Self-denial by shamans promotes perceptions of religious credibility

Abstract

Religious leaders refrain from sex and food across human societies. Researchers argue that this self-denial promotes perceptions of credibility, invoking such associations in explanations of shamanism and prosocial religion, but few, if any, quantitative data exist testing these claims. Here we show that shamans in a small-scale society observe costly prohibitions and that observers infer three dimensions of credibility-cooperativeness, religious belief, and supernatural power-from shamans' self-denial. We investigated costly prohibitions on shamanic healers, known as sikerei, among the rainforest horticulturalist Mentawai people of Siberut Island. We found that shamans must observe permanent taboos on various animals, as well as prohibitions on sex and food during initiation and ceremonial healing. Using vignettes, we evaluated Mentawai participants' inferences about self-denial, testing three mechanisms of religious credibility: cooperative costly signaling, credibility-enhancing displays, and supernatural otherness. We found support for all three mechanisms: Mentawai participants infer self-denying shamans to be (1) cooperative, (2) sincere believers in the religious rules, and (3) dissimilar from normal humans and with greater supernatural powers. These data provide novel, quantitative evidence that self-denial develops to promote perceptions of practitioners' religious credibility.

Full text

Self-denial by shamans promotes perceptions of religious
credibility
Manvir Singh* and Joseph Henrich
Department of Human Evolutionary Biology, Harvard University, Cambridge, MA 02138
*Corresponding author: manvirsingh@fas.harvard.edu
3 July 2019
Abstract
Religious leaders refrain from sex and food across human societies. Researchers argue that this
self-denial promotes perceptions of credibility, invoking such associations in explanations of
shamanism and prosocial religion, but few, if any, quantitative data exist testing these claims.
Here we show that shamans in a small-scale society observe costly prohibitions and that
observers infer three dimensions of credibility – cooperativeness, religious belief, and
supernatural power – from shamans’ self-denial. We investigated costly prohibitions on shamanic
healers, known as sikerei, among the rainforest horticulturalist Mentawai people of Siberut
Island. We found that shamans must observe permanent taboos on various animals, as well as
prohibitions on sex and food during initiation and ceremonial healing. Using vignettes, we
evaluated Mentawai participants’ inferences about self-denial, testing three mechanisms of
religious credibility: cooperative costly signaling, credibility-enhancing displays, and supernatural
otherness. We found support for all three mechanisms: Mentawai participants infer self-denying
shamans to be (1) cooperative, (2) sincere believers in the religious rules, and (3) dissimilar from
normal humans and with greater supernatural powers. These data provide novel, quantitative
evidence that self-denial develops to promote perceptions of practitioners’ religious credibility.

p. 2
Introduction
Religious self-denial, whether enforced or apparently voluntary, is ubiquitous. The renunciation
of sex, valuable food, or social contact has been practiced by Catholic priests, Buddhist and Jain
clergy, Taoist monks and nuns, Christian monastic sects, and many other practitioners who
devote themselves to communing with the divine (Bell & Sobo, 2001; Gross, 1992).
Importantly, religious self-denial extends beyond world religions. For example, Winkelman and
White (1987) coded ethnographic information about trance practitioners, or shamans
(Hultkrantz, 1993; Lewis, 2003; Singh, 2018), in a representative sample of human societies.
They found that shamans abstained from food, sex, or social contact in 83% of the societies
coded.
Why should religious leaders observe such costly prohibitions, especially given that they
can exploit their position to devise self-serving rules (Singh, Glowacki, & Wrangham, 2016;
Singh, Wrangham, & Glowacki, 2017)? One common proposal is that leaders’ self-denial
promotes religious credibility (Henrich, 2009; Norenzayan et al., 2016; Singh, 2018; Sosis &
Alcorta, 2003), although such claims remain largely untested. By “credibility”, we mean an
individual’s believability. Credibility is contingent on factors such as trustworthiness and
perceived expertise.
In this research, we tested three hypotheses connecting self-denial to credibility:
cooperative costly signaling (Sosis & Alcorta, 2003), credibility-enhancing displays (Henrich,
2009), and supernatural otherness (Eliade, 1964; Mauss, 2001). Each of these hypotheses posits
that observers infer from religious self-denial some trait about the practitioner – respectively,
cooperativeness, religious belief, and supernatural power.
According to the cooperative costly signaling hypothesis, costly behaviors serve as reliable
signals of an actor’s cooperative intent (Bulbulia & Sosis, 2011; Gintis, Smith, & Bowles, 2001;
Irons, 2001; Sosis & Alcorta, 2003). For example, costly behaviors might indicate that an
individual is committed to cooperating with members of a particular group (Posner, 2000; Sosis,
2006) or that they believe in a religious system which includes cooperative doctrines (Henrich,
2009). Researchers have examined whether behaviors like possession and donating to charities
act as costly signals of cooperativeness (Hall, Cohen, Meyer, Varley, & Brewer, 2015; Power,
2017a, 2017b), but few if any quantitative studies have tested whether this hypothesis explains
the ascetic practices of religious leaders.
Aside from broadcasting cooperativeness, self-denial may also indicate the sincerity of
one’s belief. Developing this intuition, researchers have modeled how costly actions, termed
credibility enhancing displays (CREDs), coevolve culturally with certain beliefs (Henrich, 2009;
Wildman & Sosis, 2011). CREDs function as cues to social learners that a cultural model (i.e.,
someone from whom social information is learned) genuinely subscribes to the beliefs he or she
espouses. For example, consider a model who tells a learner that it is safe to consume some
mushroom. In response to a learner’s reasonable apprehension, the model can enhance their
credibility by taking a bite of the mushroom – that is, by engaging in a behavior that would be
sensible only if the model genuinely believes the beliefs they espouse.
Researchers argue that CREDs culturally evolve to ensure the spread of other beliefs. For
instance, recent work suggests that cultural evolution may have favored costly, religious displays
to instill stronger beliefs in a powerful, moralizing god (Norenzayan et al., 2016). By this logic,

