Language-Informed Synthesis of Rational Agent Models for Grounded Theory-of-Mind Reasoning On-The-Fly

TL;DR

This paper introduces LIRAS, a framework that combines language and visual inputs to improve social reasoning in agents, outperforming existing models in tasks inspired by cognitive science experiments.

Contribution

LIRAS is a novel framework that integrates multimodal language models with Bayesian inverse planning for context-specific social inference.

Findings

LIRAS outperforms state-of-the-art models in social reasoning tasks.

The framework effectively combines linguistic and visual information.

LIRAS captures human judgments more accurately across multiple domains.

Abstract

Drawing real world social inferences usually requires taking into account information from multiple modalities. Language is a particularly powerful source of information in social settings, especially in novel situations where language can provide both abstract information about the environment dynamics and concrete specifics about an agent that cannot be easily visually observed. In this paper, we propose Language-Informed Rational Agent Synthesis (LIRAS), a framework for drawing context-specific social inferences that integrate linguistic and visual inputs. LIRAS frames multimodal social reasoning as a process of constructing structured but situation-specific agent and environment representations - leveraging multimodal language models to parse language and visual inputs into unified symbolic representations, over which a Bayesian inverse planning engine can be run to produce granular probabilistic judgments. On a range of existing and new social reasoning tasks derived from cognitive science experiments, we find that our model (instantiated with a comparatively lightweight VLM) outperforms ablations and state-of-the-art models in capturing human judgments across all domains.