Save

Trance, Dissociation, and Shamanism: A Cross-Cultural Model

In: Journal of Cognition and Culture
View More View Less
  • 1 Center for Mind and CultureBoston, MABoston University School of TheologyBoston, MA
  • | 2 Virginia Modeling, Analysis, and Simulation Center, Old Dominion UniversitySuffolk, VA
Full Access

Abstract

Religious practices centered on controlled trance states, such as Siberian shamanism or North African zar, are ubiquitous, yet their characteristics vary. In particular, cross-cultural research finds that female-dominated spirit possession cults are common in stratified societies, whereas male-dominated shamanism predominates in structurally flatter cultures. Here, we present an agent-based model that explores factors, including social stratification and psychological dissociation, that may partially account for this pattern. We posit that, in more stratified societies, female agents suffer from higher levels of psychosocial trauma, whereas male agents are more vulnerable in flatter societies. In societies with fewer levels of formal hierarchy, males come into informal social competition more regularly than in stratified contexts. This instability leads to a cultural feedback effect in which dissociative experiences deriving from chronic psychosocial stress become canalized into a male religious trance role. The model reproduces these patterns under plausible parameter configurations.

Abstract

Religious practices centered on controlled trance states, such as Siberian shamanism or North African zar, are ubiquitous, yet their characteristics vary. In particular, cross-cultural research finds that female-dominated spirit possession cults are common in stratified societies, whereas male-dominated shamanism predominates in structurally flatter cultures. Here, we present an agent-based model that explores factors, including social stratification and psychological dissociation, that may partially account for this pattern. We posit that, in more stratified societies, female agents suffer from higher levels of psychosocial trauma, whereas male agents are more vulnerable in flatter societies. In societies with fewer levels of formal hierarchy, males come into informal social competition more regularly than in stratified contexts. This instability leads to a cultural feedback effect in which dissociative experiences deriving from chronic psychosocial stress become canalized into a male religious trance role. The model reproduces these patterns under plausible parameter configurations.

1 Introduction

Among the many varieties of worldwide religious expression, trance and possession practices – such as Tungus shamanism, Korean muism, or North African zar spirit cults – stand out for their eye-catching ecstatic practices, emphasis on altered states of consciousness, and apparent antiquity in human evolutionary history. Numerous researchers have even suggested that, phylogenetically, shamanism and related trance phenomena represent the oldest type of human religiosity (Eliade, 2004; Le Barre, 1972) as well as the earliest definable specialized profession (Le Barre, 1972). Phylogenetic reconstructions of hunter-gatherer cultures indicate that animism and shamanism probably emerged before other common religious traits, such as ancestor worship and belief in high gods (Peoples et al., 2016). Thus, religious practices based on trance are thought to be deeply rooted in human behavior and culture.

However, shamanism and trance cults are not a uniform category (Lewis, 1971). For example, while both southern African !Kung num cult practitioners and Korean mansin shamans enter trance states during ceremonies, local interpretations and phenomenology of these trance states differ greatly. Korean shamans, for instance, are typically possessed by external spirits during trance performances (Kendall, 1987), whereas the !Kung healers do not experience possession, but instead become charged with a spiritual healing energy, num (Katz, 1982). Moreover, Korean shamans are overwhelmingly female, while !Kung num trancers are male. These differences exemplify an important pattern in cross-cultural data worldwide: structurally complex societies are more likely to host female-dominated trance cults that are focused on voluntary spirit possession, whereas trance practices in flatter and less complex societies tend to lack spirit possession and are often biased more toward male participation and leadership (Lewis, 1971; Bourguignon, 1973; Wood & Stockly, 2018).

This paper presents an agent-based computational model that explores the possible causal relations behind this striking pattern. Specifically, we simulate the social structural or hierarchical conditions under which different gender balances may come to characterize subpopulations of trance practitioners within larger populations of interacting individuals. This question is more complex than it seems. Previous studies have suggested that, to the extent that women participate disproportionately in possession cults, this is in part a compensation for serious gender inequalities in structurally complex and stratified societies (Lewis, 1971; Bourguignon, 2004). Not all scholars agree with this explanation (e.g., Boddy, 1989), yet nearly all scholars agree that predominance of women in possession cults is striking, arouses curiosity, and calls for theoretical clarification. However, given the fact that women are more active than men in religious life across nearly all societies (Stark, 2002; Trzebiatowska & Bruce, 2014), the predominance of males in religious trance practices in small-scale societies could be seen as equally worthy of explanation. The present model therefore builds from the assumption that any robust explanatory account of female dominance in trance cults in complex societies must also account for male dominance of trance practices in less-segmented societies.

Briefly, empirical data indicate that social stressors, particularly including social competition and defeat, are important triggers for dissociative psychological responses (Björkqvist, 2001; Marmot, 2005; Price et al., 1994; Yamasaki et al., 2016). In some cultural contexts, such responses may be canalized by training and cultural feedback into shaman or spiritual trance leadership roles (Bourguignon, 2004; Van Duijl et al., 2010). Both men and women can respond to social stress and trauma by spontaneously entering dissociative trances, yet neurologically and hormonally, men react more strongly to certain social stressors than women do (Matud, 2004; Stephens et al., 2016). Our model therefore bases its dynamics largely on emotional responses to social hierarchy and psychosocial stress. We posit that, in highly structurally segmented or hierarchical societies, the burden of social stress and marginality will fall more on women, who tend to possess lower formal and informal social status (Lewis, 1971). In certain circumstances, however, men in structurally flatter societies may suffer from relatively high levels of social stress, due precisely to the lack of fixed or formal hierarchy (Chaudhary et al., 2016). Our model thus simulates the conditions under which social competition and hierarchical dynamics may negatively impact men and women in different ways, leading to different levels of dissociative and somatic expressions and canalization into culturally stabilized trance practitioner roles.

