SOURCE-LINKED INTELLIGENCE

Testing the Construct Validity of a Functional Valence Axis in LLM Agents

arXiv · AI, language, vision and robotics · article · Sep 19, 2026 · UTC

Contrastive activation directions are often interpreted from what they decode or how strongly they steer behavior. But what evidence is sufficient to identify the construct represented by such a direction, rather than a correlated feature of the contrast used to extract it? We study this question for a good--bad outcome direction in a maze task, using controlled interventions that separate the realised outcome from the informational history through which it became known. Across multiple LLM checkpoints, directions fitted on one explicit outcome encoding transfer well to another, indicating tha

Read original source ↗ Open in workspace

recordType

paper

region

Global