Some people do hate our language. My stepdad was always close to tears if he heard someone cuss. Oh, that was hurting his god.

@DenoPenno Your stepdad would have been very uncomfortable around me

Why would you "hate" a Thing, especially a nonsensical thing?
Seems like a big fat waste of time to me.......

How did I know that it was you who had commented, before I even opened up the post?

Your "theory" of the evolutionary utility of religion is interesting, but I remain skeptical. Here's why: biological evolution operates by natural selection eliminating unfit individuals from a population. This only works because elimination of the individual means they can't pass on the heritable factors (genes) responsible for their unfitness. More fit individuals enjoy higher rates of reproduction, and over time the gene pool (and phenotypes in the population) change. Your idea implies that there is a religion gene (or set of genes), and there is no evidence for that. Religiosity is not an inherited condition. It is a cultural artifact, something that is imprinted on individuals after birth. In many cases, the choice to profess a religion (or not) IS a matter of life or death: being accused of heresy can be hazardous to one's health. It may be an advantage to "believe," but it's not an evolutionary advantage, because it's not a heritable trait.

@Flyingsaucesir

You knew because you know me to be a passionate supporter of science over popular opinion.

It's not "my" theory. Google it. Help me find a single scientific paper, by someone who studies that particular behavior, which attributes the development of religion to something other than the various mechanisms of evolution. I post paper after paper here describing the current scientific consensus on the matter and constantly invite everyone to post papers to the contrary, but so far, no one has provided one. YOU COULD BE THE FIRST!!

( RESTRICTIONS APPLY: Paper must have been written this millennium, and not just an opinion piece, but a science-based paper with references, etc. similar to this: [core.ac.uk] )

@skado In a very broad sense, every human behavior can be attributed to "the various mechanisms of evolution." We're here, we're products of evolution, therefore everything we say or do can be attributed evolution. But that does not necessarily mean religion exists because it confers an evolutionary advantage.

I am not at all inspired to go chasing rainbows. But if you can come up with an article by some eminent evolutionary biologist, a Dawkins or Gould or Mayr, someone with real gravitas, that treats the subject, I will read it.

@Flyingsaucesir
To my knowledge, none of those three scientists specialized in the study of religious behavior. Opinions maybe, but did they publish scientific studies on it? There are few 20th century scientists with more gravitas in the science world ( it's not a popularity contest ) than E. O. Wilson. Check him out.

Did you read the link I supplied above?
.

@skado I like E.O. Wilson.

@skado This is what that article looks like on my phone (the only device I have with Internet access).

@Flyingsaucesir

"In his new book The Social Conquest of Earth (2012), naturalist E. O. Wilson argues that our best chance at understanding and advancing morality will come when we “explain the origin of religion and morality as special events in the evolutionary history of humanity driven by natural selection.” This is a bold claim, yet a familiar one for Wilson, who has been advocating something like this approach to human morality ever since his landmark 1975 work Sociobiology."
[thenewatlantis.com]

"Wilson's parents were Southern Baptists though he was also raised by conservative Methodists. He abandoned Christianity before college, later describing himself as a "provisional deist" and agnostic. In On Human Nature, he argued that belief in God and rituals of religion are products of evolution.
[ffrf.org]

@Flyingsaucesir

761© Springer Science+Business Media Dordrecht 2015
T. Heams et al. (eds.), Handbook of Evolutionary Thinking in the Sciences,
DOI 10.1007/978-94-017-9014-7_36
Chapter 36
Origins and Evolution of Religion
from a Darwinian Point of View:
Synthesis of Different Theories
Pierrick Bourrat

Abstract The religious phenomenon is a complex one in many respects. In recent
years an increasing number of theories on the origin and evolution of religion have
been put forward. Each one of these theories rests on a Darwinian framework but
there is a lot of disagreement about which bits of the framework account best for the
evolution of religion. Is religion primarily a by-product of some adaptation? Is it
itself an adaptation, and if it is, does it benefi ciate individuals or groups? In this
chapter, I review a number of theories that link religion to cooperation and show that
these theories, contrary to what is often suggested in the literature, are not mutually
exclusive. As I present each theory, I delineate an integrative framework that allows
distinguishing the explanandum of each theory. Once this is done, it becomes clear
that some theories provide good explanations for the origin of religion but not so
good explanations for its maintenance and vice versa. Similarly some explanations
are good explanations for the evolution of religious individual level traits but not so
good explanations for traits hard to define at the individual level. I suggest that to
fully understand the religious phenomenon, integrating in a systematic way the dif-
ferent theories and the data is a more successful approach.

1 Introduction
Over the last 15 years, the religious phenomenon has sparked an increasing interest
among evolutionists. Although there is only one Darwinian Theory of evolution,
there is a myriad of theories proposing an evolutionary and Darwinian explanation
of the origin and evolution of religious beliefs and practices. An obvious reason to
this is that religion is an extremely complex phenomenon which can be carved not
only in different cultural traits that can themselves be studied independently, but
also from different points of view. Indeed, one can study rituals, beliefs in super-
natural entities or the economical aspects of religions which all belong to the
762
phenomenon of religion. Matching with these different aspects of religion, some
theories focus more on the cognitive aspect of religion while others see the phenom-
enon from a larger perspective. Yet, the complexity of the phenomenon is not the
only reason for the numerous Darwinian theories of religion. Another reason is
sociological. Although all of these theories claim to be Darwinian, they have been
developed in different fields. Some come from evolutionary biology, others from
psychology, cognitive sciences, anthropology or economics. Those fields compre-
hend Darwinism in different ways. For example, a postulate of evolutionary
psychology is that many of the human psychological adaptations have been shaped
in an ancestral environment which was very different from our modern setting. This
would have led some of our behaviors to be mismatched with the modern environ-
ment. However, this is not a postulate of human behavioral ecology or evolutionary
anthropology. Indeed, in these fields it is postulated that our modern behaviors are
optimal from an adaptive point of view (relatively to some tradeoffs an organism
cannot avoid). These different ways of conceptualizing evolutionary mechanisms
and Darwinism have subsequently led to different paradigms when religion has
started to be a hot topic in evolutionary sciences.

