Think Socially via Cognitive Reasoning

LLMs trained for logical reasoning excel at step-by-step deduction to reach verifiable answers. Yet, this paradigm is ill-suited for navigating social situations, which require an interpretive process for ambiguous cues that rarely yield a definitive outcome. To bridge this gap, we introduce Cognitive Reasoning, a paradigm modeled on human social cognition. It structures the interpretive process as an adaptive cognitive flow of interconnected cognitive units (e.g., observation or attribution) for effective social thinking and responses. We then propose CogFlow, a framework that instills this capability in LLMs. CogFlow first curates a dataset of cognitive flows through tree-structured simulation of the associative and progressive nature of human thought. After instilling the basic cognitive reasoning capability via supervised fine-tuning, CogFlow adopts reinforcement learning to enable the model to improve itself via trial and error, guided by a multi-objective reward that optimizes both cognitive flow and response quality. Extensive experiments show that CogFlow improves the social cognitive capabilities of LLMs and supports humans for more effective social decision-making, while generalizing promisingly to diverse reasoning tasks.