Content uploaded by Benjamin Grant Purzycki
All content in this area was uploaded by Benjamin Grant Purzycki on Oct 21, 2022
Content may be subject to copyright.
Full Terms & Conditions of access and use can be found at
Religion, Brain & Behavior
ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/rrbb20
Perceptions of moralizing agents and cooperative
behavior in Northeastern Brazil
Montserrat Soler, Benjamin Grant Purzycki & Martin Lang
To cite this article: Montserrat Soler, Benjamin Grant Purzycki & Martin Lang (2022) Perceptions
of moralizing agents and cooperative behavior in Northeastern Brazil, Religion, Brain & Behavior,
12:1-2, 132-149, DOI: 10.1080/2153599X.2021.2006285
To link to this article: https://doi.org/10.1080/2153599X.2021.2006285
View supplementary material
Published online: 06 Apr 2022.
Submit your article to this journal
Article views: 49
View related articles
View Crossmark data
Perceptions of moralizing agents and cooperative behavior in
, Benjamin Grant Purzycki
, and Martin Lang
Ob/Gyn and Women’s Health Institute, Cleveland Clinic, Cleveland, OH, USA;
Department of the Study of Religion,
Aarhus University, Aarhus, Denmark;
LEVYNA: Laboratory for the Experimental Research of Religion, Masaryk
University, Brno, Czech Republic
Evolutionary theories suggest that gods of world religions are associated
with moralizing qualities and impartial behavior toward co-religionists,
and that secular authorities approximate this eﬀect. However, there is a
lack of theorizing and experimental studies regarding the inﬂuence of
local religions on inter-personal conduct. In the current research, we
obtained data on beliefs regarding the moralizing qualities of the
Christian god, a local god (Ogum), and police in a sample from
Northeastern Brazil (n= 193). We used these beliefs as predictors of
behavior in Dictator Games where players distributed endowed money
between anonymous individuals belonging to local and distant
communities. We used subtle reminders of the Christian god, Ogum,
and police to investigate their inﬂuence on game behavior. The
correlational and priming results are mostly in agreement, revealing
that: (a) the Christian god is perceived as most moralizing, but (b) has
only limited impact on game behavior, while (c) adherence to Ogum is
associated with ingroup favoritism, as is (d) priming with secular
authority. These results illustrate the diﬀerential eﬀects of belief in
moralizing and local deities on extended prosociality but show that in
speciﬁc contexts, secular authorities may emulate the eﬀects of local
rather than moralizing deities.
Received 6 March 2020
Accepted 25 June 2020
Brazil; Candomblé; dictator
game; economic games;
moralizing gods; prosocial
behavior; secular authority
The explicit connection between belief in supernatural agents and human moral behavior is a cen-
tral feature of the world’s major religions. This observation has sparked considerable theoretical and
empirical work to explain how and why humans have developed beliefs in supernatural beings that
punish and reward followers (Johnson, 2015; Johnson & Bering, 2006; Norenzayan, 2013; Noren-
zayan & Shariﬀ,2008; Roes & Raymond, 2003; Shariﬀ& Norenzayan, 2011). Such beliefs are pre-
dicted to reduce cheating and promote intra-group cooperation in large groups where monitoring
of others is diﬃcult, thereby facilitating the emergence of complex societies (Johnson, 2005; Nor-
enzayan, 2013; Roes & Raymond, 2003). Empirical work has shown that a greater degree of belief
in a moralizing, punishing god is associated with prosocial attitudes and behaviors (Atkinson &
Bourrat, 2011; Johnson et al., 2013; Shariﬀ& Norenzayan, 2011). Priming experiments that utilize
moralizing god concepts have obtained similar results, suggesting a causal connection between such
beliefs and prosociality (Ahmed & Hammarstedt, 2011; Gervais & Norenzayan, 2012a; McKay et al.,
2011; Shariﬀ& Norenzayan, 2007).
© 2022 Informa UK Limited, trading as Taylor & Francis Group
CONTACT Montserrat Soler email@example.com, firstname.lastname@example.org; Martin Lang email@example.com
Supplemental data for this article can be accessed at https://doi.org/10.1080/2153599X.2021.2006285
RELIGION, BRAIN & BEHAVIOR
2022, VOL. 12, NOS. 1–2, 132–149
Gods that interfere in human aﬀairs tend to possess “hyper-mentalizing”features (e.g., omnipre-
sence, omniscience) that go beyond typical abilities to infer others’mental states. However, ethno-
graphic evidence indicates that not all deities display an equal concern with behavior toward
anonymous others. Supernatural agents in some religious systems either lack these abilities or dis-
play no interest in enforcing moral norms (e.g., Chalupa, 2010; Purzycki, 2013; Purzycki & Sosis,
2011; Whitehouse, 1996). Empirical work in such settings has been scarce, but some studies suggest
that trust and cooperation are preferentially directed toward those who demonstrate adherence to a
common group identity through costly commitment signals (Purzycki & Arakchaa, 2013; Soler,
2012); that is, these systems rely on face-to-face communication of their members rather than
on internalized beliefs in moralizing deities. In line with this argument, McNamara et al. (2016)
have shown that primes that invoke local gods (i.e., gods that are closely associated with circum-
scribed populations and do not explicitly care about moral issues) can also increase parochialism
in experimental economic games. Similarly, cross-cultural studies using such games suggest that
belief in local gods’moralizing abilities does not predict greater generosity toward distant coreligio-
nists, as belief in punitive and omniscience deities does (Lang et al., 2019; Purzycki et al., 2016).
A corollary of this line of reasoning is related to the degree of secularity observed in Western
countries, where the last few decades have witnessed a dramatic decline in religiosity (Zuckerman,
2007). A classic interpretation of this trend is that religious belief and ritual behavior intensify when
there is uncertainty or danger (Barber, 2011; Gmelch, 1974; Malinowski, 1948; Sosis & Handwerker,
2011). Uncertainty may also motivate individuals to turn to civil institutions as a means of re-assert-
ing control over their circumstances (Kay et al., 2008,2010). Thus, the monitoring role of moraliz-
ing gods may be substituted by eﬃcacious judicial and criminal systems that enforce social norms
among strangers (Gervais & Norenzayan, 2012b; Shariﬀ& Norenzayan, 2007). Indeed, priming
subjects with secular reminders such as “police”and “jury”has been shown to have a similar
eﬀect on increasing cooperation on economic game behavior as moralizing god primes (Shariﬀ
& Norenzayan, 2007).
Notably, previous research reporting a relationship between beliefs in gods and how they corre-
spond to the breadth of generosity or honesty has been limited to variation of speciﬁcally moralistic
deities where the primary contrast is whether or not such gods punish and are omniscient, not
whether non-moralistic gods have similar eﬀects. In fact, precedent studies either do not specify
recipients’religiosity (McNamara et al., 2016) or restrict recipients to the moralistic deity tradition
or religious outgroups (Lang et al., 2019; Purzycki et al., 2016). In other words, participants who
have engaged in these experiments made economic decisions between themselves, local or distant
co-religionists of the moralistic tradition, and members of religious and/or ethnic outgroups. Reci-
pients have not been deﬁned as subscribing to local traditions. Here, our focus supplements this
research by focusing on co-religionists of more locally-bound beliefs.
We investigate the interaction between local-god communities among a population in North-
eastern Brazil that practices Candomblé, a religion of the African Diaspora. While Candomblé
has a strong presence in the area, the majority of the population practices Catholicism or, increas-
ingly, Evangelical and Pentecostal Christianity. The gods of Candomblé are characterized as non-
moralizing, but most adherents see no conﬂict in professing a concurrent belief in the Christian
god. In addition, decades of economic and political turmoil in the country have resulted in mistrust
of civil institutions (Power & Taylor, 2011). Thus, this population is embedded in multiple layers of
religious belief and moral authority, making it an ideal setting to assess the relationship between
religious beliefs, secular authority, and moral behavior. Using ethnographic interviews, economic
games, and object primes, we assess the impact of religiosity and secular authority on prosociality.
