When Judgements Gain Control

With raw scene text fixed, changing only supplied consequence judgement becomes<br>action-effective substantially earlier under T3 and reaches a much larger final gain: the correct<br>high-minus-low intervention is positive by step 48 under T3 and reaches +3.48 pp, versus<br>a late, weak +0.54 pp effect under T1. The learned route is not a clean ( field, item, value )<br>variable. At T3 step 240, a wrong-item value retains +2.63 pp of active-action effect, while a<br>wrong-field value yields −0.71 pp and redirects probability toward measurement and evidence-<br>collection actions.

We therefore propose a control-routing view of training: optimization<br>does not merely learn what to output; it determines which information is routed to which<br>downstream reader, how much gain that route receives, and how precisely the route is typed.<br>Functional field routing can emerge before exact item binding, while endogenous report states<br>may remain outside the action path. These results separate representation learning, routing<br>learning, and binding learning as distinct stages of policy formation.</p>