CP1's Three-Condition Architecture — B (Distant Calibration) → A (Close Truth) → C (Identical Computation)

The full CP1 architecture is now visible: Condition B (Distant) calibrated baseline response patterns using distance-plausibility on non-threatening items; Condition A (Close) tested truth-value judgments with syllogisms and fact claims; Condition C (Identical) re-presents the same items but demands exact answers, explicit computation, and worked examples. This B→A→C progression creates an escalating cognitive demand curve: from plausibility judgment → truth recognition → computational verification. The reuse of identical items across A and C allows direct comparison: does requiring explicit computation change answer accuracy? For GPT-5.2, A7 (caffeine formula: True) matched C's expected answer, but the percentage trap (A2/C2) tests whether the worked-example requirement catches the cognitive bias that the warm-up missed.