THE THEORY OF MODES
OF RELIGIOSITY
II
04-024 Ch 04 2/5/04 8:25 AM Page 61
04-024 Ch 04 2/5/04 8:25 AM Page 62
I
T HAS LONG BEEN RECOGNIZED that religion encompasses two very different
sets of dynamics: Max Weber (1930, 1947) distinguished routinized and
charismatic religious forms; Ruth Benedict (1935) contrasted Apollonian and
Dionysian practices; Ernest Gellner (1969) explored the opposition between literate
forms of Islam in urban centers and the image-based, cohesive practices of
rural tribesmen; Jack Goody (1968, 1986) developed a more general dichotomy
between literate and nonliterate religions; Victor Turner (1974) distinguished fertility
rituals and political rituals as part of an exposition of the contrasting features
of what he called "communitas" and "structure"; I. M. Lewis (1971)
juxtaposed central cults and peripheral cults; Richard Werbner (1977) contrasted
regional cults and "cults of the little community"; Fredrik Barth (1990) distinguished
"guru" regimes spread by religious leaders, and "conjurer" regimes in
which religious revelations inhere in collective ritual experiences. These are just a
few of the many attempts to characterize a fundamental divergence in modalities
of religious experience and practice (Whitehouse 1995, chapter 8; Peel 2004;
Laidlaw 2004). At the root of all such dichotomous models is a recognition that
some religious practices are very intense emotionally: they may be rarely performed
and highly stimulating (e.g., involving altered states of consciousness or
terrible ordeals and tortures); they tend to trigger a lasting sense of revelation and
to produce powerful bonds between small groups of ritual participants. By contrast,
certain other forms of religious activity tend to be much less stimulating:
they may be highly repetitive or "routinized," conducted in a relatively calm and
sober atmosphere; such practices are often accompanied by the transmission of
complex theology and doctrine, and also tend to mark out large religious communities
composed of people who cannot possibly all know each other (certainly
The Theory of Modes of Religiosity 4
63
04-024 Ch 04 2/5/04 8:25 AM Page 63
not in any intimate way). But all the great scholarship so far devoted to understanding
these contrasting sets of dynamics suffers from two major shortcomings.
The first is that none of the theories advanced in the past was sufficiently comprehensive;
each theory focused on just a few aspects of the two modes of religious
experience and action. The second major shortcoming is that none of the
existing theories explained adequately why we get two contrasting forms of religious
experience in the first place.
This chapter introduces the theory of divergent modes of religiosity, which I
term "doctrinal" and "imagistic" (Whitehouse 1995, 2000a). The aim of the
modes of religiosity theory is to tie together all the features of the two modalities
of religious experience that other scholars have already identified and to explain
why these contrasting modalities come about in the first place. This theory advances
a set of hypotheses amenable to empirical investigation, concerning the
causal interconnections between a set of cognitive and sociopolitical features.
Modes of Religiosity and Memory
In order for particular religions and rituals to take the form that they do, at least
two things must take place. First, these religious beliefs and rituals must take a
form that people can remember. Second, people must be motivated to pass on these
beliefs and rituals. If people cannot remember what to believe or how to do a ritual,
these beliefs and rituals cannot be passed down from one generation to the
next, and so the religious tradition would not be able to establish itself. Equally, if
people do not think that particular beliefs and rituals are important enough to
pass on, the beliefs will mutate or become extinct. That being said, memory and
motivation have the potential to present far bigger problems than one might suppose.
Some religious activities are performed very rarely. Unless some very special
conditions apply, there is a real risk that people will forget the details of what
these activities mean and even forget how to perform them correctly. A potential
solution to this problem is to have a very repetitive regime of religious transmission.
One advantage of such a strategy is that a substantial corpus of complex cosmology
can be reproduced in this fashion. People can learn difficult concepts,
dogmas, and stories--and will remember these in the long run--if they repeat
them frequently. But this can produce problems of motivation. Continually listening
to sermons and performing the same rituals over and over might become
extremely boring. And if people are bored, there is a danger they won't continue
to follow, or pass on, the religion. There are solutions to all these potential problems,
and these solutions have profound consequences for the forms that religion
can take. But before we can go into that, we need to grasp the general nature of
memory functions (see figure 4.1).
64 CHAPTER 4
04-024 Ch 04 2/5/04 8:25 AM Page 64
There are basically two kinds of memory--implicit and explicit (Graf and
Schachter 1985). Implicit memory deals with things we know without being aware
of knowing (such as the varied forms of procedural competence required in successfully
riding a bicycle).1
Explicit memory deals with things we know at a conscious
level, and can be further subdivided into two types--short-term and
long-term.2
Short-term memory enables us to hold onto concepts for a matter of
seconds (e.g., a new phone number, which we might remember just long enough
to write down before forgetting). Long-term memory enables us to hold onto concepts
for hours--and in some cases for a whole lifetime. Long-term memory can
also be subdivided into two types--semantic and episodic.3
Semantic memory
consists of general knowledge about the world (e.g., how to behave in restaurants,
or what is the capital city of France, etc.). We can seldom recall how or when we
acquired this sort of knowledge. By contrast, episodic memory consists of specific
events in our life experience (e.g., our first kiss, the death of a beloved relative, the
day war broke out, etc.). These types of memory are activated somewhat differently
in doctrinal and imagistic modes of religiosity.4
And these differences go a
long way to explaining the divergent sociopolitical features of the two modes.
The Doctrinal Mode of Religiosity
In the case of the doctrinal mode of religiosity, ritual action tends to be highly routinized,
facilitating the storage of elaborate and conceptually complex religious teachings
in semantic memory, but also activating implicit memory in the performance of
THE THEORY OF MODES OF RELIGIOSITY 65
Figure 4.1. Types of Memory
04-024 Ch 04 2/5/04 8:25 AM Page 65
most ritual procedures. These cognitive features are linked to particular social morphology,
including hierarchical, centralized institutional arrangements, expansionary
potential, and dynamic leadership. The specific hypotheses enumerated below are
summarized in figure 4.2.
