JESPER SØRENSEN
The Question of Ritual
A Cognitive Approach
Introduction
Why does ritual continue to be an issue in religious studies and in anthropology? Why do we keep writing books and having conferences devoted
to this apparently elusive subject? In this paper I will propose a cognitive
approach to rituals, focussing on those aspects of rituals that are distinct
from other types of actions, together with what cognitive responses these
differences provoke. It will be argued that rituals violate basic causal assumptions and by doing so, trigger off cognitive processes in order to
ascribe purpose and meaning to the action. In conclusion, this will be related to findings in ethology and evolutionary theory, arguing that ritual
as a behavioural category plays an important role in the formation of symbolic thinking.
However, I would first like to discuss why rituals persistently provoke
such heated scholarly debate. Why are rituals interesting? A pragmatic answer to this could be that it is our most important raw material. Besides texts,
behaviour, and notably ritual behaviour, constitutes the primary source
material for the study of religion. As a mode of activity, ritual provokes a
search for explanation, for a rationale underlying the apparently irrational
and non-instrumental behaviour performed and observed. Thus one of
the primary features of ritual action is exactly the opacity of the behaviour
in question. It is not obvious what reasons underlie the behaviour, and
how its purpose is related to the actions performed. Thus when the functionality of an action recedes into the background, it is recategorised as an
instance of ritualised behaviour motivated by other factors both formally
and substantially.
This peculiar character of ritual action is recognised by both participants seeking out ritual and by religious experts searching for an answer
to the meaning and purpose of particular rituals, as well as by scholars
looking for the meaning and function of particular rituals and of ritual in
general. In the study of religion, religious ritual as a behavioural category
is of course more circumscribed than the broad folk-theory just referred to:
208
JESPER SØRENSEN
it concerns interactions with hidden and superempirical agents, it is prescribed by tradition, and it is a collective mode of action, even when performed in private. I will claim nevertheless that our intuitions about what
constitutes ritual behaviour are informed by the folk-theory according to
which ritual is a type of behaviour not easily ascribed to rational or instrumental causes. The reasons for ritual behaviour must be found elsewhere.
Why is ritual such an important category in the study of religion? I
think the reason is that ritual is the primary and most accessible indication
that people entertain religious beliefs — beliefs subsequently used as the
explanation of the actions observed. Like other types of human action, rituals are explained by reference to the belief-states of the agents. They are
seen as indicating that people entertain certain beliefs, as they are believed
to act upon these. However this seems to be a circular argument. On the
one hand, ritual actions are used to claim the existence of religious belief
motivating the actions (actions pointing to the existence of beliefs), and on
the other hand these purported beliefs are used to explain the ritual actions observed (beliefs explaining actions). This is of course a standard
mode of inference in explaining ordinary, non-ritual behaviour, but as we
shall see in the following, the special character of ritual actions questions
the utility of this approach. Problems arise when people do not refer to
beliefs when asked why they perform a particular ritual and what it means
— at least these beliefs will not explain the form, content or structure of
actions performed. The circle is broken and the ritual can no longer be seen
as an index of specific and commonly held underlying beliefs, nor can the
rationale underlying ritual performance be explained by reference to such
beliefs. As we shall see later, this does not imply that there is no relation
between ritual and belief-states, but merely that we need to move beyond
the circular argument in order to describe how they are connected.
Thus traditional explanations of rituals point to the underlying beliefs
and explain ritual actions as social expressions of more or less conscious
but culturally shared beliefs. A way out of this circularity connecting individual belief-states and ritual action is to claim that rituals do not find
their raison d'être in individual beliefs, but somewhere else. Even though
more sociologically inclined explanations have not fundamentally challenged the premise of individual beliefs underlying ritual actions, they
have added another explanatory level. It is argued that rituals form a necessary part of social life, ensuring the continuation and persistence of the
group by expressing more or less consciously held fundamental values
and structures of the group. People perform rituals motivated by beliefs,
but the real function of rituals can be found elsewhere, namely in the effect
the rituals have on social cohesion through their employment of common
symbolic structures. Through in-depth analysis, the observer is believed
to be able to "crack the code" and relate the symbolic elements found in
THE QUESTION OF RITUAL
209
the ritual to a general symbolic system that in some aspects corresponds to
external social structure.
