1. OUT OF AFRICA:
LESSONS FROM A BY-PRODUCT OF EVOLUTION
PASCAL BOYER
Dr Faust: Habe nun, ach! Philosophie,
Juristerei und Medizin,
Und leider auch Theologie
Durchaus studiert, mit heißem Bem.ühn.
Da steh ich nun. ich armer Tor!
Faust. I. 1
What happens in religion can be explained in terms of human propensities
that would be there, religion or not. The study of religion is
like that of politics, poetry, gardening or mass murder. Such a study
simultaneously deflates and reduces those cultural phenomena. It shows
that they are only particular illustrations of more general processes
of human behavior and also shows that they are explained by these
processes. Indeed, one great merit of cognitive approaches is to show
how the human notions and norms and behaviors we usually call
"religion" can be explained in terms of mental processes and social
dynamics that are potentially present in all human beings and present
in a variety of other cultural domains. This is also why the dull
business of demarcating what is "religion" from what is not is better
left to lexicographers; it should not unduly trouble scholars.
Whether accounts of religion are of interest depends, not on where
they place its boundaries but on how they account for the observed
behavior they purport to explain. Cognitive studies of religion, inaugurated
by such works as Rethinking Religion, offer such causal accounts.
My aim here is to recapitulate some of the achievements of this
novel enterprise and to point to some possibly important, though as
yet unexplored consequences.
28 PASCAL BOYER
Organized and Traditional
We commonly call "religions" two rather disparate sets of objects.
There is a set of so-called "world religions" or doctrines of great
diffusion, such as Judaism, Buddhism or Islam. There is also what
anthropologists, when they are in the mood for categorization, call
"traditional" religions: systems of beliefs and practices that are firmly
rooted in some local social relations, with little explicit theology' and
no corporation or guild of religious officers. For many understandable
reasons, scholars of religion have generally established their basecamp
in a thorough knowledge of "world religions." From this starting
point they then tried to climb all the way up to a general understanding
of religion in human kind. This however was not always
very successful, despite the many new and fascinating vistas opened
to scholarly exploration. This was probably unavoidable. Studying
doctrinal religions is all too likely to lead one onto a false trail, as
far as religion in general is concerned. This is because doctrinal, socalled
"world" religions are a secondary, derivative development of
a much more general and deeply human tendency to imagine important
supernatural agents and to entertain precise descriptions of their
powers. Without an understanding of this general mental disposition
one does not understand much of the special case of "religions"
armed with an official personnel, some theologians, an important
economic role and an affinity for political power. So bongo drums
(as a metonym for religion in general; are very likely to explain a
lot that we need to know about crucifixes (standing for corporate,
doctrinal religion) rather than the other way around.
The focus on what we are familiar with--those highly doctrinal
phenomena people call "world religions"--is the source of many a
confused \iew about religion. For instance, it is in my experience
exceedingly difficult to convince most people of straight facts that
are familiar to any anthropologist: that most religion is not about
the creation of the world, that it is rarely about God, that it is very
seldom about the salvation of the soul. More important and even
more difficult to impress upon most people: most religion has no
doctrine, no set catalogue of beliefs that most members should adhere
to, no overall and integrated statements about supernatural agents.
Most religion is piecemeal, mostly implicit, often less than perfectly
consistent and, most importantly, focused on concrete circumstances. People
OUT OF AFRICA 29
use their religious concepts to account for their uncle's death or their
child's illness or their neighbor's good fortune, not to explain the
persistence of evil or the existence of the universe. So most religion
in the world has nothing to do with what our common explanations
of religion would assume.
Another error is just as common and even more damaging. It
consists in thinking that some societies or groups happen to have
"world" or "organized religion" and others have what would be generally
called "traditional religion," in the same way as some people
are nomadic and others sedentary. In this view, the Kwaio people
in the Solomon Islands for instance have a religion that happens to
be of the "traditional" type (no set doctrine, no interest in cosmic
conundrums, no explicit discourse about transcendence), while that
of Christian Americans is of the "organized" type. This way of seeing
the contrast between those two genres of religious activity may
seem plausible enough. It seems to be accepted, if tacitly, in most
general presentations of the subject, student textbooks and scholarly
reflections. It is in fact terribly misguided and the origin of many
confusions in the study of religious thought.
