informationstate

Information state is a term used in linguistics, philosophy of language, and related fields to denote the information that is currently assumed to be common ground between interlocutors in a discourse, as well as the information actively under discussion. It represents what is known or taken for granted in the present conversational context and guides how new utterances are interpreted.

In dynamic semantics and pragmatics, information state is updated as conversations proceed. An assertion adds content to the information state; presuppositions establish assumptions that must hold; questions introduce new information to be learned. The result is a revised common ground that shapes subsequent interpretation and inference.

Information state differs from information structure, though they interact. Information structure concerns how propositions are packaged (topic vs. focus and givenness) in an utterance, while information state concerns the evolving repository of shared knowledge and assumptions that underpins discourse.

Formal models often treat information state as a context, sometimes a set of propositions or a structured information record, that a sentence transforms. In dynamic semantics, the meaning of a sentence is a function that maps an input information state to a new state and a truth value, reflecting the update caused by the sentence.

Applications include discourse analysis, natural language understanding and generation, and AI modeling of agent knowledge. The concept extends to multiple-agent settings and may be implemented with possible-worlds representations or more general information-state structures.