p. 3
religious leaders abstain from food and sex because it makes them more effective as transmitters
of the faith.
A growing body of research provides evidence for CREDs in different informational and
behavioral domains (e.g., Kraft-Todd, Bollinger, Gillingham, Lamp, & Rand, 2018; Willard &
Cingl, 2017). When betting on the validity of stories, subjects who witness others put down
money are more likely to do the same (Willard, Henrich, & Norenzayan, 2016). US internet-
users who recall their caregivers acting according to religious virtues are more likely to believe in
God and be certain of that belief (Lanman & Buhrmester, 2017). Despite these advances, no
work has tested whether a CREDs model explains the severe, religious restrictions of religious
leaders.
Crucially, some behaviors can serve as both CREDs and cooperative costly signals
(Bulbulia & Sosis, 2011), such as if a person demonstrates that they believe in a moralistic,
punishing god. However, not all cooperative costly signals are CREDs – because a behavior
might signal cooperativeness through channels other than belief – and not all CREDs are
cooperative costly signals – because, among other reasons, there are many beliefs that a person
can hold (such as that a mushroom is safe to eat) that do not make the believer more cooperative.
A third explanation for why practitioners observe costly prohibitions is what we refer to
here as supernatural otherness. According to this hypothesis, observers infers that someone is
self-denies is different from normal humans. This supposed difference makes it more conceivable
that the self-denial has superhuman powers, such as healing or divine contact (Mauss, 2001;
Singh, 2018). Researchers have not specified how or why observers should understand self-
deniers to be distinct from normal humans, but potential mechanisms include perceptions that
deniers have strange preferences or unique cognitive abilities enabling their abstention – or that
they undergo some kind of transformation as a result of the denial, such as becoming purer
(Smith, 2007).
To test whether any or all of these hypotheses explains why religious leaders observe
costly prohibitions, we investigated taboos on shamans among the Mentawai of Siberut Island
(Indonesia). Examining these dynamics among the Mentawai offers at least two advantages
compared to research in industrialized populations with Abrahamic religions. First, as a belief
system including, among other attributes, shamanism and an animist worldview (Loeb, 1929b,
1929c; Schefold, 1988), Mentawai religion shares characteristics with the traditional religions of
small-scale societies around the world, especially those of hunter-gatherers (Peoples, Duda, &
Marlowe, 2016). Studying Mentawai religion thus permits generalizations to a diversity of
contexts, most of which are less relevant when examining the centralized religions of complex
societies (Boyer & Baumard, 2016). Second, religious systems include many practices and beliefs,
some of them recent variations due to cultural drift, others more functionally important and
stable over time (Currie & Mace, 2014; Rogers & Ehrlich, 2008). The unique arrangement of
Siberut Island, which contains many cultural regions that differ slightly in their practices and
beliefs, allows us to identify those taboos on shamans that are shared across cultural regions and
thus more likely to be functionally important.
Mentawai and sikerei
The Mentawai of Siberut Island are forest-dwelling sago-horticulturalists who live in river valleys

p. 4
separated by hilly expanses of forest (Tulius, 2012) (Fig. 1A). At least 11 major valleys cover the
island, each hosting a set of communities who speak their own dialect and decorate themselves
with unique tattoo motifs. Throughout the rest of this paper, we refer to the set of communities
who reside in the same valley and share a dialect as a cultural region.
Fig. 1. (A) Siberut Island, the largest island of the Mentawai Archipelago (Indonesia). Colored
dots represent different study sites; the legend specifies the villages surveyed with the cultural
region in parentheses. Indonesia is colored light gray in the inset, while other countries are in
dark gray. (B) A Mentawai shaman and his wife.
Missionaries and government programs have transformed the religious lives of people
living elsewhere in the Mentawai Archipelago, but dense tracts of forests have hindered these
efforts on Siberut (Schefold, 1988). This, in combination with struggles by Mentawai clans to
resist these programs, has allowed the traditional religious system to survive relatively intact in
the interior of Siberut, bolstered by a strong shamanic institution. Nevertheless, tourism, the
spread of Islam, settlement agendas by the government, and the expansion of formal education
are rapidly transforming Mentawai social and cultural life (Delfi, 2013, 2017; Hammons, 2010),
making ethnographic investigations invaluable.
Mentawai shamans are a class of men believed to possess the unique ability of seeing
otherwise invisible spirits (Loeb, 1929c). These spirits include the ghosts of ancestors, deities
that cause sickness (e.g., sikaoinan, a water-dwelling spirit, sometimes described as a crocodile,
who punishes stinginess), and human souls, whose departure from the body manifests as illness.
As healers, shamans are experts in herbal medicine and the special songs used for communicating
B A

p. 5
with various spirits. In some cultural regions, shamans are marked by their continued use of the
loincloth and their full-body tattoos (Fig. 1B).
Shamans and their wives are both known as sikerei; other individuals are referred to as
simata (uncooked, unripe, immature). Because only male sikerei are believed to see spirits and
invited to heal illness during healing ceremonies – and because the term is commonly used to
describe those individuals who provide these services – we use sikerei to refer to male sikerei
unless otherwise specified.
Sikerei treat illness in pabetei, healing ceremonies that last from one day to a week.
Families of sick people invite one to six shamans to administer treatments, which can include
providing herbs, sweeping away evil spirits, beckoning a sick person’s soul, and summoning the
punitive water-spirit and removing it from the house. Shamans enter trance during a special
nighttime treatment, known as lajok simagre, during which several shamans dance and summon
beneficent spirits, some of whom possess the practitioners.
A man hoping to become a shaman can start by erecting and inhabiting a small house in
the forest (pulaeat), observing taboos while dedicating himself to raising chickens and pigs. After
days or weeks there, the novice finds a shaman-guru (sipauma) who typically demands pigs,
durian trees, coconut trees, and sago. In return, the guru teaches the novice the herbal remedies
and songs required for healing while treating the trainee’s eyes to help him see spirits. Some
initiates do not move to a forest house; instead, their training begins with a severe, untreatable
illness, perceived by others as a sign that a person must become a shaman.
The sikerei must constantly observe taboos, or keikei (Loeb, 1929b; Schefold, 1988). As
with shamans around the world (Eliade, 1964; Narby & Huxley, 2001), the Mentawai sikerei
must abstain from sex and various food items during initiation and ceremonial periods, in
addition to permanently refraining from several hunted animals. The severity of sex taboos in
particular is captured in the frequent remark that, in becoming a sikerei, one’s wife becomes one’s
sister.
We used costly prohibitions on Mentawai shamans (keikei sikerei) to evaluate whether and
how religious self-denial promotes religious credibility. We first documented those taboos that
apply to shamans and investigated the costliness of a subset of those prohibitions. We then used
vignettes to probe participants’ inferences about self-denying shamans and test the three
hypotheses reviewed earlier.
Study 1: What are the taboos on shamans?
We first documented the activities and food items that are tabooed to shamans during initiation,
during healing ceremonies, and permanently. By examining which prohibitions are shared across
river valleys, we identified the taboos that are most resilient to change and which thus seem most
functionally important (Currie & Mace, 2014).
Methods
We interviewed 88 participants about taboos on shamans across four cultural regions of southern
Siberut (see Figure 1a; Sabirut: n = 20; Sarereiket: n = 27; Silaoinan: n = 21; Taileleu: n = 20),
asking about temporary taboos during initiation and healing as well as permanent taboos that

p. 6
apply through a shaman’s lifetime. One participant was excluded from the permanent dietary
taboo condition because of admitted ignorance. All participants provided informed consent
before the study. The Harvard University Committee on the Use of Human Subjects approved
this study and all others described in this paper.
We collected initiation and healing taboos using free-lists. We collected permanent
dietary taboos, in contrast, using a checklist of 14 items. We developed the checklist after
administering pilot interviews in three cultural regions (Sareireket, Sabirut, Simatalu). Thirteen
of the 14 items in the checklist were those mentioned by more than one respondent during pilot
interviews. The fourteenth (Mentawai langur, Presbytis potenziani) is commonly said to be freely
consumed by shamans and was included to both confirm participant comprehension and to
discourage participants from assuming that all items in the list were tabooed.
We excluded instances when participants specified that an item on the checklist was
tabooed to shamans only during special periods. If a respondent mentioned that an item was
permanently tabooed to a shaman but also temporarily tabooed during initiation and healing, we
only included it as a permanent prohibition. If a participant listed a permanent dietary taboo but
then specified that it only applied during healing ceremonies, we included it only as a periodic
taboo during healing. We categorized taboos at two levels, first grouping similar responses and
then aggregating those taboos within super-ordinate categories, such as taboos pertaining to sex,
eating, and grooming.
We did not use inferential analyses on the free-list data; instead we present the raw
frequencies of commonly cited taboos (see Figure 2 and Supplementary Table S1). The checklist
data, on the other hand, was analyzed with cultural consensus analyses using the AnthroTools
package in R (Purzycki & Jamieson-Lane, 2017).
Results: Taboos during healing and initiation
Figure 2 shows those items that respondents mentioned in at least three cultural regions for at
least one of the two categories of prohibition.