The contrast is misguided because so-called "organized" religion,
with explicit doctrine and specialized personnel, never displaces the
other kind; it supplements it. It is an add-on. an extra layer, an additional
growth. What anthropologists usually describe as "traditional"
religion is based on ways of thinking--about supernatural agents,
about their interest in moral action, about their responsibility in
human misfortune, etc.--that we find in all human groups. True, in
some societies (including the ones most readers of this book belong
to; there is also a totally different, integrated, explicitly argued version
of religion, produced and fostered by specialists gathered in corporate
associations. But this, the evidence suggests, does not really
change most people's intuitive adherence to the more common ways
of thinking. Many Christians probably think that, as far as religion
is concerned, their minds are filled with Christian doctrine, or at
least with notions and norms derived from Christian teachings. Most
professionals of religion, as it were, priests and ministers and theologians,
maintain a similar belief. Many students of religion also
believe that. But nothing could be further from the truth. Doctrinal
religion is a veneer that certainly covers, often conceals non-doctrinal
concepts but which would not hold without the underlying material.
30 PASCAL BOYER
Indeed, people's adherence to the doctrine actually requires intuitive,
generally unconscious ways of thinking that stray far from the doctrine
and in some cases contradict it.
Out of Africa: Religious Thinking in Generic Human Beings
To understand why this is so, it may be of help to recapitulate the
key conclusions of cognitive studies of religious thought and behavior.
Though the research program is still developing, it already provides
us with a broad sketch of general features of religious concepts
and ritual, as well as detailed studies of few--too few- empirical
cases. Here are the essential points in my view:
* The way we acquire, store, organize religious concepts is to a large
extent inaccessible to conscious inspection.
This is not so surprising, since the mechanics of concept acquisition
and maintenance in general, just like other cognitive mechanics, are
outside conscious access. We do not know or experience how our
visual cortices translate retinal images into the illusions of 3-D scenes.
We do not know or experience how other cortical networks produce
grammatical sentences. In the same way. we do not know or experience
the processes whereby we attribute agency to unobserved
agents, or moral judgments to those same imagined agents. The way
to reveal how this takes place is not, or not just, to ask people what
their "beliefs" are--for people do not believe what they believe they
believe. The only way is to run experiments, test models of cognitive
structure, measure how well these models account for observed
religious behavior.
* Most religious concepts are parasitic upon mental systems that
would be there, religion or not.
Cognitive studies reach a similar conclusion in several distinct domains
of religious thought. It turns out that having religious concepts does
not require specific mechanisms in the mind, dedicated to religion.
Compare with vision for instance, or language comprehension, or
the understanding of other people's emotions, which all require specific
functional structures. Religion seems to be parasitic in the sense that
all the systems involved in its acquisition and its mental effects would
be there, religion or not. Naturally, there is nothing especially derogaOut
OF AFRICA 31
tory in saving that religion is parasitic in this precise sense, since
aesthetic pleasure for instance is parasitic in that way too. Let me
recall some of the domains in which this parasitic nature of religion
is best illustrated:
* Religious ritual is parasitic upon action representations.
The pattern for cognitive explanations of religious behavior was set
by Rethinking Religion. That volume demonstrated that the processes
whereby human beings represent actions in general (any action from
the humble to the sublime, from the quotidian to the heroic; are
sufficient to account for the structure of religious ritual. But Religion
Explained also pointed to two features that were found in many otherdomains.
First, what explained the apparently specific features of the
religious domain was a minor modification or "tweaking"" of the
more common processes. Second, these common processes are spontaneously
activated in people, they do not require cultural transmission;
only the specific features that "tweak" them are socially
transmitted. Beyond ritual action, one can observe similar features
in other domains of religion:
* Religious agency (gods, spirits, ancestors, etc.) belongs to a larger
repertoire of supernatural agents defined as violations of intuitions
about agents.