1.1 Gender Differences in Trance Religions

In a seminal set of cross-cultural studies, Bourguignon (1973; see also Bourguignon & Evascu 1977) found that societies worldwide were more likely to exhibit voluntary possession trance practices if they also had high levels of structural or jurisdictional hierarchy, large population size, and a history of slaveholding. Thus, more hierarchical and densely settled societies were more likely to have trance cults in which participants became intentionally possessed by alien spirits. Greenbaum (1973) found evidence that, in Africa, spirit possession cults were particularly associated with structural rigidity above and beyond structural hierarchy. This combination of hierarchical segmentation and role rigidity corresponds roughly to what anthropologist Mary Douglas (1996, 1999) called high grid, high group cultures, or cultures with high levels of social control and group boundaries.

These findings take on new relevance in light of the fact that possession cults are widely associated with, participated in, and led by women (Bourguignon, 2004; Lewis, 1971; Sered, 1996). In a recent analysis of combined data from the Standard Cross-Cultural Sample (Murdock & White 1969) and the HRAF Probability Sample (Naroll 1967), Wood and Stockly (2018) found that specifically female-led possession trance was positively associated with several indicators of social complexity, including population size (β = .38), levels of jurisdictional hierarchy (β = .37), and existence of high god beliefs (β = .28). Thus, while men can and do participate in spirit possession trance in many societies (Kua et al., 1986; Schmidt, 2010), women play a much larger role in possession trance cults than would be expected by chance. Numerous scholars have attempted to explain this pattern by arguing that possession cults offer, variously, an avenue for “protest” against male domination in patriarchal societies (Lewis, 1971); an outlet for parody and female empowerment (Boddy, 1994); and an “escape valve” for women’s social pressures in structurally complex societies (Bourguignon, 1973). Despite the wealth of posited explanations, however, the question of why women are so disproportionately involved in many spirit possession religions is still an open one, and no single theory enjoys consensus (Schmidt, 2010).

Yet another question has gone largely unasked: why do men so often predominate in religious trance practices that do not feature spirit possession? Given the worldwide tendency of men to dominate structural hierarchies, the predominance of men in non-possession shamanic practices appears unsurprising. Yet there are at least two reasons to think otherwise. First, women are generally more religious in nearly all societies and religious contexts (Stark, 2002; Trzebiatowska & Bruce, 2014). Thus, the existence of a religious institutional form that is monopolized by men is worthy of interest demographically. Second, and importantly for this paper, trance and related practices appear to offer many affinities to, and to be rooted partly in the capacity for, dissociative psychological experience (Bhavsar et al., 2016). Dissociation is a pattern of breakdown in patterns of psychic unity, often characterized by odd sensory experiences, selective forgetting of or inattention to experience, or breaks in the continuity of consciousness, as in absorptive or fugue states (Spiegel et al., 2013; Sar, 2017). Persons high in dissociative tendencies show better facility in processing emotions than non-dissociators, suggesting that one component of a personality prone to dissociation may be emotional sensitivity (Oathes & Ray, 2008). In its pathological forms, dissociation often manifests in response to psychological or physical trauma, particularly social trauma (Ng & Chan, 2004).

Trance practitioners and leaders of possession trance cults may also be more prone to somatization, or the tendency to express psychological distress through bodily symptoms, than non-practitioners (Seligman, 2005). Unsurprisingly, somatization and dissociation positively covary with one another across individuals (Nijenhuis, 2001). In light of a body of empirical evidence (albeit inconsistent) that women may be more prone to somatization (Barsky et al., 20001; Delisle et al., 2012; Kroenke & Spitzer, 1998; Matud, 2004) and possibly dissociation (Akyüz et al., 1999; Lipsanen et al., 2000; Sar, 2011), at least some of the burden of explanation lies on the question of why men come in many societies to dominate roles that are defined by altered states of consciousness with affinities to dissociative trance.

Here, is important to emphasize that sex differences in the dissociation are probably due to cultural contexts and, most importantly, to differential exposure to trauma rather than biology (Sar, 2011; Spitzer & Freyberger, 2008). Plainly speaking, women on average experience more trauma than men do – especially social trauma (Olff et al., 2007). This may be one reason why they may suffer more from dissociative symptoms than men, if in fact they do (Spitzer et al., 2003). In societies with religious roles based on trance, dissociative symptoms may be incorporated into culturally specific schemas for self-presentation and identity, which in turn can canalize people with such symptoms into particular social roles. The etiology of shamans and other trance practitioners is therefore often rooted in culturally constructed interpretations of “anomalous experiences,” which are often products of psychological dissociation (McClenon, 1993). Thus, where women are exposed to greater trauma and social marginalization and where the surrounding culture offers trance leadership roles, women may come to be associated with such roles through long-term processes of cultural feedback. By a parallel logic, where sufficient numbers of men suffer from psychosocial trauma, they may also become the targets of similar cultural selection and canalization processes.

By no means does this chain of reasoning imply that shamanism or religious trance practices are merely pathological or symptomatic of pathology, or that shamans suffer from mental illness at higher rates than the general population. The hypothesis that shamans are simply mentally ill (e.g., Devereux, 1961) has been roundly challenged over the past half-century, and is now mostly discredited (Waida, 1983). Nepalese shamans, for example, have been found to suffer from no higher rates of dissociative disorders or other psychological illnesses than the general population (van Ommeren et al., 2004). However, the debunking of the “shaman as neurotic” trope does not negate the fact that shamanism and religious trance are often emically or indigenously associated with particular kinds of mental or emotional distress, including “spirit sickness” that presages entrance into the shaman role (Clark, 2006; Harvey, 1979). According to emic accounts, becoming initiated is claimed to catalyze recovery from psychological distress for the new shaman or medium (Waida, 1983). Thus, while shamanism and religious trance are not pathological in themselves, they often refer to – and are culturally and experientially associated with – expressions of psychosocial trauma. Dissociative responses to trauma may serve as a gateway for many individuals into trance practitioner roles, which in turn help resolve those responses by integrating them into a culturally normative framework (Luhrmann, 2007).

