Modular Cognitive Architecture Emerges in Large Language Models
Source
Pengrui Han, Jacob Andreas, Evelina Fedorenko, Andrea Gregor de Varda
Author
Pengrui Han, Jacob Andreas, Evelina Fedorenko, Andrea Gregor de Varda
Published
Why it matters
If task families recruit distinct neuron populations, that is a handle for interpretability work, targeted editing, and reasoning about which capabilities interfere. It reframes model internals as modular rather than uniformly entangled.
articleDiagnostic Foundation for Evaluating LLMs' Research Integrity as Co-ScientistsYash Tripathi, Silu Sharma, Sai Sidhanth Manoharan Jayanthi, Shivank Garg, Lin Li
articleThe Devil Is in the Interface: Evaluating How Tool Architecture Shapes Coding Agent BehaviorXiangzhe Xu, Hamidreza Saghir, Qianhui Wu, Marc-Alexandre C\^ot\'e, Tong Wang, Kiran Lakkaraju, Kexin Pei, Xiangyu Zhang
articleAI Evaluation Should Work With HumansJan Kulveit, Gavin Leech, Tom\'a\v{s} Gaven\v{c}iak, Raymond Douglas