This is the clearest example of the tweaking process. The material
of religion does not in this respect differ from that of folklore. There
is a small repertoire of possible types of supernatural characters, most
of whom are found in folktales and other minor cultural domains,
though some of them are the important gods or spirits or ancestors
of "religion."" Most of these agents are explicitly defined as having
counter-intuitive properties that violate general expectations about
agents. They are sometimes undetectable, or prescient, or eternal.
The way people represent such agents activates the enormous but
inaccessible machinery of "theory of mind" and other mental systems
that provide us with a representation of agents, their intentions and
their beliefs. All this is inaccessible to conscious inspection and requires
no social transmission. On the other hand, what is socially transmitted
are the counter-intuitive features: this one is omniscient, that
one can go through walls, another one was born of a virgin, etc.
* Interaction with religious agents is parasitic upon cognitive systems
for social interaction.
32 PASCAL BOYER
Social interaction requires the operation of complex mental systems:
to represent not just other people's beliefs and their intentions, but
also the extent to which they can be trusted, the extent to which
they find us trustworthy, how social exchange works, how to detect
cheaters, how to build alliances, and so on. Again, these mental systems
are largely inaccessible, only their output is consciously represented.
Now interaction with supernatural agents, through sacrifice,
ritual, prayer, etc., is framed by those systems. Although the agents
are said to be very special, the way people think about interaction
with them is directly mapped from their interaction with other people.
* Religious morality is parasitic upon non-religious moral intuitions.
Developmental research shows the early appearance and systematic
organization of moral intuitions: a set of precise feelings evoked by
the consideration of actual and possible courses of action. Although
people often state that their moral rules are a consequence of the
existence or of the decrees of supernatural agents, it is quite clear
that such intuitions are present, independent of religious concepts.
Moral intuitions appear long before children represent the powers
of supernatural agents, they appear in the same way in cultures
where no one is much interested in supernatural agents, and in similar
ways regardless of what kind of supernatural agents are locally
important. Indeed, it is difficult to find evidence that religious teachings
have any effect on people's moral intuitions.1
Religious concepts
do not change people's moral intuitions but frame these intuitions
in terms that make them easier to think about. For instance, in most
human groups supernatural agents are thought to be interested parties
in people's interactions. Given this assumption, having the intuition
that an action is wrong becomes having the expectation that
a personalized agent disapproves of it. The social consequences of
the latter way of representing the situation are much clearer to the
agent, as they are handled by specialized mental systems for social
interaction. This notion of gods and spirits as interested parties is
1
Most students of religion take fur granted that religiously coded morality must
have an effect on people's moral intuitions. I have yet to come across any actual
evidence for this effect. True, people's explicit discourse, whereby they justify the intuitions,
is certainly affected by local religious concepts. But that kind of discourse is
a posteriori. We haw no evidence that it modifies the intuitions themselves. On the
contrary, cross-cultural evidence shows a great convergence in moral intuitions
despite great differences both in explicit moral codes and in supernatural beliefs.
OUT OF AFRICA 33
far more salient in people's moral inferences than the notion of these
agents as moral legislators or moral exemplars.
* Notions of ritual specialists are based on non-religious notions of
causal essence.
In most human groups some people are thought to be in a privileged
position to interact with supernatural agents. A clan's patriarch
is evidently the best interlocutor for the ancestors, a local shaman
has what it takes to negotiate with wayward spirits. People think of
such ritual specialists as having some internal, vaguely defined quality
that sets them apart from the common folk. Learning to perform
the rites, or acquiring the secret anti-witchcraft recipes, are secondary;
what matters most is possession of that internal capacity, construed
in quasi-biological terms. This is where, once again, what may have
seemed a specifically religious phenomenon is derived from common
cognition. The notion of a hidden causal essence that cannot be
observed yet explains outward form and behavior, is a crucial feature
of our spontaneous, intuitive way of thinking about living species.
Here it is transferred upon a pseudo-natural kind, as it were: a subkind
of human agents with different essential characteristics.
