Singh's article attempts to explain the cross-cultural persistence and similarity of professionalized shamanic roles as a function of cultural evolution acting within the constraints of evolved cognitive biases. At the same time, Singh purports to account for variation in the specific forms of shamanism. His choice, however, to reduce the operational definition of shamans to “practitioners who enter trance to provide services” (sect. 1, para. 1) substantially weakens the potential for true comparison among discrete forms of shamanism. And there are discrete forms of shamanism: Shamanic or trance practice takes predictably different forms in distinct societies in relation to several crucial axes of variation, particularly gender roles, social structure, and type of trance (Bourguignon Reference Bourguignon1973; Winkelman Reference Winkelman1986a).

Moreover, ecstatic states such as trance are not limited to professional shamans. Hayden (Reference Hayden2003) distinguishes shamans from mystics who enter trances but do not function as helpers or healers and, conversely, shamans from magicians who effectively harness spiritual power but do not use trance states. Therefore, focusing on competitive advertising as the function of trance within shamanic cultural forms misses something about both shamanism and trance.

Using a sample from Murdock's original Ethnographic Atlas, Bourguignon (Reference Bourguignon1968) found that societies with higher levels of structural complexity were more likely to exhibit possession trance, whereas less complex societies were more likely to have trance without possession (see also Bourguignon & Evascu Reference Bourguignon and Evascu1977; Winkelman Reference Winkelman1986a). Greenbaum (Reference Greenbaum and Bourguignon1973) found that, within sub-Saharan African societies, structural rigidity, that is, highly prescribed, hierarchical social roles, predicted the existence of possession trance above and beyond social complexity. Similarly, using Murdock and Provost's (Reference Murdock and Provost1973) societal complexity variables, Winkelman (Reference Winkelman1986a) found that the training process for shamanic healers was more likely to involve spirit possession in more complex societies, especially those with high political integration, high population density, and high levels of social stratification.

Another crucial layer of variability not acknowledged in Singh's article is that trance cults focused on spirit possession are disproportionately headed by female priests or shamans or attract a predominately female following (Lewis Reference Lewis1971; Sered Reference Sered1994). This is no trivial detail, even if the sole focus remains on professionalization and credibility displays. A new aggregated database that one of us (Stockly) has collaborated in constructing and validating may be especially well suited for investigating both the variability between types of trance (spirit possession trance and trance without possession) and the distribution of such techniques along gender and sex lines (Stockly et al. Reference Stockly, Arel, DeFranza, Wildman and McNamara2017). The Sex Differences in Religion Dataset (SDRD) compiles data from existing databases and original variables coded from several ethnographic accounts into a single dataset, enabling statistical analyses using data that previously had existed only in isolation. The SDRD focuses especially on data relevant for women and gender roles within spiritual and religious traditions, including variables regarding the status of women, non-binary gender roles, marriage residence patterns, religious and cultural rituals, domestic violence, and social development for a representative worldwide sample of 215 different cultures. The sample encompasses both the Standard Cross-Cultural Sample (Murdock & White Reference Murdock and White1969) and the HRAF Probability Sample (Naroll Reference Naroll1967). For the present analysis, in conjunction with new SDRD codes for spirit possession, we recoded and supplemented Snarey's (Reference Snarey1996) codes on high gods and used data originally coded by Justinger (Reference Justinger1978), Lagacé (Reference Lagacé1977), and Huber et al. (Reference Huber, Linhartova and Cope2004).

Using the SDRP database, we conducted new analyses focused on the relationships among social structure, trance type, and gender and sex roles. Here, we report preliminary findings. For these analyses, skewness and kurtosis of all variables were within acceptable limits (±2.00) for parametric tests. Sample sizes differ as a function of variable overlap between source datasets.

Corroborating Bourguignon's earlier findings, the presence of female-dominated possession cults exhibited a significant positive Pearson product-moment correlation with larger societal population (r=.383, n=35, p=.023); more layers of institutional hierarchy within stateless societies (r=.366; n=34, p=.033); and belief in moral high gods (r=.284, n=102, p=.004). The latter association bears on recent findings in cultural evolution indicating that, as societies grow in population size and complexity, religious systems may converge toward veneration of moral high gods and monotheism (Norenzayan et al. Reference Norenzayan, Shariff, Gervais, Willard, McNamara, Slingerland and Henrich2016; Purzycki et al. Reference Purzycki, Apicella, Atkinson, Cohen, McNamara, Willard, Xygalatas, Norenzayan and Henrich2016). The fact that possession cults not only thrive in such contexts but are typically dominated by women's participation calls for explanation (Lewis Reference Lewis1971). Perhaps shedding some light on this question, we found that female-dominated possession cults are also positively associated with relative economic deprivation (r = .343, n = 34, p = .047) and low frequency of premarital sex (r = .411, n = 35, p = .014).

Each of the above variables indexes one or more important features of what Douglas (Reference Douglas1970; Reference Douglas1999) characterized as “high-grid, high-group” societies – cultures that exert high levels of hierarchical social control over strictly bounded populations. Meanwhile, female-dominated possession cults were inversely correlated with the overall social status of those who enter possession trances (r = –.365, n = 105, p < .001). This is unsurprising, given that they are women, but it further emphasizes the complex and intriguing association between strict social control and possession trance, as in Korean shamanism (Kendall Reference Kendall1987).

We agree with Singh that trance states are not anthropological esoterica (Bourguignon Reference Bourguignon1973, p. 11). Fertile opportunities for theoretical progress are missed, however, when researchers neglect the interactions among social structure and religious forms (Douglas Reference Douglas1970). The questions of whether a shaman in a given society is a man or woman (or a third gender; e.g., Callender & Kochems Reference Callender, Kochems and Blackwood1986) and whether that shaman's trance is characterized by the bodily intrusion of culturally posited supernatural agents or the retention of personal agency are directly relevant for Singh's agenda of mapping the strategic affordances that influence the spread of shamanic practices. Clearly, something about complex, hierarchical societies bends cultural selective pressures toward female-dominated possession trance. Why? Singh's preference for collapsing these varied distinctions is, unfortunately, a step away from, rather than toward, greater knowledge. If shamanism is nothing but “cheesecake,” then it comes in far more than one flavor.