We test predictions of this idea with scenarios that pull compensation & deterrence apart.
For example: vary the harm to the victim while holding the offender's gain fixed. Pure deterrence predicts no effect. Yet consistent with compensatory concerns, demands rose with the victim's loss.