Note that, so far, we have considered the most important domains
of religious thought and behavior--supernatural agency, ritual action,
morality, misfortune, ritual specialists -without mentioning what would
be to some people the sine qua non of religion. There has been no
mention of transcendence, of infinite power, of cosmology, of howsouls
get saved or why evil exists. This is because such questions arc
blithely ignored by most people in most places in the world, and
have been so for most of human history, as far as evidence can tell.
Religion does not exist because of the need to answer such questions,
far from it. Such questions are a special, local development
that arose in societies where religion had become the affair of a
guild of specialists. If by religion we understand the thoughts and
actions of actual people in actual human groups, these doctrinal
questions seem to be a late, minor, derivative phenomenon.2
2
That religious guilds and their doctrines then have enormous political and social
effects should nut make us forget that essential point. Too often in the study of
religion, what is said to count is an abstract model of the doctrine itself, rather
than the thoughts or actions of the people who embrace it. But if we want to
explain actual historical development, the latter is what matters.
34 PASCAL BOYER
Most religious thought has no doctrine for any of these questions,
because it has no doctrine at all. If we understand by doctrine a
minimally integrated set of coherent assumptions about supernatural
agents, their powers, the justification for rituals, the reasons why
some people interact with the gods rather than others, etc., it is quite
clear that in most groups in the world one can find no such thing.
This has in some quarters fuelled a long-lasting misunderstanding
between scholars of religion and anthropologists. The former assumed
that people outside organized religion must have some doctrine, only
a rather esoteric one, or a mytho-poetic one, or an enacted one;
anthropologists tried hard to show that in most human groups coherent
religious behavior is combined with vague, fragmentary, idiosyncratic
and often less than perfectly coherent accounts of supernatural
agents.
Indeed, one must remember that, as far as anthropological and
cognitive evidence can guide us, the situation is quite similar in
groups where there is some official religious doctrine. Again, this is
one of those anthropological findings that some students of religion
do not seem to register, or whose import they fail to see. In places
where a doctrine is available (indeed where people are taught that
doctrine), and believe themselves to hold the beliefs typical of the
doctrine, there is large and converging evidence that their actual
thoughts and intuitions are similar to what I described so far, rather
than to the doctrine itself. People may well be taught and repeat
that religious agents are transcendent, vet they see their gods as veryclose
interlocutors; they are taught and repeat that gods are omniscient
and in fact assume that the gods have cognitive limitations,
for instance, cannot attend to many things at the same time, like
any human being. They arc taught and repeat that a statue is just
a symbol of the god, yet assume that the actual artifact is endowed
with special powers.
To sum up, most religious thinking seems to proceed in a way
that requires no special doctrine: only an explicit mention of the few
"tweaks" to add to ordinary cognition. Indeed, religious systems of
most societies throughout history have worked in that way. This is
going on even when there is a doctrine, and more surprisingly even
when people themselves believe they believe in the doctrine.
OUT OF AFRICA 35
Out of the Fertile Crescent: Religious Thinking by Professionals
As I said above, doctrinal religion does not displace the general,
intuitive kind, but it adds to it an extra layer of explicit concepts
and norms. This new layer is characterized by explicit and coherent
links between the various concepts, a definition of a domain of "religion"
as specific, sui generis (something that is missing in most human
groups), the presence of an organized group of religious scholars or
specialists.
Why and how does that happen? Why is it the case that some
human societies have that extra accretion of concepts and social relations
centered on supernatural concepts? Why is there organized religion
at all? The way this process is described by religious groups
themselves is in terms of a myth of origin or an epic. The narrative
says that a new doctrine appeared, that it gradually convinced more
and more people, that the doctrine stemmed from a set of important
texts or from a revelation, that proper transmission and maintenance
of the doctrine and rites required an organized group of scholars or
priests. So the tenets of doctrine were there first and their social
effects were among the many consequences of people's adherence to
those articles of faith.
