The roles of the analogy with natural selection in B.F. Skinner's philosophy.

Beginning in the 1950s, B.F. Skinner made increasing reference to an analogy between operant conditioning and natural selection. This analogy is the basis of an argument that, in contrast to Skinner's other critiques of cognitive science, is neither epistemological nor pragmatic. Instead, it is based on the claim that ontogenetic adaptation is due to a special mode of causation he called "selection by consequences." He argued that this mode of causation conflicts with explanations that attribute action to an autonomous agent with reasons for acting. This argument dismisses ordinary explanations of action, and has implications not only for cognitive science but also for morals. Skinner cited the latter implications to counter objections to the application of behavior analysis to the reform of society and its institutions. Skinner's critique, however, rests upon empirical assumptions that have been criticized by other behavior analysts. Although for Skinner the major role of the analogy was to propose an empirical thesis, it also can play a metaphysical role-namely, to demonstrate the possibility of ontogenetic adaptation without reference to agents who have reasons for acting. These two roles, empirical and metaphysical, are the mirror image of the empirical and metaphysical roles of the computer analogy for cognitive science. That analogy also can be (and has been) interpreted as an empirical thesis. Its empirical implications, however, have been difficult to confirm. It also, however, has played a metaphysical role-namely, to demonstrate the possibility that a physical process could perform logical operations on states having propositional content. Neither analogy provides a well-confirmed, general answer to the question of how to explain the process of ontogenetic adaptation. But together they show there are two metaphysically coherent, but conflicting, answers to this question. Depending upon one's epistemology, the analogy with natural selection may provide a useful point of departure for a strategy of research. Such a pragmatic grounding for a research strategy does not, however, provide sufficient reason to abandon for purposes of ethics the concept of persons as autonomous agents.