Frege’s theory of meaning, for all its sophistication, relied on an unsatisfactory account of thoughts as abstract objects. The Tractatus did not have to deal with such a problem, because it treated meaning—and language altogether—independently of the ways in which language is actually used by human beings. Less than 10 years after the work’s completion, however, Wittgenstein came to believe that this dimension of language is of paramount importance. Without some account of it, he now thought, the entire system of the Tractatus would collapse like a house of cards. In writings and teachings from 1930 on, accordingly, he emphasized the connections between words and practical human activities. Words are animated, or given meanings, by such activities—and only by them. In the variety of little stories describing what he calls “language games,” Wittgenstein imagined people counting, calling for tools, giving directions, and so on. Comparing the meaning of a word to the power of a piece in chess, he insisted that it is only in the context of human activity that meaning exists. By conceiving of language apart from its users, therefore, the Tractatus had overlooked its very essence. The slogan accordingly associated with Wittgenstein’s later work is that “Meaning is use,” though he himself never expressed this view in such an unqualified form.

One of Wittgenstein’s principal themes is the open-ended or open-textured nature of linguistic dispositions. Although it may seem, especially to philosophers, that word usage is determined by the application of distinct and definite rules—and thus that knowing the meaning of a word is the same as knowing the corresponding rule—careful examination of actual speech situations shows that in no case can a single rule account for the countless variety of uses to which an individual word may be put. Wittgenstein asks, for example, what rule would explain the great variety of things that may be called a game. When one looks for something that all games have in common, one finds only “a complicated network of similarities overlapping and criss-crossing: sometimes overall similarities, sometimes similarities of detail.” The different games seem to be united only by a vague “family resemblance.” The usage of the word, therefore, is determined not by a complicated rule or definition—even one applied unconsciously—but only by a fairly relaxed disposition to include some things and to exclude others. If there is any rule involved at all, it is a trivial one: call games only those things that are games. Thus, knowledge of word meaning, and membership in the linguistic community generally, is not a matter of knowing rules but only of sharing dispositions to apply words in something like the way other people do. There is no conceptual foundation for this activity: the concept is generated by the usage, not the usage by the concept.

This means in particular that word usage cannot be founded in Lockean ideas. Wittgenstein’s refutation of this view is one of the most devastating short proofs in philosophy. He first poses the problem of how someone can understand the order to bring a red flower from a meadow: “How is he to know what sort of flower to bring, as I have only given him a word?” One possibility is that the hearer associates the word red with an idea (a mental image of red) and then looks for a flower matching the image. Wittgenstein says,

But this is not the only way of searching and it isn’t the usual way. We go, look about us, walk up to a flower and pick it, without comparing it to anything. To see that the process of obeying the order can be of this kind, consider the order “imagine a red patch.” You are not tempted in this case to think that before obeying you must have imagined a red patch to serve you as a pattern for the red patch which you were ordered to imagine.

The most-celebrated passages in Wittgenstein’s late masterpiece Philosophical Investigations (1953) attempt to unseat the notion of private experience. Their interpretation is endlessly controversial, but the basic idea is that objects of thought cannot include elements that are purely “private” to a single individual—as sensations, for example, are supposed to be. For if there were private objects of thought, then there could be no distinction, in what one says about one’s own thoughts, between being right and merely seeming to be right. Objects of thought, therefore, must be essentially public, checkable items about which one can in principle converse with others.

Not only experience and observation but also reason and logic are transfigured in Wittgenstein’s later philosophy. For Frege and Russell, the propositions of logic and mathematics are pristinely independent of sense experience, depending for their truth only on the structures of the abstract world they describe—a world made accessible to human beings through the light of pure reason. This vision was later somewhat compromised by the logical positivists’ assimilation of logic and mathematics to tautology and convention. In the later Wittgenstein, however, the entire distinction between logical and empirical truth becomes unclear. Logic, for example, is a set of practices and therefore a language, perfectly in order as it stands; what counts in logic as a correct application of a term or a permissible inference, therefore, depends only on what logicians do. As with word meanings in more-ordinary contexts, what matters are the settled dispositions of those who use the language in question. Because these dispositions may change, however, meaning is not—at least in principle—fixed and immutable. The rules reflecting common usage, including even fundamental physical principles and the laws of logic themselves, may change, provided enough of the relevant linguistic community begins using old words in new ways. The securest and most certain of truths may be coherently rejected, given that the rules underlying them have changed appropriately. There are no “higher” rules by which to evaluate these changes.

An uncomfortable vision opens up at this point. The very idea of truth seems to presuppose some notion of correctness in the application of words. If one calls a hippopotamus a cow, except metaphorically or analogically, then presumably one has gotten something wrong. But if the rule for applying the word cow is derived entirely from linguistic practice, what would make this case merely a mistake and not a change in the rule—and thus a change in what the word cow means? An adequate answer to this question would seem to require some account of what it is for a rule to be “in force.” Wittgenstein suggests in some passages that there is no substance to this notion: in normal times, everyone dances in step, and that is all there is to it. This suggestion is made with particular force in the discussion of rule following in the Philosophical Investigations. It is clear nevertheless that Wittgenstein believed that the distinction between mistake and innovation could be made.