The Site: Cachoeira. The Recôncavo Baiano extends from the littoral around the city of Salvador
da Bahia to surrounding regions which were once the center of the sugar cane industry. The town of
Cachoeira (pop. 15, 000), where the research was conducted, sits about 75 miles inland. Up until the
1800s, it held an important strategic position linking wealthy sugar plantations to the coast through
the Paraguassú River. Today it is better known for its charming but decaying Baroque architecture
RELIGION, BRAIN & BEHAVIOR 133
which makes it a popular day-trip destination for tourists. Thus, residents have modest but steady
sources of wage labor through daily tour buses and seasonal festivals that attract large numbers of
locals from surrounding areas. Other widespread economic activities include small-scale com-
merce, day labor, and domestic work. While the town has a small number of prominent families,
most study participants were low-income and worked in the informal sector. Given its history as a
central hub of the country’s enslaved labor force, Cachoeira is also well-known as a center of Afro-
Brazilian culture. Candomblé houses or terreiros are present in practically every neighborhood.
Thus, even non-devotee residents have some familiarity with the religion.
Fifteen years ago, Cachoeira was a sleepy maze of narrow cobble-stone streets. Since then,
global and local events have ushered in noticeable changes. Although the Northeast is one of
the poorest regions of the country, Brazil’s economic boom (2010–2013) brought a period of
prosperity. The commercial area in the city center grew, a large outdoor market was rehabi-
litated, and the volume of cars and motorcycles signiﬁcantly increased. The use of cell phones
and social media are now widespread. The opening of a regional branch of the federal univer-
sity has attracted students from all over Northeast Brazil. However, not all changes have been
positive. The inﬂux of students has meant that prices for housing and commodities around the
town have risen, pricing out locals. Whereas anything other than petty crime was practically
unheard of in the recent past, drug-related violence is on the rise (two young men were mur-
dered during Soler’s most recent two-week stay in 2015). More dramatically, the spectacular
crash of the Brazilian economy in 2014 had dire consequences. By August 2015, when the pre-
sent study was conducted, Brazilian currency had lost a third of its value in the course of eight
months. Many study participants mentioned recent job losses and held a bleak outlook of their
The political fallout from the country’s economic crisis was not yet evident at the time of the
research, but stability has never been a hallmark of Brazilian institutions. Corruption is endemic
and distrust of a government traditionally dominated by oligarchs is widespread (Power & Taylor,
2011). For low-income citizens, the most immediate representative of state authority is the police.
There are various law-enforcement bodies in the country but the best-known is the Polícia Militar
or PM. An ancillary of the military, the PM is ostensibly responsible for street patrols and public
safety at the local level. However, the violent exploits of the organization in the slums or favelas
of large cities are well-documented (Ahnen, 2007; Caldeira, 2002). Raids, retaliatory violence,
and extra-judicial executions, particularly against poor Afro-Brazilians, are routine practices.
Thus, as a representation of secular authority, the PM is unlikely to evoke positive attitudes or sub-
stitute the eﬀects of a moralizing god.
Brazil’s Religious Landscape. Brazil is home to a complex mix of Christian faiths, indigenous
traditions, and various religions of African origin. Among the latter, the best-known is Yoruban-
inﬂuenced Candomblé, which shares beliefs and rituals with other religions of the African Diaspora.
Candomblé centers on the cultivation of axé, the life-force of the universe. Although there is a dis-
tant creator god (Olorun or Olódùmarè), worship is primarily directed at a group of deities called
orixás (also santos or saints) and various other supernatural entities unique to Brazil. Each orixá is
associated with a speciﬁc realm of life, such as contagious disease (Obaluâe) or sacred herbs and
plants (Oxossí). Adherents must ﬁnd out which orixá is their dono da cabeça (owner or master
of their head), which makes them the symbolic son or daughter of that deity. Sons and daughters
share the personality traits of their dono da cabeça, rely on that orixá for supernatural intervention,
and must abide by the speciﬁc taboos and rituals associated with that entity (e.g., sexual abstinence
on certain days, food restrictions, and speciﬁcoﬀerings). Thus, an orixá that is particularly impor-
tant for one person, may be signiﬁcantly less so for another.
Candomblé worshippers are organized around autonomous groups called terreiros (other terms
are axé, roça, or casa). The terreiro represents a physical space for worship as well as a community of
followers led by a charismatic priest or priestess (pai or mãe-de-santo). Although each leader may
set his or her own norms, there are suﬃcient commonalities across terreiros to recognize adherents
134 M. SOLER ET AL.
as belonging to the povo-de-santo (literally, “people of the saint”which refers to all followers of
Candomblé beliefs and rituals are concerned with attaining happiness, health, and prosperity in
this life rather than accumulating good deeds or securing rewards in the afterlife. This pragmatism
has led to descriptions of the religion as “amoral”(Amaral, 2002; Chesnut, 2003; Prandi, 1991).
More recently, there has been resistance to this characterization and a recognition of an inherent
ethical system centered on the sacred responsibilities (obrigações) of each adherent, which are con-
tingent on their role within the hierarchy of the terreiro and their relationship with the orixá that
“owns”their head (Berkenbrock, 2017; Rabelo, 2016). In addition, Candomblé adherents do not live
in an isolated bubble; they are fully integrated in broader Brazilian society. Thus, the degree to
which ethical or moral dictates emerge from Candomblé or from the surrounding socio-cultural
environment is diﬃcult to ascertain.
For much of its history, Candomblé existed in the shadow of Roman Catholicism. Perhaps as a
cover for their beliefs, Candomblé adherents at some point adopted some aspects of Catholicism.
Orixás are associated with corresponding Catholic saints (e.g., Iansá with Santa Barbara, Iemanjá
with the Virgin Mary) and some feasts correspond with important dates in the Catholic calendar.
There are even Catholic festivals where Candomblé followers play a central role. The feast of Our
Lady of the Good Death in Cachoeira, for example, involves processions and a Catholic mass, but is
organized by a confraternity of women who are Candomblé adherents. Thus, there is little conﬂict
between being a Candomblé follower and believing in the Christian god or other deities. A compli-
cating factor is the growing presence of Evangelical and Pentecostal churches. Evangélicos, as fol-
lowers of these religions are called in Brazil, tend to subscribe to a conservative version of Protestant
Christianity. There is active hostility from some of these churches towards Afro-Brazilian religions,
which are often portrayed as superstitious, dangerous, or evil. Despite this, most Evangélicos are
familiar with Candomblé, may have family members who are Candomblé adherents, or even con-
verted from Candomblé into their current religion. Thus, a notable feature of this site is the reli-
gious heterogeneity of the population and the familiarity of any given person with various
General Predictions. If the gods of world religions are indeed more moralistic, punitive, and
knowledgeable of human behavior, and reminders of such deities can harness their general
eﬀects of extending prosociality beyond the conﬁnes of family and one’s local community, then
exposure of the Christian god should be more associated with prosociality beyond local commu-
nities (i.e., less parochial game behavior). While we generally expect people to favor anyone closer
to themselves (i.e., individual > local religious community > distant co-religious community > dis-
tant religious outgroup), when individuals entertain the belief in a punitive, omniscient deity that
cares about morality, they should be more generous toward geographically distant co-religionists
and religious outgroup members than those who do not endorse such beliefs. In other words,
such deities should bring more investment to others beyond participants’immediate lives. How-
ever, as Candomblé deities are locally important spirits, they should be associated with increased
parochialism in game behavior; when reminded of such deities, people should be more willing to
favor their local communities at the expense of co-religionists who live elsewhere and religious out-
group members. Moreover, while secular judicial systems and institutions might generally function
in a similar way to the moralistic gods of the world religions, given the aforementioned climate of
police corruption in Brazil, we posit that the police will be associated with the lowest expression of
prosociality towards others in this context. We tested these predictions using a dictator game and
four priming conditions.