Aside from their number, another remarkable fact of the Darwinian theories of
religion is that they are often presented as incompatible or mutually exclusive.
For example, according to different theories, the beliefs in supernatural agents are
either a byproduct of our social evolution or an adaptation. And among the adap-
tive theories, these beliefs are either individual adaptations with the function of
maintaining one’s reputation or group adaptations with the function of reinforcing
social cohesion. This kind of divergence is, I have already noted, partially explained
by the different origins of the Darwinian theories of religion. Yet, in many cases,
I will show that these oppositions are fallacious. In fact, it is possible to integrate
the different theories if one pays attention to what explanatory level, level of orga-
nization and temporal scale they operate at. At the end of the chapter, I will argue
that each of these oppositions is often the result of a lack of precision in the phe-
nomena explained the theories. While being rarely expressed, some theories explain
the origin of some religious phenomena while other theories explain some reli-
gious phenomena in an actual context. Similarly, the different theories study one
and the same phenomenon using different scales, whether spatial or temporal.
More generally, this chapter will be an integrative synthesis of the different
Darwinian theories of religion. The integration I will propose will have the benefi-
cial effect to palliate different problems encountered by each separate theory while
still keeping their explanatory power. Although some research has been conducted
in order to treat several levels of organization at once with religion (see for example
Johnson and Bering 2009; Roes and Raymond 2003; Sosis and Alcorta 2003; Sosis
et al. 2007), more work needs to be done in order to understand the articulation of
these different levels. My aim in this chapter will be an attempt to make this articu-
lation. For each theory I present, I will ask the following questions “Does this
theory give an explanation of a phenomenon encountered in religions or does it
provide an explanation of the stabilization of this phenomenon?” or “Can this
P. Bourrat
763
theory be treated independently from any other or does it relate to (even implicitly)
parts of other theories?”
This synthesis, however, will not be exhaustive. There are at least two reasons for
that. First, disagreements within each discipline exist and presenting them would
exceed the scope of this chapter. Second, among the different perspectives one can
adopt on religion, my synthesis will mostly focus on cooperation. Indeed, many
theories propose that one of the evolutionary functions of religion is to increase
cooperation between individuals. The problem of cooperation is one of the most
important in evolutionary sciences and focusing on religion from this point of view
is up to now the most fertile research program involving Darwinian Theory.
However, it would be inaccurate to claim it is the only one.

The theories I will examine will be classified in three categories. The first
category is the by-product theory of religion (Barrett 2000 ; Boyer 2001 ; Boyer and
Ramble 2001 ) which is undeniably the most popular and also the strongest from an
empirical point of view. According to this theory, the religious phenomenon, or
more precisely the beliefs in supernatural agents, is not an adaptation. Rather, its
proponents believe it is a by-product of other adaptations to sociality. Another cor-
pus of theories, on the contrary, seeks to explain the religious phenomena as past or
actual adaptations. These theories are the theories stemming from Evolutionary
Psychology (Dennett 2006 ), the fear of supernatural punishment theory (Johnson
and Bering 2009 ), the costly signaling theory of religion (Alcorta and Sosis 2005 ;
Cronk 1994 ; Sosis and Alcorta 2003 ; Sosis et al. 2007 ) and the kleptocracy theory
(Diamond 1997). Finally, a third type of theories sees religion from a multilevel and
adaptive point of view. In particular, they consider the group level as a privileged
level upon which selection is acting (Roes and Raymond 2003 ; Snarey 1996 ; Wilson
2002 , 2005 ). Following the three types of theories I have just pointed out, the
remainder of this chapter will be divided into three sections. The first section will
focus on the by- product theory of religion, while sections two and three will focus
on adaptive theories of religion that regard the individual and the group as the privi-
leged level of selection respectively. In sections two and three, the theories I will
present are solutions to the problem of cooperation. I will show that schematically
there are two potential solutions involving religion with the problem of cooperation.
These solutions match with the theories focusing on the individual and those focus-
ing on the group. The first solution proposes that an individual cooperate because
their reputation is involved. If they do not cooperate, their reputation and conse-
quently their fi tness might suffer from it. The second solution focuses on the idea
that selection operates at the level of the group. Individuals cooperate because it
increases group fitness, sometimes at the expense of their own fitness. If they do not,
their group might get extinct or be outcompeted by another one. The division
between these two forms of adaptive theories will be very schematic. Indeed, soon
I will show that in some theories the level upon which selection acts will be difficult
to delimitate; it will be possible to conceive that individuals cooperate both for
themselves and their group.
36 Origins and Evolution of Religion from a Darwinian Point of View…
764
2 The By-product Theory of Religion
I will first present the by-product theory of religion, which was originally developed
within the field of cognitive sciences. It proposes that the beliefs in supernatural
agents are principally the consequence of selective pressure cognitive traits which
are (or were) critical for the survival of individuals in their environment, rather than
an adaptation. Buss et al. (1998 , p. 537) define by-products as “[…] characteristics
that do not solve adaptive problems and do not have to have functional design. They
are carried along with characteristics that do have functional design because they
happen to be coupled with those adaptations.” In the present, case with belief in
supernatural agents, two traits or characteristics have been distinguished.

2.1 Two Critical Traits
The fi rst of these two traits is what Justin Barrett (2000 ) calls Hypersensitive Agency
Detection Device (HADD). This expression comes from an argument developed by
Stewart Guthrie who proposes that humans are naturally predisposed to detect
agents in their environment when such agents do not actually exist (Barrett 2000 ;
Guthrie 1993 ; Tremlin 2006 ). One classical example given by Stewart Guthrie
( 1993 ) is the clouds one can sometimes see shaped as faces, a phenomenon which
gave the title to his book. According to Guthrie, such a perceptual device (although
biased) could have been adaptive in our evolutionary past. Indeed, in an uncertain
environment failing to detect an agent could have had much more dramatic conse-
quences for survival than detecting an agent where there was none. Let us imagine,
for example, that this agent is an enemy or a predator: failing to notice its presence
in time could be fatal. An analogy between this type of cognitive bias and a fire
alarm can be made. If a fi re alarm is very sensitive and goes off very often, even
when there is no fire, it will always be a better alarm than an alarm which does not
go off when there is a fi re. Because of this asymmetry, one should expect a fire
alarm to be designed in a way that it goes off more often than there are fires.
Following the same reasoning but applying it to an evolutionary context, one should
expect that humans (and other organisms) detect agents in their environment more
often than there actually are. Beliefs in supernatural agents would be the result of a
manifestation of this cognitive bias.

