Seems a bit like squishing high-order categories onto biological phenomena. It seems a bit far fetched that the existence of cross-cutting categories like fluffy would be applied to all the different conceptions of what are basically tools or obstacles, as I've learned, that the brain tends to categorize into. If fluffy did exist as a sort of meta-category, though, perhaps it would exist in the touch circuitry as a bit of meta data associated with that specific usable object.

In a nutshell, my contention is that the brain does not have a library of categories it uses to efficiently map objects like some kind of database with tags, but rather builds an individual association for each object, of course in a web of relations, and it is this individual association that comes to represent the object as a tool or obstacle.

Have you noticed how your brain will be able to "tell" you exactly how it would feel to lick any thing you look at? Try...

When using iTunes, I can hover over a song and I can hear (in my mind's ear) it start to play.

But what happens if you try to lick the song?

How does this square with research showing uncategorized priors? https://arxiv.org/abs/2007.09200 comes to mind.