Towards A Holistic Landscape of Situated Theory of Mind in Large Language Models

In this position paper, we seek to answer two road-blocking questions: (1) How can we taxonomize a holistic landscape of machine ToM? (2) What is a more effective evaluation protocol for machine ToM? Following psychological studies, we taxonomize machine ToM into 7 mental state categories and delineate existing benchmarks to identify under-explored aspects of ToM.We argue for a holistic and situated evaluation of ToM to break ToM into individual components and treat LLMs as an agent who is physically situated in environments and socially situated in interactions with humans. Such situated evaluation provides a more comprehensive assessment of mental states and potentially mitigates the risk of shortcuts and data leakage.

While LLMs have demonstrated some capability of inferring communicative intentions, beliefs, and desires (Andreas, 2022; Kosinski, 2023; Bubeck et al., 2023), researchers also reported concerns regarding a lack of robust agency in LLMs for complex social and belief reasoning tasks (Sap et al., 2022; Shapira et al., 2023a) and in-context pragmatic communication (Ruis et al., 2022). Emerged or not emerged, that remains a question (or may not even be the central question to ask). In our view, existing evaluation protocols do not fully resolve this debate. Most current benchmarks focus only on a few aspects of ToM, in the form of written stories.