The second trait, which together with HADD makes belief in supernatural agents
possible, is the theory of mind module. Theory of mind is defined as the ability for
one individual to impute mental states to others (Premack and Woodruff 1978 ). For
obvious reasons, this ability is crucial for humans. Although there is no or little
research made on this topic, it seems that people who do not have it or possess it in
a very limited way, such as autistic people (Baron-Cohen 1995 ; Bloom 2004 ; Pinker
1997 ), develop beliefs in God which are very different from the majority of us.
These beliefs do not imply the existence of God as a supernatural agent but more as
P. Bourrat
765
a force or general principle (Atran 2002 ; Bering 2002 ; Tremlin 2006 ). The key point
with the theory of mind and its relation to supernatural agents is that it suggests that
humans are predisposed to conceptualize anthropomorphically the supernatural
agents in which they believe, in other words, with human traits and preoccupations.
For example, Barrett and Keil ( 1996 ) have shown that when students who claim not
to have an anthropomorphic understanding of God, were asked to answer very
quickly to questions about its supernatural powers, they were shaping their answer
anthropomorphically. In fact, although the participants of this experiment had a
conception of God as omnipotent (i.e. non-anthropomorphic) in their religion, when
they had to answer quickly and intuitively to some questions about God, their
answers showed that they were not conceptualizing it as omnipotent as they were
claiming. Rather they thought of God as being able to execute only one action at a
time (i.e. anthropomorphic).
Boyer ( 2001 ), from a rather anthropological background, and Barrett (2004 ),
from an experimental psychological background, both propose a detailed account of
the by-product theory of religion.

2.2 Religion Is Too Complex to Be Only a By-product
Although very young, the Cognitive Science of Religion from which the by-product
theory of religion emerged, could become extremely fertile in the near future
because it proposes new approaches to study religion. Yet, one can detect a limit to
this approach and more specifically to the by-product theory: whereas it can provide
an explanation of the emergence or the origins of some religious behaviors and
beliefs, its explanatory power plummets when one tries to explain why supernatural
agents, religions and rituals can sometimes be so elaborated. Let’s recall the defini-
tion of a by-product from Buss et al. (1998 ) given earlier. If one follows this defini-
tion and subscribes to the by-product theory, beliefs in supernatural agents are only
a by-product of HADD and of the Theory of Mind module. However, it is legitimate
to question why, on top of their anthropomorphic characteristics, those agents have
other characteristics such as being all-powerful, benevolent or having created the
universe etc. The proposition that religion is only a by-product of the architecture of
our mind (directly inherited from our evolutionary past) implicitly commits us to
deny that from the emergence of those by-products, natural selection did act upon
them. Yet, it seems a priori reasonable to think that some processes of natural selec-
tion whether biological or cultural did happen since then. Some characteristics of
supernatural agents are observed with regularity and transculturally, and they do not
seem to have no direct links either with the HADD or with the Theory of Mind
module. It is moreover difficult to conceive how these characteristics could only be
cognitive by- products. The theories that I will examine in sections two and three
propose precisely that the different patterns one can observe are adaptations or at
least have an adaptive role in human societies. Thus, by-product theory – although
an important hypothesis for the origin of religion and a starting point for other
36 Origins and Evolution of Religion from a Darwinian Point of View…
766
evolutionary theories of religion – does not allow for a complete explanation of the
beliefs in supernatural agents, let alone the religious phenomenon in its totality.

3 Adaptive Theories at the Individual Level
3.1 Religion as an Ancestral Adaptation
Which Is Not Anymore: “Sweet-Tooth”
Theories of Religion
“Sweet-tooth theories of religion” is a label given by Daniel Dennett (2006 ) to a
body of evolutionary theories on religious beliefs and practices. The main idea
behind this label is that religion has incorporated some characteristics which were
adaptive in our evolutionary past, but nowadays they are not anymore. Dennett com-
pares religions to the sweet and fatty foods most of us like, which were probably
rare for our ancestors. There is little doubt that such food was providing a selective
advantage to the individuals who had access to it. Yet, in our modern environment,
this kind of food is not rare anymore and have deleterious effects (such as hypercho-
lesterolemia, diabetes etc.) if consumed in excessive quantities. In spite of this mis-
match, it is still attractive for us. Mismatch theory is a classical concept of
evolutionary psychology. A mismatch occurs when an individual is not adapted to
its environment any more. The cause of the mismatch is that the environment has
evolved too quickly for new adaptive solutions to be found. We can use the idea that
we are not adapted to our modern environment rich in sweet and fatty food, in
regards to religion. In fact, it is possible to conceive religious beliefs and practices
as costly or neutral (from an evolutionary point of view) nowadays, when they incite
individuals to cooperate more. An individual may incur net costs by cooperating in
an anonymous society where no one else cooperates. However, cooperating through
the means of religion may have been evolutionarily advantageous when conditions
where different. Advantages may have been, for example, to allow for a better coop-
eration within small groups of individuals, so that beliefs and practices which made
cooperation easier would have been selected. Subsequently, according to sweet
tooth theories, because the environmental conditions have suddenly changed, the
beliefs and religious practices lost their adaptedness (that is, their function in social
cohesion) but remained attractive for the human mind. Thus, under this view,
Christianity, Judaism or Islam could be seen as cultural phenomena which have
been invented and selected by humans because they reunite a combination of attrac-
tive ideas, yet they would have no evolutionary function nowadays. To push our
analogy between food and religion a little further, our sweet tooth and taste for fatty
food pushed humans to create and eat fast food which has a lot of success all around
the world in spite of its deleterious effects. Similarly, our spiritual intuitions which
were adaptive in the past might have had a great influence on our modern religions
and would simply be a very good synthesis of the different traits which were one
day adaptive.
P. Bourrat
767
This idea fits perfectly with the framework of Evolutionary Psychology and
would deserve further developments more. Yet, Dennett does not point out any pre-
cise study confirming his ideas. Indeed, he remains rather vague on the potential
function religion might have had. Although it might be difficult to test hypothesis
about the past, it is worth verifying the compatibility of those hypotheses with the
different evolutionary scenarios of human evolution. But even if research could be
done, it would have to show that religion does not have any influence on fitness
nowadays. The same demonstration would also need to be done with the by-product
theory of religion. But in the latter case it would be less problematic since the by-
product theory makes almost no assumptions on the ancestral environment in which
beliefs in supernatural agents did appear except that it was a social one. Finally,
although it seems quite probable and reasonable to think that some religious traits
have been adaptive in our evolutionary past, it seems equally reasonable to think, as
I already mentioned in the previous section, that the same or new religious traits are
adaptive today. Indeed, some authors consider the by-product and sweet-tooth theo-
ries as insufficient to fully explain the religious phenomena, and argue that some
religious traits do have an adaptive role nowadays.

