Rotting with their rights on: why the criteria for ending commitment or restraint of liberty need not be the same as the criteria for initiating commitment or restraint of liberty, and how the restraint may sometimes justifiably continue after its prerequisites are no longer satisfied.

That coerced treatment must end when the criteria for initiating coerced treatment cease to apply appears to be universally accepted by courts and commentators.2 Moreover, the consensus appears to be justified by a steel-trap argument. If coercion is justified only when the patient is mentally ill and incapable, because then the patient lacks autonomous capacities, or lacks practical reasoning abilities that undercut autonomous capacities, then these justifications have no force when the patient either is not mentally ill or is capable. A parallel claim holds for civil commitment. This received wisdom, or in = out thesis, rests upon a conceptual confusion: a failure to distinguish the criteria for initiation of intervention, those for cessation of intervention, and the purpose of the commitment or coerced treatment. If the criteria for commitment were mental illness and dangerousness, and the criteria for release were the same, then the purpose of commitment would be to restore persons to the point where they are either just barely not mentally ill, or just barely not dangerous. That is a silly and self-defeating purpose for that large class of patients who, because of lack of insight, or otherwise, do not become treatment compliant until they are substantially healthier than being barely not mentally ill or barely not dangerous. It sets them up to become revolving-door patients. The purpose of commitment is rather to maximize the patient's mental health, and minimize her dangerousness without unduly burdening her liberty. If society is going to violate a patient's liberty, it should do so in a way that will resolve the problem that justified the restriction on liberty in the first place, so long as the restriction of liberty is not too great in relation to the expected gains from the intervention. The criteria for releasing a patient from commitment are in this way responsive to the purpose of the commitment. For some revolving-door patients, this entails that the criteria for their release from commitment should be stricter than the criteria for initiating commitment in the first place. The criteria for release from commitment for revolving-door patients should be that the criteria for initiation for commitment is not met plus it being more likely than not that the patient will be treatment compliant after release, assuming the additional restriction on liberty is less than the gains from the additional restraint, and the restriction is not unduly burdensome. Spelling this out, the criteria for release should be either not mentally ill, or else not dangerous, or capable, and more likely than not to be treatment compliant after release. For those patients for whom such a test is overly optimistic, we might substitute that there is a reasonable probability of treatment compliance after release, or that the probability of treatment compliance has been enhanced. These criteria are to be thought of as rough and ready rules of thumb, and not as analytically precise tests.