Obviously, such a narrative belongs to the register of fairly tales
rather than serious scholarship. What the historical evidence says is
both more complex and more plausible. Complex polities originated
in a few regions of the world, a few millennia ago and became states,
small kingdoms empires or city-states. Their complex economies and
embryonic markets meant that many activities became the province
of specialized groups, craftsmen in particular. These groups or guilds
worked as cartels, often maintaining an exclusive grip on the delivery
of particular goods or services. They organized training, often kept
a minimis clausus of new practitioners, sometimes arranged uniform
prices and generally guaranteed a certain quality of service. This
happened in most trades and crafts, for the intensification of agriculture
meant that most people were far too busy to practice these
activities and that enough surplus was generated to feed specialists.
The provision of religious services is no exception to this trend.
Together with guilds of merchants or blacksmiths or butchers there
appear groups of ritual officers and other specialists of the supernatural.
They generally operate a monopoly, with an exclusive right
to perform particular rites. They form centralized organizations that
36 PASCAL BOVER
maintain a strict a strict control over new candidates. They try to
bind as closely as possible with sources of political power. They, naturally,
work on the assumption that they do provide something and
that what they provide could not be obtained elsewhere.
However, the special nature of the commodity they provide means
that religious guilds cannot operate entirely in the same way as craftsmen's
associations. First, there is of course no objective way to determine
whether any religion is better than any other, whereas people
can always observe that the trained and experienced cobbler makes
better shoes than they themselves would. So, however strongly the
guild may claim that its rites are the only way of obtaining particular
results, people are fickle and may at any moment decide that
some cheaper, home-made recipe is just as good. Second, precisely
because of that elusive quality of supernatural services, there is always
some competition. In most complex polities, there is an organized guild
of religious practitioners as well as a whole variety of informal providers,
local shamans, wizards, healers, inspired idiots and ominous dreamers.
Some of these competitors are the shamans and other local specialists
found in most human groups. Their claim to efficacy is based
on local reputation, on apprenticeship with a famed specialist, on
supposed connections to local supernatural agents, in general on their
own individual characteristics.
In most cases, the guild uses whatever political clout it can garner
to dissolve this competition, demote it. relegate it to unimportant
or local rituals, hinder its operation or the transmission of its
recipes. This is bound to fail in the long run, for the strength of
informal practice is precisely that it is informal and can therefore
be started anew at very low cost. As all religious specialists know,
the war against what they tend to call superstition is never-ending
Religious guilds, being cartels of specialists, tend to unify the provision
of services: that is, they try to promote the notion that, to
some extent, the same service will be provided by any member of
the guild. They also try to promote the complementary notion, that
no one outside the guild could provide this service. This has consequences
for the presentation of religious specialists. Local specialists
like shamans and diviners are authoritative only in a particular
place; the guild potentially covers any territory. Local specialists are
supposed to be different by internal nature from other people; the
guild describes its members as specially trained. A religious guild
promises to deliver a stable, uniform kind of service that only it can
provide, but also a service that any member of the guild will proOUT
OF AFRICA 37
vide in the same way. Proper service depends not on the personal
qualities of the specialists but on their being similar to any other
member of the guild. Naturally, a group like that will claim connection,
not to local spirits and ancestors but to larger-scale supernatural
agents with whom the guild proposes to interact with in the
same way, regardless of the particular place and customers.
The differences also extend to the concepts put forward in the
guilds. It is quite natural for a shaman to construe his locally recognized
powers as a link to local supernatural agents. By contrast,
specialists who endeavor to operate on a large market, indeed on
any market that is available, naturally think of themselves as interacting
with highly abstract, delocalized, cosmic gods. A local shaman
tends to interact with social groups: a family, a lineage. His interventions
are said to protect the bones of the lineage or restore a
family's defiled honor. By contrast, guilds generally tend to garner
help from central political power and consequently address not local
groups but the individual. Hence their insistence on such notions as
the individual soul, one's personal merit, one's salvation.