p. 7
Fig. 2. Prohibitions on shamans during initiation and healing ceremonies, according to free-lists
by respondents in four cultural regions of southern Siberut. Rows correspond with responses
from the regions of Sabirut (SAB), Sarereiket (SAR), Silaoinan (SIL), and Taileleu (TAI).
White cells occur when no participants in a cultural region reported a taboo, dark blue cells occur
when all participants reported a taboo, and transitional shades denote intermediate frequencies.
The free-list response columns only include those taboos that were reported in at least three
cultural regions for at least one domain. “Fast intermittently” is labeled with an asterisk because it
is a prescription rather than a prohibition. The five aggregated columns refer to super-ordinate
categories that contain the responses on the left and others; for example, “Cooking/work”
includes “Clear brush for gardening”, “Cut/break”, “Plant”, “Work (general)”, and other work-
related prohibitions that were reported in low frequencies. Raw frequencies appear in
Supplementary Table S1.
Participants reported 54 taboos on shamans during initiation, 13 of which were
mentioned in at least three cultural regions. Seven prohibitions appeared across all of the sites:
eating without self-control, eating sour foods, committing adultery, having sex with one’s spouse,
having sex with anyone, and doing any kind of work (including hunting, repairing a house, or
tending to one’s gardens). To eat with self-control means to eat only when sitting down in a
house with others, ideally when the food being served was prepared at an earlier time. To eat
without self-control, in contrast, is to eat while walking or casually sitting, or to eat food that has
been foraged and immediately prepared, like freshwater fish or taro leaves.
Taboos during healing ceremonies are similar to those observed during initiation.
Participants mentioned 52 prescriptions that apply to shamans during healing. No taboos were
mentioned in only three cultural regions; nine were mentioned in all study regions. As with
taboos on shamans during initiation, taboos during healing ceremonies center on work, food, and
sex. Figure 2 displays the frequencies with which participants listed different prohibitions along
with aggregated frequencies (see Supplementary Table S1).
FREE-LIST RESPONSES AGGREGATED
Clear brush fo r
gardening
Cook
Cut/break
(e.g.,
bamboo, firewood)
Display anger or yell
Eat fern
Diplaziu m
esculentum
Eat sour
Eat without self
-
control
Fast
intermittently*
Have sex
(adultery)
Have sex
(spouse)
Have sex
(general)
Plant
Shower
Work
(general)
Cooking/wo rk
Disturb household
Eat
Groom
Have sex
SAB
INITIATIONHEALING
SAR
SIL
TAI
TAI
SAB
SAR
SIL

p. 8
Results: Permanent dietary taboos
Using cultural consensus analyses, we determined that five animal species are tabooed in all four
cultural regions: eels (Anguilla bicolor), flounders (Pleuronectiformes), gibbons (Hylobates klossii),
the white morph of the simakobu monkey (Simias concolor), and three-striped squirrels (Lariscus
obscurus) (see Fig. 3 and Supplementary Tables S2 and S3). Confirming respondents’ honesty
and comprehension, only one participant of eighty-seven replied that Mentawai langurs are
tabooed, and they specified that this was a special case that required unique circumstances.
Fig. 3. Permanent dietary taboos on shamans (A) and individuals’ preferences for those food
items (B, C, D). Panel A: Participants in four cultural regions of southern Siberut (Sabirut,
Sarereiket, Silaoinan, and Taileleu) answered “yes” or “no” to whether fourteen food items are
permanently tabooed to shamans. White cells occur when no participant reported that an item
was prohibited, dark blue cells occur when all participants reported a prohibition, and
transitional shades denote intermediate frequencies. Cultural consensus analyses identified five
food items as being tabooed across all four regions, indicated in colored boxes. The Mentawai
langur was included as a control because it is freely and commonly consumed by shamans. Panels
B, C, and D: Because of methodological limitations, different tasks were administered to
measure how the prohibited foods ranked in people’s dietary preferences. Panel B shows
respondents’ preferences for twenty-four foraged animals, including the three non-aquatic
species consistently prohibited to shamans (III: gibbon; IV: simakobu monkey [white morph]; V:
three-striped squirrel). Panels C and D show the number of times different river (C) and ocean
(D) animals were named as the most preferred and frequently consumed items; the items
SIL
TAI
SAR
Eel
Anguilla bicolor
Fern
Diplazium
esculentum
Flounder
Pleuronectif ormes
Gibbon
Hyloba tes
kloss ii
Green broadbill
Calyptomena
viridis
Mentawai
squirrel
Callosciurus
melanogaster
Mushroom
Unknown
poliphore
Palm hearts
{e.g. , sago , coconut}
Simakobu
(white morph)
Simias
concolo r
Tar o leav es
Colocas ia
esculenta
Three
-striped squirrel
Lariscus
obscurus
Tur m er ic
Cucuma
longa
Tur t l e
Heosemys
spinosa
Mentawai
langur*
Potenz iani
presbytis
SAB
VIV
III
III
0
4
8
12345678910 11 12 13 14 15 16 17 18 19 20 21 22 23 24
PREFERENCE
0
10
20
30
24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46
FREQU ENCY NAMED
0
20
40
47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71
FREQU ENCY NAMED
III IV
V
I
II
NON-AQUATIC ANIMALS RIVER ANIMALS OCEAN ANIMALS
A
B C D

