Now I know they’re trying to say that the neurons they’re recording are infact coding for the vector from the origin to the exit; but how can you know that? I mean, it just gets very complicated when you start ascribing purposes to behavioral-neural corelations (or cognitive-neural in this case).

I suppose the idea is like, if you recorded from some neurons and they spiked violently when you showed them a banana, that doesn’t tell you if the cells are recognizing the yellow of the banana, the edge of the banana or responding to a hunger cue, however if you lesion the area you’re in, and the monkeys still see bananas, and still eat them but can no longer tell yellow from blue, then you’ve got your answer.

So someone needs to design a very clever cognitive/behavioral test to see what this cognition-neural correlation is actually coding for.

BTW, since Bayle and I are at institutions with site licenses, we often post articles that others without institutions willing to pay $$$ may not be able to access. I apologize for that and encourage readers who find accessible copies of the articles (eg. where the author has retained rights to put the article on his/her own website for free download) to post links in the comments section, so that others can benefit from their legwork!