3.2 Religion as a Contemporary Adaptive
Phenomenon at the Individual Level
3.2.1 The Fear of Supernatural Punishment Theory
The fear of supernatural punishment theory, mainly defended by Jesse Bering and
Dominic Johnson (Johnson 2005 ; Johnson and Bering 2009 ), also has its roots in
cognitive sciences. The Theory of Mind and HADD have, according to this theory,
strictly the same role that in the by-product theory. However, this theory takes into
account a new constraint on evolutionary dynamic which is different from all the
constraints the by-product theory assumes, namely language. This constraint is the
most important pillar of the fear of supernatural punishment theory. Bering and
Johnson propose that together with the birth of human language and Theory of
Mind, reputation becomes an essential characteristic of humans. According to them,
language permits a rapid dispersal of ideas and information of the type “who did
what”. The emergence of human-like language would hence impose nolens volens
new selective pressures on individuals through the reputation they acquire. A bad
reputation is extremely disadvantageous and cooperation might be a strategy which
would mitigate this problem by preventing an individual from acquiring such a rep-
utation. Although it can be costly to cooperate, the net balance between the costs
and benefi ts of cooperation is often positive in contexts where reputation is
important.

Starting from these hypotheses, Johnson and Bering propose that beliefs in
supernatural agents are means to avoid the consequences of a bad reputation by
increasing individual’s cooperation. Let us recall that HADD and the Theory of
36 Origins and Evolution of Religion from a Darwinian Point of View…
768
Mind module are responsible for the concepts of anthropomorphic supernatural
agents. The HADD erroneously recognizes agents in the environment and the
Theory of Mind gives these agents human-like intentions. Let us now posulate that
an individual believes that an ancestor, or any other supernatural agent they believe
in, will punish them if they act antisocially within their community. Johnson and
Bering consider that, on average, such an individual will acquire a better reputation
relatively to another individual who does not have these beliefs since the former,
because of their fear in supernatural punishment, will have an incentive to have less
antisocial behaviors. This will lead our individual to have a fitness—classically
defi ned as the product of the survival and reproduction— ceteris paribus superior to
an individual who does not have such beliefs. If those beliefs are heritable, they will
be transmitted to the next generations and selected by natural selection because of
the selective advantage they procure: if an individual believes they are constantly
observed by someone, as it could be the case if they believe in the existence of a god
who can punish them, their incentive (whether conscious or unconscious) to be
prosocial1 will be strong since they will feel monitored by this god. However, such
constraint is not imposed on individuals who do not have beliefs in supernatural
agents, and although they might overall be prosocial, they will be more likely to be
caught while committing acts of selfish behavior which will damage their reputation
and, as a result, reduce their fi tness.

The fear of supernatural punishment theory seems at first glance promising and
contrary to sweet-tooth theory easily testable. As a matter of fact, Johnson and
Bering claim that this theory has an empirical support (Bering et al. 2005; Johnson
2005). Yet, it has many limits and its testability is an extremely delicate matter.
Enumerating these limits here would be beyond the scope of this article, and for that
reason I will restrain my analysis to the most important of them (for more details on
these limits see: Bourrat et al. 2011). One of the most important limits of Johnson
and Bering’s ideas is that they are presented as a theory in which the only important
level of selection is the individual level. Yet, reputation is only a strong constraint for
a given individual with the members of the group she interacts with. In fact, it intui-
tively seems much more problematic when your neighbor has a bad opinion about
you rather than someone you do not know and who lives 3,000 km away. Once this
remark is taken into account seriously, it becomes clear that the fear of supernatural
punishment theory can be of great help for understanding beliefs in moralizing
supernatural agents from a Darwinian point of view, but only in the context of
unique or isolated groups. Nowadays, the context is different, numerous groups of
individuals with different beliefs exist and virtually none of them is isolated. The
fear of supernatural punishment theory as proposed by Bering and Johnson is insuf-
ficient to give an account of all the characteristics linked to cooperation and the
beliefs of the different supernatural agents one can observe in different groups. This
theory can hardly explain why different human groups have beliefs in different
supernatural agents and punishments or why the supernatural punishments believed
in one group are not believed as threatening in other groups. At best, the fear of
1 By “prosocial” here I mean “having a propensity to cooperate with others”.
P. Bourrat
769
supernatural punishment theory is part of a larger explanation of the evolution of
beliefs in supernatural agents. That said, it is more refined when compared to the
by-product theory because it provides a partial explanation of why humans often
believe in supernatural agents who have moral virtues and are inclined to punish.
Another limit to this theory comes from its mitigated empirical results. Johnson
( 2005 ) has tested it transculturally using a database of 186 societies and with more
than 2,000 variables. Some of them were linked to religious and supernatural beliefs
while other being were to cooperation. As he claims, the results reinforce the
hypothesis of fear of supernatural punishment as a mechanism explaining (at least
partially) the different levels of cooperation, whether it occurs directly between
individuals (e.g., not robbing your neighbor) or are institutionalized (e.g., the exis-
tence of a currency or a police as part of the society). Bourrat et al. (2011 ) have
tested this hypothesis again using the same database, but they started with the prem-
ise that the fear of supernatural punishment hypothesis should be valid not only for
supernatural agents which are not gods or ancestors (while Johnson was only con-
sidering high gods) but also for any other forms of supernatural punishment, as
Johnson and Krüger (2004 ) initially proposed it. Indeed, no criteria in the theory can
justify a special treatment for religious supernatural agents . Bourrat and colleagues
also underlined that the theory can only account for the direct cooperation between
individuals and not for institutional cooperation. The results they obtained could not
confi rm the fear of supernatural punishment hypothesis. That said, they consider
that the anthropological database they used is not fine-grained and precise enough
to detect personal beliefs. Atkinson and Bourrat (2011 ) have therefore tested this
hypothesis yet again, using a database reporting personal beliefs. Their result seems
to support the fear of supernatural punishment hypothesis in its broader version, that
is, which is not limited only to supernatural agents.