1. Frequent repetition activates semantic memory for religious teachings.
One of the most conspicuous features of the doctrinal mode is that the
transmission of religious teachings is highly routinized (i.e., frequently
repeated). A great advantage of frequent repetition is that it allows the
establishment of a great deal of explicit verbal knowledge in semantic
memory. Doctrines and narratives that would be impossible to learn and
remember if they were rarely transmitted can be effectively sustained
through repetitive sermonizing. Repetition, however, can lead to reduced
levels of motivation. In detailed empirical studies of this phenomenon, I
have labelled this the "tedium effect."5
But many routinized religions are
successful at holding onto their followers through a variety of mechanisms,
including supernatural sanctions (such as eternal damnation) and, more
66 CHAPTER 4
Figure 4.2. The Doctrinal Mode of Religiosity
04-024 Ch 04 2/5/04 8:25 AM Page 66
positively, incentives (such as eternal life and salvation). Of course, the
power of these mechanisms depends on people believing the religious
teachings. In order for people to believe in a set of doctrines, the doctrines
have to be cast in a highly persuasive fashion.This is commonly achieved, at
least in part, by special techniques of oratory established over time through
processes of selection. Routinized religions tend to be associated with
highly developed forms of rhetoric and logically integrated theology,
founded on absolute presuppositions that cannot be falsified.6
All of this is
commonly illustrated by poignant narratives that can easily be related to
personal experience.7
Additionally, the heavy repetition of explicit beliefs
increases their accessibility and relevance in everyday settings.8
2. Semantic memory for religious teachings and the presence of religious
leaders are mutually reinforcing features.
Where religious ideas are expressed in words (e.g., transmitted through
oratory), it is likely that the orators themselves will rise above the common
herd. Most religious traditions of this sort have celebrated leaders who
may take the form of gurus, messiahs, prophets, divine kings, high priests,
mediums, visionaries, disciples, or simply great evangelists or missionaries.
The very fact that there are so many different types of, and terms for,
religious leadership is an index of how widespread and important the
phenomenon is. Partly through their skills as orators, these leaders become
marked out as special. But, at the same time, their pronouncements (real
or attributed) provide the central tenets of a belief system, and their deeds
become the basis for widely recounted religious narratives, transmitted
orally. Both forms of knowledge are stored primarily in semantic memory.
3. The presence of religious leaders implies a need for orthodoxy checks.
Where religious leaders are upheld as the source of authoritative
religious knowledge, their teachings must be seen to be preserved intact.
At the very least, the credibility of any such tradition depends on its
adherents agreeing what the teachings are, even if other traditions hold to
alternative (and perhaps conflicting) versions. We might call this the
principle of agreement. Agreement depends partly on effective detection
of unauthorized innovation, and then on its effective obstruction and
suppression. Religious routinization contributes to both detection and
suppression by conferring a selective advantage on standardized/orthodox
forms over nonorthodox ones. The link between routinization and
detection is especially straightforward. Frequent repetition of a body of
religious teachings has the effect of fixing it firmly in people's minds. In
literate traditions, the teachings might also be written down in sacred
texts, and thereby fixed on paper (at least to some extent). But the crucial
THE THEORY OF MODES OF RELIGIOSITY 67
04-024 Ch 04 2/5/04 8:25 AM Page 67
thing is that standardized versions of the religious teachings become
widely shared and accepted through regular public rehearsal and
reiteration.9
Once this has happened, the risks of innovation going
undetected become remote. Rather more complex is the role of
routinization in the obstruction of unauthorized innovation, to which we
now turn in points four and five.
4. Frequent repetition leads to implicit memory for religious rituals.
So far, we have considered only the effects of frequent repetition of
religious teachings; but what about the effects of routinized ritual
performances? Rituals that are performed daily or weekly rapidly come to
be processed, to a considerable extent, in procedural/implicit memory.10
There can be little doubt that at least some Christians, for instance,
spend significant portions of church services simply going through the
motions. This is not a slur on people's religious commitments. It is
simply a psychological reality that repetitive actions lead to implicit
behavioral habits that occur independently of conscious thought or
control. Although potentially accessible to conscious representation (e.g.,
for the purposes of teaching a child or newcomer how to behave in
church), liturgical rituals may not, in the normal pattern of life, trigger
very much explicit knowledge at all.
5. Implicit memory for religious rituals enhances the survival potential of
authoritative teachings stored in semantic memory.
To the extent that people do participate in routinized rituals "on
autopilot," this reduces the chances that they will reflect on the meaning
of what they are doing. In other words, frequent repetition diminishes
the extent to which people come up with personal theories of their
rituals.11
And they are more likely to accept at face value any official
versions of the religious significance of their rituals. The processing of
routinized rituals as implicit procedural schemas really opens the way for
religious authorities to tell worshippers what to believe, especially when it
comes to the meanings of their rituals. At the same time, the provision of
a standardized orthodoxy tends to limit individual speculation. The
causal role of routinization in the suppression of unauthorized
innovation is, here again, governed by principles of selection. It is not
that frequent enactment of rituals prohibits exegetical innovation, but it
tends to reduce the volume and elaborateness of exegetical reflection,
leading to relatively low rates of unauthorized innovation across
populations of religious adherents. The question "relative to what?" will
be answered later in this chapter through an examination of processes of
exegetical reflection and independent innovation in the imagistic mode.
68 CHAPTER 4
04-024 Ch 04 2/5/04 8:25 AM Page 68
6. The need for orthodoxy checks encourages religious centralization.
Not all innovation is a bad thing.The principle of agreement simply
requires that innovation is seen to originate from authoritative sources and is
accepted or observed by all loyal followers. Routinization may have the effect
of insulating orthodoxies to some extent from unintended innovation, but it
does little to obstruct the determined heretic.The problem here is clearly
one of policing. As soon as a routinized religion becomes well established,
we tend to see the emergence of a central authority and some sort of
ranked, professional priesthood.12
It becomes the task of delegated officials
to police the orthodoxy across the tradition as a whole, and there will often
be a proliferation of sanctions for unauthorized innovation and heresy
(ranging from excommunication and ostracism to torture and execution).
7. Semantic memory for religious teachings leads to anonymous religious
communities.