1.2 Physiological Pathways to Dissociative Trance

Physiological pathways link trance to psychosocial trauma, stress, and dissociative responses. These particularly include the the hypothalamic-pituitary- adrenal (HPA) axis, which governs the release of stress hormones and the fight-or flight response. Cumulative exposure to psychosocial stress – such as social defeat – can even produce psychosis through dysregulation of the HPA and dopamine systems (Luhrmann, 2007; Van Winkel et al., 2008). Dysregulation of the HPA axis also contributes to the etiology of dissociative disorders (Bob et al., 2008; Simeon et al., 2007). Importantly, the HPA axis appears to be particularly responsive to psychosocial stressors, especially those that directly bear on the ego’s position or status within a social matrix (Kajantie & Phillips, 2006). The sensitivity of the HPA axis to social triggers may partly explain the relationship between dissociative symptoms and social stress or psychosocial trauma (Ng & Chan, 2004; Yamasaki et al., 2016).

It is of interest, then, that men appear to respond with greater HPA axis activation and subsequent cortisol reactivity to social stress than women (Kudielka & Kirschbaum, 2005; Stephens et al., 2016; Zorn et al., 2017). The difference between the sexes in HPA response to social stress declines after menopause, when women’s hormonal profiles come to more closely resemble men’s (Kajantie & Phillips, 2006). Additionally, women report more stressors related to family, while men report more stressors related to relationships and work (Matud, 2004). These findings hint that social stressors are likely to have a significant effect on hormonal and neurological functioning for men, especially those having to do with competitive achievement. In social settings where achievement is highly variable – that is, where there are large gaps in achievement or production between men – social stress may be significant.

Considered in light of these findings, the economic structures of small-scale and foraging societies hint at the possibility of chronic conditions of male social stress. Among acephalous societies such as the !Kung and the Hadza, hunting success is often the primary means of achieving (informal) prestige, yet “a high proportion of animals are killed by a small proportion of men” (Woodburn, 1982, p. 440). Studying another foraging society, Chaudhary et al. (2016) found that informal status hierarchies were steeper among men than among women. Additionally, male social networks showed steeper inequalities in “relational wealth,” or access to informal networks of assistance, respect, and knowledge, than women’s networks did. Men with higher relational wealth also enjoyed better outcomes on a variety of objective measures of fitness, including fertility and food risk, than their lower-status peers – despite the lack of significant formal status distinctions in the society. Meanwhile, hunting success in politically egalitarian societies predicts better reproductive success (Smith, 2004) and social capital and popularity predicts polygyny and reproductive success among hunter-gatherers (Chaudhary et al., 2015).

In other words, despite lacking formal political leadership, acephalous societies (1) exhibit significant informal social inequality that (2) has empirical impacts on fitness-related and psychosocial outcomes and which (3) appears to be steeper and more consequential for men than for women. By definition, such informal hierarchies offer few offices, or defined roles insulated from competition or social shifts. Where inequalities in resources or formal social rank are minimal, positions of respect or influence are thus often “fluid, open, and contested” (McCaffree, 2015, p. 97). As a result, informal social hierarchies that result may be comparatively unstable, characterized by frequent (if subtle) rearrangements or competitive interactions to determine precedence. Such conditions may be optimally set up to elicit chronic hormonal stress responses and, potentially, HPA axis dysregulation among men who are particularly sensitive to psychosocial stress, or who lack opportunities or the ability to acquire informal prestige.

This postulate sheds light on the fact, long noted by ethnographers, that entrance into the role of shaman or trance specialist may afford some socially marginal people with a practical avenue into otherwise inaccessible social prominence or respectability; for example, among the relocated Sakhalin Ainu of Hokkaido, Japan, male shamans are typically “men barred from access to regular routes for political success” (Ohnuki-Tierney, 1980, p. 211). In this sense, trance leaders may be socially marginal or anti-structural “outsiders” (Turner, 1996). Thus, while males’ monopolization of resources and physical abuse ensure that life is overall more stressful for women than men across cultures (Mayor, 2015), some contexts, including in certain acephalous societies, may be highly stressful for some males. Males may then come to monopolize or dominate the social roles that are associated with displacement from social structures, and with psychosocial stress and dissociative or trance behavior. By contrast, as has been discussed extensively in previous literature (Boddy, 1989; Bourguignon, 1973; Lewis, 1971), women are often oppressively subordinated in highly stratified societies, leading to their general predominance in trance cults in such contexts. Below, we present a model that explores these postulates and their consequences.

2 Model Design and Implementation

Based on the theoretical foundation provided above, we designed and built an agent-based computer model using the software AnyLogic 8 (Borshchev & Filippov, 2004).1 In agent-based models (ABMs) autonomous agents interact and make decisions based on rules and states that update at each time step. Agent-based models are good tools for examining hypotheses regarding the emergence of nonlinear and other effects from low-level rules and interactions between individuals. In this case, we attempted to replicate, or “grow,” high-level patterns of gendered participation in religious trance practices out of individual interactions (Bonabeau, 2002; Epstein & Axtell, 1996).

Time steps in the model each represent one month. At initialization, all agents are assigned a gender (male or female), a shaman status (yes/no), and a boolean attribute of predisposition to somatize (yes/no). Agents also receive a social rank, a somatization threshold, and a shaman signal, as well as, for males, a likelihood of competing that is inversely proportionate to the number of social ranks in the simulated society (see below). To simulate the hierarchical dynamics of formally structured societies, we include an attribute of rank, which runs arbitrarily from the lowest possible rank of 1 to the highest rank of 10. Conceptually, this rank attribute represents formalized or legitimated status differences between individuals, such as the difference between an officer and a private, or between a noble and subjects. The number and distribution of ranks by gender can be made to vary (see below). These agent attributes are set according to the parameters listed in Table 1. Rank, somatization threshold, and shaman signal are determined by triangular, or quasi-normal, distributions across the population bounded by minima and maxima. The peak of the triangular distribution is the mode, or most common value for that attribute.