It is quite natural for a local specialist to use flexible, highly variable
ritual recipes, using his personal knowledge of situations and
customers. A guild by contrast, trying to make most of its members'
interchangeability, is bound to insist on highly codified, inflexible ritual
recipes. Because of all these trends, members of religious guilds
generally use literate codes and other texts to maintain uniform provision
of religious services. Given that such guilds only appeared in
complex politics and that these very often had some writing system,
it is not surprising that the guilds also used writing. A great advantage
of writing is that it facilitates the uniformity of service and practice
that is the main selling point of such professional groups. So
religious guilds that set great store by literate sources, written transmission
and the kind of systematic argument made easier by writing,
are more likely to subsist than groups that ignored the technology
of writing. Conversely, given that uniformity and substitutabilitv are
important assets of the guild, any appeal to personal charismatic features
or shamanistic revelation are actively discouraged.3
3
Incidentally, to say that guilds act against the competition, exert some coherent
political influence, or maintain their predominance through the use of particular
concepts does not mean that these social groups are agents. All it means is that
most members of such groups tend to adopt a strategy of coalitional solidarity with
the guild; social and political effects stem from these aggregated strategies.
38 PASCAL BOYER
These common social factors--the constitution of a cartel of religious
specialists, its requirements in terms of uniformity- and stability--explain
the convergent features of many such religious groups:
their insistence on cosmic questions rather than particular misfortune
as the foundation of religious behavior, their notions of personal salvation
as opposed to collective security, and more generally, the idea
that religion requires a doctrine, that it is based on a doctrine, that
its outward manifestations are consequences of the doctrine: all statements
that make sense as the self-serving discourse of professionals,
but should not bamboozle students of religion, whose job it is to
explain religious thought and behavior as they occur, not as the guild
wishes.
A Darwinian By-Product
All these historical facts show how quite a few theories of religion
have got things diametrically wrong. If you assume that the source
of religion lies in metaphysical questions for which human kind needs
answers--like the destiny of the soul or the creation of the universe
or the origin of evil--you mistake a recent and regional development,
itself limited to professional associations, for a general feature
of human minds. Religion certainly is old, but it is not about that
at all. It is about daily occurrences and interaction with imagined
agents of a counter-intuitive nature.
That religion in this sense is '"old" is not disputed, although what
is old by historical standards is quite recent in the archaeological
record. It is very likely that by the time modern humans came out
of Africa, they had the kind of supernatural imagination that founds
religious concepts. It is quite certain that this imagination was active
by the time of what is generally called the "cultural revolution." the
sudden explosion of cultural artifacts that show both great innovations
and the beginnings of cultural style. It is also clear that dead
bodies were the object of much special preparation by that time,
being left adorned, accompanied with various artifacts or laid down
to rest in special positions. We do not know whether those people
also associated dead bodies with concepts of supernatural agents, a
frequent feature of modern religion. To sum up. the kind of concepts
and practices that we find the world over seem to appear right
at the same time as all the mental capacities that are typical of the
modern human mind.
OUT OF AFRICA 39
However, the appearance of religion is not an important evolutionary
event, because it is only a fairly predictable by-product of
mental capacities that would have appeared, religion or not. As I
said above, most features of religion seem to be fairly simple "tweakings"
of ordinary conceptual and inferential capacities that we find
in all human minds. Now those capacities are an outcome of evolution
and their appearance was a major event. For instance, the capacity
to represent non-actual states of affairs and to draw consequences
for such representations grounds people's perception of their past as
well as their deliberation about future action. Developing such a
capacity is a major evolutionary event; as a minor consequence, it
also allows one to imagine supernatural agents. In a similar way. the
extraordinary complexity of human "theory of mind" one's intuitive
explanation of others and own behavior in terms of intentions and
beliefs1
was a major development that resulted in uniquely complex
social interaction; it also allowed people to entertain complex thoughts
about interaction with imagined agents. Modern humans also have
an instinctive fear of invisible contaminants (like the pathogens of
rotting bodies, blood, feces, etc.) and an intuitive notion of invisible
contagion; such a cognitive adaptation is of great value. It also allows,
as a by-product, the development of notions of invisible power "the
sacred." "taboo." "pollution," etc.) that we find in religious imagination.
