The evolved capacity for third-party punishment is considered crucial to the emergence and maintenance of elaborate human social organization and is central to the modern provision of fairness and justice within society. Although it is well established that the mental state of the offender and the severity of the harm he caused are the two primary predictors of punishment decisions, the precise cognitive and brain mechanisms by which these distinct components are evaluated and integrated into a punishment decision are poorly understood. Using fMRI, here we implement a novel experimental design to functionally dissociate the mechanisms underlying evaluation, integration, and decision that were conflated in previous studies of third-party punishment. Behaviorally, the punishment decision is primarily defined by a superadditive interaction between harm and mental state, with subjects weighing the interaction factor more than the single factors of harm and mental state. On a neural level, evaluation of harms engaged brain areas associated with affective and somatosensory processing, whereas mental state evaluation primarily recruited circuitry involved in mentalization. Harm and mental state evaluations are integrated in medial prefrontal and posterior cingulate structures, with the amygdala acting as a pivotal hub of the interaction between harm and mental state. This integrated information is used by the right dorsolateral prefrontal cortex at the time of the decision to assign an appropriate punishment through a distributed coding system. Together, these findings provide a blueprint of the brain mechanisms by which neutral third parties render punishment decisions.

The law typically treats an actor who knows that he will likely cause serious harm, and who does cause that harm, as a very culpable wrongdoer, meriting a severe sanction. Consider a simple illustration: Clara (a speeding driver who acts with individualized knowledge) Clara is running late for a plane to a critically important business meeting. Speeding along a narrow and deserted mountain road, she suddenly sees a drunk young man, barely moving, lying on the road in the path of the car. Rather than stop for him, she slows down but continues, striking the man as her car passes over him. She realizes that she is very likely to injure him, which she does. 1 Clara is likely to be liable in tort for a knowing battery, with ensuing punitive as well as compensatory damages, and for criminal assault and battery as well. And if she realizes that she is likely to kill the victim, and does kill him, she will very likely be guilty of murder. But the Anglo-American legal system does not treat actors so severely in cases where the knowledge is (what I will call) "statistical" rather than, as in Clara's case, "individualized." 2 Such comparative leniency is ordinarily the correct approach, as we will see. Yet it is remarkably difficult to explain and justify. Consider two examples of actors who possess statistical knowledge that they will cause harm: Agatha (a careful builder who acts with statistical knowledge) Agatha is building an enormous tunnel. 3 Based on past experience with projects of this size and scope, she can confidently predict that even with the highest level of care currently attainable, a dozen of the thousands of workers who will be employed over the ten year period of the project will Simons, Stat.Knowl Decon 11 3/3/11 suffer serious injuries, and, as a statistical matter, there is a 90% chance of at least one death. Agatha does employ a high level of care. A dozen workers do suffer a serious injury, as predicted. One dies. Knowingly injuring a person counts as battery under both criminal law and tort law. Is Agatha guilty of a dozen counts of criminal law battery? Is she liable to the dozen victims for tortious battery? For punitive damages? Knowingly causing a death usually counts as murder. Is she guilty of murder of the worker who died? Under current law, the answer to all of these questions is no. Now consider this variation: Bertha (a less careful builder who acts with statistical knowledge) Bertha is building an enormous tunnel, of the same size and scope as Agatha's. But Bertha takes less extensive precautions than are customary, based on advice that these lesser precautions are almost as effective as the usual ones, and are less costly and time-consuming, permitting the tunnel to be completed much earlier, with resulting social benefits. The projected injury rate is approximately 30% higher than would be expected for Agatha's project. Bertha might be found negligent, or perhaps reckless, in deciding to use lesser precautions. But it is quite unlikely that she will be found guilty of a knowing or int...

This article makes six points.

deserve special thanks for their extremely helpful criticism. Jared Levy and John Mills provided valuable research assistance.

The reasonable person test is often employed in criminal law doctrine as a criterion of cognitive fault: Did the defendant unreasonably fail to appreciate a risk of harm, or unreasonably fail to recognize a legally relevant circumstance element (such as the nonconsent of the victim)? But it is sometimes applied more directly to conduct: Did the defendant depart sufficiently from a standard of reasonable care, e.g. in operating a motor vehicle, that he deserves punishment? A third version of the reasonable person criterion, which has received much less attention, asks what degree of control a reasonable person would have exercised. Many criminal acts occur in highly emotional, stressful, or emergency situations, situations in which it is often both unrealistic and unfair to expect the actor to formulate beliefs about all of the facts relevant to the legality or justifiability of his conduct. A "reasonable degree of self-control" criterion is sometimes the best criterion for embracing these contextual factors. In self-defense, for example, it is conventional to ask whether the actor believes, and whether a reasonable person would believe, each of the following facts: (a) an aggressor was threatening him with harm, (b) that harm would be of a particular level of gravity, (c) his use of force in response would prevent that harm, (d) the level of responsive force he expects to employ would be of a similar level of gravity, (e) if the force was not used, the threatened harm would occur immediately, and (f) no nonviolent or less forceful alternatives were available whereby the threat could be avoided. United States law typically requires an affirmative answer to each of these questions. Yet in many cases, an actor threatened with harm will actually have no beliefs at all about most of these matters. It would be unfair to deny a full defense to all such actors. At the same time, we should still hold such an actor to a normative standard of justifiable behavior. Specifically, this essay suggests that we reformulate the reasonableness criterion and require this type of actor to exercise a reasonable degree of self-control in response to a threat of force.