Table 1
Table 1

Model parameters and effects

Citation: Journal of Cognition and Culture 18, 5 (2018) ; 10.1163/15685373-12340042

Two agents may have precisely the same official rank (if any) in their society’s formal social structure, yet may nonetheless attempt to outshine one another. To represent the informal social status hierarchies that can be steeper and more consequential for men than for women, male agents also receive an attribute that determines their ability to win informal competitions with other male of the same formal rank. The competitions between same-ranking male agents represent jockeying for prestige, skill, or dominance rather than formal social standing or agent rank. At the beginning of each simulation run, all male agents have the same value for this attribute (“victory coefficient”). Following social competitions, winners have their victory coefficient incrementally augmented, while losers receive an incremental decrement. This positive feedback ensures that agents who lose their initial encounter will be more likely to lose subsequent social competitions, while winners will become yet more likely to win, leading to long-term sorting. In empirical studies, this “competition effect” has been found to result in part from hormonal changes following victory or defeat in interpersonal competitions (Vermeer et al., 2016).

Agents exist on a grid, which in all simulations reported here is fixed at 100 by 100 cells. Agent population can vary, but for initial calibration runs is fixed at 1000. At each time step, agents move randomly on the grid, then take stock of their Moore neighborhood (that is, the eight cells immediately surrounding their own cell). Status in relation to an immediate comparison sample or relative status within a social institution has been found to impact well-being and happiness more strongly than many other variables, including actual income and health care access (Guven & Sørensen, 2012; Marmot, 2005). Additionally, lower rank within a formal status hierarchy often entails reduced freedom and more social constraints on autonomous activity and even on cognition, which can induce serious psychological and even physiological stress (Douglas, 1986; Marmot et al., 1997). To represent such findings parsimoniously, we elected to base agent satisfaction on rank relative to immediate neighbors. To compute their satisfaction, agents compare their own formal social rank against the mean of the ranks of their Moore neighbors. Thus, an agent with a rank of 2 who has three neighbors with ranks 2, 4, and 3 would experience a satisfaction level of −1.

Following comparison of average ranks, male agents additionally interact with their Moore neighborhood by sweeping the neighborhood for same-ranked neighbors. If one is encountered, a probability function based on the total number of formal ranks in the virtual society determines the likelihood that the two agents will engage in a social competition. This function is simply p = 1/r, where p is the probability of competition with a same-ranked neighbor and r = total number of ranks in the current society. Thus, in a totally flat society all encounters between male agents would entail social competition, in line with McCaffree (2015). The minimum number of formal ranks in the model as reported here is two, reflecting the difference in formal status role often assigned to women even in acephalous or politically egalitarian societies (Boehm, 2001). If competition is initiated, victory and defeat are determined by the “victory coefficient” attribute assigned to each agent, as described above. All competitive encounters in the first time step or between male agents that have not yet competed are determined randomly with an equal probability of victory for either agent (since all victory coefficients initially = .5).

Next, agents check their somatization threshold, which represents the level of dissatisfaction at which an agent will begin to exhibit dissociative or somatoform symptoms related to psychosocial stress. Agents that have met or passed their threshold, and that have the requisite predisposition attribute, begin to “dissociate.” This is a boolean variable whose effect, if “yes,” is that other agents perceive that the suffering agent is exhibiting somatization/dissociation symptoms.

Finally, agents sweep their neighborhood and receive feedback from neighbors regarding the proper interpretation of their dissociative symptoms. All agents are assigned an attribute, the “shaman signal,” that indicates the weight of their response to a neighbor’s somatization/dissociation. Higher values for this attribute reflect a stronger social signal that the afflicted agent should become a “shaman.” (For implementation purposes, the word “shaman” in the syntax simply means religious trance practitioner of any kind.) Afflicted agents gather the sum of the shaman signals and compare this sum against the vote threshold, which is a model parameter that sets the level of social feedback agents must receive before deciding to become shamans. If the threshold is met or passed, agents “convert” to shaman status. Lower shaman signal values and higher vote threshold each represent reduced cultural salience of the role of trance practitioner. In a society with an arbitrarily high vote threshold, agents would simply never become shamans because they do not “believe” that such a role exists. In this way, we simulate societal differences in the pervasiveness of shamanic or trance practitioner concepts. Vote counts are cumulative over time steps; an agent that is currently exhibiting dissociative symptoms will recall each time other agents have “told” it that it “ought to become” a shaman. If an agent’s satisfaction level rises again above its somatization threshold, it will stop soliciting votes from neighbors.

If an agent converts to shaman status, its color changes on the visualized cell grid from black to blue if female, and red if male. Moreover, its shaman signal value converts automatically to 10, regardless of its previous value. This change indicates that the opinions of religious trance practitioners themselves carry the heaviest weight in governing the spread of, and initiation into, that role. Finally, shamans increase their social rank following conversion/initiation, but subsequently no longer calculate their satisfaction in regard to their relative social rank.

Figure 1
Figure 1

Individual agent routine

The agent routine for each individual at each time step. Solid boxes are core operations of the routine, while dotted boxes and arrows indicate the algorithm components for each operation.

Citation: Journal of Cognition and Culture 18, 5 (2018) ; 10.1163/15685373-12340042

2.1 Design of Computational Experiment

Since we were interested in the gender distributions of religious trance and shaman roles, we first calibrated the model to maximize the number of female shamans and male shamans, respectively. Model calibration entails requesting that the model produce certain outputs, and allowing it to run many permutations using systematically varied input parameters (Windrum, Fagiolo, & Moneta, 2007). For initial calibrations, we constrained the model to ensure that male agents on average outranked female agents (that is, we required the mode of male ranks to be higher than the mode of female ranks). In Table 2, we report the combinations of parameters that produced simulations with outputs that most closely matched the desired configurations. Columns 3 and 4 represent the results of calibration experiments to determine the parameter combinations that maximized the female-male and male-female ratios of shamans within the population, respectively. For these calibration experiments, environment height, width, and population did not vary. Somatization thresholds were set to be identical for males and females, and the distribution of shaman signal values was uniform (was not allowed to vary across experiments). Thus, this round of calibration experiments focused mostly on changes in the ranking structure of the population, as well as the vote threshold and percent of the population predisposed to somatization.