Obviously. the fact that people entertain religious thoughts at all
can have important consequences, that we sometimes mistake for the
explanation of religion. Once people find their imagined agents plausible,
they can use them at times to allay anxiety like more pliable
versions of real agents; once the versions of imagined agents differ
from one culture to another, they can be used as convenient ethnic
markers; once rituals are organized, a willingness to undergo gruesome
ordeals can work a signal of commitment to the group. All
this is familiar to all students of religion. It would be a mistake, however,
to mistake any of these social or personal effects of religion for
a plausible evolutionary scenario. People did not create religion to
allay their fears, first because it does not and second because people
cannot create just any convenient fantasy and find it plausible.
People did not create religion to foster good morality and group solidarity,
because such a strategy would be vulnerable to defectors and
quickly unravel.
We are left with a conclusion that many evolutionary biologists
would find unsurprising--and most students of religion unpalatable:
that religion is like dancing, music, ethnocentrism or body-ornaments:
40 PASCAL BOYER
something that most humans are very good at learning and almost
incapable of resisting, may sometimes have important consequences,
yet has no other explanation than the quirks of the way evolution
made our brains. What may make all this unpleasant or unacceptable
to some people is the belief that important phenomena should
have important causes, or at least their own, special causes. But cognitive
models suggest that religion is not really special and requires
no special mental process, no exceptional evolutionary event.
No Reason for Religion
Cognitive accounts of religion even suggest that there is no good
reason for the existence of religious thoughts and behaviors. There is
not even a single cause for them. Rather, the most plausible scenario
we have makes it a by-product of a whole variety of cognitive adaptations,
of mental systems that we have for a good reason. This
causal account clashes with most people's expectations, particularly
with those of religious adherents. We generally tend to think that
people for instance perform a particular ritual for some reason:
indeed, the first thing we always do is ask them what the reason is.
But cognitive models seem to suggest that this is not the most
profitable strategy, that the explanation for religious notions lies in
processes that people cannot be aware of, so that the explicit reasons
("we sacrifice to the ancestors because they protect us") is at
best a rationalization of thoughts and behaviors that would occur
anyway.
This is an unavoidable consequence of scientific reduction, but it
is important to understand what it really means. At this point, it
may be of help to return to Lawson and McCauley's model of ritual
action in Rethinking Religion. Some commentators have complained--
and many a student will have concurred--that the core theory of
ritual, the system of rules and principles that generate well-formed
descriptions of ritual actions, was "too formal." It explains what is
often a matter of great emotion and interest, goats beheaded and
chickens disemboweled, commandments recited and sacred texts committed
to memory, in terms of a list of abstract formulae and terms
such as "agent-slot" or '"object-filter." In my view, if there was a small
defect in the model, and any scope for improvement, it was that the
model was not formal enough, perhaps included too many commonOUT
OF AFRICA 41
sense intuitions about supernatural agents and their role. For instance,
the model assumed that supernatural agents were thought to have
"super-permanent" effects. That is to say. if people for instance
assume that the ancestors really were present during the ritual and
turned the neophyte into a proper member of the group, then this
effect cannot be undone, unless another ritual with the appropriate
structural reversal is performed. Now it would seem to most anthropologists
that the attribution of such "super-permanent" effects to
imagined agents is generally not an assumption, but rather a conjecture
that is strengthened by ritual performance. It is therefore
important to elucidate why rituals have such consequences: why they
make people think that supernatural agents were really involved.
(This crucial point is addressed in Bringing Ritual to Mind, where contrasted
ways of enhancing transmission--thought sensory pageantry
or doctrinal elaboration--also strengthen people's intuitions about
the imagined agents' real participation;.
This complaint (that the Action Representation model was overly
formal) points to a very general characteristic of cognitive models,
in this as in any other domain. They explain the occurrence of
thoughts in terms that demote the usually central role of reasons.
