but yeah, at least arguing they’ve got their own constraints (we don’t fully understand) seems like a useful direction

still in the pocket of big icon I see

oh that’s cool, I hadn’t seen this! is this your pylyshynian arc now?

like assuming the lambda notation description isn’t a world model and folk physics is, there’s probably a lot of stuff you can do with the former that don’t require latter

well, just thinking about some of the imagery debate and whether a simulation that’s similar in kind is necessary compared to a description

I feel like it’s usually intuitive physics, but I don’t think you really need that to understand lots of things about the world so idk

why would you need to model the world to do most things

some of the environmental concerns could absolutely be updated, and I might, though a lot of the recent critique as far as I have been tracking is about the cost of querying rather than pretraining itself (which presents free rider problem and is my focus)

paul bloom felt like the only one who actually likes nativism

Also this Friday in 111 NBS

ILST: 1:30 PM w/cookies

Does Not Compute? The Development of Real-time Negation Processing
Victor Gomes, PhD, Psych, Harvard

victorgom.es

@notnaughtknot.bsky.social
@pennlinguistics.bsky.social
@sas.upenn.edu