On the Optimality of Discrete Object Naming: a Kinship Case Study

2025-11-24

AI Summary

This study investigates how naming systems optimally balance semantic informativeness and low cognitive complexity, focusing on kinship terminology and relaxing conventional assumptions of an “ideal listener” and universal cross-linguistic communicative needs. Method: We develop a theoretical framework grounded in information theory and Bayesian inference, training neural agents in referential games to acquire naming strategies. We formally derive the necessary and sufficient condition for optimal trade-off: the listener’s decoder must be equivalent to the speaker’s Bayesian decoder. Results: Empirical experiments demonstrate that this optimal strategy emerges spontaneously in emergent communication, yielding kinship systems that are both informationally efficient and structurally well-formed. Our work establishes a novel paradigm bridging language evolution research and the design of artificial communication systems, offering principled insights into how semantic categories self-organize under communicative pressure.

Technology Category

• Natural Language Processing: Lexical Semantics and Morphology
• Game Theory and Economic Paradigms: Imperfect Information
• Multiagent Systems: Agent Communication

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems
• Economics, Online Markets and Human Computation: Uses of LLMs and GenAI for marketplace design, bidding, and strategic interactions

Abstract

The structure of naming systems in natural languages hinges on a trade-off between high informativeness and low complexity. Prior work capitalizes on information theory to formalize these notions; however, these studies generally rely on two simplifications: (i) optimal listeners, and (ii) universal communicative need across languages. Here, we address these limitations by introducing an information-theoretic framework for discrete object naming systems, and we use it to prove that an optimal trade-off is achievable if and only if the listener's decoder is equivalent to the Bayesian decoder of the speaker. Adopting a referential game setup from emergent communication, and focusing on the semantic domain of kinship, we show that our notion of optimality is not only theoretically achievable but also emerges empirically in learned communication systems.

Problem

• Modeling trade-off between informativeness and complexity in naming systems
• Addressing limitations of optimal listeners and universal communicative needs
• Investigating optimal naming systems using kinship semantic domain

Innovation

• Introduces information-theoretic framework for discrete naming systems
• Proves optimal trade-off requires Bayesian decoder equivalence
• Demonstrates empirical emergence in learned kinship communication