In your example, I agree that almost everyone would choose the second choice, but my point is that they will be worse off because they make that choice. It is an act of altruism, not an act which will increase their own utility. (Possibly the horror they would experience in making choice 1 would outweigh their future suffering, but after the choice is made they are definitely worse off having made the second choice.)
I say that the cube cannot be part of P’s utility function, because whether the cube exists in this example is completely decoupled from whether P believes the cube exists, since P trusts the oracle completely, and the oracle is free to give false data about this particular fact. P’s belief about the cube is part of the utility function, but not the actual fact of whether the cube exists.
Summary: I’m wondering whether anyone (especially moral anti-realists) would disagree with the statement, “The utility of an agent can only depend on the mental state of that agent”.
I have had little success In my attempts to devise a coherent moral realist theory of meta-ethics, and am no longer very sure that moral realism is true, but there is one statement about morality that seems clearly true to me. “The utility of an agent can only depend on the mental state of that agent”. Call this statement S. By utility I roughly mean how good or bad things are, from the perspective of the agent. The following thought experiment gives a concrete example of what I mean by S.
Imagine a universe with only one sentient thing, a person named P. P desires that there exist a 1 meter cube of gold somewhere within P’s lightcone. P has a (non-sentient) oracle that ey trusts completely to provide either an accurate answer or no information for whatever question ey asks. P asks it whether a 1 meter gold cube exists within eir lightcone, and the oracle says yes.
It seems clear that whether the cube actually exists cannot possibly be relevant to the utility of P, and therfore the utility of the universe. P is free to claim that eir utility depends upon the existence of the cube, but I believe P would be mistaken. P certainly desires the cube to exist, but I believe that it cannot be part of P’s utility function. (I suppose it could be argued that in this case P is also mistaken about eir desire, and that desires can only really be about one’s own metnal state, but that’s not important to my argument). Similarly, P would be mistaken to claim that anything not part of eir mind was part of eir utility function.
I’m not sure whether S in itself implies a weak form of moral realism, since it implies that statements of the form “x is not part of P’s utility function” can be true. Would these statements count as ethical statements in the necessary way? It does not seem to imply that there is any objective way to compare different possible worlds though, so it doesn’t hurt the anti-realist position much. Still, it does seem to provide a way to create a sort of moral partition of the world, by breaking it into individual morally relevant agents (no, I don’t have a good definition for “morally relevant agent”) which can be examined separately, since their utility can only depend on their map of the world and not the world itself. The objective utility of the universe can only depend on the separate utilities in each of the partitions. This leaves the question of whether it makes any sense to talk about an objective utility of the universe.
So, does anyone disagree with S? If you agree with S, are you an anti-realist?