Because we can say words without meaning them, or say we believe something without believing it, we are tempted to say that meaning, and believing are mental acts that accompany speech. But just because there are some cases of saying words without meaning them does not mean that in other cases meaning the words cannot be a matter of simply saying them. We can say that certain gestures or facial features often accompany instances of believing, but none of these gestures or facial features must be present. Often there is nothing characteristic that goes on when I say, "I believe x" that means I actually believe x. Rather, the surrounding circumstances of my utterance make me believable.
There is no particular feature that defines a friendly face. Rather, there is a family of features: certain shapes of eyes, a certain kind of smile, certain lines on the face, etc. We might even say of a certain face, "it's the eyes that make it friendly." And yet, another face might have the same eyes, but because of lines or other features, the face as a whole might seem unfriendly.
If I say "red" while looking at a red object, I am inclined to say that the word "red" came to me in a particular way, different from if I had just uttered some other random word. I am then tempted to say I have a particular feeling when I say "red," and that this is what genuinely meaning something amounts to. But just because saying "red" might be accompanied by a distinct feeling does not mean that I have the same distinct feeling every time I call something "red."
We may say the same thing about "willing" as we do about such words as "thinking," "believing," and "reading." We can think of a paradigmatic case of willing, where I will myself to lift a heavy weight. However, we are then tempted to think that this clear-cut act of volition must be present in all other instances of intentional actions, and that these instances differ from unintentional actions precisely because there is an act of volition at work in them. In many cases, for example in voluntary versus involuntary speech, it is hard to determine whether and what kind of act of volition exists.
We might test someone's understanding of different words by running off a series of words—some familiar, some technical, some in another language—and have him say "Yes" when he understands a word and "No" when he does not. Wittgenstein runs us through a series of examples that show there is no characteristic sensation of understanding or not understanding.
The theme of fixity of meaning should be familiar to us by now. One of Wittgenstein's principal objects of criticism is the idea that words have fixed meanings, or some essential feature common to all uses of a particular word. In this passage, he suggests that we might be convinced of the fixity of meaning because some paradigmatic examples are so persuasive. Wittgenstein uses the word "will" as an example. We think of lifting a weight as a paradigm of voluntary action and then try to claim that all cases of voluntary action fit this paradigm. We think, "If I walk down the street, I am moving voluntarily and exercising my will in the same way I do when I lift a weight." Our comparison is shoddy, however, because you are not as aware of the movement of your legs as you are of the movement of your body when you lift a heavy weight. The word "will" does not mean precisely the same thing in both cases.
Wittgenstein explains the distinction between voluntary and involuntary movement by giving an example of an involuntary movement such as raising an arm to maintain balance as you lean away from a wall. Muscle spasms or epileptic seizures are also cases of involuntary movement. When we say other actions are voluntary, we are distinguishing them from involuntary actions because voluntary actions involve control of movement. Because the movement of my legs when I walk is voluntary, we assume this means I must be in control of this movement in the same way that I am in control of my muscles as they lift a heavy weight.
The problem with using paradigmatic cases is perhaps clearer in reference to belief or meaning than to volition. We are most clearly aware that we mean what we say when someone questions our sincerity. In these cases, we might be aware of a deep-felt conviction. This leads us to think that in all cases of meaning what we say there must be a similar inner conviction or emphasis, but that at some times we are more aware of it than at others. However, this sense of conviction might not always be present. When I say "the cup is on the table" and mean it, it is likely that I do not feel any conviction or emphasis accompanying these words.
Wittgenstein wants us to recognize that there is not some external marker that tells us when we are and are not using a word properly. When we talk about meaning what we say, we are not referring to an inner feeling that may or may not be present. We cannot prove the claim "you didn't mean it" by pointing out that a certain inner feeling was absent.
In this passage, Wittgenstein seems to contradict his own philosophy by applying the voluntary/involuntary distinction universally. He seems to suggest that all speech and all movement is either voluntary or involuntary. J. L. Austin pointed out this inconsistency in his paper "A Plea For Excuses." Austin is almost certainly not thinking of Wittgenstein in that paper, but his remarks apply. Wittgenstein would probably be sympathetic with Austin's analysis, for he seems to have slipped in this passage dividing all motions in the voluntary/involuntary categories.