The complexity of a cultural toolkit is highly dependent on the size and interconnectedness of the group that holds that toolkit (Derex et al. Reference Derex, Beugin, Godelle and Raymond2013; Kline & Boyd Reference Kline and Boyd2010; Powell et al. Reference Powell, Shennan and Thomas2009). If a society shrinks, it can experience loss of cultural knowledge because of low-fidelity transmission (Henrich Reference Henrich2004). Even at its best, social learning is imperfect, and details of tools, practices, and beliefs can be lost from one generation to the next. For example, if a society has a highly effective but complex fish hook design, it is unlikely that everyone in that society will master its construction. Given more minds, the probability that at least one learner will learn everything necessary to recreate this technology increases; furthermore, given a larger number of people to learn from, the greater is the probability that any specific learner will learn all of the key pieces of knowledge (Muthukrishna et al. Reference Muthukrishna, Shulman, Vasilescu and Henrich2013). Small variations in transmission can lead to innovation on a tool or piece of knowledge, rendering something like a fishhook more effective than the previous version (Aoki et al. Reference Aoki, Lehmann and Feldman2011; Eerkens & Lipo Reference Eerkens and Lipo2005). These innovations will be passed on to the next generation through the preferential learning from these experts (e.g., Chudek et al. Reference Chudek, Heller, Birch and Henrich2012; Henrich & Gil-White Reference Henrich and Gil-White2001).

Singh discusses this in the context of a whole society losing shamanism, but not in the context of how these mechanisms affect the shaman's toolkit itself. Professionalization of shamanism reduces the number of teachers and learners capable of transmitting the shaman's knowledge. If only a few shamans and a few apprentices hold a specific type of knowledge, then we should expect lossy transmission. Even if we assume that the fidelity of transmission is extremely high at an individual level – each apprentice might train for years to perfectly learn the entire toolkit – the limited number of minds this knowledge iterates through will decrease the likelihood of new innovations accumulating. This type of high-fidelity vertical transmission is conservative and is unlikely to produce much variation (Cavalli-Sforza & Feldman Reference Cavalli-Sforza and Feldman1981). Even under conditions of perfect fidelity, the best we can hope for is that the shaman's toolkit remains static and simple. Under conditions of low fidelity, it will need to be largely reinvented in each generation.

In many of the domains in which a shaman works, this is unimportant. If the shaman's primary role is to control the weather, then a lack of fidelity and innovation will do no harm. There is no real causal connection between the details of the toolkit and its alleged effects. Therefore, the probability of this knowledge ever becoming functional is infinitesimal. For other domains, this possibility could pose real problems. Medicine and food acquisition can have tangible impacts on a society's ability to survive and prosper. Small innovations in these domains can accumulate over time to create effective tools that far surpass anything individuals could come up with on their own (cf. Cavalli-Sforza & Feldman Reference Cavalli-Sforza and Feldman1981; Henrich & Henrich Reference Henrich and Henrich2010). Conversely, the failure to innovate could make a society less likely to thrive in its environment and less able to compete with other groups.

For shamanism to avoid becoming maladaptive in these cases, other mechanisms need to be used to increase the probability of transmission and innovation. Competition among multiple shamans could increase the probability that functional practices arise, but if we follow Singh's claim, the selection pressures here are for more elaborate displays rather than functional skill sets. Even if this is not the case, the number of shamans is likely to remain too low for effective innovation in many societies. Domains such as medicine can be maintained by dispersed parallel traditions outside of shamanism. As effective innovations accumulate in an area such as medicine, the shaman's purview should become increasingly restricted to only the perceived spiritual aspects of the illness (Hultkrantz Reference Hultkrantz1985).

Paying attention to these dynamics allows us to make more specific predictions about how shamanism might change or disappear as societies become more complex. The role of shamans in domains where functional knowledge is possible should become more restricted as these more effectual practices develop. In addition to the suggestions in Singh's article, shamanism may disappear because the early professionalization of shamanism reduces the chances a shaman's toolkit can compete with the increasingly complex technology of a growing society.

Further, given imperfect fidelity, the pieces of a shaman's toolkit that are maintained should become simpler and easier to transmit. Traits like redundancy through repetition of steps can increase the probability of this transmission (Acerbi & Tennie Reference Acerbi and Tennie2016). This type of routinization would make supernatural practices more ritualized over time (cf. Humphrey & Laidlaw Reference Humphrey and Laidlaw1994). In some cases, shamanism's ecstatic trance states may thus be replaced with more easily transmitted ritualized action. If easy-to-transmit mechanisms such as repetition and ritualization become part of what is passed on, they will begin to look increasingly like what we expect from religions (Whitehouse Reference Whitehouse2004). Thus, another way shamanism may disappear is by changing into something that looks increasingly like religion. This has been the case with the religious traditions that followed many of prophets and religious founders Singh offers as examples of shamanism.

When we compare the roles of shamans with the roles of religious professionals as a broad category, the likelihood of this transition becomes more plausible. Priests and pundits are also privileged communicators with the supernatural world in domains of uncertainty (Weber Reference Weber and Fischoff1969), but these religious professionals are more likely to work in routinized ritual practices than the elaborate trance states of shamans (cf. Whitehouse Reference Whitehouse2004). Even in our increasingly technological world, the problem of uncertainty remains. Increases in safety and security have led to some decline in religious belief in some parts of the world (Norris & Inglehart Reference Norris and Inglehart2004), yet religion worldwide has shown little decline (Stark Reference Stark1999; Stark & Bainbridge Reference Stark and Bainbridge1985). The problem of coping with uncertainty remains, and managing this uncertainty frequently stays in the hands of religious professionals.