Theories focussing on the symbolic character of ritual actions have been
challenged on several points. In Rethinking Symbolism, anthropologist Dan
Sperber (1975) argues against the notion that ritual actions should be seen
as expressions of symbolic systems in need of interpretation. He argues
that the relation between ritual actions and underlying explanatory systems, whether internal beliefs or external semiological structures, are much
more complex than hitherto acknowledged. The symbolic interpretation
given by the observer is, according to Sperber, a mere extension of the activity found in the ritual itself, and symbolic interpretations will therefore
never explain why a ritual is performed nor what it does, but merely add to
the ever-growing numbers of local exegeses.
More recently, Caroline Humphrey and James Laidlaw (1994) have argued that the relation between intentions entertained by the agents and
the actions performed are radically transformed in ritual actions, in comparison with ordinary action. When engaged in ritual performance, people only entertain intentions in relations to the whole ritual sequence,
whereas the constituent actions are stipulated by tradition and therefore
intentionally underdetermined. In ritual, there is no direct feedback between the result of actions performed and how the ritual is performed.
Ritual as a special behavioural category
Above, we found two points in which ritual actions are to a significant
degree distinct from ordinary actions: 1) The performance of rituals cannot be explained by reference to underlying symbolic and explanatory systems or belief-states; 2) the relation between participants' intentions and
the actions performed is radically altered. In this paper I wish to point to a
third important feature of ritual actions. On a cognitive level, rituals imply
that intuitively held assumptions and domain-specific expectations regarding the causal properties of the entities involved are disconnected (Sorensen
2000). In recent years cognitive psychology has established evidence that
human cognition is constrained to a certain extent by domain-specific categorisation. Humans entertain different intuitive and unconscious expectations regarding different domains of reality, such as physical objects, animals, theory of mind, and social categories. The exact number of these
domains is disputed, as is their basis, whether innate or acquired, but this
need not concern us in the present context (for discussion see Hirschfeld
and Gelman 1994; Sperber, Premack and Premack 1995; Whitehouse 2001).
Contrary to ordinary actions, in which a large number of such intuitively held assumptions guide our expectations, rituals explicitly downplay
14
210
JESPER SØRENSEN
or violate some of these assumptions and thereby provoke a search for
meaning through either perceptual characteristics or more or less reified
symbolic interpretations. When stones can think and act, birds are spirits
and human souls can leave the body, ordinary and automatic domain-specific inferences are violated. This not only enhances memorability and transmission, as argued by Pascal Boyer (1994, 2001), but it also provokes the
invocation of alternative hermeneutic strategies available in order for participants to extract the meaning or purpose of the ritual sequence. It is
these alternative strategies I will discuss in the following.
The difference between ordinary action and ritual action can be illustrated by means of two simple models depicting the two types of action as
event-frames (Fillmore 1982; Sørensen 2000). An event-frame is an idealised, mental construct of a given action-sequence containing the relevant
elements and their internal relation. In the models described below, an
event-frame is depicted by an analytic distinction between three phases of
the actions involved. A phase before the commencement of the action (conditional space), the proper action itself (action space), and the effect of the
action (effect space). A premise of the following argument is that agents
performing an action will have a broadly equivalent representation of an
action, its conditions, and its consequences.
Figure 1 depicts the non-ritual action of breaking a window by throwing a stone through it. In the conditional space, all sorts of intuitive background knowledge about windows, human agents, and interactions between these are present. Windows are physical objects that do not move
by themselves but are subject to physical causation; humans can act on
their own volition and can interact with most physical objects through their
body. In short, it contains knowledge derived through a combination of
experience and intuitive ontological assumptions. This might seem trivial
and so it is in the sense that it is not present in our consciousness in normal
circumstances. But it has crucial effects on the way we represent actions
involving the window: that it can be broken by a physical object, that it is
subject to physical force, that humans can perform certain movements and
thereby influence the window etc. Thus the action of breaking a window is
informed by knowledge about windows and human agents and knowledge about what type of actions can interfere with windows, i.e. the causal
expectations and assumptions relating action and the conditional space
through what can be called a diagnostic process.
