¹ Williams, Bernard, ‘Wittgenstein and Idealism’, in Understanding Wittgenstein, Royal Institue of Philosophy Lectures, Vol. 7, 1972–73, Vesey, G. (ed.) (New York: St Martin's Press, 1974).Google Scholar

² The remarks on solipsism in the Tractatus are exceedingly compressed, even for the Tractatus. Some thoughtful interpreters of those remarks believe that Wittgenstein was not actually stating a solipsistic position. I disagree; but also I apologize for my peremptory treatment of the matter, my excuse being that it is not my topic in this essay. The bare way in which I have sketched the position may make it appear that Wittgenstein's adoption of solipsism was ridiculous and dogmatic. I believe it was ingenious and profound.

The most fundamental step in the development of solipsism in the Tractatus is the observation that ‘There is no thinking, representing subject’ (5.631). In the Tractatus Notebooks (p. 80)Google Scholar, this idea is expressed in the remark that ‘The I is not an object’. A basic doctrine of the Tractatus is that each significant proposition is ultimately analysable into a grouping of ‘names’, each of which designates an object. To say that ‘The I is not an object’ means that the word ‘I’ is not a name and so does not appear in the final analysis of any sentence in which it occurs. For example, the pronoun ‘I’ will not appear in the final analysis of ‘I am angry’, nor in the final analysis of any first-person psychological sentence. What about third-person psychological sentences, such as ‘He is angry’? It is sufficiently obvious that if ‘I’ is not a name, neither is ‘He’. The sentences ‘I am angry’ and ‘He is angry’ differ in meaning; but the difference cannot consist in the two pronouns designating numerically different objects, i.e. numerically different ‘thinking, representing subjects’.

Where then can the difference in meaning of those two sentences lie? It must lie in the predicates. The predicate ‘angry’ must differ in meaning in the firstperson and third-person sentences. Let us say that in the first-person sentence ‘angry’ refers to something genuinely ‘mental’, while in the third-person sentence it refers to physical behaviour. Thus, ‘I am angry’ is equivalent to ‘There is anger’, whereas ‘He is angry’ is roughly equivalent to ‘The behaviour of that body is similar to the behaviour of this (my) body when there is anger’. According to this model, first-person psychological sentences do truly assert the existence of thoughts, feelings, sensations, whereas third-person sentences only assert the existence of physical behaviour. This is a form of solipsism, although a no-self solipsism. Let us call this psychological solipsism.

How do we get from psychological solipsism to world solipsism? There is an easy transition by way of phenomenalism, which is the doctrine that propositions about physical things and events are analysable in terms of propositions about sensations and sense-impressions. There is evidence from Wittgenstein, 's Philosophische Bemerkungen (pp. 51, 84)Google Scholar, and also from Waismann's record of conversations with Wittgenstein (Wittgenstein und der Wiener Kreis, 45)Google Scholar that at an early stage Wittgenstein accepted phenomenalism. The conjunction of psychological solipsism and phenomenalism yields world solipsism, i.e. ‘I am the world’, which of course cannot be said but shows itself.

The foregoing is a very skimpy outline of the case for interpreting the Tractatus as presenting a no-self solipsism. I derived this interpretation of the Tractatus from reading the doctoral dissertation of my friend and Cornell colleague, Professor Richard Miller. Miller's interpretation has been recently published: Miller, Richard W., ‘Solipsism in the Tractatus’, Journal of the History of Philosophy 18, No. 1. (01, 1980).CrossRefGoogle Scholar