Live data from Hacker News

Why concepts are (probably) vectors [pdf]

colala.berkeley.edu

1–2 of 2 posts

Re: Why concepts are (probably) vectors [pdf]

#2
The response to the compositionally issues is pretty hand wavy. Fodor & McLaughlin long ago pointed out that merely encoding compositional representations is not the issue.

I think this is a more general problem with these fields that prefer short papers, though. You can't really make a convincing argument for concepts being anything in 10 pages.