THE QUESTION OF RITUAL
211
Conditional space
Action space
Effect space
`Broken
window' /
Figure 1. Non-ritual event-frame
Similarly, a prognostic process relates the action to representations of the
effect of the action, in this case the window being broken. Again, all sorts
of causal expectations are present, for instance that it is the physical force
of the stone that breaks the window, that the shattered glass will fall down
etc. Even though these assumptions are not present in our consciousness,
they guide our behaviour. This is evident in cases when an action fails to
produce the desired effect. The failure of the stone to break the window
will not be represented as caused by the window moving in response to
my throwing a stone, or the stone "deciding" not to hit the window. Rather
I will infer that I need to pick a larger stone, use more force, and/or move
closer in order to break the window, all inferences based on windows being classified as a specific type of physical object.
The purpose of this sketchy analysis of an ordinary action is to highlight some aspects of what happens when actions become ritualised, i.e.
what distinguish ritual from non-ritual actions. I have already mentioned
how the relation between the agent's intention and the actions performed
is radically transformed in ritual actions (Humphrey and Laidlaw 1994).
Here I will focus on how ritual actions can be characterised as based on a
similar radical transformation or even violation of intuitively held causal
expectations.
If we compare ritual actions to the non-ritual actions described above,
we find a significant transformation that can be illustrated by figure 2:
JESPER SØRENSEN
212
Conditional space
Action space
Effect space
Figure 2. Ritual event-frame
In the model, the non-ritual action of breaking a window with a stone is
changed into the ritual action of effecting a transition from a state of impurity to a state of purity by means of a performance of a specific ritual. The
most important change is that the intuitively held and domain-specific inferences relating the condition space to the action space, and the action
space to the effect space are severed. Participants have little or no domainspecific intuition about how the states of affairs preceding the ritual are
related to the actions performed in the ritual, or about how these ritual
actions produce the desired effects. Instead, two strategies for relating the
action to the condition and the effect are utilised. The first consists in the
numerous iconic and indexical relations that it is possible to establish between aspects of the condition, action and effects spaces (depicted by the
dotted line connecting the spaces). The obvious ecamples are usually described as "magical rituals", as when the Azande attempts to cure a man
suffering from leprosy by means of a ritual involving a creeper shedding
its "skin" as a part of its natural process of growth (Evans-Pritchard 1937).
As the plant survives and prospers through this loss of outer extremities,
so the man suffering from similar loss of outer extremities should survive
and prosper. In this case, the iconic relation between the perceptual characteristics of the man and the creeper is highlighted, and in fact motivates
the ritual procedure. However, rituals not usually described as magical
also contain this feature, as when blessing involves physical contact or the
ingestion of specific objects. In these cases, the action can be interpreted as
THE QUESTION OF RITUAL
213
a metonymic relation between the status of the person performing the blessing, the actions performed, possible objects utilised and its purported effect (see Sørensen 2000 for an analysis of magical rituals).
Thus we find both the so-called laws of sympathetic magic described
by James Frazer — not restricted to magic but found in some form in most
rituals. The violations of domain-specific expectations and inferences
prompt a cognitive search for other available clues to the meaning of the
action, and relations of similarity and contact are easily accessible. In the
words of the German ethologist Hans Kummer, the strong causal inferences given by domain-specific intuitions are replaced by weak causal relations based on perceptual similarity and spatio-temporal contiguity
(Kummer 1995). We shall return to ethology and its importance for the
study of ritual below.
