Interchange

What we have here is an argument to the effect that American practical values are inferior to Continental intellectualist values, and as Skinner’s position is a product of these inferior values, it too is inferior. As Vonèche’s chapter itself emanates from the Continental intellectualist tradition, it provides a working example of the kind of thought generated by this alternative value system. To a considerable extent the chapter speaks for itself and requires no critique beyond caveat emptor. But I cannot resist cataloging a selection of its more egregious errors. The shorter section, headed ‘The Logical Argument’, receives the main part of my attention.

Since Skinner is well-known for his rejection of Bridgemanian operationism, it is hard to tell just what Vonèche is asserting when he claims that Skinnerian behaviorism represents operationism’s most rigorous application to psychology. If this claim is true, then it is true only in regard to Skinner’s own special sense of operationism. This does not appear to be Vonèche’s point.

Skinner’s ‘criterion for the evaluation of knowledge’ has to do with one’s effectiveness at getting along in that aspect of the world reflected in some putative domain of knowledge. One knows how to ride a bicycle if one can ride a bicycle, regardless of anything the ‘logical analysis of scientific discourse’ might have to say about it.

Contrary to Vonèche’s claim, Skinner does not replace causality with correlation. What makes Skinner’s views on causality somewhat distinct from much contemporary scientific thought is his rejection of mechanistic reductionism. But phenomena can be shown to stand in a causal relation even when not mechanically linked one to another.

The old chestnut that scientific laws must be ‘unlimited and universal’ is often rolled out in service of such arguments as that no social science can ever be real science, or that the theory of evolution is not a scientific theory. In fact it does not matter much if Skinner’s or anybody else’s laws are unlimited or universal. All a valid generalization must do is generalize over its own domain. If some philosophers of science want to restrict the appellation ‘scientific law’ to generalizations over universal domains, that is a matter of convention. It has nothing to do with the usefulness of law-like generalizations whose established domains fall somewhat short of the fringes of the universe.

Vonèche quite uncontroversially asserts that descriptive propositions (e.g., ‘the apple is red’) can be evaluated only in a context (I assume he means to include both a linguistic context—the English language in this case—and a material context). Vonèche then claims that this truism is inconsistent with Skinner’s epistemology! No reasons are given for drawing this conclusion; indeed, none can be given, as such a conclusion is entirely spurious.

On the whole Vonèche’s discussion of automata theory is correct. However, in consequence of both misinterpreting Skinner’s position and failing to note a subtlety of automata theory, his conclusions are incorrect. Automata theory has convincingly demonstrated that any adequate learning model must incorporate an ‘internal’ state concept analogous to the state table of a Turing machine. However, it must be noted that this notion of ‘internal’ state is a formalism, and does not rely on any particular mentalistic interpretation of such states. Contrary to Vonèche’s claim, Skinner’s ‘exclusive use of external predicates’ does not form ‘a function independent of internal state’, in the necessary formal sense of internal state. ‘A state is simply an element of a state-space, which is characterizable purely mathematically in set theory’, as Nelson puts it (1975, pp. 257–8). Particular interpretations placed on it are not relevant to the force of the formalism. It happens that Skinner’s model interprets this formalism in terms of an environmental history rather than in terms of a current mental state. The nature of that interpretation is not at all critical to the claims of automata theory, however. Yet even with that said, Skinner does acknowledge that ‘what an organism does will eventually be seen to be due to what it is, at the moment it behaves…’ (Skinner, 1974, p. 249). If that does not count as an acknowledgment of internal states, I do not know what would. This and related issues are explored in greater depth in Schnaitter (1986). Also see Bealer (1978) and Nelson (1984).

Skinner does not attempt to explain all behavior through learning. He makes frequent reference to phylogenetic contributions to behavior.

Skinner does not rely heavily on reflexes in accounting for learning. (Even when he does they are dull American reflexes, never the ‘brilliant’ ones that apparently shine in Geneva.) One would think it beyond question that Skinner’s major emphasis is on the modification of emitted (not elicited) behavior by its reinforcing consequences.

Radical behaviorism is not committed to the view that behavior cannot be modified through ‘mere exposure’ (e.g., observationally, or imitatively).

Skinner is not opposed to ‘interpretive discourse’, and quite explicitly and self-avowedly engages in it. The book About Behaviorism is almost exclusively interpretive.

Skinner does not claim that ‘theory duplicates reality’, whatever that might mean.

About ‘The Sociological Argument’, I have but two points to make. First, for those whose hearts swell to such refrains, the tune has been much better sung by others (e.g., Mishler, 1975; Schwartz and Lacey, 1982). Second, this kind of thing is going to give the ad hominem a bad name.