Various anthropologists have described cultures in which sincere beliefs about witchcraft and sorcery apparently function as a mechanism for enforcing social commitments. For example, Elizabeth Colson's ethnography of the Gwembe Tonga describes a highland-dwelling population who managed to survive periods of severe drought by sharing resources. Individuals often did so reluctantly, but in Gwembe society, it became expected that no one would deny a request to share. As Colson (Reference Colson1960) explains:

Variations on this basic mechanism conceivably could enforce a wide range of social commitments. Even if you are fairly sure that your neighbour doesn't himself practice witchcraft, it is advisable to repay your debt, or else he might employ a shaman to smite you. Commitment problems are pervasive in social life, so we would expect this system to become established especially in societies with a standing belief in supernatural intervention and an absence of formal enforcement.

Notice that this commitment hypothesis involves a different model of cultural evolution from that outlined in the target article. Singh proposes that shamanism evolves by a process of cultural diffusion. In this view, variation and selection occur at the level of competing social scripts. By contrast, the commitment hypothesis proposes that selection occurs at the level of biological individuals or cultural groups. A biological individual might receive reciprocal benefits from the widespread fear of shamanic retribution. Alternatively, this belief might motivate altruistic behaviour, as seems to be the case for the Gwembe Tonga. Although all three proposals see shamanic beliefs as culturally transmitted, they disagree about the underlying causal process. Specifically, they disagree about which kind of fitness difference (among social scripts, biological individuals, or cultural groups) is the causally potent factor.

Singh argues that only diffusion explains the cluster of properties typically associated with shamanism across cultures. He does not consider in detail, however, how the commitment hypothesis might also explain this pattern. Take the tendency for shamans to have jurisdiction over important events, such as disease and crop failure. As Singh points out, these events are periodic, humanly uncontrollable, and fitness relevant. The fact that they are also disruptive makes them effective threats for motivating compliance.

Why might it be important for shamans to practice self-denial? Singh argues that this tendency must evolve through competition among shamanic scripts to appear credibly non-human. Alternatively, ritualized asceticism might function as a signal of self-restraint. To see why this might be important, we need only place ourselves within the supernatural belief system. Genuine shamans would wield Gyges-like power over us mortals. Our instinctive response would be to expel them from the community out of fear of abuse. Evidence that shamans lack interest in worldly pleasures would help placate such worries. Singh's strongest objection to this proposal, as I understand it, is that ritualized asceticism is not an honest signal of self-restraint. This objection, however, overlooks an interesting feature of this signalling system: Because supernatural forces are ultimately fictional, a shaman can't actually cheat. What matters is that members of the community tolerate the shaman's presence long enough to reap the fitness benefits without feeling threatened by the inequality in perceived power. Thus, although asceticism probably wouldn't guarantee against an actual threat of shamanic abuse, it provides adequate protection against an illusory one.

Why do shamans engage in trance? Perhaps this is interpreted as a signal of non-humanness, as Singh suggests. A simpler explanation, however, is that it contributes to the credibility of a shaman by serving as evidence of supernatural contact. The commitment model would predict that trance is most exaggerated in societies in which claims to shamanic authority are most in question, for example, when a culture comes under the influence of a competing religious system.

Perhaps the biggest challenge for the commitment model surrounds professionalization. If the threat of supernatural retribution helps enforce commitments, then why restrict access to a small class of professionals who have satisfied an initiation ceremony? The first thing to note is that cultures vary in the extent to which ordinary people are believed to have access to supernatural power. Evans-Pritchard, in his ethnography of the Azande, describes a population in which witchcraft was democratically accessible and covertly practiced. The result was a fairly unstable social system. People routinely accused community members of casting spells out of malice or personal gain. This gave rise to a second level of supernatural access, where oracles became the arbitrators of supernatural justice. As he explains:

Such anecdotes suggest that democratic, covert access to supernatural power is not a stable enforcement mechanism because it often is suspected of being manipulated. Restricting supernatural access to a limited number of third parties, whose power is overtly displayed in ceremony, might help mitigate the level of perceived abuse.

To be clear, my aim in defending the commitment hypothesis is to make a broader methodological point: Rival models of cultural evolution often make overlapping predictions about the sorts of traits that will be cross-culturally shared. Thus, no amount of evidence in support of those universals could possibly serve to discriminate among the alternatives. A more promising strategy is to first determine the distinct patterns of cultural variation predicted by alternative hypotheses, and then review the anthropological literature to test those competing predictions (e.g., Linquist Reference Linquist2016).