Another hermeneutic strategy used to relate the condition, action and
effect spaces is that of symbolic interpretations. In figure 2 this is depicted
as the curved arrows substituting the severed domain-specific causal connections. In symbolic interpretations of ritual actions, the prognostic and
diagnostic processes are uphold by symbolically expressed links between
the spaces. The most obvious example is of course established dogmas
connecting certain states to certain rituals purported to effect specific
changes. At the other end of the spectrum, it also covers idiosyncratic interpretations made "on the fly" to explain the meaning of some ritual. In
between these extremes we have the more or less creative use of established cultural models used by participants to make sense of the actions
they engage in. The vast terminology employed by New Age groups is an
excellent example of such free-floating cultural models used to interpret
ritual actions.
To summarise: when people engage in ritual actions, such as the consummation of bread and wine insufficient to satisfy hunger, and attempts
to inflict pain on an opponent by means of manipulation of a doll, we intuitively know that something special is going on, i.e. that the action performed is a special type of action. Usually we just name such actions "symbolic" in order to point to their special features. I will argue that this definition is premature. The primary mistake lies in the confusion of the sign
itself, in this case the ritual action, with the way the actions is interpreted.
The ritual elements are not in themselves symbols. They can be interpreted
symbolically, but this is not the only way ritual actions are interpreted or
understood. Iconic and indexical connections play a significant role in most
ritual actions, facilitating a more direct and less contextually informed interpretation based on the recognition of relations of similarity and contact.
Besides these two hermeneutic strategies, there is a third strategy, in which
the ritual action is not interpreted at all, but only performed because "our
ancestors did so". However, even in such cases, the ritual is performed on
214
JESPER SØRENSEN
certain specified occasions, and the failure to perform the ritual often effects representations of the dire results of not performing the action.
These hypotheses concerning the special properties of ritual action (as
an ideal type) favour a procedural approach to the ongoing construction
of ritual meaning and purpose by participants. Whereas the purpose, intention and thereby meaning of ordinary actions are processed by a combination of domain-specific and cultural knowledge, ritual actions radically downplay intuitive, domain-specific processing. This leads to the
application of three alternative hermeneutic strategies:
(a) Use of perceptual characteristics of the actions, notably relations of similarity (icons) and contagion (indexes).
(b) Symbolic interpretations, from idiosyncratic to culturally reified models.
(c) Contextual interpretations based on the broader context in which the
ritual action takes part.
A result of this analysis is that instead of arguing whether rituals are in
themselves symbolic and expressive or rational and instrumental, we can
focus on the aspects of ritual action that differ from ordinary actions and
outline possible cognitive processes carried out by ritual participants. Even
though some rituals might actively exploit one or other of the hermeneutic
strategies available to participants, all rituals can be interpreted by all three
strategies, and I believe that investigations into the history of ritual practices will expose a constant flux between the different hermeneutic strategies.
In the following, I will restrict myself to a discussion of the first two
strategies and their internal relation.
From iconic and indexical to symbolic interpretations
Above I described two strategies, that of interpretation by means of iconic
and indexical relations and that of symbolic interpretation, as two equally
applicable strategies utilised by participants to make sense of ritual actions. However, this is not an accurate description as the two modes of
interpretation are unequal in a number of ways. In short, one can say that
iconic and indexical relations are more easily and more automatically processed; they utilise very fundamental cognitive processes used in basic level
categorisation; and they do not require a significant amount of background
knowledge. In contrast, symbolic interpretations are slower and less automatic; they use higher level and more complex cognitive processing; and
they require a significant amount of background knowledge, as the sym-
THE QUESTION OF RITUAL
215
bolic interpretation relates elements of the ritual to symbolic structures
found outside the ritual, for instance in myths or dogmatic systems. In
order to make this difference more explicit, let me use an example from a
classic study in ethology, Konrad Lorenz's Das sogenannte Böse (1963). Lorenz
argues that ritualisation in animals can be defined as the redirection of
instinctual actions from their former function to that of communication.
