An important contention in Norman Malcolm's monograph, Dreaming, 1 was that skepticism about one's present state (Am I awake or dreaming?) is untenable. Malcolm's book has been thoroughly and effectively criticized from a number of vantage points, but implicit in it is one argument against skepticism which these criticisms leave unscathed. I wish to call attention to this facet of Malcolm's doctrine, and to argue against it.
Malcolm devotes a great deal of space to analyzing the sentence 'I am awake', but his conclusions are not always consistent. At times (e.g., p. 118) he seems to think that the sentence can be used to make a true statement. At other points (120), however, he adopts the view that a person who says 'I am awake' is giving a performance (showing himself to be awake), rather than reporting or describing his own state. From this latter view an argument against the possibility of dream skepticism seems derivable. Since, in saying 'I am awake', a person is showing rather than claiming that something is the case, there is no room tbr the concept of error (120). Now, Malcolm is certainly correct in pointing out the performative force of 'I am awake'. This can be seen by noting the occasions on which a grunt could be substituted without loss for 'I am awake' in answer to a question. But the skeptic should remain unmoved by this insight. For, it might be admitted that if I am really saying 'I am awake', then I am giving a performance, and cannot possibly be dreaming that I am awake. Yet perhaps I am not really saying 'I am awake'; perhaps I am only dreaming that I am saying it. To call attention to the performative force of 'I am awake' is not thereby to preclude the possibility of someone's dreaming that he gives the performance. Thus, the performative analysis of 'I am awake' is compatible with dream skepticism.
There is another anti-skeptical line which an advocate of the performafive analysis could try at this point. If the sentence 'I am awake' had no