Few-shot Language Coordination by Modeling Theory of Mind

TL;DR

This paper explores few-shot language coordination by modeling theory of mind, enabling agents to adapt quickly to diverse partners and improve communication efficiency in referential and navigation tasks.

Contribution

It introduces a ToM-inspired approach for neural agents to model partners' beliefs, enhancing their ability to coordinate with diverse, unseen interlocutors.

Findings

ToM modeling improves communication success rates.

Agents better predict partner reactions and generate concise instructions.

Explicit mental state modeling enhances few-shot language adaptation.

Abstract

$\textit{No man is an island.}$ Humans communicate with a large community by coordinating with different interlocutors within short conversations. This ability has been understudied by the research on building neural communicative agents. We study the task of few-shot $\textit{language coordination}$: agents quickly adapting to their conversational partners' language abilities. Different from current communicative agents trained with self-play, we require the lead agent to coordinate with a $\textit{population}$ of agents with different linguistic abilities, quickly adapting to communicate with unseen agents in the population. This requires the ability to model the partner's beliefs, a vital component of human communication. Drawing inspiration from theory-of-mind (ToM; Premack& Woodruff (1978)), we study the effect of the speaker explicitly modeling the listeners' mental states. The speakers, as shown in our experiments, acquire the ability to predict the reactions of their partner, which helps it generate instructions that concisely express its communicative goal. We examine our hypothesis that the instructions generated with ToM modeling yield better communication performance in both a referential game and a language navigation task. Positive results from our experiments hint at the importance of explicitly modeling communication as a socio-pragmatic progress.