Where religious beliefs and practices are frequently repeated, we have
seen that at least part of this religious knowledge is organized in semantic
memory. This means that the knowledge itself becomes separate from
particular episodes in which it is acquired. For instance, a Christian may
believe certain things (e.g., about the significance of the crucifixion) and
may do certain things (such as participating in weekly church services), but
that is not the same as remembering how and when all this knowledge was
acquired. In other words, many of the beliefs and acts that define a
person's identity as a Christian are not remembered as special episodes.13
In consequence, many aspects of what makes somebody a Christian are
really abstracted properties that, in principle, could be ascribed to
anybody. And, in fact, they do get ascribed to anonymous others. To
understand why, it is useful to think about the issues in a concrete way. If
you ask a regular church-going Christian to tell you what happened at a
service or mass three years ago, he or she wouldn't be able to remember the
actual event. That person could tell you, though, what happened, because
it would have been the same thing that always happens. In other words,
what makes a particular episode distinctive gets forgotten. This of course
includes the makeup of the congregation: people in the congregation come
and go, people die, they move in and out of the area, and there may be
visitors who come and go. If it is a big congregation, there may be many
people there who do not know each other personally. Thus, one's
memories for Christian rituals are not memories for a particular group of
people. What it means to be a regular churchgoer is not to be part of a
particular group, but to participate in a ritual scheme and belief structure
that anonymous others also share.14
Of course, the anonymity principle
THE THEORY OF MODES OF RELIGIOSITY 69
04-024 Ch 04 2/5/04 8:25 AM Page 69
only comes into operation if the religious community is large enough to
ensure that no individual follower could possibly know all the other
followers. And it turns out that there are factors at play in routinized
religions that encourage rapid spread, and therefore large-scale religious
communities. One of the most important of these is the emphasis on
oratory and religious leadership.
8. The presence of religious leaders is conducive to the religion spreading
widely.
The fact that the religious teachings are expressed in oratory, on the
part of great leaders (or their deputized representatives), means that these
teachings are readily transportable. Only one or a few proselytizing
leaders or good evangelists are required to spread the Word to very large
populations.15
In sum, the doctrinal mode of religiosity consists of a suite of mutually reinforcing
features. When these features coalesce, they tend to be very robust historically
and may last for centuries and even for millennia. At the root of all this is a
set of cognitive causes deriving from the ways in which frequently repeated activities
and beliefs are handled in human memory.
The Imagistic Mode of Religiosity
The sorts of practices that lead to the coalescence of imagistic features are invariably
low frequency (rarely enacted). They are also, without exception, highly
arousing. Examples might include traumatic and violent initiation rituals, ecstatic
practices of various cults, experiences of collective possession and altered states
of consciousness, and extreme rituals involving homicide or cannibalism. These
sorts of religious practices, although taking very diverse forms, are extremely
widespread.16
Archaeological and historical evidence suggests they are also the
most ancient forms of religious activity.17
As with the doctrinal mode, the coalescence
of features of the imagistic mode derives its robustness from the fact
that these features are causally interconnected or mutually reinforcing. Once
again, this claim rests on a series of testable hypotheses, depicted in figure 4.3
and enumerated below.
1. Infrequent repetition and high arousal activate episodic memory.
Rarely performed and highly arousing rituals invariably trigger vivid and
enduring episodic memories among the people who participate in them. It
appears to be a combination of episodic distinctiveness, emotionality, and
consequentiality that together result in lasting autobiographical memories.18
70 CHAPTER 4
04-024 Ch 04 2/5/04 8:25 AM Page 70
These memories can be so vivid and detailed that they can take the form of
(what some psychologists call) flashbulb memories.19
It is almost as if a
camera has gone off in one's head, illuminating the scene, and preserving it
forever in memory.The effects of infrequent performance and high levels
of arousal should be thought of in terms of processes of selection.
Religious practices that are rarely performed, but which elicit low levels of
arousal, are unlikely to be passed on: people will rapidly forget the
procedures, and especially their meanings, during the long gaps between
performances;20
even if they could remember some aspects of the rituals,
their lack of thought about these practices for long periods would not be
conducive to high motivation. In short, rarely performed religious practices
that survive tend to involve high levels of arousal, and this is due to the
triangular nexus of causes indicated in figure 4.3.
THE THEORY OF MODES OF RELIGIOSITY 71
Figure 4.3. The Imagistic Mode of Religiosity
04-024 Ch 04 2/5/04 8:25 AM Page 71
2. Activation of episodic memory triggers spontaneous exegetical reflection,
leading to expert exegetical frameworks stored in semantic memory.
The combination of infrequent repetition and high arousal may
provide excellent conditions for remembering the details of religious
procedures, such as ritual actions. But it does not seem to help people to
remember verbally transmitted information, such as doctrines and
narratives.21
It turns out that this needn't matter. In fact, the meaning and
salience of rare, climactic rituals usually lies in their capacity to trigger
spontaneous exegetical reflection (SER)--often experienced as personal
inspiration or revelation. The key to understanding this lies in the fact
that episodic memory is a type of explicit memory. This means that rare,
climactic rituals are processed at a conscious level. Not surprisingly,
people tend to reflect extensively on these experiences, and speculate
about their significance and meaning.22
This eventually results in
elaborate, if idiosyncratic, exegetical knowledge stored in semantic
memory. An important factor here is that elevated arousal is occasioned
typically by sensory stimulation (often using a variety of channels--
auditory, visual, kinesthetic, olfactory, etc.). This in turn encourages
people to draw associations between different images evoked in religious
ceremonies, which are rooted in the way perception is organized
(McCauley 2001). Two points need to be borne in mind here. The first is
that rare and climactic rituals evoke abundant inferences, producing a
sense of multivalence and multivocality of religious imagery, experienced
as personal and unmediated inspiration. The second requires a separate
hypothesis, illustrated in point three.
3. SER leads to a diversity of religious representations.
The personal experiences and revelations triggered by rare, climactic
rituals tend to be quite unique. They may converge on certain themes and
central ideas, but there is nothing resembling the kind of uniformity of
belief that characterizes doctrinal orthodoxies. The principle of
agreement, if it is invoked at all, applies only to the ritual procedures
themselves and not to their meanings.23
If exegesis is verbally transmitted,
it is restricted to "experts" whose adherence to the principle of
agreement may well be asserted but seldom demonstrated.24
4. SER and representational diversity inhibit dynamic leadership.
If a fertile and compelling array of religious beliefs and interpretations
is generated independently through personal reflection, dynamic
leadership is almost impossible to establish. If a leader tried to come
forward at rare, climactic rituals to advance an intricate and coherent
body of doctrine, people might listen. But they would very rapidly garble
72 CHAPTER 4
04-024 Ch 04 2/5/04 8:25 AM Page 72
or forget what they had been told and, at least in the long run, their own
inspirational ideas are likely to be more compelling than the content of a
single oration. In such circumstances, admittedly, the possibility remains
open for an individual, group, or class to be elevated socially, and for this
to be expressed in the structure and choreography of rituals and the
accordance of ritual precedence to persons of high standing. But
leadership of this sort is primarily symbolic rather than dynamic.25
5. Lack of dynamic leadership, lack of centralization, and lack of
orthodoxy are mutually reinforcing.