Table 2
Table 2

Initial calibration experiment results

Citation: Journal of Cognition and Culture 18, 5 (2018) ; 10.1163/15685373-12340042

As Figure 2 illustrates, a flatter ranking structure led to a higher ratio of male trance practitioners to female trance practitioners. However, several assumptions of the initial calibration experiments may not be realistic. Specifically, males and females are assumed to have identical distributions of the somatization threshold value. However, as outlined above, men appear to respond more powerfully via HPA axis and cortisol (stress hormone) response to psychosocial stress than women (Stephens et al., 2016). The fact that sex differences in actual dissociative symptoms appear to be largely context-driven (Spitzer & Freyberger, 2008) is accounted for with the somatization predisposition parameter, which determines what proportion of the population has “the type of brain” that generates recognizable dissociation/somatization symptoms if subjected to sufficient stress and does not differ by agent gender. The actual intensity of the stress response itself, however, does differ between the sexes, and so we ran two additional calibration experiments to maximize the female/male and male/female ratio of shamans/trance practitioners while allowing the somatization threshold parameters to vary across both sexes (Table 3).

Figure 2
Figure 2

Figure 2a shows the rank distributions by gender that are associated with the highest ratio of female shamans to male shamans in calibration runs in which somatization threshold parameters are constrained to be stable and equal between genders. Figure 2b shows the rank distributions by gender that are associated with the highest ratio of male shamans to female shamans in calibration experiments with the same constraints. Vertical axis represents social rank. Overall, flatter and more gender-equal social structure is associated with more male shamans.

Citation: Journal of Cognition and Culture 18, 5 (2018) ; 10.1163/15685373-12340042

Table 3
Table 3

Secondary calibration experiment results

Citation: Journal of Cognition and Culture 18, 5 (2018) ; 10.1163/15685373-12340042

A simulation run using the parameter settings highlighted in the third column of Table 3 produced a male:female shaman ratio of slightly less than 7:1 (86.5% male) over a 20-year simulated time period for a population of 1000. Overall, 60.7% of the male population is a practitioner of religious or healing trance at the end of this period, while only 8.2% of female agents have become trance practitioners. Note that female agents have, on average, a higher tolerance for psychosocial stress than males, as empirical evidence indicates is the case (Stephens et al., 2016).

A simulation run using the parameter settings in the second column of Table 3 produces a female:male shaman ratio of slightly less than 7:5 (56.2% female). Notably, there are no female agents left that are not shamans at the end of the run. Additionally, 82.7% of males are also shamans. Thus, this configuration of input parameters produces a “shaman-drenched” society.

Although the majority of males are indeed trance practitioners in some societies (Katz, 1981) and a substantial proportion of the female (and male) population participates in, for example, the Vodou possession trance religion in Haiti (Bourguignon, 1976), it is probably more common for specialized trance professionals to comprise a relative minority of the total population that hosts them. One desiderata would therefore be to determine whether the model could produce both a given male:female ratio and a majority of non-shamans among both genders, under a realistic configuration of parameters. Further calibration testing with constraints relaxed does point to configurations that meet these criteria, for both high male:female and high female:male shaman ratios (Table 4).

Table 4
Table 4

Parameter configurations for high male:female shaman ratio and low overall shaman population

Citation: Journal of Cognition and Culture 18, 5 (2018) ; 10.1163/15685373-12340042

We calibrated the model to produce the highest possible male:female shaman ratio while eliminating constraints on the relative rank of genders. The resulting configuration produced a society in which more males than females were shamans, but where most agents were not shamans. However, it was unrealistic in that females somewhat outranked males politically. We then reduced the ranking structure of the society to two ranks – one for females and a higher one for males – while holding all other parameters derived from the previous calibration test constant. We consider this the simplest possible realistic political structure for most societies, since political egalitarianism among acephalous societies typically does not include political equality between men and women (Boehm, 2001). The resulting simulation runs produced a society in which male shamans were much more common than female shamans (males constituted 92.5% of shamans), but in which only a small minority of males (12.5%) occupied the shaman role (Figure 3). This pattern better matches many real-world acephalous societies, in which the shaman role is often professionally distinct and not open to everyone (Le Barre, 1972).

Figure 3
Figure 3

Simulated environment for highly egalitarian society

This image demonstrates the ABM simulation interface. Black agents are nonshamans. Red agents are male shamans. Blue agents are female shamans. In a society with realistic parameters governing sex differences in gender (Column 3 of Table 4), only 12.5% of men and 1% of women are shamans at the end of a 20-year simulated period.

Citation: Journal of Cognition and Culture 18, 5 (2018) ; 10.1163/15685373-12340042

Next, we manually created a realistic, relatively shallow social hierarchy by raising the minimum male rank to 3 and the maximum to 5, while raising the maximum female rank to 2. For both genders, the rank mode was the lowest rank, representing the common pyramidal distribution of true social hierarchies. Once again, we held all other parameters constant. This configuration produced a simulated society in which 54.6% of shamans were female, and 12.8% of females were shamans. Overall, 88.1% of the society was non- shamans at the end of a 20-year simulated run. These patterns are a relatively good match for a society such as Haiti, which is characterized by social inequality and hosts a possession trance cult, Vodou, that attracts practitioners of both genders but which is generally slightly tilted toward female participation (Bourguignon, 1976). The difference in social ranks leading to male versus female predominance among trance practitioners matches established data and appears to result from realistic parameter configurations.