Materials and methods
The study took place in January 2015 and August 2015. Data collection included a preliminary
phase to collect data for ethnographic interviews. Study participants played four dictator games
RELIGION, BRAIN & BEHAVIOR 135
in four randomly assigned treatment conditions that ostensibly should have allowed us to calculate
the eﬀect of religious and secular primes on prosocial behavior. Players responded to ethnographic
interviews before and after the games.
Preliminary Data Collection. Preliminary questionnaires were conducted to determine which
deities were widely known and which moral traits were viewed as important at the study site.
These data were collected at a public plaza with 20 locals that did not take part in the ﬁnal
study. Findings were used to construct ethnographic interviews for the study. Each of the intervie-
wees was asked to list and rank ﬁve spirits or deities that “you think people here believe in”. By far
the most commonly cited moralizing god was the Christian God (Deus in Portuguese). Choosing a
local god presented more diﬃculties. Several orixás received an equally high ranking and number of
mentions. In addition, some orixás are connected to notions of fairness or the afterlife (e.g., Xangô,
Nanã), some are believed to lead people to bad outcomes, including death (i.e., Exú), and some are
widely syncretized with Catholic ﬁgures (e.g., Iemanjá). Taking these issues into consideration,
Ogum, a male orixá syncretized as St. George, was chosen to represent the local god. Among the
at least 12 orixás recognized in Candomblé, Ogum is one the best-known and most widely wor-
shipped. Ogum’s origin story identiﬁes him as a fearsome warrior, and as such he is associated
with ﬁghts, struggles, and conquering adversity. Qualities linked with Ogum are courage, arrogance,
anger, and determination, but there is nothing inherently moralizing about him (i.e., these qualities
may be channeled for whatever purpose suits a devotee). Interviews also included a free-list of ﬁve
qualities that “make someone a bad or immoral person”. We used these data for validation checks
(see Results section).
Recruitment. Participants were 56 males and 128 females (n= 193; 9 participants have missing
data for sex). All participants spoke Portuguese and the vast majority reported living in the town
where the study was conducted (99.5%). Descriptive variables of the sample are shown in Table
1. The goal was to recruit members of the povo-de-santo, that is, devoted Candomblé followers,
in order to explore extended prosociality among followers of a local god. However, responses to
questions about engagement with Deus and Ogum suggest lower levels of adherence to Candomblé
than expected. These questions were “How often do you think about Deus/Ogum?”and “How often
do you perform activities or practices to talk to or appease Deus/Ogum?”and were answered on a
ﬁve-point frequency scale; all interview questions are available online at https://osf.io/epkbw/. All
participants except two reported thinking daily about Deus and 85% of the sample reported daily
performance of activities to talk to or appease Deus. In contrast, 61% reported thinking about
Ogum at least a few times a year (24% did so daily), while 55% reported performing activities to
talk to or appease Ogum a few times a year (16% did so daily). That is, despite recruitment through
personal visits at local terreiros to maximize participation of devoted adherents, we sampled partici-
pants with varying commitment to Ogum, from fully initiated members to many who simply enjoy
attending feasts and ceremonies. To limit collusion, potential subjects were invited to attend a
Table 1. Sample descriptive variables.
Mean SD Range
Age 34.8 12.3 18 to 75
Yrs of formal education 8.6 4.1 0 to 20
Number of children 1.9 1.9 0 to 10
Household size 4.1 2.0 1 to 17
Yrs lived in a large urban area
3.7 7.4 0 to 36
Co-religionist closeness 1.5 1.0 1 to 5
Outgroup closeness 1.6 1.1 1 to 5
Police evaluation 0.3 1.1 −2to2
Freq. thinking about Deus 4.0 0.3 0 to 4
Freq. thinking about Ogum 1.7 1.7 0 to 4
Freq. ritual Deus 3.8 0.7 0 to 4
Freq. ritual Ogum 1.4 1.5 0 to 4
136 M. SOLER ET AL.
morning or afternoon session at a centrally located community center over the course of eight days.
However, Cachoeira is a small town and word travels fast. As people found out from friends and
neighbors that study participants earned cash, it is likely that not all attendees considered them-
selves as belonging to the povo-de-santo. One woman, for example, did not complete the study
because she identiﬁed as evangélica and felt uncomfortable giving money in the games to a Can-
domblé adherent. To control for the possibly varying attachment to Candomlé (co-religionists)
and Evangelical (outgroup) communities, we included in our statistical models questions about
emotional closeness to co-religionists and outgroups (see Table 1).
Ethnographic Interviews. There were four sets of interviews used in the study: a demographic
questionnaire, a material uncertainty scale, an income and household wealth questionnaire, and
a religiosity/moral values interview which included questions about the Christian moralizing god
(Deus), the local god (Ogum), and the secular authority (the police). Only data from the demo-
graphic and religiosity/moral values interviews are presented here. Questions from the religios-
ity/moral values interview were answerable as either “yes”or “no”(coded as dummy variables)
or through a 5-point frequency scale (0 = never, 1 = once a year, 2 = once a month, 3 = once a
week, 4 = every day or several times a day). The basic predictor variables pertaining to a god’s pun-
ishment and omniscience were constructed from four binary questions asking whether a god (1)
can punish people for their behavior; (2) inﬂuence what happens after death; (3) know what people
think and feel; and (4) see what people are doing in São Paulo. Answers to these questions were
averaged, creating a Punishment-Monitoring scores for Deus and Ogum.
Economic Games. To measure prosocial behavior, we employed modiﬁed versions of the dicta-
tor game (DG). In its simplest form, the DG consists of a player who receives an economic allot-
ment to split between herself and an anonymous recipient in any manner she chooses. The
amount allocated by the player to the recipient is a baseline measure of prosociality. Here, each sub-
ject played four games in random order: local coreligionist vs. distant coreligionist, themselves vs.
