vii. The questions we are inclined to ask about a certain phenomenon depend heavily on the picture we have of that phenomenon and the use we put it to. To talk about the mind as giving words meaning is a picture, like saying that the carbon atoms in benzene form the shape of a hexagon. We have not described a matter of fact, but have rather provided a picture for looking at the matter. These pictures often direct us toward certain questions and uses, and can lead us astray if we are not careful.
viii. When we tell someone they will feel a certain sensation if they move their arm in a certain way, we are not conjecturing what inner feeling they will have. Our talk about sensations is not about inner feelings as much as about the common experiences that contribute to these feelings.
ix. Fear can be called a state of mind, but the words, "I am afraid," are not necessarily a description of that state of mind. They can be used for all sorts of purposes. I would not say I am describing an inner state when I say these words, but for all that, other people can take them as a description.
x. Moore's paradox examines the peculiarity of the first person present indicative. We can say, "It was raining, but I didn't believe it," or "it is raining, but he doesn't believe it," but not, "it is raining, but I don't believe it." We can observe behavior and describe a belief in other people, but in the first person case I am not describing but asserting.
Unlike Part I, Wittgenstein never arranged the remarks in Part II into a form he felt suitable for publication. He leaves us with fourteen sections (the eleventh taking up more than half of Part II), each of which gathers together a somewhat organized series of remarks on a particular question or topic. Many of the sections deal with similar themes, and continue the preoccupation of Part I in sorting out the grammar and meaning of various verbs like "believe," "mean," and "fear," that seem to hold some uncertain connection to the mind.
A theme that arises in several of these sections is the peculiarity of first person ascriptions of belief, meaning, fear, and so on. When talking about what other people believe or mean, I am trying to observe or describe their experience, based on what I can gather from their outward expression. When talking about what I believe or mean, there is no parallel observation or description. I cannot be wrong, nor can questions of knowledge even arise, when I say that I believe it is raining. Nonetheless, just as I can take other people's first person claims about belief, etc., to be descriptions of an inner state, so can other people take my first person claims. What I say is part of the data that other people use to infer my experience: it serves for them as a description of my inner state.