Pragmatism insists that, whatever philosophical propositions be true, one class of propositions must certainly be false -- all those, namely, which either assert or imply that human intelligence has no part, or no distinctive part, in the control of physical events and bodily movements, in the modification of environment, or in the actual determination, from moment to moment, of any of the content of reality. That man is a real agent -- and that the distinctive quality of his agency consists in the part played therein by the imaginative recovery and analysis of a physically non-existent past and the imaginative prevision of a physically non-existent future -- these are the first articles of any consistently pragmatic creed. Such a creed is simply a return to sanity; for these two theses are the common and constant presuppositions of the entire business of life. Never, surely, did a sillier or more self-stultifying idea enter the human mind, than the idea that thinking as such -- that is to say, remembering, planning, reasoning, forecasting -- is a vast irrelevancy, having no part in the causation of man's behavior or in the shaping of his fortunes-a mysterious redundancy in a cosmos which would follow precisely the same course without it. Nobody at a moment of reflective action, it may be suspected, ever believed this to be true; and even the composing and publishing of arguments for parallelism is a kind of reflective action.But Lovejoy finally wrote a full-length essay on his argument in response to an essay "Is Thinking Merely the Action of Language Mechanisms?" by the behaviorist John B. Watson published in The British Journal of Psychology 11 (1920): 87-104. You can read it online here. Watson's presentation is eerily similar to more recent defenses of eliminative materialism. For example, at one point (page 94) he contrasts the "introspectionist," that is, someone who believes that we can learn something about our thoughts by thinking, with the behaviorist who believes everything we can learn about our thoughts comes from the observation of the external effects -- the responses to stimuli, the external behavior.
The introspectionist hopes for a solution of the metaphysical problem through some mystic self knowledge. The behaviourist believes in no such transcendental human power. He himself is only a complex of reacting systems and must be content to carry out his analysis with the same tools which he observes his subject using. I cannot, therefore, agree with Mr Thomson that there is a mind-body problem in behaviourism. It is a serious misunderstanding of the behaviouristic position to say, as Mr Thomson does -- "And of course a behaviourist does not deny that mental states exist. He merely prefers to ignore them." He 'ignores' them in the same sense that chemistry ignores alchemy, astronomy horoscopy, and psychology telepathy and psychic manifestations. The behaviourist does not concern himself with them because as the stream of his science broadens and deepens such older concepts are sucked under, never to reappear.Eliminativists say much the same: they claim their position is just the result of taking science seriously, and any denial of their position is because people don't want to give up some sacred aspect of life. Just like Watson, they compare their position with astronomy and chemistry and the denial of their position with astrology and alchemy. For an example, see here.
Lovejoy responded in "The Paradox of the Thinking Behaviorist" published in The Philosophical Review 31 (1922): 135-47, which you can read online here, although you'll have to scroll down to page 135. To summarize: the behaviorist claims that unvocalized thought is incoherent: laryngeal movements are all that occur. "Perceiving a thing, in short, is identical with the motion of the muscles involved in uttering its name." All "thoughts," therefore, can be completely explained in terms of the physical motions involved in speaking, without any reference to their contents, what the thoughts are about. And this would obviously include the behaviorist's thoughts about behaviorism. In this case, the behaviorist (along with everyone else) hasn't actually said anything, he's just made sounds with no meaning. In which case, the behaviorist thesis has not actually been put forward. We can only affirm behaviorism by tacitly presupposing that behaviorism is false, since it is only if it is false that the behaviorist thesis can have any meaning.
One objection raised against Lovejoy's argument, in that same issue of Philosophical Review is "Awareness and Behaviorism: Apropos of Professor Lovejoy's Critique" by Howard Warren (pp. 601-605). He suggests that Lovejoy begs the question by presupposing that the only way to make sense of the behaviorist's claims is if behaviorism is false. But of course, the behaviorist claims that there is another way to make sense of his thoughts, a way that is consonant with behaviorism. Again, this is very similar to one of the main objections that eliminativists give to the claim that eliminativism is self-defeating. The counter-objection is, as Nagel puts it in The Last Word,
the appeal to reason is implicitly authorized by the challenge itself, so this is really a way of showing that the challenge is unintelligible. The charge of begging the question implies that there is an alternative -- namely, to examine the reasons for and against the claim being challenged while suspending judgment about it. For the case of reasoning itself, however, no such alternative is available, since any considerations against the objective validity of a type of reasoning are inevitably attempts to offer reasons against it, and these must be rationally assessed. The use of reason in the response is not a gratuitous importation by the defender: It is demanded by the character of the objections offered by the challenger.So I think Lovejoy's argument that behaviorism is self-defeating is sound. One can only accept behaviorism is true if one ultimately presupposes that behaviorism is false. As such, behaviorism defeats itself.
(cross-posted at Quodlibeta)