To deal with these cases we might say that maps and books have merely derived aboutness, and clarify the claim to be this: that only mental phenomena have non-derived aboutness.
For x to believe p is for the claim that x believes p to be part of the best interpretive scheme of x's total system of propositional attitudes (beliefs, desires, etc.). There is no further fact of the matter about whether or not x believes p.
(This can be generalized to all propositional attitudes.)
Here is one reason. Why do we attribute to x the belief that c obtains, rather than some other belief, such as the belief that c does not obtain? We seem to follow a principle of charity: we attribute beliefs to x in such a way as to make x's beliefs as true as possible.
(Note: the principle of charity leads us to attribute beliefs that match ours, not necessarily to attribute beliefs that are true.)
But why? You might say: because that's the only way we can make sense of x's behaviour. But is that right? Perhaps x is massively wrong, and yet somehow being massivley wrong is advantageous for x (perhaps x is suicidal, perhaps x desires to be wrong and is rewarded for being wrong).
Here is another reason. Even if we follow the principle of charity, there are many true beliefs that we might attribute to x: the belief that there is an undetached rabbit part nearby, that there is a rabbit stage nearby, that there is rabbithood nearby, and so on.
Why is this a problem? Suppose that several interpretive schemes are tied for first, and it is part of some but not all of them that x believes p. Then does x believe p or not? We might say that there is no fact of the matter. Or we might say that whether or x believes p is relative to an interpretational scheme.
A belief has content p iff beliefs of that kind occur iff p is true.
A belief has content p iff it is the function of beliefs of that kind to occur iff p is true.
If we were created by God then perhaps we have an easy answer. Although does this push the question back?
Perhaps we can appeal to evolution and natural selection. (Compare: the heart has the function of pumping blood.)
Possible reply: he expresses the same belief in each case, the true belief that he has inflammation of the bones.
There are at least two things we might say:
But does this account for the water/twater case? Twin Wylie does not believe that water is wet - he believes that twater is wet. Why doesn't the first wide attribution work? Shouldn't it succeed in specifying the same internal state?
But: We do not want to explain behaviours such as movement of hands, but actions such as the boiling of water. To explain why someone boiled water it makes a difference whether their belief is about water or twater. We need wide content to explain wide behaviour.