p. 9
tabooed to shamans were mentioned second most frequently (I: eel) and not at all (II: flounder).
Raw frequencies appear in Supplementary Table S2.
Study 2: How costly are shamans’ dietary taboos?
We found that shamans across southern Siberut observe periodic prohibitions on sex and
unconstrained eating, while they are permanently prohibited from five foraged food items. The
periodic taboos, especially those on sex, appear decisively costly, but the costliness of the
permanent dietary taboos is more unclear. We thus used ranking tasks to establish the relative
costs of abstaining from eels, flounders, gibbons, white simakobu monkeys, and three-striped
squirrels.
Methods
To detect whether and how costly permanent dietary taboos are for shamans, we conducted two
tasks probing Mentawai participants’ dietary preferences.
First, we investigated the costs of giving up non-aquatic, tabooed animals. We assembled
as exhaustive a list as possible of Mentawai wildlife by consulting the appendix of a conservation
plan printed by the Indonesian Ministry of Forestry (PHPA, 1995), collected photographs of
each species mentioned in the report, and convened with focus groups of Mentawai participants
to determine the edibility and local name of each animal. The compiled list excluded all aquatic
species, representing only iba sibara ka leleu (meat of the jungle). After incorporating edible
insects and mollusks, we concluded with a list of 77 consumed, non-aquatic species (74 species
excluding the items commonly tabooed to shamans).
To determine how the three tabooed, non-aquatic food items ranked in comparison with
other non-aquatic, foraged animals, we presented non-shamans (n = 40) with photographs of 24
animals, 21 randomly drawn from the list of 74 animals alongside the three tabooed species
found on land (Hylobates klossii, Simias concolor [white], Lariscus obscurus; see Supplementary
Table S4 for a list of the twenty-four animals). All participants gave informed consent. The
photographs were presented randomly in a grid of 4 photographs by 6 photographs. Participants
divided the 24 items into two groups of 12, one including animals they would never eat again
and one including animals they would continue to eat. They sub-divided these groups again, and
so on and so forth, until they produced eight piles, producing a ranking of foraged items from
“most willing to give up” (ranking = 1) to “least willing to give up” (ranking = 8). We calculated
each item’s mean ranking.
We targeted willingness-to-give-up because this integrates people’s preference for an
animal with its availability. For example, denying oneself a high-quality animal that is rarely
encountered is less costly than rejecting an item that is of slightly lower quality but frequently
encountered. Willingness-to-give-up captures this asymmetry, and participants appeared to
consider both their preference for an item and its availability when making their decisions. In
contrast, questions about which items participants enjoy the most or find the most delicious are
easily swayed by highly-favored items that are frequently tabooed or rarely encountered,
distorting estimates of the cost of prohibition.
Because we could not obtain a list of consumed, aquatic species with accompanying
photographs, we administered a different task to determine the costs of renouncing eels and

p. 10
flounders. We asked the same subjects to name six aquatic, foraged animals, selecting three from
the river and three from the ocean. We specifically asked participants to consider both the
frequency with which they ate those items and their preference for those animals in making their
selections. We identified synonyms and determined each item’s ranking by counting the number
of times it was named.
Results
The items vary considerably in people’s self-reported willingness to give them up (Fig. 3B-D).
Among the 24 foraged land animals, items tabooed to shamans ranked as tenth (Hylobates
klossii), sixteenth (Simias concolor [white]), and twenty-first (Lariscus obscurus) in people’s dietary
preferences (see mean rankings and standard deviations in Supplementary Table S4).
In naming the river species that they enjoyed the most and consumed most frequently,
participants listed the item tabooed to shamans, Anguilla bicolor, second, preferring it to all
freshwater species aside from shrimp. In marked contrast, no participant mentioned flounders
when naming favored saltwater species.
Study 3: What do observers infer about self-denying shamans?
We have established that Mentawai shamans observe costly periodic prohibitions, as well as
permanent dietary prohibitions with more ambiguous or varied costs. We therefore conducted an
experiment in the field to probe whether self-denial by shamans promotes any dimension of
religious credibility, specifically testing the predictions of the cooperative costly signaling,
CREDs, and supernatural otherness hypotheses.
Methods: Participants and procedure
Participants (n = 96) were opportunistically recruited in two villages in the interior of Siberut
Island. All participants gave informed consent.
We presented each participant with two shamans introduced as pretend characters. We
provided background details about the characters (family life and accomplishments) and said that
both characters observed all of the shaman taboos. One shaman was described as refraining from
either sex or valuable food items, while the other was said to have innocuous food preferences.
We counterbalanced the characters’ images and details and randomized the category of self-
denial (food/sex). In the following example, the first character (Aman Dong Dong) refrains from
foraged food items (counter-balanced information is italicized; treatment information is
underlined):
Self-denying character: Aman Dong Dong, here, is a shaman. He follows all of the shaman
taboos. He has two children; he is knowledgeable in making canoes. He does not eat pangolins, Pagai
Island macaques, and flying foxes.
Control character: Aman Paule, here, is also a shaman. He follows all of the shaman taboos. He
has three children; he has a lot of gardens. He does not like to eat chili; he likes to eat cassava leaves.

p. 11
We specified that the shaman abstains from eating pangolins, Pagai Island macaques, and
flying foxes because these animals, like two items tabooed to shamans (gibbons, white simakobu
monkeys) and other large animals (e.g., wild pigs, other primates), constitute a class of hunted
animals afforded special reverence, known as matei keccak (dead soul).
We then asked participants two comprehension questions that also served to prime the
relevant information: (1) Food condition: Who does not eat flying foxes? // Sex condition: Who does
not sleep with their wife every day?; (2) Who does not like to eat chili? If a participant failed, we re-
read the character descriptions. We then asked participants questions about the characters’ belief,
cooperativeness, power, and similarity to other humans (see Supplementary Materials for the full
list of questions). The questions were administered in one of four randomized orders, after which
we again asked the comprehension questions.
Methods: Analysis
We conducted the task with 96 participants. To ensure the quality of the data reported, we only
analyzed the responses of respondents who passed a series of comprehension and attention
checks. First, we excluded subjects who failed the final comprehension check (n = 12). Second,
we excluded participants whose answers contradicted for at least three of four pairs of questions
targeting the same inference (questions beli1 and beli3; coop1 and coop2; powe1 and powe2;
simi1 and simi2 in the Supplementary Materials) (n = 11). Third, we excluded participants who,
in their responses, alternated between the characters for at least 13 of the 14 questions (n = 9)
(e.g., naming the first character, then the second, then the first, and so on). Twenty-six
participants failed at least one check (six failed more than one). Two more participants were
removed from analyses for experimenter error, yielding a final sample size of 68 respondents. See
Supplementary Table S5 for results when including excluded participants; all results remain
significant at a level of p<0.0001, although effect sizes are smaller.
We conducted all statistical analyses in R (R Core Team, 2015). The internal reliability
was high for three of the four traits (Cronbach’s a > 0.75 for belief, cooperativeness, and power),
but lower for similarity (Cronbach’s a = 0.59). We created indices for each of the four traits that
took values from -1 to 1, the magnitude representing the extent to which a participant inferred
the self-denying shaman to exhibit the given trait. Using the lmer function of the lme4 package
(Bates, Maechler, Bolker, & Walker, 2015), we ran a linear mixed regression in which the
domain of the trait inference was used to predict the magnitude of the score with sex,
counterbalanced information, and category of self-denial (food or sex) included as covariates and
random effects for participant ID. We used the lsmeans package (Lenth, 2016) to (1) produce
estimated means; (2) conduct a Wald test (two-sided) testing whether these means differ from 0,
adjusting for multiple comparisons using the Holm-Bonferroni method; and (3) test whether the
means differed from each other, again using the Wald Test (two-sided) and adjusting for
multiple comparisons using Holm-Bonferroni. Significance was defined at an alpha level of 0.05.
To test whether the category of self-denial (food or sex) had an effect on the trait
inferences, we ran a second linear mixed regression, identical to the first except with the category
of self-denial interacting with the domain of trait inference. We used the lsmeans package to
compare means between the estimated trait inferences in both categories of self-denial, adjusting

