Arguing against the infallibility thesis, Churchland (1988) suggests that we make mistakes in our introspective judgments because of expectation, presentation, and memory effects, – three phenomena that are familiar from the case of perception. As an example where expectations come into play, he offers the case of a captured spy whose interrogators have repeatedly tortured him by briefly pressing a hot iron against his back. What would happen if, after 19 times with the hot iron, the torturers surreptitiously use an ice cube instead? Since the spy strongly expects to feel pain, Churchland suggests that the spy’s immediate reaction to the ice cube will not differ significantly from the reactions that he had to the hot iron, i.e., he will mistakenly think he is feeling pain. (See also Warner 1993.) Likewise, Churchland argues that when a sensation is presented to us for a very short duration of time, mistakes are not just likely but inevitable. Finally, he asks us to consider someone who suffered neural damage at a young age and has subsequently not felt pain or any other tactile sensation for 50 years. Then suppose that her neural deficits were somehow overcome. In such a situation, Churchland argues that it would be quite implausible to suppose that she would be able instantly and infallibly to discriminate and identify all of her newly regained sensations.

Churchland’s criticisms of the infallibility thesis in some ways echo worries raised by James almost a century earlier. As James noted, “Even the writers who insist upon the absolute veracity of our immediate inner apprehension of a conscious state have to contrast with this the fallibility of our memory or observation of it, a moment later.” He concludes that “introspection is difficult and fallible; and that the difficulty is simply that of all observation of whatever kind.” (James 1890/1950)

Another line of objection to the claim of infallibility derives from some remarks of Wittgenstein (1958). In the course of offering his private language argument, he worries about how an individual in isolation would be able to develop a language to refer to her own sensations. The problem is that in such cases there “is no criterion of correctness. One would like to say: whatever is going to seem right to me is right. And that only means that here we can’t talk about ‘right.’” Armstrong (1963) fleshes out the objection as follows (see also Wright 1989):


If introspective mistake is ruled out by logical necessity, then what sense can we attach to the notion of gaining knowledge by introspection? We can speak of gaining knowledge only in cases where it makes sense to speak of thinking wrongly that we have gained knowledge. In the words of the slogan: ‘If you can’t be wrong, then you can’t be right either.’ If failure is logically impossible, then talk of success is meaningless.

In the empirical domain, work in a variety of areas provides important evidence for the fallibility of introspection. Influential studies by Nisbett and Wilson (1977) suggest that we often misdescribe our own reasoning processes. In one study, subjects were presented with four pairs of stockings and asked to indicate which pair had the highest quality. The leftmost pair was preferred by a factor of almost four to one. However, unbeknownst to the subjects, all four pairs of stockings were identical. Though position effects were clearly playing a role in the subjects’ choice, none of them identified position when asked to explain their reasoning, and those who were asked explicitly whether position played any role in their reasoning process all denied it. The evidence from this and other studies thus suggests that people often form mistaken beliefs about what reasoning processes they are utilizing; as Nisbett and Wilson conclude, the evidence is “consistent with the most pessimistic view concerning people’s ability to report accurately about their cognitive processes.”