{"slug": "llms-know-the-constraint-but-do-not-use-it-activation-bottlenecks-in-pragmatic", "title": "LLMs Know the Constraint But Do Not Use It: Activation Bottlenecks in Pragmatic Constraint Reasoning", "summary": "A new arXiv study (2608.12321v1) finds that large language models (LLMs) often fail to use implicit feasibility constraints despite encoding them internally, identifying a routing problem rather than a knowledge gap. Testing 14 models, the authors show that probes on two open-weight models decode constraints with over 88% accuracy, yet activation patching repairs one model (+6.4 nats) but not the other (-0.07 nats). No prompted intervention fully mitigates the failure, as all inflate conservative bias through a single pathway: prerequisite mention.", "body_md": "arXiv:2608.12321v1 Announce Type: new\nAbstract: When a salient surface cue competes with an implicit feasibility constraint, LLMs often fail -- but aggregate accuracy conflates genuine constraint inference with conservative defaulting. We formalize the distinction as conditional constraint activation: the constraint is internally encoded (Knowledge) symmetrically across constraint-present and -absent prompts (Symmetry), yet only sometimes routed into the decision (Routing) and repairable by a donor activation (Repair). A quartet diagnostic over 14 models reveals two failure modes; probes on two open weights decode the constraint above $88\\%$, yet activation patching repairs one ($+6.4$ nats) and not the other ($-0.07$). On a mitigation frontier, no prompted intervention reaches the repair corner: all inflate conservative bias through a single mediation pathway -- prerequisite mention. Hidden-constraint failure is a routing problem, not a knowledge problem.", "url": "https://wpnews.pro/news/llms-know-the-constraint-but-do-not-use-it-activation-bottlenecks-in-pragmatic", "canonical_source": "https://arxiv.org/abs/2608.12321", "published_at": "2026-08-14 04:00:00+00:00", "updated_at": "2026-08-14 04:12:01.318273+00:00", "lang": "en", "topics": ["large-language-models", "ai-research", "ai-safety"], "entities": ["arXiv"], "alternates": {"html": "https://wpnews.pro/news/llms-know-the-constraint-but-do-not-use-it-activation-bottlenecks-in-pragmatic", "markdown": "https://wpnews.pro/news/llms-know-the-constraint-but-do-not-use-it-activation-bottlenecks-in-pragmatic.md", "text": "https://wpnews.pro/news/llms-know-the-constraint-but-do-not-use-it-activation-bottlenecks-in-pragmatic.txt", "jsonld": "https://wpnews.pro/news/llms-know-the-constraint-but-do-not-use-it-activation-bottlenecks-in-pragmatic.jsonld"}}