distant coreligionist, distant coreligionist vs. outgroup member, and themeselves vs. outgroup
member. Recipient categories were described to subjects as “an individual from the povo-de-
santo in Cachoeira”(local-coreligionist member), “an individual from the povo-de-santo in Salva-
dor”(distant co-religionist), and “an Evangelical person in Salvador”(outgroup member). In other
words, recipients were identiﬁed as either local or distant Candomblé followers, and the outgroup
members were geographically distant Evangelicals. We used the nearby city of Salvador as a refer-
ence because it has many Candomblé terreiros and it is large enough to hold many anonymous co-
religionists. Subjects were provided with $10 Brazilian Reais (BR) (approximately $2.85 USD at the
time) in coins of $1BR for each round of the DG. Players were instructed to divide a $10BR during
each game in whatever manner they chose using labeled pairs of covered cups with a slit to place
coins. It was explained that coins in the “self”cup would be taken home by the player at the end of
the game and that coins in the other cups would be given to individuals belonging to the recipient
categories. Prior to the beginning of play, the game was described in detail to individual participants
and test questions were used to ensure full understanding of the procedure. Upon ﬁnishing the data
collection, coins allocated to local and distant co-religionists and to outgroups were donated to the
Object Primes. Subjects were randomly assigned to one of four possible conditions: a control,
moralizing god, local god, and secular authority. In the control condition, subjects played the econ-
omic game on an empty table. In the other three conditions, object primes were placed on one side
of the table where they were clearly visible. The moralizing god prime was a Christian Bible, the
local god prime was a blue bead necklace easily recognizable as representing Ogum, and the secular
prime consisted of a cap used by the military police with a prominent seal representing the
General Procedure. All parts of the experiment were carried out at a local community center
which had several rooms with doors that aﬀorded necessary privacy. Participants waited in a class-
room as they arrived. Groups of four or ﬁve were then led into another room where the lead author
RELIGION, BRAIN & BEHAVIOR 137
explained that the task involved an economic game and a series of questionnaires. Subjects were
assured that all their choices in the game would be conﬁdential and that they were free to stop
the activity at any moment they chose without forfeiting a show-up fee of $10 BR. After providing
consent, each subject ﬁlled out an identiﬁcation card with basic contact information that randomly
assigned each person to one of the four possible prime treatments and a card with 4 digits that rep-
resented the order in which the games would be played (also distributed randomly). Subjects were
then individually led by a ﬁeld assistant to one of two rooms where they could play the games in
complete privacy. Each of these rooms had a table and chair facing a blank wall. Object primes
were placed on one side of the table. All pictures and posters were either removed from the
walls or covered with white sheets of paper. A ﬁeld assistant provided the subject with the ﬁrst
set of cups and coins, explained the game in detail, ran sample runs of the game, and asked testing
questions to ensure a full understanding of the procedure. Assistants then left the room, closed the
door, and waited outside. After each game, subjects alerted the assistant who then provided the next
pair of cups while the lead author picked up the cups with the allocated coins inside. This was
repeated until the subject completed the four games. Following the games, subjects moved to an
outside area where two ﬁeld assistants conducted the demographic and religiosity interviews. In
the meantime, the lead author recorded each game oﬀer and organized pay-outs behind closed
doors in another room. After the ﬁrst interviews were over, each subject entered this room
where the lead author conducted a household wealth interview and a market integration interview.
Following this, subjects were thanked and provided cash pay-outs in an envelope. All methods, pro-
cedures, and data collection forms were reviewed and approved by the Institutional Review Board
of Montclair State University.
Predictions.The cup dyads follow most of our general predictions (see above). We expect a gen-
eral pattern of allocations beneﬁtting players, followed by their local communities, followed by dis-
tant co-religionist communities, with distant outgroup members receiving the least (i.e., S>L>D
>O). However, getting primed with a Bible or expressing higher ratings of God’s omniscience and
punishment should increase allocations to distant co-religionists in the local vs. distant co-religio-
nist game and in the self vs. distant co-religionist game, as well as increase allocations to the reli-
gious outgroups in the distant co-religionist vs. outgroup game, and in the self vs. outgroup game
(Prediction #1). This prediction reﬂects the fact that while co-religionist cups were designated to
Ogum followers, these followers are mostly members of the Catholic church; hence, belief in the
Catholic god (Deus) should positively aﬀect allocations to Evangelicals (outgroups) who believe
in the same god. In contrast, getting primed with the Ogum prime or expressing higher ratings
of Ogum’s omniscience and punishment should boost parochialism such that participants will pre-
fer the self and local co-religionist cups (Prediction #2). In other words, belief in and priming with
Ogum should aﬀect cup preferences in the opposite direction than belief in and priming with Deus.
We should see a similar pattern in the police condition; if the police are generally held to be a hostile
and ineﬀective secular institution, we should see people withholding resources from relatively more
socially distant players (Prediction #3).
Analysis. We performed all analyses in R, version 3.5.3 (R Core Team, 2019). For each of the
four games, we predicted the number of coins in the less parochial cups: a distant co-religionist
cup in the games that pitted self and local co-religionist against distant co-religionist; and an out-
group cup in the games that pitted self and distant co-religionist against outgroup. Since the Dic-
tator Game data were censored at 0, we used the Tobit regression (McDonald & Moﬃtt, 1980),
which proved to ﬁt the data better than a negative binomial regression that could also account
for the data-censoring at 0. Data were analyzed using the tobit function from the AER package
(Kleiber & Zeileis, 2008) and plots were created using the package ggplot2 (Wickham, 2016).
We divide the main analyses into two parts—modeling the relationship between belief in omnis-
cient and punitive aspects of Deus and Ogum, and allocations in the Dictator Game; and testing the
eﬀects of priming with Deus, Ogum, police, and control primes on Dictator Game allocations. That
is, models in the ﬁrst part tested Predictions 1–3 using self-reported beliefs in Deus, Ogum, and
138 M. SOLER ET AL.
opinions about police, and models in the second part tested Predictions 1–3 using the priming
materials. In the ﬁrst set of models, we ﬁrst included the predictor variables mapping the monitor-
ing and punitive characteristic of Deus and Ogum. In the second step, we adjusted these models for
demographic variables pertaining to age, sex, years of formal education, the number of children,
household size, years lived in a city, and for emotional closeness to co-religionists and outgroups.
In the priming analyses, we ﬁrst compared the diﬀerences between priming with Deus and the other
primes, and in the second step adjusted these eﬀects for the same demographic variables as in the
To investigate the diﬀerential eﬀects of belief in Deus and Ogum on behavior in the Dictator Game,
we ﬁrst assessed whether participants believed that Deus was indeed more moralizing compared to
Ogum and the police. Subsequently, we tested whether this purported diﬀerence in moralization
predicted game behavior, and, ﬁnally, tested whether priming with diﬀerent god concepts aﬀected
Preliminary Analyses. By way of a validation check, we ﬁrst assessed participants’attributions to
the Christian god, a Candomblé deity, and the police in order to demonstrate that the Christian god
is the most morally concerned and knowledgeable of the three agents. As expected, responses
demonstrate that Deus is ascribed with both moralizing concerns and punishment-monitoring abil-
ities to a higher degree than Ogum (Figure 1, top row). While the diﬀerences between Deus and
Ogum were relatively subtle for punishing, the belief in Deus’monitoring abilities were twice as
high compared to Ogum. In fact, these beliefs reached ceiling levels. Furthermore, there was no per-
son in the whole sample who answered ‘No’to all of the four questions related to Deus. Figure 1,
middle row, also revealed stark diﬀerence between Deus and Ogum in rewarding abilities: the for-
mer again reaches ceiling levels while the latter hovers around the mean of the reward question.
Recall that in the preliminary interviews, we asked participants about what constitutes a bad or
immoral person. The most frequently mentioned items were deceitfulness, envy, wickedness, alco-
hol and drug use, and stealing. These items were translated based on Soler’s own ethnographic
experience (see Soler, 2012,2016) and include synonymous terms. Thus, frequently mentioned
variations on the word falso (fake or phony) included falsedade (deceitfulness or phoniness) or
even mentiroso (liar) or mentira (lie), but only falsedade was included in the ﬁnal ethnographic
interviews. Alcohol use is frequent in this population, but alcohol and drug use (rather than
abuse) were ranked high as moral failings. Envidia and colloquial expressions such as olho grande
(literally, “large eye”which refers to coveting what others have) were translated as envy. Maldoso
refers to someone who is wicked, mean, or cold-hearted. Finally, ladrão and roubar were subsumed
under stealing. The term ladrão translates literally as “thief,”but is used more broadly to refer to
young men involved in crime, particularly drug-traﬃcking. The usage is similar to “thug”or “hood-
lum”in the United States and can reﬂect socio-economic and racial prejudice. Drawing from these
data, we asked participants how much Deus, Ogum, and police cared about these four immoral
behaviors (last row in Figure 1). Deus was perceived as most concerned with all moral items.
Ogum was believed to be more concerned than police with deceitfulness, envy, and wickedness,
but less concerned about the use of alcohol and drugs. Together, these summary statistics show
that Deus was indeed perceived as a relatively more moralizing deity compared to Ogum and police.
