THE OPTIMIZATION RESEARCH-PROGRAM - STUDYING ADAPTATIONS BY THEIR FUNCTION

We develop as a scientific research program (sensu Lakatos, 1978) the essential assumptions used in applying optimization techniques to the study of adaptation. The assumptions and hypotheses of the research program are of two kinds: (1) hard core and (2) strategic. Any prediction is derived from a conjunction of both hard-core and strategic hypotheses. The hard-core hypotheses define evolution as the change in the frequencies of strategies (rather than, for example, alleles or genotypes). Consequently, fitness is a property of strategies rather than genetic entities or individuals. Strategic hypotheses describe how a strategy's fitness depends on various inputs (e.g., mating success and energetic intake), and on how the feasible set of inputs to fitness is constrained. Strategies are described as being supervenient (sensu Rosenberg, 1985) on genotypes because while a given genotype specifies a strategy, a given strategy does not specify any finite set of genotypes. The supervenience of strategies lends credence to functional explanations of the occurrence of strategies. In the Optimization Research Program the function of a strategy is functional only in that it influences the per-capita growth rate of the strategy. A refuted prediction leads to modifying the strategic hypotheses rather than the hard-core ones. In a progressive research program, the new, modified hypotheses avoid being characterized as ad hoc by virtue of the fact that they can generate new testable predictions. Lakatos proposed that science progresses through a competition between different research programs as defined by their hard cores. We suggest that alternative hardcores might include the assumption of studies in quantitative genetics to the effect that genetics generally constrain adaptation at an evolutionarily stable equilibrium.