The fact that each person experiences inspiration as coming directly
from the gods or ancestors, rather than being mediated by leaders or
priests, means that there is no place here for centralized authority. And
there is no orthodoxy over which such an authority might preside.
6. High arousal fosters intense cohesion.
The high arousal involved in the imagistic mode tends to produce
emotional bonds between participants. In other words, there is intense
social cohesion.26
People who are bound together in this way tend to
form rather small and localized communities.
7. Intense cohesion and episodic memory foster localized, exclusive
communities.
Where rituals are remembered episodically, each participant remembers
who else went through the rituals with them. Ritual groups are based on
memories for shared episodes, in which particular coparticipants feature.
Consequently, religious communities tend to be exclusive: you cannot be a
member unless people remember you as part of a previous cycle of
religious activities; and, by the same token, you cannot very easily be
excluded once you are in (i.e., your participation cannot be easily forgotten).
This tends to give rise to fixed and exclusive ritual groups in which there is
no easy way of adding to, or subtracting from, the established membership.
8. Localized/exclusive communities and lack of dynamic leadership inhibit
spread/dissemination.
Unlike the beliefs and practices of the doctrinal mode, traditions
operating in the imagistic mode do not spread widely.27
Since religious
understandings are inspired by collective ritual performances, the unit of
transmission is the entire ritual group (not a small number of talented
orators). It follows that the spread of such traditions would be inefficient
and costly: either the local group must perform its rituals with
neighboring groups, or the local group must be mobile (i.e., migratory or
nomadic). But, either way, the practices are likely to mutate as soon as
they get passed on.28
In part, this is because of the lack of leaders and
THE THEORY OF MODES OF RELIGIOSITY 73
04-024 Ch 04 2/5/04 8:25 AM Page 73
religious hierarchies capable of policing an orthodoxy, and in part it is
because each ritual community is likely to be fiercely exclusivist (and
therefore will tend to emphasize local distinctiveness over regional unity).
The Nature and Origins of Modes of Religiosity
The key features of doctrinal and imagistic modes of religiosity stand in stark
contrast to one other, as represented in table 4.1. It will be observed that these
contrasting features are of two types. First, there are cognitive features, concerned
with differences in the way religious activities are handled psychologically. Second,
there are sociopolitical features, concerned with contrasts in social organization
and politics at the level of groups and populations. This clustering of sociopolitical
features has been widely recognized for quite a long time, but what is new
about the theory of modes of religiosity is the way it places these features together
in a single model, and then explains the clustering of features in terms of a set of
cognitive or psychological causes.
The theory advanced here operates on principles of selection. Modes of religiosity
constitute attractor positions around which ritual actions and associated
religious concepts cumulatively tend to cluster. Innovations remote from these attractor
positions cannot survive.29
For instance, a new prophet might discourse on
his elaborate personal revelations, and audiences might be eager to listen. But if
that discourse is to crystallize into a stable body of teachings, it must be subjected
to regular reiteration and safeguarded by a system of effective policing. If not, it
74 CHAPTER 4
Table 4.1. Contrasting Modes of Religiosity
Variable Doctrinal Imagistic
Psychological Features
1. Transmissive frequency High Low
2. Level of arousal Low High
3. Principal memory system Semantic schemas and implicit Episodic/flashbulb
scripts memory
4. Ritual meaning Learned/acquired Internally generated
5. Techniques of revelation Rhetoric, logical integration, Iconicity, multivocality,
narrative and multivalence
Sociopolitical Features
6. Social cohesion Diffuse Intense
7. Leadership Dynamic Passive/absent
8. Inclusivity/exclusivity Inclusive Exclusive
9. Spread Rapid, efficient Slow, inefficient
10. Scale Large-scale Small-scale
11. Degree of uniformity High Low
12. Structure Centralized Noncentralized
04-024 Ch 04 2/5/04 8:25 AM Page 74
will be garbled or simply forgotten. Likewise, a new ritual might be invented to
mark the effects of a rare event, such as a solar eclipse. But if that ritual is to establish
the basis for a new religious tradition, it must be sufficiently arousing,
shocking, and personally consequential to drive subsequent revelations based on
SER. If not, it too will fail to stabilize as a tradition. History is obviously littered
with such failures.
Religious practices commonly satisfy at least one or other of the two sets of
psychological conditions specified in table 4.1.The activation of these conditions
provides the underlying causes of the distributed (population-level) effects depicted
as the "sociopolitical features" of religion. But it would not make sense to
try to single out any one of the psychological causes as somehow prior to any of
the others. There is no independent variable driving the rest, only a set of conditions
that some patterns of human activity manage to satisfy, thus accounting for
their cultural success.
Modes of Religiosity in the Real World
Anybody who has studied a particular religious tradition in any detail will know
that religions are neither doctrinal nor imagistic in terms of the features identified
in my model. In some cases, a religious tradition that incorporates all the elements
of the doctrinal mode also exhibits some of the features of the imagistic mode.
At the same time, this religion may embrace a large population of lay adherents
who have little or no access to the tradition's complex body of revelatory knowledge,
and so could hardly be said to be motivated by it. Some rituals might be low
in frequency and elicit low levels of arousal. Other rituals might be neither frequent
nor particularly rare and instead are scattered across an intermediate range
of performance frequencies. Some frequent rituals may be completely lacking in
known exegesis, and some rarely performed rituals might be associated with quite
an elaborate and standardized exegetical corpus. These kinds of scenarios might
seem to disprove the claims of the theory of modes of religiosity. And if they do
not, then what counts as falsification?
In the first place, as noted in the previous section, modes of religiosity are attractor
positions. They do not specify a set of law-like rules for building individual
behavior.The claim is not that all instances of ritual action conform to one or
other mode of religiosity. Indeed, that would be impossible by definition, since
roughly half the variables with which the theory is concerned itself (the sociopolitical
features of table 4.1) relate to distributed population-level attributes rather
than particular instances of thought/behavior (even though it is the latter that cumulatively
cause the former). So we cannot say that a particular ritual, for instance,
is doctrinal or imagistic.We can only say that its long-term reproduction through
THE THEORY OF MODES OF RELIGIOSITY 75
04-024 Ch 04 2/5/04 8:25 AM Page 75
the innumerable thoughts and actions of many people results in the coalescence
of features specified by the modal theory. These features, in other words, are discernible
only as marked tendencies within a religious tradition, taken in the round.
