We often say that humans struggle to produce random sequences… but what if that's not the whole story? In a new paper, @dorsaamir.bsky.social, @samiyousif.bsky.social & I show that context matters (a lot!). When playing strategic games, humans can produce near-random behavior osf.io/preprints/ps...

Show Post

My bet is that no vision encoder can solve this by itself (at least not the current crop even after scaling up). As I understand it human shape perception relies on much more than ventral stream and possibly on "deliberative" systems (so PFC, etc. involved) — sounds like a complex network.

Show Post

Cool stuff :) I'm curious if shape-biased models close the gap meaningfully; see the examples in www.nature.com/articles/s42... The tested networks are quite biased towards texture (or at least not as strongly biased towards shape as humans are); I wonder if that alone is enough said divergence.

Show Post

Il y aura, dans le prochain La Recherche en kiosque le 17 septembre un mini-dossier sur l’IA (14 pages) avec notamment des contributions de @monniauxd.bsky.social Rattrapé par l’actualité, je viens de mettre sur notre site un article sur le grabuge ambiant www.larecherche.fr/math%C3%A9ma...

Show Post

Nice paper. Congrats!

Show Post