Summary
People can say the same thing for different reasons, and listeners can understand the same words but learn different things. The authors introduce a model where speaking is both a way to give information and to change what someone else thinks or does. This model treats conversations as ongoing processes where speakers and listeners update their understanding and plans together. Unlike previous models, it tracks multiple separate updates, such as guessing what the other person is thinking, learning about them, deciding when to clarify, and habits about speaking. These ideas help explain how people adapt to each other during conversations.
What this means in practice
- •For conversational ai developers: Build dialogue systems that better model speaker-listener interactions by separating evidence gathering from intervention during conversations.
- •For human-robot interaction designers: Design robots that adapt their speech strategies to users by inferring user-specific parameters and managing clarifications over multiple turns.
Tested on simulated data.
Abstract
Identical utterance choices can arise from different communicative causes, and identical interpretations can leave different traces in what a listener learns. Rational Speech Act (RSA) models treat interpretation as inference over speaker meaning, but standard one-shot RSA does not intrinsically distinguish these causal update targets. We develop a coupled active-inference model of dialogue in which a listener's likelihood for an utterance is the policy distribution attributed to the speaker, placing the speaker's expected free energy within the listener's variational free energy. Each utterance is therefore both evidence about and an intervention on a partner. The model separates four updates that RSA approaches typically collapse: inference about the partner's current pragmatic state; learning of partner-specific parameters; prospective evaluation of clarification or repair; and a policy prior shaped by habit and a context-sensitive price of time. Under one-step, exact-inference restrictions, the model recovers the RSA speaker and listener, with RSA as the restricted single-turn limit of the coupled process. Outside these restrictions, the updates obey distinct rules and timescales, predicting which adaptations persist, remain partner-specific, or transfer. Worked examples show audience design reversing after clarification, self-confirming misunderstanding in which both interlocutors have low free energy while disagreeing about reference, and rational closing before uncertainty is resolved. Consistent with critiques of equating natural language with communication, the model treats communication as a downstream use of linguistic structure and recasts production and interpretation as coupled inference: speaking is both an intervention on a partner and an epistemic action that samples evidence for the speaker's model of that partner.