Among greylag geese, a mating pair strengthen their mutual bonds by
performing in unison the same sequence of actions normally used to fight
an enemy, but without the presence of any enemies. According to Lorenz,
the aggressive action is redirected so as to communicate reciprocal bonding between the two geese and it is thereby transformed from a direct instrumental function of fighting off an adversary to a communicative function of confirming the bond between the mates. Now, the concept of communication can be misleading as it implies a conscious attempt to transfer
information. In this case, it is the direct and most likely unconscious exchange of signals in contrast to the intentional exchange of symbolic sign
prototypical of human communication. The example illustrates how ordinary actions, when deprived of their ordinary function (in this case aggression), will provoke a search for other perceptible features that can give
functional purpose to the action. It is the iconic structure of the actions
performed in unison by the two geese that becomes highlighted when the
enemy is absent. By performing the actions in unison, the bonding is directly acted out rather than symbolically expressed. The strong causality
involved in the aggressive behaviour addressed against an enemy is transformed into the weak causality involved in the perceptual similarity of the
actions and spatial contiguity of the bonding pair.
This is not intended as a mere analogy from the animal kingdom. I believe there is a strong affinity and possibly direct evolutionary connection
between the ways in which animals and human beings process and construct iconic and indexical relations, and how these become highlighted in
ritualised behaviour. I also believe that there is a direct relation between
the ways animals and humans ritualise otherwise functional behaviour,
and that thorough studies of human rituals will expose a ground of "deep
meaning" based on iconic and indexical features (cf. Eibl-Eibesfeldt 1989;
Burkert 1979, 1996). This evolutionary origin might explain why iconic and
indexical interpretations are more easily and more rapidly processed than
symbolic interpretation. They have been with us for a long while and are
therefore deeply embedded in our cognitive architecture. I shall return to
this question below.
But there is one aspect that distinguishes the way humans use and interpret rituals from that of other animals: the tendency to apply more or
less consistent symbolic interpretations to the actions. Even though such
symbolic interpretation can be understood as a "side-effect" not intrinsic
216
JESPER SØRENSEN
to ritual itself (Staal 1979), this side-effect has had crucial importance for
the survival and grounding of ritual actions in the larger social and symbolic fabric of society. When actions are ritualised, they not only facilitate
the relative highlighting of iconic and indexical interpretations but also
the application of new or already existing symbolic interpretations. Such
symbolic interpretations are loosely constrained but not determined by
the iconic and indexical features just described. This implies that it is possible to extract a basic meaning structure from the actions performed that
will constrain, but not determine, subsequent symbolic interpretations.
When the bread of the Eucharist is consumed, it is difficult to ignore that
something is taken into the body and that it is a transformation by "eating", even though this need not be highlighted in the symbolic interpretation chosen, or can be given very divergent symbolic interpretations. The
rapid change of interpretations of the same ritual found in the history of
most religions testifies to this non-determinacy. This does not imply that
there are no constraints imposed on new interpretations by the actions
performed, but as the constraints are relatively loose, extrinsic factors will
enter into the development of symbolic interpretation, e.g. dogmas and
narratives.
This flux, and even struggle, characterising interpretations of ritual actions begs the question of why rituals have such a dominant place in most
religious traditions. Why perform rituals if they notoriously provoke conflicting interpretations? This is of course a complex question, to which several answers have been provided during the last 150 years. To make a crude
summary, two basic positions crystallise within the functionalist approach
to ritual and religion. The first conceives rituals as a conserving force in
society and in the history of religion. Rituals are relatively stable through
history and their primary function is to address and solve social crisis,
while most other aspects of religions can change. The second position conceives ritual as an innovative force, as a means of transcending social structure, as when the creative or subjunctive mood replaces the indicative, to
paraphrase Victor Turner (1990). Both positions have important insights.
