{"pk":49865,"title":"Understanding Task Representations in Neural Networks via Bayesian Ablation","subtitle":null,"abstract":"Neural networks are powerful tools for cognitive modeling due to their flexibility and emergent properties. However, interpreting their learned representations remains challenging due to their sub-symbolic semantics. In this work, we introduce a novel probabilistic framework for interpreting latent task representations in neural networks. \nInspired by Bayesian inference, our approach defines a distribution over representational units to infer their causal contributions to task performance. Using ideas from information theory, we propose a suite of tools and metrics to illuminate key model properties, including representational distributedness, manifold complexity, and polysemanticity.","language":"eng","license":{"name":"","short_name":"","text":null,"url":""},"keywords":[{"word":"Artificial Intelligence; Psychology; Representation; Bayesian modeling; Neural Networks"}],"section":"Papers with Poster Presentation","is_remote":true,"remote_url":"https://escholarship.org/uc/item/51v69326","frozenauthors":[{"first_name":"Andrew","middle_name":"","last_name":"Nam","name_suffix":"","institution":"Princeton University","department":""},{"first_name":"Declan","middle_name":"I.","last_name":"Campbell","name_suffix":"","institution":"Princeton University","department":""},{"first_name":"Tom","middle_name":"","last_name":"Griffiths","name_suffix":"","institution":"Princeton University","department":""},{"first_name":"Jonathan","middle_name":"","last_name":"Cohen","name_suffix":"","institution":"Princeton University","department":""},{"first_name":"Sarah-jane","middle_name":"","last_name":"Leslie","name_suffix":"","institution":"Princeton University","department":""}],"date_submitted":null,"date_accepted":null,"date_published":"2025-01-01T18:00:00Z","render_galley":null,"galleys":[{"label":"PDF","type":"pdf","path":"https://journalpub.escholarship.org/cognitivesciencesociety/article/49865/galley/37827/download/"}]}