3.2.2 The Costly Signaling Theory Applied to Religion
The costly signaling theory applied to religion focuses on another specific aspect of
the religious phenomenon, namely rituals. However, this theory has been developed
from the perspective of cooperation, similarly to the theories I have discussed
earlier.
The costly signaling theory has been developed mostly by evolutionary biolo-
gists and has its roots in the work of John Krebs and Richard Dawkins (Dawkins
and Krebs 1978 ; Krebs and Dawkins 1984 ) on manipulation of signals, and of
Amotz Zahavi (1975 ) on the handicap principle. Krebs and Dawkins propose that
signals should be seen as attempts to manipulate individuals rather than attempts to
inform them. This idea is one of the pillars of Dawkins’ concept of extended pheno-
type (Dawkins 1982 ), in which the phenotype of an individual is not only the direct
expression of their genes, but also the way this individual modifies their environ-
ment including others’ behaviors. In a revised version of the manipulation of signals
theory, Krebs and Dawkins (1984 ) defend the idea that the use of signals should be
regarded as manipulative or cooperative attempts depending on signaler and receiver
36 Origins and Evolution of Religion from a Darwinian Point of View…
770
individuals’ interests. When signalers and receivers do not have overlapping
interests, receivers should undergo strong selection pressure to detect and resist the
attempt of manipulation from the signaler. In return, this should create a selective
pressure on signalers to develop new strategies of manipulating the receivers and so
forth ad infi nitum . This phenomenon is typical of arm races such as described by the
Red Queen Hypothesis (Van Valen 1973 ). Krebs and Dawkins argue that when sig-
nalers and receivers have overlapping interests, the signal is expected to be simpler:
manipulation is not in the evolutionary interests of the signaler since it is also in the
interests of the receiver to cooperate.

The handicap principle proposes, among other things, that males, in a context of
sexual selection, should display their qualities to females by inflicting a cost to
themselves (Zahavi 1975 ; hence the name of the principle). For example the long
peacock tail displayed by males is very costly to have. Nevertheless, because a long
and colorful tail increases the chances for an individual to be predated and is costly
to produce and maintain, a peacock male with such a tail signals to females that in
spite of this handicap it has been able to survive. A female choosing to mate with
this male consequently selects his qualities.

In the framework of Krebs and Dawkins, the handicap principle matches with a
case of cooperation between signalers and receivers (hence no arm race between
them is expected), but in a context of possible defection. Indeed, Krebs and
Dawkins draw their attention to many cases of signals where imitation is impossi-
ble, because there is an indispensable link between the signal and the underlying
necessary condition for it to exist. This is the case, for example, between body size
and frequency of vocal signals. In this case, no arm race between signalers is
expected since no “cheap” imitation will be convincing. Yet, as suggested by
Zahavi (1975), even when such a link between the signal and the underlying condi-
tion for this signal does not exist naturally, signals which are costly to produce
would be honest signals for the receivers and would allow them to distinguish good
quality signals from bad quality ones. In those conditions, a signal will be honest
if and only if the cost of imitation is as or more important than the benefits that
would be received from that imitation (Grafen 1990) and an arm race between
signalers is expected until a point where only some individuals can pay the costs.
This idea precisely fits the case of the peacock tail. There is no direct trait naturally
measuring the propensity of a male to avoid predators. Yet, the cost of producing a
long tail becomes a reliable signal for females who will choose the males with the
longest tails.

Several scholars have proposed to apply these ideas to human rituals (Cronk
1994; Irons 1996; Sosis and Alcorta 2003). They suggest that the function of ritu-
als is to promote cooperation within a group, community or society which will
only be possible if the rituals are costly to exercise. For example, we can find
frightening or painful rites of passage ceremonies, including tattooing and scarifi-
cation in numerous societies. These rituals, as well as many others, can be inter-
preted as signals displaying a commitment to the community or wider society.
Individuals undergoing them show that they truly want to be part of the community
P. Bourrat
771
since they are willing to pay a price for it. By doing so they demonstrate that in
future they will not free ride when it will be possible since the cost of free riding
would be superior to the cost undergone during the ceremony or ritual. For
instance, Sosis et al. (2007) have proposed that the tattoos, genital mutilations and
scarification that some men undergo during different rituals signal the belonging
to a group. When a man accepts to be marked by tattoos, scarification and mutila-
tions, it will be impossible for him to move from the group to another one in
future. Indeed with the permanent mark of the group he belonged initially it would
be very hard for him to be accepted in another group in full trust. Hence, these
definitive marks represent reliable signals of future cooperation since being banned
from a group would be extremely costly. However, a signal can be costly in differ-
ent ways: the same absolute cost can be undergone in one shot, as it is the case
with initiations, or undergone over time, as it is the case with rituals such as catho-
lic masses (cost in terms of time spent). These two “types” of costs perfectly cor-
relate with the two forms of religiosity one can find transculturally: “doctrinal”
and “imagistic”(Atkinson and Whitehouse 2011; Whitehouse 2004). In a doctrinal
mode, religiosity is acquired through the practices of frequent but low arousal ritu-
als such as masses. In the imagistic mode, religiosity is acquired through the prac-
tices of highly arousing but very infrequent rituals, such as initiations which
individuals remember all their life.

The costly signaling theory applied to rituals has a promising explanatory
power and there is already interesting research suggesting its value. This is the
case with the work of Richard Sosis and his colleagues (Sosis et al. 2007) on scari-
fications and on American communities in the nineteenth century (Sosis 2000;
Sosis and Bressler 2003). Furthermore, the costly signaling theory can be applied
to a vast range of human behaviors and in different contexts. In the field of reli-
gion, for example, Terence Deacon (1997) proposes that the commitment of
women to religion such as Christianity or Judaism, in some societies, could be an
honest signal of their fidelity. The rationale behind this idea is that pious women
should be, in principle, more reluctant to be unfaithful than other women. The fear
of supernatural punishment could be, in this case, the underlying mechanism of
some honest religious signals. As there is no direct observable expression of fidel-
ity, piety can be one proxy for it. Hence, demonstrations of piety are expected to
be stronger and more explicit when the fidelity of women is under selection pres-
sure. This prediction is supported by the research of Boster et al. (1998) on cer-
tainty of paternity which is superior with Jewish priests. The authors argue that the
laws of sexual purity to whish Jewish women are committed is the reason of this
result. Such commitment can be translated in terms of costly signals as Sosis and
Alcorta (2003) do. That said, this theory which comes directly from behavioral
ecology and mostly focuses on the biological individual has some limits. For
example, it is hard to conceive why some cultural traits increasing the fitness of
individuals on the long term (such as rituals) would be so widespread in human
societies without invoking a form of cultural group selection. I will come back to
this problem in the fourth section.