p. 12
for multiple comparisons using the Holm-Bonferroni method.
Results
Fig. 4 displays the estimated mean scores of the four traits. Respondents rated shamans who self-
denied as being stronger believers in Mentawai religious beliefs (regression coefficient [CR] =
0.59, df = 237.89, 95% CI = [0.41, 0.76]), more cooperative (CR = 0.46, df = 237.89, 95% CI =
[0.29, 0.64]), more supernaturally powerful (CR = 0.42, df = 237.89, 95% CI = [0.25, 0.59]), and
less psychologically and physically similar to normal humans (CR = -0.40, df = 239.27, 95% CI =
[-0.57, -0.22]). We conducted pairwise comparisons among all coefficients to test for
differences. The effect size for similarity was significantly different from all inferred traits, which
is to be expected given that it is negative. Otherwise, there were no significant differences among
the coefficients (see Supplementary Table S6), although the estimated effect for belief is nearly
50% higher than the effect for supernatural power.
Fig. 4. Estimates of inferences about self-denying shamans. Scores of 1 indicate that participants
infer self-denying shamans to always to exhibit some trait, whereas scores of -1 indicate that
participants infer self-denying shamans to never exhibit that trait. Error bars represent 95%
confidence intervals. Asterisks denote p-values; *** <0.0001.
We found no effect for the category of self-denial (see Supplementary Table S7). In other
words, respondents made similar inferences about shamans who refrained from sex as they did
about shamans who refrained from eating various food items.
Discussion
Mentawai taboos impose enduring costs on shamans, barring them from having sex during
SIMILARITY
POWER
COOPERATIVENESS
BELIEF
−1.0 −0.5 0.0 0.5 1.0
***
***
***
***

p. 13
healing ceremonies and permanently forbidding them from consuming valued items, among
many other prohibitions. We found that participants infer shamans who abstain from food and
sex to be more cooperative, sincere in their belief, supernaturally powerful, and psychologically
and physically dissimilar. These findings are consistent with the hypothesis that self-denial by
religious leaders develops to bolster their credibility.
This research is significant for at least three reasons. First, it represents the first
quantitative examination of why religious leaders engage in costly self-denial. Although
evolutionary and psychological researchers have studied practices like possession, ritual scarring,
and church attendance (Hall et al., 2015; Power, 2017a; Sosis, Kress, & Boster, 2007), and
although they commonly posit functional hypotheses for leaders’ celibacy and asceticism
(Henrich, 2009; Norenzayan et al., 2016; Singh, 2018; Sosis & Alcorta, 2003), no study to our
knowledge has tested these accounts. Second, this study provides the first experimental evidence
that costly behaviors can simultaneously garner perceptions of supernatural power and difference
from normal humans. This is consistent with wisdoms about the metamorphic nature of pain
and hardship (Glucklich, 2001; Hoffman & Trawalter, 2016), but it suggests that, whether or
not pain and denial actually change a person, people’s psychological biases predispose them to
interpret those behaviors as transformative. Lastly, this research advances our understanding of
the origins of institutionalized leadership in human societies. Shamans represent the only
professional class in many societies, and their religious authority often translates into leadership
roles beyond the supernatural, such as in organizing economic activity or arbitrating conflict
(Singh, 2018). This research identifies self-denial as a potential mechanism by which shamans
and other practitioners maintain their authority, even in small-scale societies.
Many evolutionary scholars emphasize the costliness of prohibitions (Barker, Power,
Heap, Puurtinen, & Sosis, 2019; Sosis, 2006), but our research indicates that other factors might
influence prohibition. In line with the emphasis on costliness, we found that shamans are
tabooed from sex and freely eating during healing ceremonies and initiations. But we also found
that for permanent dietary taboos, cost seems tangential to whether an item is prohibited.
Consider eels and flounders, both of which are prohibited to shamans. Eels ranked as the second
most frequently-consumed and favored river species. The communities surveyed live along rivers
and regularly fish in them, so this suggests a real, appreciable cost. In contrast, no participant
named flounders when listing their preferred food items. This is not surprising. The Mentawai
living in the interior of Siberut never come into contact with flounders, and in follow-up
conversations, several participants admitted to never having eaten one.
Rather than banning especially valuable food items, these permanent dietary taboos
instead may target peculiar animals. People regard atypical entities as impure or sacred (Douglas,
2002; Leach, 1964; Sperber, 1996), so rejecting them may enhance perceptions of a practitioner’s
difference. In fact, all of the foods tabooed to shamans are known as makatai (bad, broken, evil),
and, with the exception of the three-striped squirrel, all are categorical anomalies. Gibbons are
likened to humans (Schefold, 1972, 1982). The flounder is literally regarded as a split fish (laitak
katsila). The white simakobu is the only white primate on the island. And eels are described as
slippery and, in myths, are compared to snakes (Loeb, 1929a).
What explains prohibitions on both beneficial acts (sex, easy eating) and strange animals?
There are at least two interpretations. First, different prohibitions may evoke different
inferences. For example, the permanent withdrawal from strange animals may serve to indicate
an inherent difference from normal people, enhancing perceptions of supernatural power, while

p. 14
the withdrawal from sex and freely eating might indicate another dimension of religious
credibility, such as cooperativeness or belief. Alternatively, different prohibitions may evoke
similar inferences that vary along a single dimension of prohibition. For instance, the permanent
withdrawal from strange animals may produce the same kinds of inferences as does withdrawal
from sex during healing ceremonies.
Our experimental results support the second interpretation: Different prohibitions evoke
similar inferences. The evidence is that we found no differences in the inferences participants
made about a shaman who refrained from having sex compared to the inferences they made
about a shaman who denied himself hunted animals. Moreover, many participants spontaneously
referred to the self-denying character as makeikei (tabooed), regardless of his taboo. In doing so,
they likened the novel forms of self-denial to other taboos, considering the shamans along a
single dimension of prohibition. These results suggest that participants’ inferences were not
unique to the specific items but instead constituted more generalizable perceptions of people who
self-deny.
Readers should be aware of at least two limitations of our field experiment. First, the self-
denying character engaged in voluntary self-denial beyond the typical restrictions observed by
shamans. All shamans are expected to permanently reject eels, gibbons, and so on, as well as sex
and uncontrolled eating during healing ceremonies – but the treatment character went beyond
these restrictions, abstaining either from other hunted animals or completely from sex. Thus, our
experiment risks capturing participants’ inferences about additional self-denial rather than
inferences about typical prohibitions. Second, and relatedly, the experimental characters rejected
items different from those normally forbidden to Mentawai shamans. Rather than rejecting eels,
gibbons, and so on, the characters in the food-prohibition condition abandoned flying foxes,
pangolins, and Pagai Island macaques. How confident can we be that the inferences people made
about these hypothetical forms of self-denial are equivalent to the inferences they make about
actual abstention?
These points are important to consider, and future research should further probe how
these variables bias people’s inferences. Still, our basic interpretation is justified. The most
conservative interpretation, given these limitations, is that Mentawai participants infer several
dimensions of religious credibility from shamans who self-deny, regardless of whether the
shamans refrain from sex or from eating several notable, large animals.
In sum, we have provided evidence that religious leaders in a small-scale society observe
permanent and periodic prohibitions and, in doing so, garner religious credibility and perceptions
of supernatural powers. This project helps elucidate why figures across time and space, from the
shamans of rainforest horticulturalists to the memorialized prophets of the world’s major
religions, have rejected sex, food, social contact, and other necessities on their journeys toward
religious leadership and apparent divinity.
Acknowledgments: We thank Boroi Oggok (Rustam) Sakaliou, Cameron Curtin, and Steve
Worthington at the Harvard Institute for Quantitative Social Science for their assistance. Luke
Glowacki and Martin Lang provided helpful comments on earlier drafts of this manuscript.
Author Contributions: MS and JH designed the study. MS collected and analyzed data. MS
wrote the manuscript, and both authors edited it.

