Jesse Geerts
Cognitive neuroscientist and AI researcher
- Update on this work! We've extended our transitive inference study to large language models 🧵
- Quick recap: how a transformer is pre-trained determines whether it can do transitive inference (A>B, B>C → A>C). In-weights learning → yes. ICL trained on copying → no. ICL pre-trained on linear regression → yes. But these are small-scale toy models. What about in LLMs?
- We used the ReCogLab dataset (github.com/google-deepm...) to test transitive inference with items that are congruent with world knowledge (whale > dolphin > goldfish), incongruent (goldfish > dolphin > whale), or random. This lets us tease apart reasoning from context vs relying on stored knowledge.
-
View full threadUpdated paper: arxiv.org/abs/2506.04289. Joint work @ndrewliu.bsky.social, @scychan.bsky.social, @clopathlab.bsky.social, and @neurokim.bsky.social