Extending rational models of communication from beliefs to actions
Not on the map. Why?
Abstract (arXiv)
Speakers communicate to influence their partner's beliefs and shape their actions. Belief- and action-based objectives have been explored independently in recent computational models, but it has been challenging to explicitly compare or integrate them. Indeed, we find that they are conflated in standard referential communication tasks. To distinguish these accounts, we introduce a new paradigm called signaling bandits, generalizing classic Lewis signaling games to a multi-armed bandit setting where all targets in the context have some relative value. We develop three speaker models: a belief-oriented speaker with a purely informative objective; an action-oriented speaker with an instrumental objective; and a combined speaker which integrates the two by inducing listener beliefs that generally lead to desirable actions. We then present a series of simulations demonstrating that grounding production choices in future listener actions results in relevance effects and flexible uses of nonliteral language. More broadly, our findings suggest that language games based on richer decision problems are a promising avenue for insight into rational communication.
Key facts. Cited by 8 papers in the map (data through 7 Oct 2026). Builds on 2 of its 27 references that are in the map. First posted on arXiv 25 May 2021.
Builds on
2 of 27 references.
References · 2
- 2020Emergent Multi-Agent Communication in the Deep Learning Eracited by 168
- 2017A Survey on Dialogue Systems: Recent Advances and New Frontierscited by 162
Built on by
Citing papers per year: 2022: 4; 2023: 2; 2024: 1; 2025: 0; 2026: 1.
8 papers.
- 2022Inferring Rewards from Language in Contextcited by 50
- 2023Say What You Mean! Large Language Models Speak Too Positively about Negative Commonsense Knowledgecited by 32
- 2022How to talk so AI will learn: Instructions, descriptions, and autonomycited by 22
- 2024Toward In-Context Teaching: Adapting Examples to Students' Misconceptionscited by 12
- 2026Value of Information: A Framework for Human-Agent Communicationcited by 10
- 2022Towards Pragmatic Production Strategies for Natural Language Generation Taskscited by 7
- 2022Linguistic communication as (inverse) reward designcited by 6
- 2023Characterizing tradeoffs between teaching via language and demonstrations in multi-agent systemscited by 2
- 2026Value of Information: A Framework for Human-Agent Communication10 Jan 2026
- 2024Toward In-Context Teaching: Adapting Examples to Students' Misconceptions7 May 2024
- 2023Characterizing tradeoffs between teaching via language and demonstrations in multi-agent systems19 May 2023
- 2023Say What You Mean! Large Language Models Speak Too Positively about Negative Commonsense Knowledge10 May 2023
- 2022Towards Pragmatic Production Strategies for Natural Language Generation Tasks23 Oct 2022
- 2022How to talk so AI will learn: Instructions, descriptions, and autonomy16 Jun 2022
- 2022Linguistic communication as (inverse) reward design11 Apr 2022
- 2022Inferring Rewards from Language in Context5 Apr 2022