772
3.2.3 The Kleptocracy Theory of Religion
The kleptocracy theory, although being very general, is not a theory of religious
beliefs and practices that one can apply to any society, unlike the theories we exam-
ined thus far. Rather, it focuses on the evolution of religion in agrarian societies, the
only ones in which an important social stratification is possible. The concept of
kleptocracy has its roots in the Marxist argument proposing that religion is created
by the societies’ elites in order to materially exploit the people who only receive the
secondary or derived benefits of a low but constant level of security and productivity
(Atran 2002 ).
According to Jared Diamond (Diamond 1997 ), the “true” religions only emerge
when a central authority assimilates the beliefs in the supernatural in order to estab-
lish a pyramidal scam. Supernatural agents, in this case, can be conceived as a mean
of reinforcing the power of kleptocrates (the ruling elite). The vast majority of indi-
viduals is exploited by them, but this system of exploitation is cooperative.
Kleptocrates maintain the social cohesion of the group by force directly, but also
and more importantly most the people truly believe that the ruling elite have a direct
link with their god(s). Thus, this justifies their privileged position. It is almost cer-
tain that in past or present societies, the elite are healthier than other individuals (see
Diamond 1987 ), especially in agrarian systems where food storing is the rule.
Another demonstration of elite’s privileged position is Laura Betzig’s research
( 1986 ), which demonstrated transculturally with the help of anthropological data
that the level of despotism is positively correlated to the size of harems in a society.
Hence, we can imply that the fitness of the elite is usually superior to the fitness of
the people in those societies.

Let us note that kleptocracy theory is compatible with the fear of supernatural
punishment theory. This will be the case if one given individual believes that they
will be punished by a god if they do not obey their monarch or sovereign. However,
the fear of supernatural punishment is only one of the possible mechanisms that can
reinforce obedience and submission, since cooperation between individuals might
as well be the result of threats and direct punishments from the elite. Likewise, the
enforcement of rituals might increase cooperation between individuals, as it has
been recently argued: the synchronization of movements one can observe during
rituals would have such an effect on people (Wiltermuth and Heath 2009 ). These
two examples, which show how the elite can exploit the people for them to cooper-
ate using religion as a justification, are however far from being exhaustive.
Although kleptocracy theory is very general and potentially embeds some theo-
ries I already examined, one question remains unanswered. If a kleptocracy
decreases the fitness of most individuals but the elite, why were there so many klep-
tocracies in human history and why have they been so successful? One possible
answer I propose is that the fitness of an individual in a kleptocratic society, in spite
of being lower when compared to the fitness of an individual in a non-kleptocratic
society, will become higher when the two societies will be in competition with each
other. Indeed, a centralized power such as the one we can encounter in kleptocratic
societies allows for the cooperation of a higher number of individuals which are
P. Bourrat
773
more specialized (e.g. army, farmers etc.) than in a non kleptocratic society. These
two parameters are critical when there is an intergroup competition (Alexander
1987 ). I will come back later to this point in the next section.
One of the strengths of kelptocracy theory is that it can allow the integration of
many other theories from different fields, whether cognitive sciences, ecology or evo-
lutionary sciences, but it is also one of its weaknesses since it makes it extremely hard
to test precisely and in its totality. Another important point to note is that although this
theory can give an account of the origin of “true religions” as defined by Diamond,
the theory cannot explain why a kleptocratic society will be successful over time if
one does not consider at least two levels of selection: the individual level and the
group level. Diamond himself accounts for such possibility and proposes that a form
of group selection could explain the success of kleptocratic societies (Diamond 1997).

4 Adaptive Theories at the Group Level
Individual level selection explanations have been privileged by evolutionary biol-
ogy for more than 30 years (Wilson and Wilson 2007 ). However, many authors
agree that major transitions took place in evolution. A major transition in evolution
can be defi ned as the emergence of a new property that has dramatic consequences
on the course of evolution. One family of major transitions is evolutionary transi-
tions in individuality. An evolutionary transition in individuality is the result of the
emergence of a new kind of individual from the cooperation of lower- level individu-
als (Maynard Smith and Szathmáry 1995 ; Michod 1999 ; Okasha 2006 ). The emer-
gence of human language and culture is sometimes considered as the most recent
evolutionary transition in individuality since it seems possible that the human coop-
eration leads to the formation of news “cultural” individuals in competition at the
level of societies, namely human “superorganisms”.

Until now, all the theories I have examined propose an explanation of the religious
phenomena at the individual level, which implicitly or explicitly rests upon the con-
cept of gene. Yet, we have seen the limits of these theories to explain the diversity
of religions, whether it is diversity in supernatural agents, in rituals or supernatural
punishments, which can be very different from one society to the other. To fill the
gap between these gene-centered theories and the diversity of religious phenomena,
it seems necessary to postulate some mechanisms of cultural evolution and selection
in order to get more complete picture of the evolution of religion. This is exactly
what David Sloan-Wilson (Wilson 2002 , 2005 ) proposes. According to him, the
evolutionary function of religion is essentially to favor the production of common
goods within group and to diminish their costs through the altruistic cooperation of
individuals belonging to one and the same group. This leads to the emergence of an
intergroup competition when more than one group can be considered (direct com-
petition if the groups are in contact and indirect if they are not). According to
Wilson, it is hence perfectly legitimate to consider whether actual forms of religion
are cultural adaptation at the group level.