p. 15
Financial Support: This research was funded by a National Science Graduate Research
Fellowship, a Sheldon Traveling Fellowship from the Harvard Committee on General
Scholarships, and a grant from the Mind, Brain, and Behavior Initiative at Harvard University.

p. 16
References
Barker, J. L., Power, E. A., Heap, S., Puurtinen, M., & Sosis, R. (2019). Content, cost, and
context: A framework for understanding human signaling systems. Evolutionary
Anthropology: Issues, News, and Reviews, (November 2018), 86–99.
https://doi.org/10.1002/evan.21768
Bates, D., Maechler, M., Bolker, B., & Walker, S. (2015). Fitting linear mixed-effects models
using lme4. Journal of Statistical Software, 67(1), 1–48.
Bell, S., & Sobo, E. J. (2001). Celibacy in cross-cultural perspective: An overview. In E. J. Sobo
& S. Bell (Eds.), Celibacy, culture, and society. Madison, WI: The University of Wisconsin
Press.
Boyer, P., & Baumard, N. (2016). The diversity of religious systems across history: An
evolutionary cognitive approach. In T. K. Shackelford & J. R. Liddle (Eds.), The Oxford
handbook of evolutionary psychology and religion. Oxford University Press.
Bulbulia, J., & Sosis, R. (2011). Signalling theory and the evolution of religious cooperation.
Religion, 41(September 2015), 363–388. https://doi.org/10.1080/0048721X.2011.604508
Currie, T. E., & Mace, R. (2014). Evolution of cultural traits occurs at similar relative rates in
different world regions. Proceedings of the Royal Society B, 281(1795), 20141622.
Delfi, M. (2013). Islam and Arat Sabulungan in Mentawai. Al-Jami’ah: Journal of Islamic Studies,
51(2). https://doi.org/10.14421/ajis.2013.512.475-499
Delfi, M. (2017). Local belief system, tatouage, tradition and adaptation in Mentawai. In
Proceedings of the 3rd International Indonesian Forum for Asian Studies (pp. 900–911).
Douglas, M. (2002). Purity and danger: An analysis of concept of pollution and taboo. New York:
Routledge.
Eliade, M. (1964). Shamanism: Archaic techniques of ecstasy. Princeton: Princeton University Press.
Gintis, H., Smith, E. A., & Bowles, S. (2001). Costly signaling and cooperation. Journal of
Theoretical Biology, 213(1), 103–119. https://doi.org/10.1006/jtbi.2001.2406
Glucklich, A. (2001). Sacred pain: Hurting the body for the sake of the soul. Oxford and New York:
Oxford University Press.
Gross, R. L. (1992). The sadhus of India: A study of Hindu asceticism. Jaipur, India: Rawat.
Hall, D. L., Cohen, A. B., Meyer, K. K., Varley, A. H., & Brewer, G. A. (2015). Costly
signaling increases trust, even across religious affiliations. Psychological Science, 26(9), 1368–
1376. https://doi.org/10.1177/0956797615576473
Hammons, C. S. (2010). Sakaliou: Reciprocity, mimesis, and the cultural economy of tradition in
Siberut, Mentawai Islands, Indonesia. University of Southern California.
Henrich, J. (2009). The evolution of costly displays, cooperation and religion: credibility
enhancing displays and their implications for cultural evolution. Evolution and Human
Behavior, 30(4), 244–260. https://doi.org/10.1016/j.evolhumbehav.2009.03.005
Hoffman, K. M., & Trawalter, S. (2016). Assumptions about life hardship and pain perception.
Group Processes & Intergroup Relations, 1368430215625781.
https://doi.org/10.1177/1368430215625781
Hultkrantz, Å. (1993). Introductory remarks on the study of shamanism. Shaman, 1(1–2), 5–16.
Irons, W. (2001). Religion as a hard-to-fake sign of commitment. In R. M. Nesse (Ed.),
Evolution and the capacity for commitment (pp. 292–309). New York: Russell Sage

p. 17
Foundation.
Kraft-Todd, G. T., Bollinger, B., Gillingham, K., Lamp, S., & Rand, D. G. (2018). Credibility-
enhancing displays promote the provision of non-normative public goods. Nature,
563(7730), 245–248. https://doi.org/10.1038/s41586-018-0647-4
Lanman, J. A., & Buhrmester, M. D. (2017). Religious actions speak louder than words:
exposure to credibility-enhancing displays predicts theism. Religion, Brain and Behavior,
7(1), 3–16. https://doi.org/10.1080/2153599X.2015.1117011
Leach, E. (1964). Anthropological aspects of language: Animal categories and verbal abuse. In
E. H. Lenneberg (Ed.), New directions in the study of language (pp. 23–63).
Lenth, R. V. (2016). Least-squares means: The R package lsmeans. Journal of Statistical Software,
69(1), 1–33.
Lewis, I. M. (2003). Ecstatic religion: A study of shamanism and spirit possession (Third Ed).
London and New York: Routledge. https://doi.org/10.4324/9780203241080
Loeb, E. M. (1929a). Mentawei myths. Bijdragen Tot de Taal-, Land-En Volkenkunde van
Nederlandsch-Indië, 85(1), 66–244.
Loeb, E. M. (1929b). Mentawei religious cult. University of California Publications in American
Archaeology and Ethnology, 25(2), 185–247.
Loeb, E. M. (1929c). Shaman and seer. American Anthropologist, 31(1), 60–84.
Mauss, M. (2001). A general theory of magic. London and New York: Routledge.
Narby, J., & Huxley, F. (Eds.). (2001). Shamans through time: 500 years on the path to knowledge.
New York: Penguin Putnam.
Norenzayan, A., Shariff, A. F., Gervais, W. M., Willard, A. K., McNamara, R. A., Slingerland,
E., & Henrich, J. (2016). The cultural evolution of prosocial religions. Behavioral and Brain
Sciences, 39, e1. https://doi.org/10.1017/S0140525X14001356
Peoples, H. C., Duda, P., & Marlowe, F. W. (2016). Hunter-gatherers and the origins of
religion. Human Nature. https://doi.org/10.1007/s12110-016-9260-0
PHPA. (1995). Siberut National Park Integrated Conservation and Development Management Plan,
vol. 3: Appendices. Jakarta, Indonesia.
Posner, E. A. (2000). Law and social norms. Cambridge, Massachusetts: Harvard University
Press.
Power, E. A. (2017a). Discerning devotion: Testing the signaling theory of religion. Evolution
and Human Behavior, 38(1), 82–91. https://doi.org/10.1016/j.evolhumbehav.2016.07.003
Power, E. A. (2017b). Social support networks and religiosity in rural South India. Nature
Human Behavior, 1, 0057. https://doi.org/10.1038/s41562-017-0057
Purzycki, B. G., & Jamieson-Lane, A. (2017). AnthroTools: An R Package for Cross- Cultural
Ethnographic Data Analysis. Cross-Cultural Research, 51(1), 51–74.
https://doi.org/10.1177/1069397116680352
R Core Team. (2015). R: A language and environment for statistical computing. Vienna,
Austria: R Foundation for Statistical Computing. Retrieved from http://www.r-project.org/
Rogers, D. S., & Ehrlich, P. R. (2008). Natural selection and cultural rates of change. Proceedings
of the National Academy of Sciences of the United States of America, 105(9), 3416–3420.
https://doi.org/10.1073/pnas.0711802105
Schefold, R. (1972). Religious involution: internal change, and its consequences, in the taboo
system of the Mentawaians. Tropical Man, 5, 46–81.
Schefold, R. (1982). The culinary code in the puliaijat ritual of the Mentawaians. Bijdragen Tot

