{"id":"W4402620243","doi":"10.1038/s41593-024-01758-5","title":"Semi-orthogonal subspaces for value mediate a binding and generalization trade-off","year":2024,"lang":"en","type":"article","venue":"Nature Neuroscience","topic":"Neural and Behavioral Psychology Studies","field":"Neuroscience","cited_by":28,"is_retracted":false,"has_abstract":false,"ca_institutions":"Université de Montréal","funders":"National Institute on Drug Abuse; Gatsby Charitable Foundation; Simons Foundation; National Center for Research Resources; National Institute of Mental Health; U.S. Department of Health and Human Services; National Science Foundation","keywords":"Linear subspace; ENCODE; Generalization; Orthogonalization; Encoding (memory); Value (mathematics); Subspace topology; Population; Computer science; Neuroscience; Artificial intelligence; Mathematics; Psychology; Biology; Algorithm; Machine learning; Genetics; Medicine; Pure mathematics","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0001797122,0.0001903692,0.0001488185,0.0001596486,0.0004164825,0.0001995021,0.0002491629,0.0001472279,0.000003367768],"category_scores_gemma":[0.0003907201,0.000141732,0.00006338292,0.0007528376,0.0003905217,0.0004026933,0.00007756953,0.0004229594,0.000004255982],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00001442249,"about_ca_system_score_gemma":0.00002665736,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":6.228114e-7,"about_ca_topic_score_gemma":0.00000334028,"domain_scores_codex":[0.9981713,0.00006526473,0.000162115,0.0009074761,0.0003263019,0.0003675327],"domain_scores_gemma":[0.9994633,0.0002555849,0.00004560351,0.0001264877,0.00001404073,0.00009497803],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.00001244202,0.00002170345,0.0007902475,0.0000268862,5.667617e-7,0.00004719729,0.0001272066,0.00001486621,0.9902459,0.003442341,0.001959864,0.003310751],"study_design_scores_gemma":[0.0004905303,0.000459057,0.01258603,0.00008726318,0.00005934375,0.0004342301,0.00004224983,0.01087573,0.8898818,0.002353962,0.08213151,0.0005982824],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9870248,0.00133023,0.000152226,0.00684674,0.003886476,0.0003254757,0.00008279667,0.0002154569,0.0001357877],"genre_scores_gemma":[0.9940892,0.0006221758,0.0000728599,0.004435977,0.0001896903,0.00003591513,0.00000238741,0.0000187466,0.0005330702],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.1003641,"threshold_uncertainty_score":0.5779664,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.07158509430793548,"score_gpt":0.3797692308329927,"score_spread":0.3081841365250572,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}