LANGUAGE might then be able actually to observe the processes of meaning and understanding According to one version of the mental-mechanism doctrine, to understand the meaning of a word is to call up an appropriate image in connection with it I am told ‘Bring me a red flower’ and according to this story I have to have a red image in my mind, and ascertain what colour flower to bring by comparing it with this image But that cannot be right: otherwise how could one obey the order ‘Imagine a red patch’? The theory sets us off on an endless regress (BB 3; PG 96) Suppose we replace the alleged inspection of an image with the actual inspection of a red bit of paper Surely, the greater vividness of the sample will make it even more explanatory! But no: if it is to be explained how someone knows what ‘red’ means it is equally to be explained how he knows that his sample—whether mental or physical—is red ‘As soon as you think’, Wittgenstein says, ‘of replacing the mental image by, say, a painted one, and as soon as the image thereby loses its occult character, it ceases to seem to impart any life to the sentence at all’ (BB 5) Of course, it is true that often as we talk mental images pass through our minds But it is not they that confer meanings on the words we use It is rather the other way round: the images are like the pictures illuminating a written text in a book One of the most important versions of the mistaken theory that meaning is a mental process is the thesis that naming is a mental act This idea is the target of one of the most important sections of the Philosophical Investigations: the attack on the notion of a private language, or more precisely, of the notion of private ostensive definition In the epistemology of Russell and the logical positivists, ostensive definition played a crucial role: it was where language linked up with knowledge by acquaintance But Wittgenstein insists that acquaintance with the object for which a word stands is not the same thing as knowledge of the word’s meaning Acquaintance with the object will not suffice without a grasp of the role in language of the word to be defined Suppose I explain the word ‘tove’ by pointing to a pencil and saying ‘This is called ‘‘tove’’ ’ The explanation would be quite inadequate, because I may be taken to mean ‘This is a pencil’ or ‘This is round’ or ‘This is wood’ and so on (PG 60; BB 2) To name something it is not sufficient to confront it and to utter a sound: the asking and giving of names is something that can be done only in the context of a language-game 140