Rituals constitute a conserving force by their relative stability through both
time and place. Rituals should be unchanging, as they are stipulated acts
whose efficacy is ensured by their origin. Thus the cohesive social force of
rituals should not be underestimated, not because all participants share
the same interpretation — they don't — but because they perform the same
public actions and thus iconically confirm social adhesion. Pushed to the
extreme, the question can be posed whether we are really dealing with
ritual rather than religious traditions. On the other hand, rituals function
as an innovative force exactly because their possible meanings are underdetermined by the actions performed. Rituals provoke the construction of
new interpretations. In fact, such new interpretations are the inevitable
THE QUESTION OF RITUAL
217
result of the relative stability of rituals compared to the relative flux of
social structure and history. Thus rituals are both conserving and innovating and a substantial number of schisms found in the history of religion
boil down to questions of the right performance of rituals and their correct
interpretation. As rituals provoke a search for meaning or purpose, religious traditions and cultures can either actively discourage explicit interpretations of ritual actions, as we see among the Baktaman of Papua New
Guinea (Barth 1975), or seek to control interpretations by restricting them
to specific authorities, as seen in most literate traditions. In both cases the
potential disruptive force of mutually opposing interpretations can be held
at bay, even if the innovative, creative and possibly disruptive forces of
new interpretations cannot be ruled out.
From action to meaning: How ritual action furthers
symbolic interpretations
The relation between ritual action and more or less authoritative interpretations is further complicated by the fact that rituals tend to downplay and
de-emphasise the symbolic elements they contain. Language is the most
central ritual element connected directly to a conventional symbolic system with a relatively fixed interpretation. However, many rituals tend to
downplay the symbolic reference of the words used in favour of the iconic
elements of prosody, and the indexical elements of enunciation. How words
are pronounced and who pronounces them have relatively more importance in rituals than in everyday language, at the same time as archaic and
even non-sense words flourish. In ritual, language itself is ritualised as its
direct reference is loosened. Ritual language is separated from the web of
symbolic reference that constitutes everyday language and, in a manner
similar to poetry, this enables novel interpretation.
The question naturally arises of why ritual has this function of separating elements from their ordinary symbolic reference? Why is ritualisation
an excellent method to achieve new meaning? I believe the answer can be
found in the role of ritualisation in human evolution. Rituals not only lead
to a constant reinterpretation of the basic semiotic and symbolic inventory
of religious traditions, but can, on a more fundamental level, be understood as the most important bridgehead leading from iconic and indexical
to symbolic representations. In The Symbolic Species, physical anthropologist Terrence Deacon argues that the origin of language is grounded in a
unique human ability to produce symbolic representations, and that this
ability developed in a co-evolutionary process involving the convergent
development of internal brain structure and external language structure
(Deacon 1997). In his argument, ritual plays a significant role as it contains
218
JESPER SØRENSEN
a feature necessary for the first construction of symbolic reference: the repetition of certain actions not directly related to a functional aim, in this
case, sounds. By containing constant repetition, ritualisation enables symbolic reference to emerge as concepts acquire meaning through stable relations to other concepts. (Deacon 1997: 402-10.) Thus language is a kind of
transformed ritual, in which interpretations are relatively locked by the
cross-referential structure of the system.
However, language did not eradicate ritual. Ritual persists despite the
existence of much more efficient means of communication. Ritual is in itself not an expressive and symbolic medium, but rather a type of action
that constitutes one of the necessary conditions for the development of
symbolic reference. Of importance in this context is that its role as an evolutionary bridgehead between iconic and indexical reference on the one
hand and symbolic reference on the other, can be reversed. By means of
ritualisation, conventional symbolic reference is dissolved into the constitutive parts of indexical and iconic relations, thereby facilitating the reinterpretation of otherwise fixed structures of meaning. Ritual is not only
one of the origins of symbolic structure, but also contains the possibility of
constantly reinventing, restructuring and reinterpreting the constituent
actions and structures of society by dissolving conventional symbolic reference into its iconic and indexical parts.
This is of course a very sketchy evolutionary outline, which needs to be
worked out in detail. I nevertheless believe it sheds light on some aspects
of ritual that can be of assistance in our current treatment of rituals found
in diverse religious and cultural traditions. I will end this article with a
short list of propositions summarising the preceding argument:
1) Ritual should not be seen as a kind of language, even if the two things
share certain characteristics. Language is dependent on a system of
conventional symbolic reference meant to communicate, whereas ritual
is a type of action meant to do something.