What the theory of modes of religiosity sets out to explain, then, is the tendency
for religious systems to gravitate toward divergent attractor positions. It is
only through these processes that intrinsically hard-to-acquire revelatory knowledge
can be generated and culturally transmitted. This kind of knowledge, stored
in semantic memory, might be quite unevenly distributed within a tradition. In the
case of the imagistic mode, such bodies of expert knowledge take many years to
develop and mature. Consequently, the less experienced members of the tradition
do not yet have access to the full motivating force of its revelatory knowledge (and
their participation often has to be coercively enforced by the elders/experts). In
the doctrinal mode, it is possible in principle for everybody to have access to revelatory
knowledge; that has clearly been the aim (if not the outcome), for instance,
in many post-Reformation Christian traditions. But it is also possible for the doctrinal
corpus to be largely confined to elites. In many parts of the world, the "little
traditions" (Redfield 1955) of rural tribespeople and peasantries are founded
on versions of elite religious practices that (for the laity) lack a systematic justification
in doctrine and narrative. In the absence of pedagogic support and effective
policing, we often find that lay versions of world religions migrate away from
both of our modal attractor positions and settle around more easily acquired, intuitive
concepts and practices (the cognitive optimum position) that consequently
require neither routinization nor high arousal to maintain. Yet another possibility,
and a particularly common one in some parts of the world, is for modes of religiosity
to interact in complex ways.
Religious traditions founded upon interacting modes of religiosity encompass
large populations but, at the same time, they are composed of many locally distinctive
ritual communities. The cohesion of the latter may readily be projected
onto the wider religious community, and such processes appear to have been crucial
in many large-scale and bloody religiously motivated wars (see chapter 7). In
other cases, however, the effect of the imagistic mode is not necessarily to intensify
commitment to a set of principles codified in language, but, rather, to provide
a substitute for such principles as the main source of religious motivation. It is
precisely within those populations that lack access to the authoritative corpus of
religious teachings--and so cannot be adequately motivated by these teachings--
that we find the greatest profusion of imagistic practices. Elitist discourses would
have us believe that the prominence of the imagistic mode among the uneducated
and dispossessed is symptomatic of ignorance. Expressed more precisely, and less
snobbishly, routinized religious rituals that lack a persuasive justification in dogma
(i.e., learned via instruction) will die out unless they are either naturalized
76 CHAPTER 4
04-024 Ch 04 2/5/04 8:25 AM Page 76
(through the proliferation of cognitively optimal versions) or motivated by forms
of religious experience and understanding that are, at least to some significant extent,
internally generated. A model for this sort of motivational base is provided
the world over by the ancient imagistic mode of religiosity.
Since the possibilities afforded by modal dynamics are quite numerous and
complex, it might seem as though the theory forbids nothing and is therefore unfalsifiable.That
is not the case, however. Chapter 9 attempts to set out predictions
of the theory of modes of religiosity in a way that could be systematically falsified
by empirical data from psychology, archaeology, historiography, and ethnog-
raphy.
The Origins of Modes of Religiosity
The presence of the imagistic mode almost certainly predates the emergence of
the doctrinal mode by a very substantial margin. The former appears in the archaeological
record at least as far back as the Upper Paleolithic period, whereas
the latter appears probably no less recently than the emergence of Bronze Age civilizations
(Whitehouse 2000a, chapter 8). Obviously, the first fully modern humans
had very much the same cognitive equipment as modern peoples, so why did
it take so long for both modes of religiosity to emerge, and why are not both
modes universal?
The answer almost certainly lies in the fact that processes of experimentation
with patterns of ritual behavior and revelatory thinking are not random. The features
that comprise our divergent modes of religiosity probably do not coalesce in
the absence of some kinds of triggers located outside the mode dynamics themselves.
It is hard to imagine, for instance, why a band of peaceful hunter-gatherers
with abundant resources would have occasion to experiment with the extremely
costly patterns of behavior needed to get the development of modes of religiosity
underway. Indeed, many modern-day hunter-gatherers with sufficient territory
at their disposal do not experiment in that way, even when surrounded by groups
with elaborate religious models of this kind. Instead, egalitarian foraging bands
(Woodburn 1982) seem to make do with relatively simple rituals and concepts of
the supernatural clustered around the cognitive optimum position. In order for
our distant ancestors in the Upper Paleolithic period to have adopted forms of religion
operating in the imagistic mode, there would have needed to be significant
external pressures.
A particularly obvious pressure for at least some ancestral populations would
have been the advancing ice sheets. For instance, we know that the increasingly harsh
conditions of the last ice age triggered new and often more dangerous strategies of
cooperative hunting. It is possible that, initially, low-frequency, high-arousal group
THE THEORY OF MODES OF RELIGIOSITY 77
04-024 Ch 04 2/5/04 8:25 AM Page 77
activities of this kind became linked with attempts to manipulate the environment
through the performance of rituals. Cohesive units formed through the communal
performance of low-frequency, high-arousal rituals would have been capable of wiping
out, displacing, or absorbing less cohesive bands of hunter-gatherers competing
for the same resources (under conditions, of course, of growing scarcity).This helps
to explain the creative explosion of artistic imagery apparently associated with terrifying
initiation rites in the Upper Paleolithic period (Pfeiffer 1982; Whitehouse
2000a, chapter 8). It suggests the development--perhaps for the first time--of
complex bodies of revelatory religious knowledge based on processes of SER. Such
traditions, once established, would have been very robust. Ritual experts may have
tried to communicate their elaborate revelations by word of mouth, but unless their
attempts to transmit information in that way could become routinized and centrally
policed, it is hard to see how the doctrinal mode could become established. One reason
why doctrinal religions seem to have been slow to get off the ground is that there
were no external pressures to carry ritual innovations in that direction until the emergence
of large-scale agricultural societies (Whitehouse 2000a, 169­172). Only then
did the seasonal labor cycles and the increasingly complex nature of social cooperation
foster the routinization of ritual and the centralization of religious authority,
allowing other core elements of the doctrinal mode to coalesce (Whitehouse 2000a,
chapter 8).
