A Notion of Complexity for Theory of Mind via Discrete World Models

TL;DR

This paper introduces a new complexity measure for Theory of Mind tasks based on cognitive load theory, and proposes Discrete World Models to improve LLM performance on these tasks.

Contribution

It defines a formal complexity measure for ToM tasks and develops Discrete World Models, a prompting technique that enhances LLM reasoning in social scenarios.

Findings

Complexity measure correlates with task difficulty.

DWM improves LLM performance on ToM benchmarks.

Spurious states affect task complexity assessment.

Abstract

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.