2) Ritual and ritualisation as a mode of behaviour is found among animals and humans alike and therefore forms a very basic, possibly innate, behavioural modality in humans. This explains why it is spontaneously produced and found in all human groups.
3) By violation of domain-specific expectations, rituals provoke the search
for other possible clues for the purpose or meaning of the ritual action.
4) Two strategies are involved: (a) basic perceptual features are utilised
to construct iconic and indexical relations; (b) symbolic interpretations
are formed, constrained but not determined by these basic structures.
In the case of religious rituals, symbolic interpretations tend to be drawn
from a culture's stock of religious representations and facilitate new
religious interpretations to emerge.
THE QUESTION OF RITUAL
219
5) Finally, rituals not only enable the construction of symbolic interpretation, but also facilitate the dissolution and deconstruction of already
established interpretations. Rituals can in this respect be understood
as generators of symbolic meaning, not because rituals have symbolic
meaning by themselves, but because they are actions that violate intuitive expectations and deconstruct established symbolic reference and
thereby give rise to alternative hermeneutic strategies used to construct
representations of meaning and function.
References
Barth, Frederik
1975 Ritual and Knowledge among the Baktaman of New Guinea. Oslo: Universitetsforlaget.
Boyer, Pascal
1994 The Naturalness of Religious Ideas: A Cognitive Theory of Religion. Berkeley:
University of California Press.
2001 Religion Explained: The Evolutionary Origins of Religious Thought. New
York: Basic Books.
Burkert, Walter
1979 Structure and History in Greek Mythology and Ritual. Berkeley: University
of California Press.
1996 The Creation of the Sacred: Tracks of Biology in Early Religions. Cambridge,
MA: Harvard University Press.
Deacon, Terrence
1997 The Symbolic Species: The co-evolution of language and the human brain.
London: Penguin Books.
Eibl-Eibesfeldt, Irenäus
1989 Human Ethology. New York: Aldine de Gruyter.
Evans-Pritchard, Edward E.
1937 Witchcraft, Oracles and Magic among the Azande. Oxford: Clarendon Press.
Fillmore, Charles
1982 Frame Semantics. In: The Linguistic Society of Korea (eds), Linguistics in
the Morning Calm; pp. 111-37. Seoul: Hanshin Publ. Co.
Hirschfeld, L. A., and S. Gelman (eds)
1994 Mapping the Mind: Domain Specificity in Cognition and Culture. Cambridge: Cambridge University Press.
Humphrey, Caroline, and James Laidlaw
1994 The Archetypal Actions of Ritual: A Theory of Ritual Illustrated by the Jain
Rite of Worship. Oxford: Clarendon Press.
Kummer, Hans
1995 Causal knowledge in animals. In: Dan Sperber, David Premack and Arm
James Premack (eds), Causal Cognition: A Multidisciplinary Debate. Oxford: Clarendon Press.
220
JESPER SØRENSEN
Lorenz, Konrad
1963 Das sogenannte Böse. Wien: Borotha Schoeler.
Sorensen, Jesper
2000 Essence, Schema and Ritual Action: Towards a Cognitive Theory of Magic.
Ph.D. dissertation. University of Aarhuus.
Sperber, Dan
1975 Rethinking Symbolism. Cambridge: Cambridge University Press.
Sperber Dan, David Premack, and Ann James Premack
1995 Causal Cognition: A Multidisciplinary Debate. Oxford: Clarendon Press.
Staal, Frits
1979 The Meaninglessness of Ritual. Numen 26: 2-22.
Turner, Victor
1990 Are There Universals of Ritual Performance? In: Richard Schechner and
Willa Appel (eds), By Means of Performance. Cambridge: Cambridge University Press.
Whitehouse, Harvey
2001 The Debated Mind: Evolutionary Psychology versus Ethography. Oxford:
Berg.