An alternative cognitive account of the emergence of doctrinal orthodoxies is
presented by Pascal Boyer's theory of the emergence of professionalized religious
guilds (Boyer 2001a). Like me, Boyer dates the emergence of doctrinal orthodoxies
no earlier than the rise of complex societies, approximately six thousand years
ago. His story begins, however, before the first centralized states began to take
shape. To the list of cultural traits that cluster around the cognitive optimum position
Boyer adds another set of concepts pertaining to the classification of humans
into different natural kinds.30
People all around the world tacitly presume
that the other humans they encounter can be classified according to innate properties.
Explicit versions of this kind of thinking--taking, for instance, the form
of racist discourses--may be highly variable in content. But what matters for
Boyer's theory is that almost any kind of trait that can be associated with some
people and not others is liable to trigger intuitions about inherent categorical differences
(regardless of how absurd these categories may seem by the lights of scientific
biological knowledge).There are those who speak this language rather than
that, who have these kinds of bones rather than those kinds (e.g., based on principles
of descent), and--most importantly from the point of view of religious
thinking--there are those who have particular qualities that make them better
equipped than others to deal with supernatural agents. These qualities may be
construed in local cultural registers as invisible marks or perhaps even as physical
78 CHAPTER 4
04-024 Ch 04 2/5/04 8:25 AM Page 78
appendages to internal organs. But something makes these people different, and
whatever it is finds abundant support in tacit essentialist reasoning. Long ago,
when all societies were very small-scale, religious specialists served only a group of
locals. But when larger populations came together with the emergence of citystates,
religious specialists, by virtue of their essentialized differences from ordinary
folks, were naturally compelled to form guilds.
As such guilds became professionalized, they came under pressure to protect
their share of the market.The solution, according to Boyer, was to establish a distinctive,
standardized, easily recognizable, and securely patented brand. But that
required the assistance of literacy. Texts would become the guarantee of both exclusive
truth and inclusive orthodoxy. Once this happened, the heyday of the lone
specialist dealing in local concepts and rituals had passed. The literate guilds instead
supplied general truths for all and, wherever possible, sought the backing of
powerful economic and political interests at the heart of the state machine. Religious
systems were now established around "coherent . . . generally integrated . . . apparently
deductive . . . and stable" religious doctrines (2001a, 278, emphases in
original).
With the rise of more complex and regionally standardized doctrinal systems
came a tendency to "downplay intuition" (Boyer 2001a, 278) and thus to exclude
the beliefs and practices associated with the cognitive optimum position.Thus, as
religious supermarket chains started to squeeze the corner shop specialists, the
wares that were on offer came to be more explicitly and coherently organized. According
to Boyer, preliterate religious traditions had "no systematic doctrine of
supernatural agents . . . [and] no theory of what these agents are like, what they
do, where they reside, etc." (2001a, 266).31
The wares of the local religious specialist
did not amount to "such a thing as `religion' as a special domain of concepts
and activities" (Boyer 2001a, 267).Thus, according to Boyer, the emergence
of professionalized religious guilds produced for the first time an explicit notion
of religion as a demarcated domain of thought and action.
Boyer's theory brilliantly welds Gellner'sWeberian sociology and Goody's theory
of literacy to the main findings of social and evolutionary psychology in a
fashion that is both highly original and rhetorically compelling. But there are also
problems with Boyer's account. To begin, Boyer advances a series of correlations
that run rather strikingly against the grain of available empirical evidence. For instance,
are the concepts of preliterate, localized traditions necessarily clustered
more substantially around the cognitive optimum position than, say, the belief systems
sanctioned by literate elites? Detailed ethnographic evidence on a great range
of small-scale nonliterate cultures strongly suggests otherwise. Australian Aboriginal
ideas about the "dreamtime" (or "dreaming") provide a good case in point.
Extensive direct study of these traditions has revealed the presence of dauntingly
THE THEORY OF MODES OF RELIGIOSITY 79
04-024 Ch 04 2/5/04 8:25 AM Page 79
elaborate bodies of philosophical/cosmological knowledge that require many
years of intensive contemplation to develop and mature. In attempting to convey
a sense of the scope and complexity of Aboriginal religious thought, one anthropologist
sums up the evidence as follows:
Australian Aborigines have incredibly subtle, philosophically challenging mystical
cosmologies that posit a spiritual plane of existence that was prior to the world
of sensory experience (in the "dreamtime") but now lies behind or parallel to it.
Mervyn Meggitt . . . describes how the old Walbiri man who was his spiritual
guide eventually told him gently that he, Meggitt, had reached his philosophical
depth and could follow no longer into the mysteries of the cosmos. Probably no
Westerner has ever fully penetrated these Aboriginal philosophical realms.
(Keesing 1981, 333­334)
Similar comments have been made with regard to the cosmologies of small-scale,
nonliterate societies in Amazonia (Reichel-Dolmatoff 1971), Africa (Griaule
1975), and Melanesia (Juillerat 1992).The overall impression from ethnographic
research is that nonliterate societies, in general, do not deal in religious concepts
that are closer to the cognitive optimum than those of literate societies. The cognitive
optimum is a natural attractor position (present in all societies) but countervailing
tendencies are by no means limited to complex societies with literate,
professionalized religious guilds.
The role accorded to literacy in Boyer's model also raises important issues. Boyer
rightly points to a close correlation between the presence of literacy and the homogenization
of regional traditions policed by professional guilds. A crucial question,
to which archaeology may hold the key, is whether the emergence of literacy
helps to create doctrinal orthodoxy or, as I would suggest, the emergence of doctrinal
orthodoxy creates some of the pressure to develop systems of writing. It would
help, of course, to know whether early signs of literate innovation occur alongside
the emergence of religious standardization or whether one tends to precede the
other. Recent studies of the origins of professional religious guilds promoting standardized
concepts of a doctrinally complex nature suggest that these features predate
the advent of literacy in a number of key locations (Mithen 2004; Johnson
2004). It would therefore appear that the presence of doctrinal orthodoxies favors
the subsequent development of writing systems, rather than being caused by them.
But even if Boyer's account turns out to provide a better fit with available data, we
still have to explain how the complex systems of concepts associated with doctrinal
orthodoxies come to be successfully disseminated in a population. The theory of
modes of religiosity might help to solve at least some of these problems.
