Shohei Hidaka (JAIST), Junya Morita (Shizuoka University), Mutsumi Imai
This session reviews emerging trends in Japanese cognitive science and examines how computational cognitive science is being redefined in the age of generative AI. We frame this shift as the rise of “post-learning cognitive science,” in which the field moves beyond treating learning as the primary explanatory mechanism and turns attention toward understanding-oriented cognitive processes. The JCSS Special Interest Group on Computational Cognitive Science provides an important foundation for this transition, promoting theory-driven and mechanistic modeling approaches.
Our Organized Session (OS) at JCSS 2025 addressed a central question for the next decade: Which cognitive phenomena require renewed theoretical focus because they cannot be adequately explained by learning alone? Although learning has long been a core topic, recent advances in machine learning have shifted many learning-related challenges from basic research to industrial development. This shift underscores the need to revisit cognitive processes that escape optimization-based accounts.
The OS identified several such targets—understanding, insight, creativity, ambiguity resolution, development, reasoning, intention, and integrative cognition—as domains where learning-based approaches remain insufficient. From the discussion, two major insights emerged: (1) cognitive science must strengthen its role as a theory-driven discipline, and (2) computational cognitive science should articulate mechanisms that enable models to move from “learning” toward “understanding.”
By outlining what must be studied “after learning,” the OS offers a forward-looking agenda for reorienting cognitive science in the generative-AI era.