{"slug": "formalizing-latent-thoughts-four-axioms-of-thought-representation-in-llms", "title": "Formalizing Latent Thoughts: Four Axioms of Thought Representation in LLMs", "summary": "Researchers introduced an axiomatic evaluation framework for latent thought representations in LLMs, comprising four axioms (Causality, Minimality, Separability, Stability) that measure representation quality independently of downstream accuracy. Auditing open-weight LLMs across 23 reasoning tasks, they found no candidate satisfied all four axioms, representations distinguished task types but not questions within the same task, and encoded little beyond input embeddings—a structural gap consistent across model families.", "body_md": "arXiv:2606.27378v1 Announce Type: new\nAbstract: We introduce an axiomatic evaluation framework for latent thought representations in LLMs, comprising metrics that are independent of downstream benchmark scores and reveal representational failures that benchmark accuracy masks. Existing evaluations conflate representation quality with model capacity. Therefore, failures cannot be attributed to the representation rather than to the model that processes it. We formalize four functional axioms (Causality, Minimality, Separability, and Stability) and define a quantitative measure for each, computed directly on the representation independently of downstream accuracy. We audit open-weight LLMs across 23 reasoning tasks (e.g., Spatial Reasoning, Factual QA). We find that no candidate satisfies all four axioms simultaneously, that the representations distinguish task type reliably but cannot distinguish between two questions within the same task, and that the representations encode little information beyond what is already present in the input embedding. The failure is consistent across dense, reasoning-distilled, and RL-trained model families, indicating that the gap is structural rather than a property of model size or training procedure.", "url": "https://wpnews.pro/news/formalizing-latent-thoughts-four-axioms-of-thought-representation-in-llms", "canonical_source": "https://arxiv.org/abs/2606.27378", "published_at": "2026-06-29 04:00:00+00:00", "updated_at": "2026-06-29 04:06:49.940118+00:00", "lang": "en", "topics": ["large-language-models", "ai-research", "ai-safety"], "entities": ["arXiv"], "alternates": {"html": "https://wpnews.pro/news/formalizing-latent-thoughts-four-axioms-of-thought-representation-in-llms", "markdown": "https://wpnews.pro/news/formalizing-latent-thoughts-four-axioms-of-thought-representation-in-llms.md", "text": "https://wpnews.pro/news/formalizing-latent-thoughts-four-axioms-of-thought-representation-in-llms.txt", "jsonld": "https://wpnews.pro/news/formalizing-latent-thoughts-four-axioms-of-thought-representation-in-llms.jsonld"}}