Supernatural Beliefs and Game Behavior. We analyzed four permutations of the Dictator Game
where participants divided money between themselves, local co-religionists, distant co-religionists,
and outgroups. In the games that included cups for players themselves, the modal contributions to
the distant co-religionists or outgroup players was 0. In other words, participants strongly preferred
to allocate all money to themselves. However, when allocating between two groups based on their
geographical and social proximity, the modal allocation was 5; participants often opted for impartial
divisions. Since the games containing the “Self cup”revealed more selﬁsh behaviors, belief in Deus
RELIGION, BRAIN & BEHAVIOR 139
had more space to drive allocations closer to the impartial division. Indeed, for both games where
participants were pitted against distant co-religionists and outgroups, we observed a substantial
positive eﬀect of belief in Deus’punitive and monitoring abilities on allocations to the less parochial
cup, albeit the 95% CI crossed zero (supporting Prediction #1). Believing in all four aspects of Deus’
monitoring and punishing increased the estimated positive allocations to distant co-religionists on
average by 1.8 coins in the self vs. distant game, and by 2.1 coins in the self vs. outgroup game (see
Table 2 for untransformed coeﬃcients and 95% CIs and Figure 2).
On the contrary, belief in Ogum
had no stable eﬀect on allocations in the games comprising the “Self cup”, and Prediction #2 was not
Interestingly, this trend was reversed for games where local co-religionists were pitted against
distant co-religionists and where distant co-religionists were pitted against outgroups. While belief
in Deus’monitoring and punitive abilities had variable eﬀects, belief in Ogum’s abilities predicted
Figure 1. Moralizing concerns and punishment-monitoring abilities of Deus, Ogum, and the police (means with SE). The ﬁrst row
displays the percentage of yes answers to binary questions pertaining to god’s punishment and monitoring abilities. The second
row displays god’s interest in punishing various misconduct and rewarding good behaviors. The third row displays perceived
concerns with various behaviors for Deus, Ogum, and the police (see R code for plot regarding stealing).
140 M. SOLER ET AL.
Table 2. Beta estimates with 95% CI from Tobit models for deity punishment-monitoring abilities predicting coin allocations to most geographically/socially distant recipient.
Local vs. Distant Self vs. Distant Distant vs. Outgroup Self vs. Outgroup
Deus −0.59 0.78 2.07 2.06 0.57 1.75 3.65 2.51
(−2.63, 1.44) (−1.50, 3.07) (−0.28, 4.42) (−0.45, 4.56) (−1.58, 2.73) (−0.76, 4.27) (1.04, 6.03) (−0.36, 5.39)
Ogum −0.81 −0.84 0.32 0.24 −2.22 −1.91 −0.04 −0.53
(−1.59, −0.04) (−1.73, 0.05) (−0.57, 1.20) (−0.73, 1.21) (−3.04, −1.40) (−2.89, −0.93) (−0.98, 0.90) (−1.63, 0.58)
Age −0.05 0.09 −0.03 −0.31
(−0.61, 0.51) (−0.52, 0.71) (−0.64, 0.58) (−1.01, 0.40)
Sex (1 = male) −0.78 −0.13 −0.85 −0.27
(−1.63, 0.06) (−1.05, 0.79) (−1.79, 0.09) (−1.33, 0.79)
Education 0.02 −0.14 −0.13 0.004
(−0.37, 0.41) (−0.57, 0.30) (−0.56, 0.30) (−0.49, 0.49)
Children 0.003 −0.03 0.06 −0.03
(−0.26, 0.27) (−0.33, 0.26) (−0.24, 0.35) (−0.37, 0.31)
Household −0.30 −0.19 −0.09 −0.17
(−0.58, −0.01) (−0.50, 0.13) (−0.39, 0.22) (−0.52, 0.17)
Material −0.17 −0.21 −0.34 0.39
(−0.65, 0.32) (−0.74, 0.32) (−0.88, 0.20) (−0.22, 1.01)
Years in city 0.06 0.11 0.03 0.04
(0.01, 0.11) (0.06, 0.16) (−0.02, 0.08) (−0.01, 0.10)
Police eval −0.29 −0.01 −0.28 0.04
(−0.68, 0.10) (−0.44, 0.42) (−0.71, 0.15) (−0.45, 0.53)
Corel close −0.01 −0.14 −0.34 −0.1
(−0.42, 0.40) (−0.59, 0.30) (−0.80, 0.11) (−0.62, 0.42)
Outgroup close 0.35 0.29 0.21 0.31
(−0.003, 0.71) (−0.10, 0.69) (−0.18, 0.61) (−0.14, 0.75)
Intercept 3.95 3.59 0.60 1.02 5.27 4.70 −0.34 1.08
(2.31, 5.58) (1.27, 5.91) (−1.30, 2.49) (−1.52, 3.57) (3.53, 7.00) (2.17, 7.23) (−2.35, 1.67) (−1.79, 3.95)
N177 128 180 130 180 130 178 129
Note: Coeﬃcients refer to less parochial cup allocations. The Deus and Ogum variables represent punishment-monitoring scores for the respective deity. Age, education, and material insecurity are
centered at their mean. Police eval is evaluation of police. Corel and outgroup emo is emotional closeness to co-religionists and outgroup. The diﬀerent number of participants between games
reﬂect mistakes in the game behavior of individual participants. Note that intercept is the intercept of latent uncensored variable.
RELIGION, BRAIN & BEHAVIOR 141
increased parochiality in both games (see Table 2 for untransformed coeﬃcients and 95% CIs and
Figure 2). Speciﬁcally, reporting that Ogum is both punitive and monitoring was associated with 0.8
fewer coins contributed to the “Distant co-religionist cup”in the local vs. distant game and with 1.9
fewer coins in the “Outgroup cup”in the distant vs. outgroup game (supporting Prediction #2).
Positive evaluation of police had similar trends as belief in Ogum, that is, toward more parochialism
in the local vs. distant and distant vs. outgroup games, but these eﬀects were generally small and
imprecise, suggesting ambiguous support for Prediction #3.
Regarding the demographic variables used to adjust the main coeﬃcients of interest, there were
no substantial eﬀects that would be precisely estimated across all four games. For instance, while
males were on average more parochial than females in all games, this eﬀect was precisely estimated
only in the distant vs. outgroup game. Similarly, household size predicted more parochialism, but
these eﬀects were weak and imprecise. The only precisely estimated eﬀects were for years lived in a
city, which positively predicted contributions to less parochial cups in all four games (see Table 2).
Priming Analyses. Participants were randomly assigned to one of four treatments: control (N=
45), moralizing god prime (Christian Bible) (N= 41), local god prime (Ogum beads) (N= 60), and
secular prime (police hat) (N= 47). We run the same Tobit model as in the analyses of supernatural
beliefs above, only substituting variables related to belief in Deus’and Ogum’s monitoring-punitive
abilities with the treatment variable.
Figure 2. Regression estimates from Tobit models. Note that our sample did not contain any participant who would answer “No”
to all four questions about Deus’punitive and monitoring abilities, hence the regression lines for Deus belief start at 0.25.