p. 18
de Taal-, Land-En Volkenkunde, 138, 64–97. https://doi.org/10.1163/22134379-90003482
Schefold, R. (1988). Lia: Das grosse Ritual auf den Mentawai-Inseln (Indonesien). Berlin: Dietrich
Reimer Verlag.
Singh, M. (2018). The cultural evolution of shamanism. Behavioral and Brain Sciences, 41, e66.
https://doi.org/10.1017/S0140525X17001893
Singh, M., Glowacki, L., & Wrangham, R. W. (2016). Self-interested agents create, maintain,
and modify group-functional culture. Behavioral and Brain Sciences, 39, e30, 40–41.
https://doi.org/10.1017/S0140525X15000242
Singh, M., Wrangham, R. W., & Glowacki, L. (2017). Self-interest and the design of rules.
Human Nature, 28, 457–480. https://doi.org/10.1007/s12110-017-9298-7
Smith, V. (2007). Clean: A history of personal hygiene and purity. Oxford: Oxford University Press.
Sosis, R. (2006). Religious behaviors, badges, and bans: Signaling theory and the evolution of
religion. In P. McNamara (Ed.), Where god and science meet: How brain and evolutionary
studies alter our understanding of religion (Vol. 1).
Sosis, R., & Alcorta, C. (2003). Signaling, Solidarity, and the Sacred: The Evolution of
Religious Behavior. Evolutionary Anthropology, 12(6), 264–274.
https://doi.org/10.1002/evan.10120
Sosis, R., Kress, H. C., & Boster, J. S. (2007). Scars for war: evaluating alternative signaling
explanations for cross-cultural variance in ritual costs. Evolution and Human Behavior,
28(4), 234–247. https://doi.org/10.1016/j.evolhumbehav.2007.02.007
Sperber, D. (1996). Why are perfect animals, hybrids and monsters food for symbolic thought?
Method & Theory in the Study of Religion, 8(2), 143–169.
Tulius, J. (2012). Family stories: oral tradition, memories of the past, and contemporary conflicts over
land in Mentawai - Indonesia. Leiden University.
Wildman, W. J., & Sosis, R. (2011). Stability of groups with costly beliefs and practices. Journal
of Artificial Societies and Social Simulation, 14(3), 6.
Willard, A. K., & Cingl, L. (2017). Testing theories of secularization and religious belief in the
Czech Republic and Slovakia. Evolution and Human Behavior, 38(5), 604–615.
https://doi.org/10.1016/j.evolhumbehav.2017.01.002
Willard, A. K., Henrich, J., & Norenzayan, A. (2016). Memory and belief in the transmission of
counterintuitive content. Human Nature, 27(3), 221–243. https://doi.org/10.1007/s12110-
016-9259-6
Winkelman, M. J., & White, D. (1987). A cross-cultural study of magico-religious practitioners
and trance states: Database. HRAF Research Series in Quantitative Cross-Cultural Data, III.
https://doi.org/10.13140/RG.2.1.4381.2720

Page S1
Self-denial by shamans promotes perceptions of religious
credibility
SUPPLEMENTARY MATERIAL
Manvir Singh and Joseph Henrich
Department of Human Evolutionary Biology, Harvard University, Cambridge, MA 02138
3 July 2019
Experimental Materials
Questions
Question IDs (e.g., beli1, coop1) match those in the dataset
Belief
beli1: For example, there is a ceremony. Of the two of them here, someone eats sour [the consumption
of sour is tabooed during ceremonies and believed to cause misfortune]. Who is it?
beli2: According to you, who believes more in arat sabulungan [Mentawai religion]?
beli3: According to you, who follows taboos less?
Cooperativeness
coop1: According to you, who is a thief?
coop2: According to you, who shares meat more?
coop3: For example, there is a burning house. Who goes to help?
trus1: For example, you are not here because you are working or with family somewhere far. Who do you
look for to help take care of your children here?
trus2: For example, you need personal advice on a family issue. Who do you ask?
Power
powe1: Who has weaker medicine?
powe2: Who has stronger medicine?
powe3: Who has stronger magic?
Similarity
simi1: Whose thoughts are closer to those of a non-shaman?
simi2: Whose thoughts are farther from those of a non-shaman?
simi3: Whose body is closer to that of a non-shaman?

Page S2
Supplementary Tables
Table S1. Raw frequencies with which respondents named taboos on shamans during initiation
and healing across four cultural regions. See Fig. 2 and the main text for details.
FREE-LIST RESPONSES
AGGREGATED
Clear brush for gardening
Cook
Cut/break
Display anger or yell
Eat fern
Eat sour
Eat without self-control
Fast intermittently
Have sex (adultery)
Have sex (spouse)
Have sex (general)
Plant
Shower
Work (general)
Cooking/work
Disturb household
Eat
Groom
Have sex
INITIATION
SAB
0.00
0.00
0.15
0.05
0.15
0.25
0.40
0.00
0.20
0.20
0.30
0.00
0.00
0.10
0.20
0.05
0.65
0.00
0.55
SIL
0.14
0.05
0.05
0.24
0.00
0.38
0.10
0.05
0.19
0.29
0.33
0.00
0.14
0.29
0.33
0.24
0.43
0.14
0.57
SRR
0.12
0.04
0.04
0.04
0.12
0.27
0.15
0.23
0.12
0.19
0.65
0.12
0.12
0.19
0.42
0.04
0.65
0.15
0.88
TAI
0.05
0.05
0.00
0.05
0.05
0.20
0.30
0.15
0.25
0.25
0.20
0.25
0.05
0.35
0.45
0.05
0.70
0.10
0.55
HEALING
SAB
0.05
0.00
0.05
0.15
0.05
0.10
0.40
0.00
0.05
0.15
0.35
0.05
0.10
0.05
0.15
0.15
0.50
0.10
0.45
SIL
0.24
0.00
0.00
0.29
0.00
0.10
0.05
0.00
0.14
0.29
0.24
0.05
0.00
0.48
0.67
0.29
0.19
0.10
0.57
SRR
0.19
0.04
0.08
0.08
0.15
0.12
0.15
0.00
0.19
0.08
0.38
0.04
0.00
0.15
0.46
0.12
0.50
0.00
0.65
TAI
0.15
0.00
0.10
0.05
0.00
0.05
0.15
0.05
0.15
0.20
0.45
0.15
0.00
0.15
0.35
0.05
0.30
0.00
0.75