36 Origins and Evolution of Religion from a Darwinian Point of View…
774
Wilson confirms his hypotheses using detailed case studies (Wilson 2002) and an
analysis of 35 religions (Wilson 2005) randomly chosen in an encyclopedia of
Religion. Although he recognizes the value and plausibility of the other Darwinian
hypotheses we examined so far, he dismisses them one by one showing that none of
them is able to provide a full explanation of the phenomena linked to the religious
practices and beliefs. He propels his ideas with the introduction of the notions of
cultural group selection and gene-culture coevolution (Richerson and Boyd 2005). In
order to understand the interaction between genes and cultural items one has to
notice that ultimately they have a common fate since they depend (in a different way)
on the same interactor, namely the human biological individual. However, genes and
cultural items (sometimes called memes) are fundamentally different on one point.
Genes are prisoners of their interactors (or their “vehicles”, to use Dawkins’ termi-
nology) except during an event of reproduction. Yet, that is not the case for cultural
entities which can easily pass from one individual to the other during an act of com-
munication. From this standpoint, we can understand culture from an epidemiologi-
cal point of view. Indeed, models of cultural evolution based on the idea of contagion
can be derived from models of virulence (Laland et al. 2000), since the cultural enti-
ties face the same constraints as biological parasites. For example, the practice of a
transmissible religious altruistic behavior will successfully spread within a group if
the cost of this practice (in terms of its fitness) is compensated by the number of
“infected” individuals of this practice. Now, the number of individuals infected by
the practice will depend on the number of individuals available in the population. In
an ideal (thus unrealistic) case of population with infinite size and infinite density,
biological and cultural evolutions could be strictly decoupled from each other. It
would not matter in this case that the practice leads to the death of the “infected”
individual if the number of available individuals is infinite and some of them adopt
the practice. In this case, culture would impose no constraints on individual fitness.
Yet, real populations are neither infinite nor infinitely dense. Consequently one given
individual has a limited access to other individuals. Under such conditions, humans
can be considered as a limited resource from the point of view of the cultural entities.
On the short term, a cultural practice can spread quickly in spite of an important cost
imposed to the individuals who display this practice, but it will sooner or later lead
to the extinction of the population since no more individuals will be present in the
surrounding of the “infected” individuals and the cost of the practice will not be
compensated by a high level of transmission. On the contrary, a practice that, on
average, allows for an increase in individuals’ fitness will persist and will be trans-
mitted over time since new individuals will always be available.

Once these considerations are taken into account, the relevance of multilevel
selection and, more specifically, group selection of cultural items, becomes obvious.
Ideas pass from one individual to the other within a cooperative group. Differences
in religious practices between groups can lead to the differential persistence of these
groups over time and potentially to their differential growth and “reproduction”.
Proselytism, wows of chastity, hostilities or intolerance towards other religious
groups are all potential adaptive traits that are better explained from a long term
cultural groups perspective, in spite of the short term individual cost.
P. Bourrat
775
To conclude this section, I will briefly present three studies which have directly
or indirectly tested theories of intergroup competition by considering some traits
linked to religion as cultural adaptations. The first one has been made by Snarey
( 1996 ). Using an ethnographic database of several hundreds of variables and societ-
ies, he demonstrates that the beliefs in moralizing high gods (by “high” he means
being an all-powerful and/or creator of the universe) is positively correlated with the
dryness of the environment. Snarey suggests that the fact of believing in all power-
ful and moralizing god increases the cooperation between individuals in environ-
ments where the problem of water availability existed in the past or still has an
important impact on the survival of the societies. Thus the result observed (i.e. the
correlation between dryness and the presence of high gods in the religion) would be
due to the elimination or transformation of the societies which did not have such
beliefs in high gods.

The second study was carried by Roes and Raymond (2003 ), who used the same
database and tested different hypotheses. They found that the beliefs in high gods
were on average found more often in the largest societies (usually found in the rich-
est environments). They propose that the beliefs in high gods allow societies to
reach a larger size by promoting a better cooperation between individuals and con-
sequently out-competing smaller groups since the size of a society is critical in
intergroup competition involving conflicts. The mechanism they propose to explain
the role of the beliefs in high gods and the size of societies is the following. A larger
society faces problems such as free riding, defection of its individuals as well as the
risk of splitting in a much more intense way than a smaller society where such prob-
lem can be solved at a local level. The belief in high moralizing gods would allow,
according to Roes and Raymond, an increase in cooperation between individuals in
a group, and thereby would lead to the partial solution of the free riding problem.
This consequently would decrease the risk of scission of this group. Moralizing
gods would serve, if we follow Roes and Raymond in their reasoning, as social glue.
Finally, a third study by Henrich and his collaborators (Henrich et al. 2010 ), has
tested the impact of belonging to Islam and Christianity (what they name world
religions) on the fairness of individuals in 15 populations. The study shows, among
other results, that if the individuals tested belong to one of these two religions there
is a higher probability that they will behave more fairly than when they belong to a
local religion. Henrich and collaborators measured fairness using three games clas-
sically used by economists which involve the interaction of two individuals. In the
fi rst game, named “the dictator game”, one of the two individuals receives a sum of
money that he or she can choose to share with the other player. He or she can decide
to keep everything or to divide this sum of money as he wishes. The second player
has no choice but to accept the amount of money which is given to them. This is not
the case in the second game, “the ultimatum game” where the second player can
refuse the amount of money the fi rst player intends to give them. If the second
player refuses the amount proposed, none of them will receive the money initially
given by the experimenter. Finally, Henrich and his collaborators created a game
based on the same principles that the “dictator” and “ultimatum” games, but in
which the second player can punish the first player by spending some money given
36 Origins and Evolution of Religion from a Darwinian Point of View…
776
to them beforehand if they believes that the proposition made by the first player is
unfair. From the results they obtain, namely a higher probability to play the games
fairly by individuals belonging to a world religion, Henrich and his collaborators
suggest that world religions have coevolved with societies while facilitating large
scale cooperation. The underlying idea behind this proposition is the concept of
cultural group selection.
None of these three studies outlined above clearly proposes a specific mechanism
involving religion that would reinforce individual cooperation within societies. For
example, Roes and Raymond use the term “moralizing god” which remains very
vague. The fear of supernatural punishment theory, the kleptocracy theory and some
aspects of the costly signaling theory could potentially all be involved in some phe-
nomena of moralization and of large scale cooperation observed by Snarey and Roes
and Raymond on the one hand, and Henrich and his collaborators on the other hand.
None of the theories or hypotheses proposing religion as an adaptation at the
group level is a theory that gives a historical and cognitive origin of religion. No
cognitive mechanisms are proposed either in Snarey’s study or in Roes and
Raymond’s one. Even Wilson, the great architect of the multilevel selection theory,
neglects some explanations given by the other theories at the individual level which
are sometimes simpler, more cognitively relevant and which do not involve the
group level. Yet, as we saw, some questions linked to the diversity of religions, to
some of their characteristics and to their stabilization over time will remain unan-
swered if they are considered from an individual level perspective only.