142 M. SOLER ET AL.
As stated in Prediction #1, the Deus prime was associated with largest contributions to the less
parochial cups across all four permutations of the Dictator Game (see Figure 3). However, the
diﬀerence from the control condition was on average quite low, ranging from −0.4 to 0.1 coins
across all model speciﬁcations and imprecisely estimated (see Table 3 for untransformed coeﬃ-
cients and 95% CIs). More stable diﬀerences were revealed in comparison between the Deus
and Ogum primes with diﬀerences ranging from 0.2 to 0.9 coins across all models (providing lim-
ited support to Prediction #2). The largest diﬀerences were observed for the comparison between
Deus and police primes: the estimated diﬀerences ranged from 0.1 to 1.9 coins across the four
Dictator Games (supporting Prediction #3). These results suggest that while the Deus prime pro-
duced limited and small positive eﬀects (compared to the control prime), the Ogum and police
primes exerted larger negative eﬀects on allocations to less parochial cups (except for the distant
vs. outgroup games, where the diﬀerences between Deus, Ogum, and police prime were the
Figure 3. Raw means with SE for the eﬀects of primes on Dictator Game allocations. Priming with Deus produced largest allo-
cations in all four games.
RELIGION, BRAIN & BEHAVIOR 143
Table 3. Beta estimates with 95% CI from Tobit models for priming eﬀects on coin allocations to most geographically/socially distant recipient.
Local vs. Distant Self vs. Distant Distant vs. Outgroup Self vs. Outgroup
Control prime −0.63 −0.43 −0.29 0.11 −0.12 −0.01 −0.62 0.03
(−1.59, 0.33) (−1.46, 0.61) (−1.43, 0.85) (−1.05, 1.28) (−1.24, 1.00) (−1.25, 1.23) (−1.82, 0.57) (−1.29, 1.36)
Ogum prime −1.17 −0.96 −0.69 −0.75 −1.04 −0.19 −1.18 −1.14
(−2.08, −0.27) (−1.87, −0.04) (−1.76, 0.38) (−1.78, 0.28) (−2.10, 0.02) (−1.29, 0.90) (−2.29, −0.07) (−2.32, 0.03)
Police prime −2.22 −2.04 −1.40 −1.50 −0.99 −0.15 −2.14 −1.99
(−3.18, −1.26) (−3.09, −0.99) (−2.54, −0.26) (−2.69, −0.32) (−2.11, 0.12) (−1.39, 1.10) (−3.34, −0.95) (−3.36, −0.61)
Age 0.01 0.2 −0.22 −0.3
(−0.50, 0.52) (−0.37, 0.78) (−0.82, 0.38) (−0.96, 0.37)
Sex (1 = male) −0.21 0.24 −0.62 0.24
(−0.99, 0.57) (−0.63, 1.10) (−1.54, 0.30) (−0.76, 1.24)
Education −0.01 −0.15 −0.28 −0.07
(−0.37, 0.35) (−0.56, 0.26) (−0.71, 0.15) (−0.54, 0.40)
Children 0.01 0.03 0.05 0.08
(−0.24, 0.26) (−0.25, 0.32) (−0.25, 0.34) (−0.25, 0.41)
Household −0.23 −0.2 −0.02 −0.21
(−0.48, 0.01) (−0.48, 0.08) (−0.31, 0.26) (−0.52, 0.11)
Material −0.25 −0.15 −0.43 0.35
(−0.65, 0.16) (−0.60, 0.30) (−0.91, 0.05) (−0.17, 0.88)
Years in city 0.06 0.08 0.03 0.02
(0.02, 0.10) (0.03, 0.13) (−0.02, 0.09) (−0.03, 0.08)
Corel close −0.14 0.01 −0.66 −0.18
(−0.50, 0.21) (−0.39, 0.41) (−1.09, −0.24) (−0.65, 0.29)
Outgroup close 0.23 0.12 0.32 0.2
(−0.10, 0.56) (−0.25, 0.49) (−0.08, 0.72) (−0.23, 0.62)
Intercept 4.20 4.60 2.98 3.24 5.34 5.41 3.39 3.74
(3.51, 4.89) (3.15, 6.04) (2.16, 3.80) (1.60, 4.88) (4.53, 6.15) (3.68, 7.14) (2.54, 4.24) (1.86, 5.61)
N190 142 193 144 193 144 191 143
Note: Intercept reﬂects allocation to Deus prime condition. Deus and Ogum are punishment-monitoring scores for the respective deity. Age, education, and material insecurity are centered at their
mean. Sex is females vs. males. The diﬀerent number of participants between games reﬂect mistakes in the game behavior of individual participants.
144 M. SOLER ET AL.
The descriptive results were consistent with the hypotheses that moralizing gods are believed to
possess higher mentalizing abilities and are more concerned with morality. Deus was universally
believed to possess omniscient abilities, while these were ascribed to Ogum by less than half of
the sample. Deus was also believed to possess greater ability to interfere in human aﬀairs than
Ogum, such as punishing and inﬂuencing the afterlife. These diﬀerences may reﬂect broader diﬀer-
ences between religions associated with texts that preserve the uniformity of belief, like Christianity,
and those that rely on oral transmission, like Candomblé, where theological orthodoxy is more
likely to vary according to individual levels of adherence. Diﬀerences were also found in the per-
ceived concern of deities and secular authorities for moral behaviors. Deus was believed to be
more concerned about moral failings than both Ogum and the police. Item-by-item comparisons
between Ogum and the police, however, reveal suggestive diﬀerences between moral sentiments
and moral behaviors. While Ogum cares more about the former (e.g., deceitfulness), the police
were believed to be more concerned with the latter (e.g., alcohol and drug abuse and stealing).
Most participants had no diﬃculty understanding the games after the ﬁrst explanation and one
or two sample rounds. As explained in the Methods section, ﬁeld assistants demonstrated additional
sample rounds to those who still seemed unclear and gave them an opportunity to ask their question
until it was clear that they had full comprehension of the rules. During delivery of ﬁnal pay-outs and
in conversation with the lead author, several individuals mentioned that they found the interviews
interesting and described their understanding of the games as an experiment related to generosity
and religion. Game behavior was within the typical range observed in other studies (e.g., Engel,
2011), with generally more selﬁsh and parochial choices across games. The predicted eﬀects of belief
in Deus’punitive and omniscience abilities on positive allocations to less parochial cups (Prediction
#1) were detected only in the games that included the “Self cup”. Exploring Figure 2 revealed that
while these eﬀects are quite substantial (2 coins more in the less parochial cups), they likely reﬂect
aversion toward total selﬁshness (putting 0 coins in the less parochial cups) observed for partici-
pants that expressed the lowest rating of Deus’punitive and omniscient abilities. In other words,
the moralizing aspects of Deus were associated with allocating at least some amount to distant
co-religionists and outgroups, rather than impartially dividing the endowed money. On the
other hand, in the local vs. distant and distant vs. outgroup games where allocations were generally
close to the impartial split, the ratings of Deus’abilities did not rise the allocations beyond the fair
split, as should be expected. Self-reported devotion to Ogum, on the other hand, acted on that varia-
bility of allocations in the local vs. distant and distant vs. outgroup games (Prediction #2) exactly
because there was room for making the oﬀers more parochial (while the allocations in the self-
games were already at 0).
The priming results do not support the hypothesis that moralizing gods decrease selﬁshness and
increase generosity to strangers. While players exposed to the moralizing god prime behaved less
selﬁshly than those in the control condition, these diﬀerences were only small and imprecise. How-
ever, we found substantial diﬀerences between the Deus prime and Ogum primes, which had oppo-
site eﬀects on allocations, mostly driven by the Ogum prime. As in the correlational analysis of
belief in omniscience and punishment, the priming results suggested that priming participants
with Ogum reminders increased parochiality of their allocations in all games except for the distant
vs. outgroup game (Prediction #2). This result complements McNamara et al.’s(2016)ﬁnding that
local god primes in Fiji increase ingroup favoritism under some conditions and is consistent with
previous similar work in Brazil (Cohen et al., 2018).
