Examining Social Attribution in LLM Reasoning: A Theory-Guided Probing Methodology

Large language models (LLMs) are increasingly deployed in sociotechnical systems to facilitate social interactions, in which *social attribution*, the reasoning process attributing external events to the causes and reasons of agents' social behaviors, plays a critical role. Such reasoning processes involve the judgments of social cause, responsibility, and blame/credit to agents. Although attributional models are well-studied in social psychology and social cognition, represented by Attribution Theory, it is still underexplored in the AI and computing community, and LLM social reasoning in particular. This paper aims to advance the understanding of LLM in a fundamental social aspect by providing the first systematic exploration of LLM social attribution. Our work focuses on responsibility and blame attributions, examining current LLMs' performances in making responsibility and blame judgments, as well as their internal mechanisms underlying such judgments. Specifically, guided by attribution theory, we first construct a social attribution benchmark consisting of a *Vignette* subset based on classic scenarios from attribution theory research and a *Reality* subset constructed based on real-world social narratives, yielding **7,639** responsibility/blame judgment questions. On this basis, we evaluate **32** representative LLMs and **5** basic non-LLM baselines. To further explore the internal mechanisms underlying the LLM judgment process, we develop a probing-based methodology to investigate the latent-space representations of **5** key attribution dimensions and the consistency of their influences on LLM judgments compared to those in human social attribution. Our research findings reveal that current LLMs exhibit measurable but incomplete agreement with human responsibility and blame judgments, and meanwhile, this agreement is positively correlated with model size. At a deep level, some attribution dimensions are systematically decodable from specific positions in LLM hidden states, and their influences on the final judgment are consistent with those indicated by human Attribution Theory.