Common sense suggests that we know our own thoughts directly, but that we infer the thoughts of other people. The former process is noninferential and infallible, while the latter is based on others' behavior and can always be wrong. But this assumption is challenged by experiments in psychology demonstrating that in certain circumstances young children tend to misdescribe their own thoughts regarding simple phenomena while nonetheless correctly describing those phenomena. It seems that these children have the same thoughts that adults have regarding the phenomena but are much less capable of identifying these thoughts. Some psychologists argue that this indicates that one's awareness of one's own thoughts is every bit as inferential as one's awareness of another person's thoughts. According to their interpretation of the experiments, thoughts are unobservable entities that, among other things, help to explain why we act as we do. It follows from this that we are wrong to think of ourselves as having noninferential and infallible access to our own thoughts.

Recognizing an obligation to explain why we cling so tenaciously to an illusory belief in noninferential and infallible knowledge of our own thoughts, these psychologists suggest that this illusion is analogous to what happens to us when we become experts in a particular area. Greater expertise appears to change not only our knowledge of the area as a whole, but our very perception of entities in that area. It appears to us that we become able to see and to grasp these entities and their relations directly, whereas before we could only make inferences about them. For instance, chess experts claim the ability to see without calculation whether a position is weak or strong. From a psychological perspective, we become so expert in making incredibly fast introspective inferences about our thinking that we fail to notice that we are making them. This failure leads naturally to the supposition that there is no way for us to be wrong in our identification of what we ourselves think because we believe we are perceiving it directly.

In claiming that we have only inferential access to our thoughts, the psychologists come perilously close to claiming that we base our inferences about what we ourselves are thinking solely on observations of our own external behavior. But, in fact, their arguments do not commit them to this claim; the psychologists suggest that we are somehow able to base our inferences about what we are thinking on internal cognitive activity that is not itself thought—e.g., fleeting and instantaneous sensations and emotions. The frequent occurrence of such internal activities explains why we develop the capacity to make quick and reliable inferences. Their internality makes it impossible for anyone else to make an inference based on them that contradicts our own. Thus, they are crucial in creating the illusion of noninferentiality and infallibility.