Abstract
$\textit{No man is an island.}$ Humans communicate with a large community bycoordinating with different interlocutors within short conversations. Thisability has been understudied by the research on building neural communicativeagents. We study the task of few-shot $\textit{language coordination}$: agentsquickly adapting to their conversational partners' language abilities.Different from current communicative agents trained with self-play, we requirethe lead agent to coordinate with a $\textit{population}$ of agents withdifferent linguistic abilities, quickly adapting to communicate with unseenagents in the population. This requires the ability to model the partner'sbeliefs, a vital component of human communication. Drawing inspiration fromtheory-of-mind (ToM; Premack& Woodruff (1978)), we study the effect of thespeaker explicitly modeling the listeners' mental states. The speakers, asshown in our experiments, acquire the ability to predict the reactions of theirpartner, which helps it generate instructions that concisely express itscommunicative goal. We examine our hypothesis that the instructions generatedwith ToM modeling yield better communication performance in both a referentialgame and a language navigation task. Positive results from our experiments hintat the importance of explicitly modeling communication as a socio-pragmaticprogress.