3 Discussion

The model presented here represents the changing sex ratios of practitioners of religious trance (“shamans”) based on social structure and psychosocial stress. Given realistic parameter values, we find that in general flatter social structures produce more male trance practitioners, while more hierarchical structures produce more female practitioners. Realistic model outputs appear to be a function of slight differences in agents’ response to social stress by gender, such that male agents are somewhat more sensitive to such stress. This result matches data on differential HPA axis response to psychosocial stressors by sex (Stephens et al., 2016; Kudielka & Kirschbaum, 2005). In the simulation, informal social competition among same-ranked males has a sufficiently strong effect to enable male dominance of trance practices based on cultural construal and channeling of dissociative symptoms. We therefore find that this model accurately represents the underlying theory and indicates that the posited causal processes leading to different gender ratios in trance practice are plausible.

3.1 Limitations

Despite being capable of reproducing posited dynamics under realistic constraints, this model suffers from some limitations. Most importantly, the model tends to be biased toward producing societies dominated by trance practitioners. While some societies do indeed have large numbers of trance practitioners, most do not. The model therefore appears to suffer from a structural bias that somewhat reduces its fidelity to real-world dynamics. No model is perfectly accurate (Winsberg, 2010), but we can realistically posit that some important dynamics are left unaccounted for in the model.

We propose that these missing dynamics are primarily (1) cultural feedback effects and (2) the social functions of trance practice. The present model focuses on individual agents: by becoming trance practitioners, agents find themselves liberated from dissociative symptoms that result from extreme or intractable psychosocial stress. They step outside the social structure, so to speak (Turner, 1975). However, in actual human societies trance practitioners interact with others as trance practitioners – specialists in unsolvable problems who may be asked to heal sicknesses, negotiate with spirits, divine the future, or auger auspicious or inauspicious times to hunt, plant, or marry. Shamans and other trance practitioners thus certainly play a functional role in the communities that host them, even as they are also often self-interested agents attempting to acquire social status or resources in ways otherwise not available to normal people (Kim, 2003). The specific role of shamans or trance practitioners differs by society, and, depending on the centrality of that role, the cognitive accessibility of the concept of “shaman” (or its local equivalent) may be higher, lower, or applicable only to some classes of people. For instance, in many societies men who are gender-nonconforming may be channeled into a trance practitioner role (Bacigalupo, 2007). These gender-nonconforming shamans may subsequently serve a concrete, publicly visible social function – such as using placebo effects and trance to effect healing of certain conditions among clients (McClenon, 1994) – that serves to reinforce, in the minds of the members of the society, the semantic association between men with indeterminate or undefinable social status and the shaman role. As such, a more comprehensive model of the causal pathways by which societies with different social structures produce different gender patterns in trance practices may require more complex agent cognition, governed by rules that enable positive feedback and canalization of social categories over time based on initial conditions.

Overall, the model adequately represents the underlying theory outlined above, and points to future opportunities for research and modeling. The theory itself, meanwhile, is novel in positing that equalitarian social settings in fact uniquely provide the conditions for social stress among males due to endemic competition that is not resolved through achieving settled formal ranks. While formal social hierarchies and institutionalized inequality have well-documented and often dramatic negative effects on people of both sexes (Marmot, 2005), the present model suggests that less-structured settings may plausibly have stressful impacts on certain categories of people, leading to the possibility of a longitudinal semantic association between members of those social categories and dissociative trance. Finally, the model does not attempt to account for the distinction between spirit possession trance and trance without possession. That is, agents in the present model do not take part in simulated “trance” practices; rather, they engage in social interactions within a bounded population, and those interactions generate changes in the social identity of some agents, such that they enter a shaman or trance practitioner role.

3.2 Theoretical Points

This line of inquiry is not intended to reduce shamanism or trance religions to either functionalist or pathological causes. We do not claim that trance religions are fully explicable either as products of rational social or interpersonal needs, or as merely culture-bound syndromes through which psychological pathologies or psychoses are expressed (Boddy, 1994). Neither do we claim that the religious dimensions of these practices are merely epiphenomenal on their social or psychological factors, or are somehow peripheral to their etiology or practice. In the real world, trance practitioners often report that trance religions and possession cults are primarily religious, not therapeutic, in nature, and any model of their role in human affairs must acknowledge this fact (Claus, 1979; Csordas, 1982). In other words, we did not set out to drown the phenomenology of trance cults, much less the actual facts of their practice and structure, in a bath of etic or formal explanation. Rather, we intended to investigate the causal factors behind apparent, predictable divergences in the specific forms these cults take in societies with distinct types of structural profiles, specifically in regard to gender. Our model is thus reductive, as it must be, only in regards to a delimited facet of the much broader phenomenon of trance-based religious and healing practices. We hope to generate novel explanatory accounts of puzzling, well-attested patterns in cross-cultural data, with the aim of providing ethnologists and ethnographers with new heuristic tools that they may test against their data and retain or reject as appropriate. Abstraction should not replace particularistic inquiry, but augment and enable it.

To that point, it bears emphasis that the theoretical scaffolding that lies beneath this model depends on two important conceptual tools from 20th-century anthropology that could be profitably integrated into 21st-century discourse in the cognitive science of religion. The first of these is Victor Turner’s (1975; 1996) concept of “anti-structure,” or a (temporary or permanent) state of inversion of normative social structure and its hierarchically arranged roles – often exemplified in marginality within a formal social structure with discrete roles, or (either temporal or permanent) placement between roles, as in liminal rites (Van Gennep 1961). In Turner’s schema, persons peripheral to the social structure, those of very low social status, and those temporarily between statuses (such as during an initiation rite) are characterized by a lack of formal integration into the social structure and therefore are bearers of anti-structure. Shamans and trance cult leaders are often described as anti-structural figures (e.g., Natvig, 1988), but excessive popularization of this claim – as in, for example, the widespread use of “wounded healer” tropes in countercultural discourse (Halifax, 1988) – has muddied the waters and thereby somewhat reduced the perceived utility, not to say cachet, of Turner’s typology for specialized inquiry. We wish to rehabilitate the structure/anti-structure distinction by highlighting its utility for understanding the social and neurological dynamics that shape the gendered expression of trance religion in distinct contexts. Given the centrality of social structural and hierarchical factors for influencing human behavior (Boehm, 2001; Chaudhary et al., 2016) and even health outcomes (Marmot, 2005), Turner’s focus on how religion and ritual are used to negotiate the vagaries of normative structural pressure makes for an important heuristic tool.