Even stronger results in this regard were observed for the comparison of Deus and police primes
(Prediction #3). Exposure to the secular prime, a police hat, increased parochial behavior in all
games except the distant vs. outgroup game. While the perception of the police was marginally posi-
tive in our sample (see Table 1), this positive view was associated with more parochial allocations
(see Table 2). Likewise, the police prime was associated with more parochial oﬀers. These ﬁndings
RELIGION, BRAIN & BEHAVIOR 145
challenge the idea that secular authority functions similarly to moralizing deities (e.g., Kay et al.,
2010; Shariﬀ& Norenzayan, 2007). This may be the case in wealthy Western countries where secu-
lar institutions are functional and respected, but the situation in Brazil is more complex. In exten-
sive ethnographic interviews with low-income residents of Salvador, Paes-Machado and Noronha
(2002) describe attitudes toward the police as “ambivalent”and “contradictory”. They point out that
the urban poor in Brazil are equally familiar with drug-related crime and the brutal tactics of the
police. In smaller towns like Cachoeira, residents may be familiar with narratives surrounding
police violence, but their actual experience of police authority may be quite diﬀerent. Until recently,
Cachoeira experienced very little serious crime, and the population is small enough that members of
the local police are personally known to many residents. These factors may account for the para-
doxical ﬁndings of positive perceptions of secular authority and less prosocial game behavior in
the presence of a police prime.
Most demographic variables had no eﬀect on game behavior. Women were overrepresented in the
sample, but this gender imbalance is reﬂective of a similar trend in Candomblé terreiros where
females make up the bulk of membership. More years living in a large city were associated with
increased prosociality and a decrease in ingroup favoritism. Salvador was most frequently men-
tioned as the large city where participants had lived and this was also used as the location of the
distant coreligionist and outgroup member in the games. It is possible that this triggered a sense
of familiarity in some players that, in turn, led to more generous oﬀers. An alternative interpretation
is that the experiences with anonymous strangers that are common in large urban centers decrease
parochialism, as other studies have reported (Henrich, Ensminger et al., 2010).
Much of the literature surrounding the cognitive and evolutionary origins of religious belief has
been developed around Abrahamic traditions (see Johnson, 2015; Norenzayan, 2013). As is true of
most experimental psychological work, socioeconomically diverse and non-Western populations
have been underrepresented (Henrich, Heine et al., 2010). The present study centers on a non-mor-
alizing religion in a setting that shares many characteristics with the Western world. Brazil is one of
the largest democracies in the world and the research site is a small city that is fully integrated in the
global market economy. Despite these similarities many subjects reported a simultaneous adher-
ence to a moralizing god and a local god that forms a distinct religious tradition, and have opposite
eﬀects on extended prosociality. Thus, the postulate that belief in moralizing gods is instrumental in
promoting intra-group cooperation may be complicated by the presence of a parallel belief system
with diﬀerent network structure and universalizing tendencies. For future work, it would be impor-
tant to disentangle moralizing abilities of local gods from their broader role as signiﬁers of group
identity. Here, it is diﬃcult to discern if Ogum promotes ingroup favoritism because of the mor-
alizing qualities associated with this deity or because any orixá is bound to trigger favorable behav-
ior toward other Candomblé followers. Furthermore, future work should also focus on better
sampling of Candomblé followers, which proved diﬃcult in the current project. Indeed, close to
50% of participants did not express adherence to the local god that was featured in the interviews
and games. This low number is likely due in part to the large Candomblé pantheon. Since each
devotee is considered the son or daughter of a speciﬁc orixá, those who “belong”to other orixás
are not expected to consider Ogum as equally important. In addition, some orixás are closely linked
to ideas of justice (e.g., Xangô) while others are perceived as rather sinister (e.g., Exú). While our
religious landscape interview suggested that Ogum is one of the most salient local gods, using
another Candomblé deity would be a good test of generalizability of the current results.
Finally, the role of secular institutions in regulating moral behavior merits further investigation.
In some ways, the police elicit similar reactions to the local god (i.e., both increase ingroup favorit-
ism). It is unclear what, if anything, links the police and local god in a way that predicts similar
behaviors in the Dictator Game. Perceptions of the police in this sample are paradoxical and appear
146 M. SOLER ET AL.
to encompass the “fear and longing”for a strong control system described in other studies (Cal-
deira, 2002; Paes-Machado & Noronha, 2002). Relationships between Candomblé followers and
deities are characterized by a similar combination of service provision, a sense of obligation and
appreciation, and respect tinged with fear. A pai-de-santo in Salvador explained this in the following
We have a day, every month, so adepts and initiates of the house come to take care of what is theirs. How?
Doing their oﬀerings and every year paying their obligation to the orixá. And if it’s an urgent case, they
make some oﬀerings to obtain success in what they want in life. Because orixá has a positive side, but it
also has a negative side. I do know what is the positive side of the axé, [but] of the negative side of the axé,
to this day I don’t know and I don’t want to know.
Although this is a speculative explanation, it is worth considering parallels between secular insti-
tutions and local religious beliefs in non-Western settings as a novel avenue of research.
1. To ease the interpretation, we present marginal eﬀects estimated by the Tobit model for non-zero values. The
original Tobit estimates for the uncensored latent variable (presented in Tables 2 and 3) were multiplied by an
adjustment factor, which takes into account the scale parameter computed across the mean values of predictor
variables in the model (see Wooldridge, 2016 and also the Supplementary R code for more details).
No potential conﬂict of interest was reported by the author(s).
We acknowledge funding from a research grant, “The Emergence of Prosocial Religions”from the John Templeton
Foundation, and the Cultural Evolution of Religion Research Consortium (CERC), funded by a generous partnership
grant (895-2011-1009) from the Social Sciences and Humanities Research Council of Canada.
Data availability statement
The data set and associated R code used to produce the reported results are freely available at https://osf.io/az6cg/.
Montserrat Soler http://orcid.org/0000-0002-1193-8710
Benjamin Grant Purzycki http://orcid.org/0000-0002-9595-7360
Martin Lang http://orcid.org/0000-0002-2231-1059
Ahmed, A., & Hammarstedt, M. (2011). The eﬀect of subtle religious representations on cooperation. International
Journal of Social Economics,38(11), 900–910. https://doi.org/10.1108/03068291111171405
Ahnen, R. E. (2007). The politics of police violence in democratic Brazil. Latin American Politics and Society,49(1),
Amaral, R. (2002). Xirê!: o Modo de Crer e de Viver no Candomblé. Pallas/EDUC.
Atkinson, Q. D., & Bourrat, P. (2011). Beliefs about God, the afterlife and morality support the role of supernatural
policing in human cooperation. Evolution and Human Behavior,32(1), 41–49. https://doi.org/10.1016/j.
Barber, N. (2011). A cross-national test of the uncertainty hypothesis of religious belief. Cross-Cultural Research,45
(3), 318–333. https://doi.org/10.1177/1069397111402465
Berkenbrock, V. J. (2017). O conceito de ética no candomblé. HORIZONTE-Revista de Estudos de Teologia e Ciências
da Religião, 905–928.
RELIGION, BRAIN & BEHAVIOR 147
Caldeira, T. P. (2002). The paradox of police violence in democratic Brazil. Ethnography,3(3), 235–263. https://doi.
Chalupa, A. (2010). Why did Greeks and Romans prayed aloud? Anthropomorphism, dumb gods and human cogni-
tion. Chasing Down Religion: In the Sights of History and the Cognitive Sciences,15,81–95.
Chesnut, R. A. (2003). Competitive spirits: Latin America’s new religious economy. Oxford University Press.
