SOURCE-LINKED INTELLIGENCE

Metacognitive Steering: Learning the Structure of Scientific Judgment

arXiv · AI, language, vision and robotics · article · Sep 14, 2026 · UTC

Long-horizon scientific discovery requires agents to alternate between exploration, disciplined execution, and critical reassessment as evidence changes. Current language models are trained primarily on the products of science and optimized using outcome-level signals, providing limited supervision for these process-level shifts in scientific judgment. We investigate whether such judgment can be recovered from scientist interaction traces and used to control the internal computation of a frozen frontier model. Using contrastive interventions collected during real scientific research, we identi

Read original source ↗ Open in workspace

recordType

paper

region

Global