Page S3
Table S2. Raw frequencies with which respondents reported dietary taboos on shamans across
four cultural regions. See Fig. 3 and the main text for details.
Eel
Anguilla bicolor
Fern
Diplazium esculentum
Flounder
Pleuronectiformes
Gibbon
Hylobates klossii
Green broadbill
Calyptomena viridis
Mentawai squirrel
Callosciurus melanogaster
Mushroom
Unknown poliphore
Palm hearts
{e.g., sago, coconut}
Simakobu (white morph)
Simias concolor
Taro leaves
Colocasia esculenta
Three-striped squirrel
Lariscus obscurus
Turmeric
Cucuma longa
Turtle
Heosemys spinosa
Mentawai Langur*
Potenziani presbytis
SAB
1.00
0.89
0.95
0.89
0.16
0.68
0.42
1.00
1.00
0.58
0.84
0.74
0.79
0.00
SIL
1.00
1.00
0.86
1.00
0.48
1.00
0.19
1.00
0.95
0.24
1.00
0.10
1.00
0.05
SRR
1.00
0.27
0.92
0.85
0.62
0.15
0.08
0.31
0.96
0.19
0.96
0.04
0.23
0.00
TAI
0.95
1.00
0.90
1.00
0.35
0.95
0.90
1.00
0.95
1.00
1.00
0.90
0.95
0.00

Page S4
Table S3. Probabilities that different items are tabooed in the four cultural regions according to cultural consensus analyses.
Eel
Anguilla bicolor
Fern
Diplazium esculentum
Flounder
Pleuronectiformes
Gibbon
Hylobates klossii
Green broadbill
Calyptomena viridis
Mentawai squirrel
Callosciurus melanogaster
Mushroom
Unknown poliphore
Palm hearts
{e.g., sago, coconut}
Simakobu (white morph)
Simias concolor
Taro leaves
Colocasia esculenta
Three-striped squirrel
Lariscus obscurus
Turmeric
Cucuma longa
Turtle
Heosemys spinosa
Mentawai langur*
Potenziani presbytis
SAB
>0.99
>0.99
>0.99
>0.99
<0.01
>0.99
0.47
>0.99
>0.99
>0.99
>0.99
>0.99
>0.99
<0.01
SIL
>0.99
>0.99
>0.99
>0.99
0.77
>0.99
<0.01
>0.99
>0.99
<0.01
>0.99
<0.01
>0.99
<0.01
SRR
>0.99
<0.01
>0.99
>0.99
>0.99
<0.01
<0.01
<0.01
>0.99
<0.01
>0.99
<0.01
<0.01
<0.01
TAI
>0.99
>0.99
>0.99
>0.99
<0.01
>0.99
1.00
>0.99
>0.99
>0.99
>0.99
>0.99
>0.99
<0.01

Page S5
Table S4. The twenty-four nonaquatic, foraged animals presented to subjects. Items tabooed to shamans in all four regions are shaded.
See also Fig. 3.
Latin
English
Mentawai
Mean ranking
Standard deviation
Anthracoceros coronatus
Malabar pied hornbill
kailaba
6.46
1.39
Ardea sumatrana
Great-billed heron
meccau
2.28
1.32
Calyptomena viridis
Green broadbill
luikluik
4.59
1.76
Centropus sinensis
Greater coucal
kemut
5.13
1.59
Ceyx erithacus
Three-toed kingfisher
sikoplaitak
1.28
0.60
Cuculux fugax
Hawk-cuckoo
buccit
4.28
1.39
Cuora amboinensis
Box turtle
lokipat
3.10
1.63
Dicaeum cruentataum
Scarlet-backed flowerpecker
ritdit
3.73
1.74
Dicaeum trigonistigma
Orange-bellied flowerpecker
dhadhatdhat
4.36
2.05
Geomyda spinosa
Spiny turtle
toulu
3.70
1.71
Hylarana nicobariensis
Nicobar golden-backed frog
loloakkek
4.05
1.88
Hylobates klossii
Kloss's gibbon
bilou
4.93
1.95
Lariscus obscurus
Three-striped squirrel
soksak
2.50
1.85
Macaca pagensis
Pagai macaque
obaketa
5.88
1.81
Occidozyga laevis
Puddle frog
utetsopak
6.63
1.41
Paradoxurus hermaphroditus siberu
Siberut palm civet
lamusek
3.45
2.18
Phaenicophaeus curvirostris
Chestnut-breasted malkoha
koitkot
5.83
1.41
Pycnonotus melanoleucos
Black-and-white bulbul
pusitattat
4.59
1.87
Pycnonotus plumosus
Olive-winged bulbul
rotdot
5.56
1.70
Rhynchophorus ferrugineus
Sago palm weevil (larvae)
tamara
6.40
2.37
Simias concolor
Simakobu monkey (black morph)
simakobuk
6.90
1.37
Simias concolor
Simakobu monkey (white morph)
simabulau
3.90
1.57
Sturnus sturninus
Daurian starling
ngorut
6.88
1.38
[Tadpole]
Tadpole
kalabbok
1.48
0.75

Page S6
Table S5. Comparison of coefficients with and without data exclusion. Standard errors are
displayed in parentheses. See Supplementary Table 5 and the main text for statistical details.
Asterisks denote p-values; *** <0.0001.
With data exclusion
Without data exclusion
Belief
0.586***
(0.088)
0.446***
(0.075)
Cooperativeness
0.462***
(0.088)
0.368***
(0.075)
Power
0.419***
(0.088)
0.313***
(0.075)
Similarity
-0.397***
(0.089)
-0.341***
(0.076)

Page S7
Table S6. Pairwise comparisons among coefficients (estimated trait inferences).
Estimated
difference
SE
df
t
p
Adjusted p
(Holm)
Belief-Power
0.166
0.113
200.2
1.48
0.45
1.0
Belief-Similarity
0.983
0.113
200.8
8.69
<0.0001
<0.0001
Belief-Cooperativeness
0.123
0.113
200.2
1.10
0.69
1.0
Power-Similarity
0.816
0.113
200.8
7.22
<0.0001
<0.0001
Power-Cooperativeness
-0.043
0.113
200.2
-0.38
0.99
1.0
Similarity-
Cooperativeness
-0.860
0.113
200.8
-7.60
<0.0001
<0.0001

Page S8
Table S7. Estimated differences between trait inferences when self-denial targeted food
compared to when it targeted sex.
Estimated difference
SE
t
p
Adjusted p (Holm)
Belief
-0.081
0.166
-0.49
0.63
0.94
Cooperativeness
-0.212
0.166
-1.28
0.20
0.61
Power
-0.120
0.166
-0.73
0.49
0.94
Similarity
0.243
0.168
1.45
0.15
0.59