For a start, the modes theory does not attempt to correlate particularism and
orality with the cognitive optimum position, on the one hand, and universalism
80 CHAPTER 4
04-024 Ch 04 2/5/04 8:25 AM Page 80
and literacy with cosmological complexity on the other. Consequently, it is not
threatened by ethnographic data demonstrating the tendency of many small-scale,
nonliterate traditions to drift away from the cognitive optimum position. On the
contrary, the modes theory was built with data of that sort in mind.The challenge
all along has been to show precisely how the localized religions of Papua New
Guinea managed to generate profoundly complex revelations associated with mystery
cults in the absence of either routinized forms of education or external
mnemonics (e.g., systems of writing). The solution, I have suggested, is one that
was discovered way back in human prehistory and has remained part of the cultural
repertoire of many populations ever since--from the "simple" societies of
such regions as Melanesia to the complex polities of Europe and elsewhere. It
takes the form of what I have called the imagistic mode of religiosity.32
Through elevated arousal, cognitive shocks, and the creation of consequential
events, the rituals of the imagistic mode set off trains of exegetical thinking that
are enduring and (over time) capable of generating highly elaborate semantic
knowledge. The motor driving this process is a stock of episodic memories for
traumatic ritual ordeals. These are the general conditions, I would suggest, in
which the great philosophers of Aboriginal Australia, Amazonia, Africa, and
Melanesia (as well as of other small-scale, nonliterate societies) come into existence.
But philosophers are found in complex societies as well. Some of these are
very much like the ritual experts of precontact New Guinea--the spiritual fathers
of Hellenistic Mithraic cults, for instance, or the high priests of many contemporary
pagan cults in Europe and North America. But another kind of philosopher
was also born with the advent of the doctrinal mode of religiosity. What
made this new breed of religious experts different was that their knowledge could
be transmitted verbally, via highly repetitive regimes of teaching and reminding.
And these methods of transmission opened the floodgates to processes of standardization
and the policing of emergent orthodoxies.
The independent development of the doctrinal mode requires that a number
of elements initially fall into place. For a start, of course, there must be inspired
individuals with a complex and compelling message to communicate. According
to Boyer's model, such persons only came into existence along with the establishment
of religious guilds. According to the modes theory, by contrast, such figures
have been around for very much longer than that. What they lacked throughout
most of human prehistory was a means of transmitting their knowledge by word
of mouth.They required a forum for teaching complex knowledge to attentive and
credulous audiences--a setting, in short, that allowed forms of routinized oratory
and instruction.
Many factors could have been implicated in the establishment of such forums
in different places at different times. An acceleration in the pace of ritual life,
THE THEORY OF MODES OF RELIGIOSITY 81
04-024 Ch 04 2/5/04 8:25 AM Page 81
caused by periods of drought, famine, or disease, might be a contender. The development
of more routinized forms of organized labor might be another. But
whatever the triggers,33
transmission of complex religious knowledge by means of
language required methods of sustained reiteration.
Still, this would not have been enough on its own. Even if we concede that
routinization could (help to) explain the susceptibility of audiences to particular
teachings (as will be argued at greater length in the next chapter), we still have to
explain what motivated their interest. Complex teachings would have to be persuasive,
and persuasiveness depends on the use of all those devices that Boyer associates
with literate guilds: coherence, integration, rhetoric, and plausibility. Literacy
might assist in the perfection of some of these features of the doctrinal mode,
thus helping to explain the concomitant emergence and spread of writing systems,
but it does not explain why doctrinal orthodoxies come into existence in the first
place.
This chapter has attempted merely to summarize the theory of modes of religiosity
at a very general level. As an introductory overview, it may have raised more
questions than it has answered. A much fuller account of my central claims now
follows. Focused heavily on the psychological variables driving the model, particularly
the complex relationships between memory, codification, frequency, transmission,
and arousal (chapters 5 and 6), this account is intended to fill in many
of the gaps in my earlier volumes and in the above précis.
Notes
1. The dividing lines between explicit and implicit memory are difficult to draw (for
a fine overview, see Schachter 1987), but evidence from studies of normal cognition (e.g.,
Roediger 1990) and amnesic patients (e.g., Graf, Squire, and Mandler 1984) show that
such a distinction (or a series of more fine-grained distinctions) is difficult to avoid (although
see Baddeley 1997, chapter 20). Many of these issues are discussed more fully in
chapters 4 and 5.
2. This particular distinction has a long history, and certainly predates cognitive science.
It is apparent, for instance, inWilliam James' (1890) discussion of primary and secondary
memory, and the first experimental studies of short-term memory date back to the
same period (Jacobs 1887).
3. The distinction between semantic and episodic memory was first fully developed
by Tulving (1972), and is now used by psychologists studying a wide range of phenomena,
including amnesia, aphasia and agnosia, story grammars, schemas and scripts, and
framing and modeling. For a thorough overview, see Baddeley 1997.
82 CHAPTER 4
04-024 Ch 04 2/5/04 8:25 AM Page 82
4. See Whitehouse 1992, 1994, 1995, 1996a, 1996b, 1996c, 1998, 2000a, 2000b,
2001a, 2001b.
5. See, for instance, Whitehouse 2000a, 44­46, 115, 142­143, 148, 155.
6. For the most part, dogma is interwoven by strings of logical implications. Since the
range of possible strings is far greater than those that happen to be exploited by religious
teachers, standardization necessitates frequent rehearsal (i.e., a routinized regime of doctrinal
transmission).
7. See, for instance, Whitehouse 1995, chapter 7, and 2000a, 60­63.
8. Powell and Fazio (1984), for instance, have shown that the motivational force of
explicit beliefs is at least partly a function of frequent rehearsal.
9. For a fuller discussion of this point, see Whitehouse 2000a, 151­153, 172­180.
Some criticisms of this aspect of my argument notwithstanding (e.g., Boyer 2002), I accept
that literacy is a precipitating condition (perhaps even a necessary condition) for the
independent invention of doctrinal mode phenomena (Whitehouse 2000a, 179­180),
but it is not essential for their reproduction (Whitehouse 1992).
10. See in particular chapter 5.
11. Some (admittedly preliminary) experimental support for this claim comes from a
study by Barrett andWhitehouse of spontaneous exegetical reflection (SER) generated by
repeated performances of an artificial ritual modeled on the Catholic practice of selfcrossing.
This study suggested that levels and volume of SER correlate inversely with degree
of repetition and habituation.
12. This argument is elaborated in Whitehouse 2000a, chapter 8. For a similar (and
fuller) overview of these processes, see Diamond 1998.