Second, Mary Douglas’s (1970; 1999) “grid-group” typology provides a useful lens for understanding the different dimensions in which societies or communities may exert social control and shape their members’ basic cognitive schemas. According Douglas, high “group” societies are those with strict boundaries between the in-group and other groups, while high “grid” societies are those in which individual behavior, roles, and categorization schemes are highly prescribed and enforced according to a coherent set of norms. These two dimensions often covary – high-grid societies are often high-group societies, and vice-versa – but this correlation is not universal. For example, low-grid but high-group societies, such as antinomian or countercultural enclaves, are common within modern, complex cultures (Douglas, 1999). High-grid and high-group societies are generally collectivistic in cultural psychological terms, while societies lower on both these dimensions are more individualistic (Triandis, 1995).2 Crucially, Douglas emphasized that, among more individualist, egalitarian societies (low-grid societies), the “form of organization is competitive, with dominant positions open to merit” Douglas, 1999, p. 411, emphasis mine).

It is important that the group-grid typology was originally conceived in order to make sense of cross-cultural African ethnographic data regarding ancestor cults and witchcraft beliefs. Not all African societies have the same religious profiles; indeed, as Bourguignon’s (1973) and Greenbaum’s (1973) work demonstrates, the diversity between African societies is remarkable. However, their work also shows that there are reliable patterns in this diversity of data, such that social structure, hierarchy, and prescriptive rigidity are important predictive factors for religious variables. Thus, there is good reason to think that structurally flat societies may be characterized not solely by Edenic egalitarianism, but also by persistently unresolved interspecific competition and unstable, churning informal status hierarchies. That is, in a social environment that lacks formal role assignments but which differentially rewards high informal status and competitive prowess – a description that aptly characterizes many hunter-gatherer societies – status-seekers may be obliged to continually re-assert, defend, or attempt to attain desirable positions. Douglas’s theory illuminates, and enjoins our attention to, the causal principles that underlie these patterns.

4 Conclusion

Dissociative experiences and the cultural interpretations that are overlain upon them are vital for understanding trance religions, and probably many aspects of what we would term religious life in general. Culturally mediated and interpreted, dissociative responses to psychosocial stress can become spotlights, shining on the fractures in a social order or the places where dominant cognitive schemas break down, lose their moral legitimacy, or are simply no longer useful for making sense of suffering. As Erika Bourguignon wrote,

Dissociation, in its ritualized form as possession trance – as well as in visionary trance – is a coping strategy that, as an institutionalized practice, signals the presence of critical stresses within the society.

2004, p. 571

Thus, while active shamans and other trance practitioners are generally no more likely to suffer from mental illness than average, their entrance into their professional role is likely often preceded, or even precipitated, by psychological or psychosocial traumas or stressors and the spontaneous dissociative and trance states that emerge from them (Merchant, 2012). This account of the complex relationship between psychological symptomology and trance practice is supported by the fact that dissociative symptoms differ in their relationship to trauma between trance practitioners and non-practitioners. For instance, among Israeli women with similar levels of traumatic past experience, women who were experienced spirit channelers showed both higher levels of dissociation and better psychological well-being than non-channelers (Stolovy et al., 2015). Meanwhile, non-trance spirit possession – that is, psychological disturbances that are not sought out through intentional trance and are emically attributed to the unwanted intrusion of culturally posited supernatural agents – is strongly associated worldwide with exposure to, and dissociative responses to, trauma (Hecker et al., 2015; Ross, 2011; Somer et al., 2015) Undesired spirit possession often precedes entry into the role of trance professional in many societies, and is often a component of “shamanic illness” (Boddy, 1989; Clark, 2006; Lewis, 1971). Trance religions and shamanic practices may, then, provide avenues for the normative or constructive cultural interpretation of dissociation and traumatic events, reducing the overall incidence of pathology (Luhrmann, 2007).

Dissociative trance disorder has, then, been described as “a global idiom of distress” (During et al., 2001, p. 235), and the association between non- voluntary possession, trauma, and dissociative trance suggests that this idiom can become very sharp indeed. The present model has attempted to demonstrate how exposure to serious psychosocial trauma may differ according to social configurations that represent Douglas’s (1970) grid-group typology, and by gender – and, subsequently, how such exposure can translate into different patterns for gender roles in the participation in trance and possession religions. The model undoubtedly leaves aspects of the real-world dynamics unaccounted for, yet its ability to replicate recognizable patterns under realistic settings lends the theoretical story behind it credibility. When the Durkheimian social structure, with all its moral force and structural sway, loses its hold on people, they search behind that structure for answers. Trance and possession are crucial modes of carrying out that search, whether it is men or women – or both together – that the structure carelessly pushes past their psychic limits.

References

  • Akyüz, G., Doǧan, O., Şar, V., Yargiç, L. İ., & Tutkun, H. (1999). Frequency of dissociative identity disorder in the general population in Turkey. Comprehensive Psychiatry, 40(2), 151159. https://doi.org/10.1016/S0010-440X(99)90120-7

    • Search Google Scholar
    • Export Citation
  • Bacigalupo, A. M. (2007). Shamans of the Foye Tree: Gender, Power, and Healing Among Chilean Mapuche. University of Texas Press.

  • Barre, W. L. (1972). Ghost Dance: Origins of Religion (First Edition edition). London: Allen & Unwin.