5 Conclusion
In this chapter I have proposed that different kinds of Darwinian theories could
account for the origins and evolution of religious practices and beliefs. I have distin-
guished six theories (or sets of theories) and emphasized wherever possible that
they are not incompatible since they are not always applicable to the same fields, the
same levels of organization and over the same periods of the human evolution.
For example, the by-product theory is a cognitive explanation of the origin of the
beliefs in supernatural agents . The sweet-tooth theory might permit to provide cog-
nitive explanations of the existence of deleterious religious behaviors from a mod-
ern individualistic perspective if more research was done within this framework.
The fear of supernatural punishment provides a cognitive explanation of the origins
of beliefs in supernatural moralizing agents. Kleptocracy and costly signaling theo-
ries respectively provide a pluralistic explanation of the emergence of the world
religion and a behavioral explanation of the emergence of costly rituals . Finally,
developing models of multilevel selection would certainly provide solid explana-
tions of the evolution and maintenance of the number of the different religious traits
and religions in the world. Table 36.1 summarizes all these differences between
theories. It is now time to acknowledge that the integration will probably be the only
way of providing the most thorough account on the evolution of Religion. Indeed,
P. Bourrat
777
we have seen that the fear of supernatural punishment theory to some extent rests on
the by- product theory. We have also seen that the Kleptocracy theory could be
linked both to the fear of supernatural punishment theory and consequently to the
by-product theory but also to the costly signaling theory through the enforcement of
rituals. Thirdly, the costly signaling theory could be linked to the fear of supernatu-
ral punishment which would provide at least a partial explanation of the reliability
of some signals (such as being pious). Finally, it seems that the group level theories
replaced in a multilevel framework are the best candidates to integrate those differ-
ent theories, although one needs to keep in mind that the individual level and the
cognitive constraints that it imposes need to be at the heart of a multilevel frame-
work (what current group-level theories hardly do). I showed that some explana-
tions, while resting on the individual level, only make sense when they are replaced
in a more general context involving the existence of other cultural groups. A group
of individuals is always hard to define, but it seems that if one chooses relevant
cultural traits, as Sosis, Kress and Boster did in their study on scarifications, multi-
level mechanisms integrated in the future models would provide a better general
understanding of the origins and evolution of religion. The Fig. 36.1 schematically
summarizes links which exist between the different theories we examined.
Figure 36.1 and Table 36.1 do not aspire to answer all the problems that would
need to be solved in order to fully understand the emergence and evolution of reli-
778
gion. However, I aimed at eliminating the direct oppositions and contradictions
between these theories as well as to show their limits. Religion is an extremely
complex phenomenon and the Darwinian approach to this phenomenon is still quite
recent. It seems fair to give each theory involved in its explanation the time to
develop and to fi nd a consensus with other theories wherever possible.
The by-product theory of religion provides a very good non-adaptive cognitive
explanation of the emergence of beliefs in supernatural agents. The fear of super-
natural punishment theory, in the version proposed by Bering and Johnson, rests on
the by-product theory but adds an adaptive dimension to the explanation of these
beliefs. In order to account for their maintenance, a group-level dimension is neces-
sary for this theory which has been build from an individual perspective. The same
remark can be made both with regards to the kleptocracy and costly signaling theo-
ries. These two theories could be partially anchored to the fear of supernatural pun-
ishment theory and consequently into the by-product theory. The maintenance of
the beliefs and practices entailed by those theories also implies a group-level
perspective.

@Flyingsaucesir

Well that didn't transfer well, but that's the best I can do.

Yes not a good idea to hate anything.

@skado Yes but I am probably the only one here who is interested.

@skado Firstly.
It is hard to see what relevance, the general genetic and evolutionary origins of religion have to an agnostic and atheist site. Which by definition, is for people who are concerned to address just one small, and late developing, sub set of religion, the god based, or theist religions. ( Many often praise many forms of animism, and even pantheism. ) So only the authority god based theory of religions origins is relevant. Especially since, given religion and human cultures great variety and adaptability, it is very hard to imagine that a whole range of cultural mechanisms do not exist, or could not designed to answer the same problems as it. Times change, and just as fifty thousand years ago, theist religion was either, an exceptional rarity, or probably not yet even dreamed of. So there is a real possibility that in ten thousand years, or less, religion may be something else again. Whether we will it or not.

Secondly.
What may have answered the problems of yesterday, may not be able to answer the problems of today or the future. Nor may those answers be acceptable, in a modern moral context. You yourself confessed to having modern liberal values.

Thirdly.
It may well be that modern liberal morality is, none adaptive, and can only drive those who adopt it to extinction. Eventually being overwhelmed by regressive forces of religiously inspired fascism. But if that is the case then so be it, though I shall fight as long as possible to prevent it, even knowing we may only be able to enjoy the party for a short while. Yet I will still support enjoying the pleasures of the party, as long as possible. ( By which pleasures I mean things like democracy, allowing women equal human rights to men, state welfare, and universal education. )

This is especially so since it is quite possible, indeed probable, that the whole human race will soon be faced with extinction anyway, due to forces that, our culture can not even begin to control. Or, that in a very unpredictable future, that our cultures will embrace technology such as genetic engineering, which will mean, controlling our own evolution.

Fourthly. More minor matters.

Most theories of religions origin seem to miss a big elephant in the room. Which is that the genetical pre-existing need, that religion is most designed to answer, is not that of, health and long life, living together without conflict, nor even remotely long term survival. But it exists to answer, by far the biggest and overwhelming human need of all, the need to avoid boredom. By far the greatest use that people make of religion, myth and story telling, more than all its other uses together, is simply that of amusement. Myth and stories occupy time with interest and excitement, and most religious rituals are designed primarily as theatre for amusement. Which became needed, ever more, after the agricultural revolution, caused an ape, who was adapted to the endless variations of hunting and gathering, to swap them for the drudgery of agriculture. Which often sadly, despite the glorification of kindness that many religions pretend, means that our boredom, more than anything else, drives us to extremes of cruelty, bullying, exploitation and even violence. Especially since religion, as the great provider of entertainment, makes this easy, by acting as a great enabler, (As in pimp.) making those things easier to access than finding more kind, even less self harming but more challenging, forms of entertainment.