Prediction and Accommodation in Evolutionary Psychology
Note: If you want to print this article, then there is a PDF version, which will print better.
Ketelaar and Ellis have provided a remarkably clear and succinct statement of Lakatosian philosophy of science and have also argued compellingly that the neo-Darwinian theory of evolution fills the Lakatosian criteria of progressivity. We find ourselves in agreement with much of what Ketelaar and Ellis say about Lakatosian philosophy of science, but have some questions about (1) the place of evolutionary psychology in a Lakatosian framework, and (2) the extent to which evolutionary psychology truly predicts new findings.
Lakatos, as Ketelaar and Ellis observe, conceives of research programs as having two levels: a hard core consisting of fundamental meta-theoretical assumptions and a protective belt containing auxiliary assumptions. Together, the hard core and the protective belt produce hypotheses and predictions that, ultimately, can confirm or disconfirm the assumptions in the hard core. Typically, however, failed predictions do not call into question the meta-theoretical assumptions of the hard core. This is so, for hypotheses and predictions derive from the hard core and the auxiliary assumptions of the protective belt. Consequently, given recalcitrant data, one can always place the blame on the assumptions in the protective belt, leaving untarnished the meta-theoretical assumptions of the hard core. It is only when the protective belt begins to function simply as a device for explaining away anomalies and does little by way of generating new predictions that the time comes to suspect the assumptions of the hard core.
Lakatos' main concern is with the evaluation of the assumptions in the hard core. Toward this end, Lakatos distinguishes between progressive and degenerative research programs. A progressive research program one that absorbs anomalies and generates new predictions confirms the assumptions of the hard core. In contrast, when a research program is degenerative when it dedicates its auxiliary assumptions to explaining away anomalies that otherwise would endanger the meta-theoretical assumptions and fails to yield new predictions the assumptions of the hard core become disconfirmed.
So far, we have said nothing that disagrees with Ketelaar and Ellis' description of Lakatos. But, we now come to our first question. How does Lakatos' understanding of the structure of science apply to evolutionary psychology? Ketelaar and Ellis often claim that it is evolutionary theory that is a progressive research program. Few would question this. In Lakatosian terms, the hard core of evolutionary theory would presumably include assumptions like the following: organisms evolve by natural selection, heritable traits are transmitted from parent to offspring via genes, etc. Evolutionary psychology is not part of the hard core of this research program. Rather, it lies in the protective belt. As Ketelaar and Ellis seem to recognize, evolutionary psychology is an application of the principles in the hard core of the Neo-Darwinian research program to questions about psychological processes.
Yet, if this is correct, Ketelaar and Ellis' use of Lakatos to assuage doubts about the legitimacy of evolutionary psychology is confused. If, as we suggested above, evolutionary psychology does not belong in the hard core of the Darwinian research program, then Lakatos' distinction between progressive and degenerative research programs does not help in the evaluation of evolutionary psychology. Whether a research program is progressive or degenerative matters to the confirmation of the assumptions in the hard core, but, we are assuming, evolutionary psychology is not part of the hard core of the Neo-Darwinian research program.
In short, Ketelaar and Ellis are unclear about whether their interest in Lakatos is for the purpose of lending credence to the Neo-Darwinian research program or to evolutionary psychology. If the former, one wonders why they feel the need. If the latter, they have misapplied Lakatosian philosophy of science.
We turn now to a second complementary issue. If evolutionary psychology is not to be evaluated as a part of a scientific hard core, then how should it be evaluated? Even more fundamentally: What is evolutionary psychology? In considering these questions, it pays to shift one's focus from Lakatosian philosophy of science to Kuhnian philosophy of science. Kuhn offers us the concept of normal science. Normal science is not concerned with testing the assumptions of the hard core. Rather, normal science proceeds on the assumption that the hard core is well-confirmed and consists in the investigation of puzzles and questions that have aroused scientists' curiosity. It is entirely rare for a scientist to engage in research that tests the assumptions of the hard core. Far more likely, the scientist will seek answers to puzzles that have little bearing on meta-theoretical assumptions: Why do oil tankers sink? What makes a tornado? What effect will genetically altered corn have on monarch butterfly populations? While Kuhns normative view of science is unacceptable ("As in political revolutions, so in paradigm choice there is no standard higher than the assent of the relevant community" (1970, p.94)), philosophers of science need to recognize that normal science is a knowledge-generating process in its own right. The merits of normal science rest not in its ability to confirm the hard core, but in its facility to solve puzzles. The remaining question is: How should puzzle solutions be evaluated?
