated; this is backward-looking adaptationism. Then this historical explanation could be a source of new, testable hypotheses. The important difference between them lies in their different confirmation relations, but this is no bar to their being integrated into a mutually supporting, bootstrapping structure.
The mutual articulation of forward- and backward-looking strategies shows one way in which we can confirm evolutionary hypotheses in psychology. Each strategy can suggest fruitful lines of inquiry that the other can exploit. If we develop a powerful pattern of explanation and predictive success like this we can be confident that we are on the right lines. Like forward-looking explanations in general, this bootstrapping is found outside psychology. For example, the theory that an asteroid impact explains the extinction of the dinosaurs was motivated by facts about the distribution of iridium. First, geologists and physicists worked backwards to uncover an historical explanation for the observable traces of the past that they had found. Generating the hypothesis about a meteor impact led to, in the second place, several predictions about evidence which should be present today — an impact crater, for example, and shocked quartz deposits (Alvarez, 1997). That second round of reasoning worked forward from an hypothesis about the past to the existence of present-day evidence which can confirm the hypothesis.
Both forms of reasoning, backward-looking and forward-looking, exist in the historical sciences, and one should not impugn them a priori unless one thinks that all historical science is suspect. No one should want to declare so much good science off-limits, so the onus is on skeptics about the strong claim in psychology to explain why it is impermissible in that one case. Evolutionary psychology should not be set rules different from those to which other historical sciences conform.3
I conclude that the popular “time machine” objection has been overhyped. If it is bad science to make claims about the causal influence of the past, then the objection miscarries because it would render a great deal of perfectly good science inadmissible. If the point is a narrower one, that forward-looking methods cannot be applied in psychology to produce testable results, then it seems to fall foul of the several research programs that have developed testable hypotheses and sought to confirm them and rule out competitors. Forward-looking adaptationism is not going to solve all psychology’s problems. The extent to which it can deliver useful results is obviously an empirical matter. But it is important to note that it is empirical, and that a priori objections based on the epistemic accessibility of the past can be answered.
That concludes my defense of adaptationist methods in psychology. The other face of narrow evolutionary psychology is a theory of psychological mechanisms. Unlike some critics, I think that the problems with narrow evolutionary psychology lie as much, if not more, with the psychology as with the evolution — or rather, with the connections between them. There is a persistent tendency to treat evolutionary considerations as supporting the position narrow evolutionary psychology takes on cognitive architecture. Evolutionary psychologists with a general commitment to modularity write as though the discovery of a discrete psychological capacity shows us that a domain-specific module has been discovered. This step is far too quick, as I shall now try to show. Once we appreciate this, we will have a better grasp of the limitations of evolutionary considerations in psychology.