A Notion of Complexity for Theory of Mind via Discrete World Models
This work addresses the need for better-defined complexity measures in ToM benchmarks for evaluating LLMs, though it is incremental as it builds on existing cognitive load theory and prompting methods.
The authors tackled the problem of undefined complexity in Theory of Mind (ToM) benchmarks by proposing a framework to measure task complexity based on the number of states required to solve it, and introduced a prompting technique called Discrete World Models (DWM) that improved performance on ToM tasks.
Theory of Mind (ToM) can be used to assess the capabilities of Large Language Models (LLMs) in complex scenarios where social reasoning is required. While the research community has proposed many ToM benchmarks, their hardness varies greatly, and their complexity is not well defined. This work proposes a framework inspired by cognitive load theory to measure the complexity of ToM tasks. We quantify a problem's complexity as the number of states necessary to solve it correctly. Our complexity measure also accounts for spurious states of a ToM problem designed to make it apparently harder. We use our method to assess the complexity of five widely adopted ToM benchmarks. On top of this framework, we design a prompting technique that augments the information available to a model with a description of how the environment changes with the agents' interactions. We name this technique Discrete World Models (DWM) and show how it elicits superior performance on ToM tasks.