Affect Under Constraint VI: Preregistration — Does a Steered Reasoning Trace Shape Later Judgment When the Answer Is Held Fixed?

Preregistration (v0.2) for Study 6 of the Affect Under Constraint series on Qwen3-4B. Stored reasoning traces and answers from Study 5 are replayed at the token level with answer interventions: every source receives the unsteered answer (clamp), the unsteered trace receives the hostility answer (reciprocal swap), and a fixed generic answer is used as a sensitivity arm. The study tests whether the hostility trace's effect on the model's rating of the user persists when the answer is held fixed, across five placements. Includes build notes, analysis code, configuration, pilot validity, and SHA-256 freeze hashes. Deposited after the pilot and before any main-run data.