13. The reality is a bit more complicated than that. Consider, for instance, conversion
experiences in some Christian traditions, which appear to be constructed around episodic
memories. From the viewpoint of my argument, three points are crucial to make about these
sorts of phenomena. First, where episodic memory plays a significant role in the doctrinal
mode, it is typically in relation to highly personalized rather than collectively experienced
episodes (episodes of the latter sort tend to produce something altogether different--an
imagistic domain of operation, discussed below). Second, these highly personalized
episodes tend to be subjected to such frequent verbal reiteration that they eventually give
rise to quite rigidly schematized, even stereotyped, narratives (thus dissolving into the standardized
schemas of semantic memory). Third, religious experiences encoded in episodic
memory are invariably superfluous to the doctrinal mode in the sense that the reproduction
of the doctrinal tradition in a recognizable form does not depend on their preservation. In
short, being a member of a doctrinal tradition (e.g., a Christian) minimally presumes some
level of commitment to schemas encoded in semantic memory--no more and no less.
14. See Whitehouse 1992 and 2000a, 9­12, 40­41, 50­52, 113­117.
15. See Whitehouse 1992, 1994, and 2000a, 72­80.
16. For ethnographic examples, see Lowie 1924, Kluckholn and Leighton 1974
(1946), Turnbull 1962, Meggitt 1962, Allen 1967, Strehlow 1965, Barth 1975, 1987,
Tuzin 1980, Herdt 1981, 1982.
17. See Lewis-Williams 1977 and Pfieffer 1982.
THE THEORY OF MODES OF RELIGIOSITY 83
04-024 Ch 04 2/5/04 8:25 AM Page 83
18. The evidence here is somewhat complex, but useful overviews are presented by
Conway 1995 and Christianson 1992.
19. This term was first coined by Brown and Kulik 1982, and has since been examined
in a variety of major studies (discussed at greater length in chapter 6). The role of flashbulb
memory in recall for ritual episodes has been most extensively discussed in Whitehouse
1996a, 2000a, McCauley 2001, McCauley and Lawson 2002, and Atran 2002.
20. The only cases of low-frequency, low-arousal rituals known to me are ones that use
external mnemonics and/or a compositional hierarchy of ritual elements (i.e., rarely performed
rituals composed of an assortment of more frequently performed rites). For examples,
see McCauley 2001 and Atran 2002. But such exceptions seem to prove the
rule--not only because they are hard to find but because they always constitute practices
that are inessential to the reproduction, in a recognizable form, of the doctrinal traditions
in which they occur.
21. A recent pilot study by Barrett andWhitehouse suggests that recall for rarely transmitted
verbal exegesis is extremely poor, and even more so for rarely transmitted behavioral
procedures. In this study, a class of 100 first-year anthropology students participated in
an artificial ritual requiring them to carry out a series of unusual actions. They were told
that the purpose of this was to learn about the pressures of ethnographic fieldwork, especially
the effects on stress levels among researchers of having to participate in strange
activities. Participants were instructed not to write down what they had heard. The theological
statement was delivered loudly and slowly, to maximize the chances of successful
encoding. Participants then completed a short questionnaire asking them to rate their
emotional states during the performance. Seven weeks later, participants completed a questionnaire
asking them to record both the action sequence they had performed, the stated
reasons for the experiment, and the fictitious theology they had heard. The elements and
sequence of the ritual actions were recalled more or less perfectly by the entire class. By
contrast, recall for the fictitious theology and even for the stated reasons for the experiment
was virtually nil. This particular experiment was unsuccessful, insofar as it was intended
to establish correlations between emotional self-ratings and recall for various
aspects of the artificial ritual.The lack of significant variation in recall performance made
this impossible. Nevertheless, our findings do suggest that the cultural reproduction of ritual
actions does not require very great frequency (even quite rarely performed actions sequences
will be well-remembered). By contrast, even the simplest exegetical and theological
concepts cannot survive relatively long transmissive cycles. In order to be learned in the
first place, and sustained in semantic memory in the long run, they must be repeated and
rehearsed.
22. All rituals have the potential to trigger SER by virtue of being irreducible to any
set of technical motivations (see the introduction). Nevertheless, frequent repetition can
reduce the likelihood of an internal search for symbolic motivations being initiated by
causing habituation and reliance on implicit procedural knowledge. This is not the case
with respect to low-frequency, high-arousal rituals activating episodic memory. Whenever
recall for the rituals is triggered, this will involve recall of an explicit kind that is, in turn,
eminently capable of setting off a search for symbolic motivations.
84 CHAPTER 4
04-024 Ch 04 2/5/04 8:25 AM Page 84
23. See Barth 1975 and Whitehouse 2000a.
24. See especially Whitehouse 2000a, chapter 4.
25. In other words, the position of leader (if it exists) does not afford opportunities
to transmit, shape, or direct any systematic program of belief and action.
26. See, for instance, Aronson and Mills 1959 and Mills and Mintz 1972.
27. Ethnographic evidence for this is presented in Whitehouse 2000a; historiographical
evidence is presented in Whitehouse and Martin 2004.
28. Classic ethnographic studies include Williams 1928, Schwartz 1962, and Barth
1987 (for an extended discussion, see Whitehouse 2000a).
29. There are exceptions, however. If transformed into more intuitive variants or provided
with exceptional forms of mnemonic support, certain concepts and practices remote
from our modal attractor positions can indeed survive. But, either way, they cease to contribute
to the transmission of revelatory knowledge or to the distinctive social morphology
of religious traditions. These issues are discussed in detail in later chapters.
30. For a slightly different view of how this works, see Hirschfeld 1996.
31. Boyer is here describing the case of a contemporary nonliterate religious tradition,
that of the Buid of the Philippines (Gibson 1986). But he uses this case to illustrate the
general situation that he suggests obtained prior to the rise of the first civilizations.
32. Possible factors triggering the emergence of the imagistic mode are discussed in
Whitehouse 2000a, chapter 8.
33. A detailed investigation of these issues was begun at a conference on the historiographical
and archeological evidence for modes of religiosity, held at the University of
Vermont in August 2002, funded by the British Academy and theTempleton Foundation,
and published in the present series (Whitehouse and Martin 2004).
THE THEORY OF MODES OF RELIGIOSITY 85
04-024 Ch 04 2/5/04 8:25 AM Page 85
04-024 Ch 04 2/5/04 8:25 AM Page 86