- A lot of aesthetic merit is explicitly about modeling one phenomenon in terms of another phenomenon or showing how two phenomena can be modelled based on a common prototype.
- A lot of aesthetic merit is explicitly about finding a short or elegant description that intuitively generates vast data (i.e. deeply evocative descriptions). There is also an explicit subclass of this form of aesthetic merit that is explicitly concerned with finding a short description that intuitively generates vast data that one has previously interacted with (i.e. articulating things in your experience or capturing things you encountered in the world).
- A lot of aesthetic merit is explicitly about stimuli that ‘resonate’ with many previously disconnected things you have encountered, felt or thought, and in so resonating illuminate possible connections (similarities?) between these previously disconnected things. Notably we sometimes have a very hard time verbalizing any personal-level reference points (hard to say what does the work resonate with et cetera) even when we are strongly inclined to claim that something like this has occurred. Often a verbalization comes within a week or two.
These three phenomena are at the core of our belief that art can be a form of thinking or of furthering our understanding or what have you. These three phenomena all call on seemingly related unexplained notions like ‘articulate’, ‘resonate’, ‘see connections,’ and so on, and I monomaniacally believe that the best way to cash out these notions has to be in terms taken from the world of data compression and data differentiation. (I mean the world of resource-bounded data compression and data differentiation — so, like, the A.I. research kind, not the pure theoretical computer science kind. Cf. Jurgen Schmidhuber on aesthetic merits of a kind related to but not identical to the phenomena above.)
An ideal theory along these lines is one that would use the language of data compression and data differentiation to provide something like a conceptual analysis of the core notions we employ in talking about these aesthetic phenomena, and also plausibly explain the empirical facts about what stimuli+person+context combos cause or fail to cause these phenomena to occur. Realistically, I think that concepts taken form the world of data compression and data differentiation can be used to construct nice approximations of the notions we employ in talking of these aesthetic phenomena, and that our aesthetic reactions imperfectly track data compression-type facts about our processing of things.