Suppose we say: Naive intuitions about beauty correspond to a folk theory whereby people naturally would all like the same things, except for certain deceptive conditions like:
- only liking something because you were told it's a great work of art
- not liking something which you would otherwise like because you've seen
- it so often
- liking something only through ignorance of some of its descriptive
- liking something which you would easily get sick of etc.
These are platitudes about which causal influences on aesthetic judgement
But now, pull a Kantian revolution, and say that it's not that these traits are misleading because they lead us to fail to track some antecedently natural and interesting category of the beautiful, but rather that `beautiful' refers to (roughly) "the kind of thing that people would like if not for the conditions indicated in the platitudes."
So far this sounds like familiar response-dependence theories. BUT there's one more wrinkle.
There might be enough natural variation between people for there to be no interesting class of objects which everyone is (ceterus paribus) disposed to like, when free from the deceptive conditions indicated by the platitudes above. It's a psychological matter whether getting more ideal in these respects would lead to convergence or divergence in aesthetic judgements.
So here's the complete view I want to propose (which, hopefully, handles this objection):
If there is a fairly definite class of things people (in our linguistic community) are inclined to like when free from the platiduinous bad influences then `beautiful' rigidly designates that class of objects.
If there is no such class, `beautiful' is implicitly speaker relative. In the latter case, we can think of arguments about beauty as involving the implicit assumption between both participants in the argument that their dispositions to like or dislike that objects would be the same if free from platitudinous bad influences.