That is, what explains why we have this or that concept is not what
we would ourselves come tip with if asked to explain it. So far. people
are used to this kind of explanation, for instance from popular
Freudian accounts of emotion and behavior. Freudian and other fantasies
of that kind tell people that, beneath what they think are the
reasons for their behavior, lies an entirely inaccessible domain of
other reasons. In contrast to this, scientific studies of mental phenomena
say something far more disturbing: that beneath our reasons
for having particular thoughts there is an inaccessible domain
of processes that do not consist in reasons at all. That is. each of
our thoughts is caused by processes that do not consist in "thoughts"
in the sense of explicit combinations of the concepts we can name.
We think we see an elephant for a good reason, namely that there
is one in front of us: but the study of visual perception tells us that
all this is achieved in the cortex by extremely complex neural processes,
none of which resembles the '"elephant" concept. To take an example
from higher processes, the reasons why we remember particular
stories and forget others are not really "reasons." We may well think
we recall a story because it makes a lot of sense or answers a lot
of questions we had. But these are only consequences of the fact
42 PASCAL BOYER
that we recall it in the first place, and this is caused by memoryprocesses
that do not consist in "reasons."
So religious concepts function in that way too. A believer may
well think she has such concepts because they explain a lot, or
because they are awesome and beautiful stories, or because life would
make no sense if they were false, or because it makes her happy,
or because most other people seem to accept them. All these are
real consequences of having the concepts, but non-starters as explanations
for why one acquired them in the first place and why they
appeared in human cultures at all. The poorest such explanation,
incidentally, is that people spontaneously and intuitively adhere to
religious concepts because religious concepts are true. (One comes
across this argument surprisingly often in debates about religion.
Besides solving the delicate problem of deciding which religious concepts
are true, between all the incompatible, mutually refuting versions
available, proponents of this simple explanation also have to
ignore two major facts of human history: there is no limit to the
range of false concepts people can sincerely and intuitively find plausible;
conversely, there is a vast domain of true concepts that our
minds find it exceedingly difficult to acquire, as science shows everyday.
Given the colossal evidence for both tendencies, the fact that
most humans find a particular representation is certainly no guarantee
of validity, far from it.
Whither the Study of Religion?
Like fish, innovative scholars form schools. Rethinking Religion spawned
a whole series of related works, showing that the cognitive study of
religion was a coherent and valuable research program. Indeed, the
various features summarized in the previous pages were taken for
the work from at least five different scholars, which I suppose counts
as a school if not quite a swarm. The program is bound to expand,
as more scholars find inspiration in the early models, detect their
many flaws and provide us with better accounts of varieties of religious
thought and behavior. Inasmuch as it is successful, this research
program will certainly' escape the narrow confines of the study of
religion. It is one of the inevitable, indeed desirable consequences of
cognitive models that '"religion" is shown to be no more special than
other cultural domains.
OUT OF AFRICA 43
Another predictable development is that our accounts of particular
features of religion will be grounded in much more precise accounts
of mental functioning, in particular in finer-grained accounts of brain
function. There is no need to postulate any religious organ in the
brain, no more than there is a literature organ or a social life center
in cortical structures. But uncovering the underpinnings of ordinary
concept-acquisition and inference, in terms of neural circuitry, is
bound to tell us a lot about the processes whereby the most baroque
imaginings acquire inherent plausibility. Obviously, all this goes in
the direction of further reduction and deflation, which may shock
many people who joined the study of religion to find precisely the
opposite, an antidote to a perceived excess of reduction. But such a
move towards explanation in terms of fine-grained brain processes
is both possible and inevitable. Sensitive souls who find reduction
shocking, indeed faint at the merest whiff of a causal explanation,
should not just steer clear of the cognitive study of religion: they
should avoid study altogether.
It is both a privilege and an achievement to show science that a
particular domain of reality can indeed be reduced, that is explained,
where people thought it was either impossible or undesirable. The
study of religious ritual in terms of action representation and of the
latter as a cognitive process was one of those inaugural events.
Whatever the difficulties and uncertainties of a first model, it made
it possible to think of many other domains of religious behavior,
indeed of many other domains of culture, as similarly constrained
by the operation of human cognitive machinery.