Lexical Consensus: Grounded Word Learning and Shared Meaning in Artificial Agents
Summary
This paper introduces Lexical Consensus, a framework for studying how artificial agents acquire and stabilize grounded word meanings using visual embeddings and lexical learners, finding that perceptual distance rather than semantic relatedness drives acquisition success.
View Cached Full Text
Cached at: 07/01/26, 03:43 PM
Paper page - Lexical Consensus: Grounded Word Learning and Shared Meaning in Artificial Agents
Source: https://huggingface.co/papers/2606.22207
Abstract
Grounded word learning experiments using visual embeddings and lexical learners reveal that perceptual distance, rather than semantic relatedness, determines acquisition success, with distinct patterns in naming and retrieval performance.
Artificial intelligence systems are commonly evaluated through task performance and behavioral imitation, but such evaluations leave open whether an artificial agent can acquire, stabilize, and use new lexical meanings from grounded experience. This paper introduces Lexical Consensus, an experimental framework for studying grounded word learning over a structured perceptual substrate. Using frozenDINOv2visual embeddings, Carroll-stylenonce words, and interpretablelexical learnersplus linear baselines, we test whether agents can acquire artificial labels for visual concepts, generalize them bidirectionally, and stabilize them across controlled settings. The main result is a robust perceptual-coherence gradient: native categories are easiest to learn, coherent overextensions remain learnable, mid-range disjunctive concepts degrade, and far-disjunctive concepts approach chance. A pre-registered CIFAR-100 dissociation experiment confirms that this gradient is governed byperceptual distancerather than semantic relatedness:perceptual distancepredicts acquisition accuracy (partial R^2 = 0.245, p < 1e-7), whilesemantic distanceadds no significant explanatory power (partial R^2 = 0.002, p = 0.660). Bidirectional evaluation shows that naming and retrieval are distinct:exemplar-based mechanismsoutperformcentroid prototypesinlabel-to-image retrieval, exposing a memory-fidelity dimension separate from naming accuracy. Falsification controls, homogeneous candidate-pool evaluations, and null results onrepresentational restructuringindicate that frozen perceptual geometry both enables lexical grounding and limits what can be acquired without representational adaptation.
View arXiv pageView PDFProject pageGitHubAdd to collection
Get this paper in your agent:
hf papers read 2606\.22207
Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash
Models citing this paper0
No model linking this paper
Cite arxiv.org/abs/2606.22207 in a model README.md to link it from this page.
Datasets citing this paper0
No dataset linking this paper
Cite arxiv.org/abs/2606.22207 in a dataset README.md to link it from this page.
Spaces citing this paper0
No Space linking this paper
Cite arxiv.org/abs/2606.22207 in a Space README.md to link it from this page.
Collections including this paper0
No Collection including this paper
Add this paper to acollectionto link it from this page.
Similar Articles
Lexical discovery in unknown environments orchestrated by Large Language Models
Proposes the Neuro-Symbolic Lexical Discovery (NSLD) framework where LLM-based agents autonomously develop shared vocabularies for unknown visual referents in unknown environments, enabling pre-deployment planning for autonomous exploration missions.
Emergence of grounded compositional language in multi-agent populations
OpenAI researchers propose a multi-agent learning environment where agents develop a grounded compositional language with defined vocabulary and syntax to achieve goals. The study also observes emergence of non-verbal communication like pointing when language is unavailable.
Learning to communicate
OpenAI researchers demonstrate that cooperative AI agents can develop their own grounded and compositional language through reinforcement learning in simple worlds. The agents learn to communicate by being rewarded for achieving goals that require coordination, creating shared symbolic languages to coordinate behavior.
On the Persistent Effects of Lexicality in Large Language Mod
This paper investigates how lexical overlap, rather than semantic content, influences LLM representations across layers and architectures, and demonstrates that this lexical effect persists even in models trained for semantic similarity, leading to degraded performance on downstream tasks.
Seeing Is Not Sharing: Some Vision-Language Models Overestimate Common Ground in Asymmetric Dialogue
This paper investigates a bias in vision-language models where they overestimate shared understanding in dialogue, confusing perceptual access with communicative grounding. The findings have implications for dialogue systems and VLM evaluation.