Cass Sunstein and Adrian Vermeule have posted Unrestricted Interrogation of Minors Not Yet Shown to Have Engaged in Culpapble Behaviors on SSRN. Here is the abstract:
Relatively unsophisticated analysis of the problem of unrestricted interrogation of minors who have not been demonstrated to have engaged in culpable behaviors often begins with the assumption that such potentially harsh interrogation would violate a deontic obligaton. This kind of moral objection to the harsh interrogative techniques frequently depends on a distinction between acts and omissions, but that distinction is misleading in this context, because government is a special kind of moral agent. Moral objections based on deontic obligations depend on a distinction between acts and omissions, but that distinction is inapplicable to actions by the state for which the obligation to prevent harm stands on an equal footing with the obligation not to cause harm.
We demonstrate that the act-omission distinction is a heuristic or mental short-cut, that generally work wells, but that can also lead to systematic errors when applied in the context of government action. For example, in the case of harsh interrogation of innocent children, the act-omission heuristic is especially dangerous because it may lead to systematic blindness to the possibility of harsh interrogaton-harsh interrogation tradeoffs. Such tradeoffs can occur because it may be the case that the harsh interrogation of one innocent child would result in prevention of the harsh interrogation (or equivalent suffering) of many innocent children. The state’s obligation to prevent harsh interrogaton or equivalent suffering caused by actors other than the state is morally equivalent to the state’s obligation not to directly engage in harsh interrogaton. For this reason, the state is morally obligated to consider harsh-interrogaton tradeoffs.
Such harsh-interrogatoin tradeoffs are especially salient because of recent research using functional MRI (magenetic resonance imaging) on the psychology of psychopathic terrorists. When a sample of such terrorists was asked to decide whether (i) to reveal the location of an explosive device or (ii) to refrain from revealing with the consequence that their own children would be tortured, the functional MRI imagery revealed that the brain state of the subjects was identical to brain states that typically accompany surrender and compliance behaviors. Although this research is currently inconclusive and may be subject to methodological challenge, we assume that it is correct for the purpose of this investigation.
The cognitive errors produced by the act-omission distinction in the case of torturing innocent children are of special concern because of the possibility that public endorsement of such errors will produce norm cascades. Because of the high emotional salience of child torture, observation of the act-omission distinction by the state in this case could lead to the reinforcement of the distinction in other similar situations. The emergence of a norm prohibiting intentional action by the state that constiutes torture of children could lead to the emergence of irrational norms against other intentional acts that violate superficially attractive deontic prohibitions, e.g., norms against the execution of innocent defendants, norms against unjust wars, norms against deliberate deception by government officials, and so forth. In the worst case scenario, irrational respect for human life, human rights, and transparent democratic processes might foreclose nuanced consideration of the full range of costs and benefits of torture, the killing of innocents, and deceptive propaganda. For this reason, we conclude that the optimal policy is to prevent the initiation of a catastrophic norm cascade. Given our assumptions, there is a moral obligation for the state to engage in the torture of innocent children.