An analogy will help to illustrate the standard that evolutionary psychology must meet. Newtonian physics left unexplained a small amount of rotation of the ellipse of Mercury's orbit, referred to as the precession of the perihelion. Naturally, a desperate Newtonian could have revised various parameters in his theory, or introduced new ones, in an effort to account for the unexpected precession. However, to do so would have been ad hoc it would have been an exercise in accommodation rather than explanation. It is precisely because Einstein's theories of special and general relativity predict just such a motion in Mercury's orbit given the same values for the various parameters that the Newtonians recognized that Einstein's explanation is superior. Einstein could account for mercury's precession without having to modify well-confirmed values of parameters or to introduce new parameters whose sole purpose would have been to accommodate the effect.
The difference between how Einstein did explain the precession in Mercury's orbit and how a Newtonian might have explained it corresponds with Lakatos' distinction between a progressive problemshift and a degenerative one. Progressive problemshifts are those that lead to genuine predictions; degenerative ones are those that lead to false predictions or offer mere accommodation. Thus, the basic ingredients for the evaluation of normal science are present in Lakatoss methodology, even if they are not used for that purpose.
But, is evolutionary psychology progressive or degenerative? First of all, it would be wrong to suppose that this is a simple either/or affair. Different applications of evolutionary psychology may be evaluated independently of one another. We also believe that it is too soon to judge the merit of some of these applications. In this regard, our discussion of the dispute between Newtonians and Einstein suggests some points that evolutionary psychologists do well to heed. First, in favor of Einsteinian normal science is the fact that it offers predictions that Newtonian physics does not. However, it is no easy matter to limn the predictive power of a theory. If we are to accept the explanations of evolutionary psychology as superior alternatives to non-evolutionary psychology, we must be clear about what sorts of phenomena non-evolutionary psychology is capable of predicting and, on the other hand, what sorts of phenomena non-evolutionary psychology cannot predict.
Similarly, like any successful normal science, the predictions of evolutionary psychology must go beyond mere accommodation. We do not wish to claim that any and all accommodation is degenerative. Indeed, accommodation plays a crucial role in theory development the constants in a theory are, after all, initially the product of accommodating theory to data. Nonetheless, a progressive research program is one that goes beyond accommodation that uses past accommodation to predict new and perhaps otherwise unexpected phenomena. Can evolutionary psychology do this? Perhaps, but it seems to us that the burden falls on evolutionary psychologists to show that it can. So, for instance, in their discussion of male parental investment Ketelaar and Ellis say that "selection should concomitantly favor the evolution of male strategies designed to reduce the chance of diverting parental effort toward unrelated young" (p. 17). They then cite Daly et al.'s claim that these strategies include "the use or threat of violence to achieve sexual exclusivity and control" (Daly et al. 1982, p. 11). But, is the use of violence to achieve sexual exclusivity a prediction of evolutionary psychology? It is difficult to see why. Surely there are many ways that a male parent might insure that the offspring in which he invests are his own. He might, as Daly et al. suggest, use violence to prevent his sexual partner from philandering. Alternatively, he might make his investments conditional on the fidelity of his partner, or he might sequester his partner for the duration of their courtship, or he might enlist the help of relatives to serve as a chaperon for his partner, etc.
The problem is that the hypothesis Ketelaar and Ellis invite us to consider that selection should favor the evolution of male strategies that reduce the chance of investment in unrelated young does not by itself entail anything about the use of violence to achieve sexual exclusivity. It does predict that males who invest in young will take measures to insure that these young are relatives, but it doesn't predict how they will do so. To discover which strategies males actually adopt, one must collect data. One might, as Daly et al. surmise, find that investing males use violence to enforce fidelity. But, we wonder, is this surmise a prediction, or is it accommodation to what has already been observed? Until these questions are sorted out, and this is by no means a trivial task, the value of evolutionary psychology remains uncertain.
Ketelaar, Timothy, and Ellis, Bruce J. (forthcoming): "Are Evolutionary Explanations Unfalsifiable?: Evolutionary Psychology and the Lakatosian Philosophy of Science." Evolutionary Psychology.
Kuhn, Thomas (1970): The Structure of Scientific Revolutions, Second Edition. Chicago: University of Chicago Press.