Cohen, E., Baimel, A., & Purzycki, B. G. (2018). Religiosity and resource allocation in Marajó, Brazil. Religion, Brain
and Behavior,8(2), 168–184. https://doi.org/10.1080/2153599X.2016.1267029
Engel, C. (2011). Dictator games: A meta study. Experimental Economics,14(4), 583–610. https://doi.org/10.1007/
Gervais, W. M., & Norenzayan, A. (2012a). Like a camera in the sky? Thinking about God increases public self-aware-
ness and socially desirable responding. Journal of Experimental Social Psychology,48(1), 298–302. https://doi.org/
Gervais, W. M., & Norenzayan, A. (2012b). Reminders of secular authority reduce believers’distrust of atheists.
Gmelch, G. J. (1971). Baseball magic. Trans-action,8, 39–41. https://doi.org/10.1007/BF02908325
Henrich, J., Ensminger, J., McElreath, R., Barr, A., Barrett, C., Bolyanatz, A., Cardenas, J. C., Gurven, M., Gwako, E.,
Henrich, N., Lesorogol, C., Marlowe, F., Tracer, D., & Ziker, J. (2010). Markets, religion, community size, and the
evolution of fairness and punishment. Science,327(5972), 1480–1484. https://doi.org/10.1126/science.1182238
Henrich, J., Heine, S. J., & Norenzayan, A. (2010). The weirdest people in the world? Behavioral and Brain Sciences,33
(2-3), 61–83. https://doi.org/10.1017/S0140525X0999152X
Johnson, D. (2015). God is watching you: How the fear of God makes us human. Oxford University Press.
Johnson, D., & Bering, J. (2006). Hand of God, mind of man: Punishment and cognition in the evolution of
cooperation. Evolutionary Psychology,4(1), 219–233. https://doi.org/10.1177/147470490600400119
Johnson, D. D. P. (2005). God’s punishment and public goods. Human Nature,16(4), 410–446. https://doi.org/10.
Johnson, K. A., Li, Y. J., Cohen, A. B., & Okun, M. A. (2013). Friends in high places: The inﬂuence of authoritarian
and benevolent god-concepts on social attitudes and behaviors. Psychology of Religion and Spirituality,5(1), 15–22.
Kay, A. C., Gaucher, D., Napier, J. L., Callan, M. J., & Laurin, K. (2008). God and the government: Testing a com-
pensatory control mechanism for the support of external systems. Journal of Personality and Social Psychology,95
(1), 18–35. https://doi.org/10.1037/0022-3518.104.22.168
Kay, A. C., Shepherd, S., Blatz, C. W., Chua, S. N., & Galinsky, A. D. (2010). For God (or) country: The hydraulic
relation between government instability and belief in religious sources of control. Journal of Personality and
Social Psychology,99(5), 725–739. https://doi.org/10.1037/a0021140
Kleiber, C., & Zeileis, A. (2008). Applied econometrics with R. Springer-Verlag.
Lang, M., Purzycki, B. G., Apicella, C. L., Atkinson, Q., Bolyanatz, A., Cohen, E., Klocová, E. K., Handley, C.,
Lesorogol, C., Mathew, S., McNamara, R., Moya, C., Placek, C., Soler, M., Vardy, T., Weigel, J., Willard, A.,
Xygalatas, D., Norenzayan, A., & Henrich, J. (2019). Moralizing gods, impartiality and religious parochialism
across 15 societies. Proceedings of the Royal Society B: Biological Sciences,286(1898), 20190202. https://doi.org/
Malinowski, B. (1948). Magic, science and religion. Doubleday.
McDonald, J. F., & Moﬃtt, R. A. (1980). The uses of Tobit analysis. The Review of Econoics and Statistics,62(2), 318–
McKay, R., Eﬀerson, C., Whitehouse, H., & Fehr, E. (2011). Wrath of God: Religious primes and punishment.
Proceedings of the Royal Society of London B: Biological Sciences,278(1713), 1858–1863. https://doi.org/10.1098/
McNamara, R. A., Norenzayan, A., & Henrich, J. (January 2016). Supernatural punishment, in-group biases, and
material insecurity: Experiments and ethnography from Yasawa, Fiji. Religion, Brain & Behavior,6(1), 34–55.
Norenzayan, A. (2013). Big gods: How religion transformed cooperation and conﬂict. Princeton University Press.
Norenzayan, A., & Shariﬀ,A.F.(2008). The origin and evolution of religious prosociality. Science,322(5898), 58–62.
Paes-Machado, E., & Noronha, C. V. (2002). Policing the Brazilian poor: Resistance to and acceptance of police bru-
tality in urban popular classes (Salvador, Brazil). International Criminal Justice Review,12(1), 53–76. https://doi.
Power, T. J., & Taylor, M. M. (Eds.). (2011). Corruption and democracy in Brazil: The struggle for accountability.
University of Notre Dame Press.
Prandi, R. (1991). Segredos guardados: Orixás na alma brasileira. Companhia Das Letras.
Purzycki, B. G. (2013). The minds of gods: A comparative study of supernatural agency. Cognition,129(1), 163–179.
148 M. SOLER ET AL.
Purzycki, B. G., Apicella, C., Atkinson, Q. D., Cohen, E., McNamara, R. A., Willard, A. K., Xygalatas, D., Norenzayan,
A., & Henrich, J. (2016). Moralistic gods, supernatural punishment and the expansion of human sociality. Nature,
530(7590), 327–330. https://doi.org/10.1038/nature16980
Purzycki, B. G., & Arakchaa, T. (2013). Ritual behavior and trust in the Tyva Republic. Current Anthropology,54(3),
Purzycki, B. G., & Sosis, R. (2011). Our gods: Variation in supernatural minds. In U. J. Frey (Ed.), Essential building
blocks of human nature (pp. 77–93). Springer-Verlag.
Rabelo, M. (2016). Considerações sobre a ética no candomblé. Revista de Antropologia,59(2), 109–130. https://doi.
R Core Team. (2019). R: A language and environment for statistical computing. R Foundation for Statistical
Roes, F. L., & Raymond, M. (2003). Belief in moralizing gods. Evolution and Human Behavior,24(2), 126–135.
Shariﬀ, A. F., & Norenzayan, A. (2007). God is watching you priming God concepts increases prosocial behavior in an
anonymous economic game. Psychological Science,18(9), 803–809. https://doi.org/10.1111/j.1467-9280.2007.
Shariﬀ, A. F., & Norenzayan, A. (2011). Mean gods make good people: Diﬀerent views of God predict cheating behav-
ior. International Journal for the Psychology of Religion,21(2), 85–96. https://doi.org/10.1080/10508619.2011.
Soler, M. (2012). Costly signaling, ritual and cooperation: Evidence from candomblé, an Afro-Brazilian religion.
Evolution and Human Behavior,33(4), 346–356. https://doi.org/10.1016/j.evolhumbehav.2011.11.004
Soler, M. (2016). The faith of sacriﬁce: Leadership trade-oﬀsin an Afro-Brazilian religion. Human Nature,27(4),
Sosis, R., & Handwerker, W. P. (2011). Psalms and coping with uncertainty: Religious Israeli women’s responses to
the 2006 Lebanon War. American Anthropologist,113(1), 40–55. https://doi.org/10.1111/j.1548-1433.2010.01305.x
Wickham, H. (2016). ggplot2: Elegant graphics for data analysis. New York: Springer-Verlag.
Whitehouse, H. (1996). Apparitions, orations, and rings: Experience of spirits in Dadul. In J. Major & A. Howard
(Eds.), Spirits in culture, history and mind (pp. 173–194). Routledge.
Wooldridge, J. M. (2016). Introductory econometrics: A modern approach. Thomson South-Western (6th ed.).
Zuckerman, P. (2007). Atheism: Contemporary numbers and patterns. In M. Martin (Ed.), The Cambridge compa-
nion to atheism (pp. 47–65). Cambridge University Press.
RELIGION, BRAIN & BEHAVIOR 149