  • Barsky, A. J., Peekna, H. M., & Borus, J. F. (2001). Somatic Symptom Reporting in Women and Men. Journal of General Internal Medicine, 16(4), 266275. https://doi.org/10.1046/j.1525-1497.2001.00229.x

    • Search Google Scholar
    • Export Citation
  • Bhavsar, V., Ventriglio, A., & Bhugra, D. (2016). Dissociative trance and spirit possession: Challenges for cultures in transition. Psychiatry and Clinical Neurosciences. https://doi.org/10.1111/pcn.12425

    • Search Google Scholar
    • Export Citation
  • Björkqvist, K. (2001). Social defeat as a stressor in humans. Physiology & Behavior, 73(3), 435442.

  • Bob, P., Fedor-Freybergh, P., Jasova, D., Bizik, G., Susta, M., Pavlat, J., & Raboch, J. (2008). Dissociative symptoms and neuroendocrine dysregulation in depression. Medical Science Monitor, 14(10), CR499504.

    • Search Google Scholar
    • Export Citation
  • Boddy, J. (1989). Wombs and Alien Spirits: Women, Men, and the Zar Cult in Northern Sudan (1 edition). Madison, Wis: University of Wisconsin Press.

    • Search Google Scholar
    • Export Citation
  • Boehm, C. (2001). Hierarchy in the forest: the evolution of egalitarian behavior. Cambridge, Mass.: Harvard University Press.

  • Bonabeau, E. (2002). Agent-based modeling: Methods and techniques for simulating human systems. Proceedings of the National Academy of Sciences, 99(suppl 3), 7280–7287. https://doi.org/10.1073/pnas.082080899

  • Borshchev, A. & Filippov, A. (2004). From system dynamics and discrete event to practical agent based modeling: reasons, techniques, tools. Proceedings of the 22nd International Conference of the System Dynamics Society, Oxford, UK.

  • Bourguignon, E. (Ed.). (1973). Religion Altered States of Consciousness and Social Change. Columbus: Ohio State Univ Pr.

  • Bourguignon, E. (1991). Possession. Prospect Heights, Ill: Waveland Pr Inc.

  • Bourguignon, E. (2004). Suffering and Healing, Subordination and Power: Women and Possession Trance. Ethos, 32(4), 557574. https://doi.org/10.1525/eth.2004.32.4.557

    • Search Google Scholar
    • Export Citation
  • Bourguignon, E., & Evascu, T. L. (1977). Altered States of Consciousness within a General Evolutionary Perspective: A Holocultural Analysis. Cross-Cultural Research, 12(3), 197216. https://doi.org/10.1177/106939717701200303

    • Search Google Scholar
    • Export Citation
  • Chaudhary, N., Salali, G. D., Thompson, A., Rey, A., Gerbault, P., Stevenson, E. G. J., & Migliano. (2016). Competition for Cooperation: variability, benefits and heritability of relational wealth in hunter-gatherers. Scientific Reports, 6, 29120. https://doi.org/10.1038/srep29120

    • Search Google Scholar
    • Export Citation
  • Chaudhary, N., Salali, G. D., Thompson, J., Dyble, M., Page, A., Smith, D., & Migliano, A. B. (2015). Polygyny without wealth: popularity in gift games predicts polygyny in BaYaka Pygmies. Royal Society Open Science, 2(5), 150054. https://doi.org/10.1098/rsos.150054

    • Search Google Scholar
    • Export Citation
  • Clark, R. (2006). The Multiple Natural Origins of Religion (1 edition). Oxford; New York: Peter Lang AG, Internationaler Verlag der Wissenschaften.

    • Search Google Scholar
    • Export Citation
  • Claus, P. J. (1979). Spirit possession and spirit mediumship from the perspective of Tulu oral traditions. Culture, Medicine and Psychiatry, 3(1), 2952.

    • Search Google Scholar
    • Export Citation
  • Csordas, T. J. (1987). Health and the holy in African and Afro-American spirit possession. Social Science & Medicine (1982), 24(1), 111.

    • Search Google Scholar
    • Export Citation
  • Delisle, V. C., Beck, A. T., Dobson, K. S., Dozois, D. J. A., & Thombs, B. D. (2012). Revisiting Gender Differences in Somatic Symptoms of Depression: Much Ado about Nothing? PLOS ONE, 7(2), e32490. https://doi.org/10.1371/journal.pone.0032490

    • Search Google Scholar
    • Export Citation
  • Devereux, G. (1961). Shamans as Neurotics. American Anthropologist, 63(5), 10881090. https://doi.org/10.1525/aa.1961.63.5.02a00110

  • Douglas, M. (1996). Natural Symbols: Explorations in Cosmology (2 edition). London; New York: Routledge.

  • Douglas, M. (1999). Four cultures: the evolution of a parsimonious model. GeoJournal, 47(3), 411415. https://doi.org/10.1023/A:1007008025151

    • Search Google Scholar
    • Export Citation
  • During, E. H., Elahi, F. M., Taieb, O., Moro, M.-R., & Baubet, T. (2011). A critical review of dissociative trance and possession disorders: etiological, diagnostic, therapeutic, and nosological issues. Canadian Journal of Psychiatry. Revue Canadienne De Psychiatrie, 56(4), 235242.

    • Search Google Scholar
    • Export Citation
  • Eliade, M. (2004). Shamanism: Archaic Techniques of Ecstasy. (W. R. Trask, Trans.) (Later Reprint edition). Princeton, NJ: Princeton University Press.

    • Search Google Scholar
    • Export Citation
  • Epstein, A., & Axtell, R. (1996). Growing artificial societies: social science from the bottom up. Washington, D.C.: Brookings Institute Press.

    • Search Google Scholar
    • Export Citation
  • Gennep, A. van, Vizedon, M. B., & Caffee, G. L. (1961). The Rites of Passage (Reprint edition). Chicago: University of Chicago Press.

  • Greenbaum, L. (1973). Possession Trance in Sub-Saharan Africa: A Descriptive Analysis of Fourteen Societies. In Religion Altered States of Consciousness and Social Change (pp. 5887). Columbis, Ohio: Ohio State Univ Pr.