Acts that Work: A Cognitive Approach to Ritual Agency

METHOD &THEORY in the STUDY OF RELIGION Method and heory in the Study of Religion 19 (2007) 281-300 www.brill.nl/mtsr Acts hat Work: A Cognitive Approach to Ritual Agency1 Jesper Sørensen Institute of Philosophy, Education and the Study of Religions, University of Southern Denmark, Campusvej 55, 5230 Odense M, Denmark E-mail: [email protected] [email protected] Abstract Questions of agency are central for understanding ritual behavior in general and representations of ritual efficacy in particular. Religious traditions often stipulate who are entitled to perform particular rituals. Further, representations of unobservable superhuman agents are often explicitly described as the ‘real’ ritual agents. Recent investigations into the processes underlying action representations and social cognition can help explain how these representations arise. It is argued that paying close attention to details in the cognitive processing of ordinary actions can shed light on how ritual actions activate part of these systems while simultaneously leaving other aspects unaccounted for. his has particular effects that make culturally transmitted representations of superhuman agents highly relevant. Keywords ritual, cognition, agency, action-representation, intentionality, causality, religion, magic Introduction Agency in religion in general and ritual action in particular is a complex problem that can be addressed at several levels of description and analysis. Believers’ representations of agents and agency play a pivotal role in many religious phenomena. Who authorises religious statements as having a special status, who is entitled to perform and to participate in specific types of ritual action, and who controls religious institutions are among the fundamental questions 1 Working on this article has been made possible due to grant 273-05-0348 from the Danish Research Council for the Humanities. A warm thanks to Justin Barrett and Pierre Lienard for criticism of earlier drafts. © Koninklijke Brill NV, Leiden, 2007 DOI: 10.1163/157006807X240118 282 J. Sørensen / Method and heory in the Study of Religion 19 (2007) 281-300 that must be addressed, when attempting to explain these phenomena. Further, one of the defining characteristics of religion is the persistent reference to non-observable or absent agents believed to influence the world. Ghosts, spirits, ancestors, and gods populate the world’s religions and religious behaviour is often said to be motivated by a desire to interact with these agents. he role of both human and superhuman agents and believers’ representations of their ability to ‘act in the world’ (their agency) is thus of central importance in many scholarly accounts of religious phenomena. Few of these, however, have addressed how people actually construct representations of agency and what role mental constraints impose on religious thought and behaviour. his deficit can be remedied by paying close attention to recent studies of social cognition. An agent is characterised by his or her ability to influence the world by means of an action. hus, agency can be more precisely investigated by paying close attention to how our cognitive system relates representations of actions, causal expectations to actions, and intentions of the agent performing the action on the one hand, with that of both human and superhuman agency on the other. As those processes are automatic and not accessible to conscious examination, psychological investigations are of primary importance in unearthing the processes responsible for human representations of agency. his, in turn, can help us build a picture of how representations of superhuman agency arise and why they seem to be so pervasive in cultural transmission. Rituals are basically actions and are as such processed by approximately the same cognitive systems as the ones elicited for processing information about ordinary actions (cf. McCauley & Lawson 2002). However, rituals are actions tweaked in a particular way and this tweaking has certain consequences central for understanding the relation between ritualised behaviour and representations of superhuman agency. In short I will argue that ritual behaviour, in itself, makes representations of superhuman agents highly relevant as these solve two problems potentially arising in people’s comprehension of ritual performance: (1) who specifies the actions performed, i.e. why those actions; and (2) how are the actions related to their purported result, i.e. why is efficacy ascribed to the actions? In order to address these questions, I will first build a model of the processes involved in processing ordinary actions. Based on this, it will be possible to show more precisely how ritual actions differ from ordinary action and how this difference affects cognitive processing. J. Sørensen / Method and heory in the Study of Religion 19 (2007) 281-300 283 Agency in Non-Ritual Action Rituals, in the sense investigated here, are not just actions, but social actions, i.e. actions that are performed either with or in the presence of other people. Rituals are not only performed but also observed, and many rituals contain an elaborate sequencing of both positions (cf. Houseman 1993). Investigations into cognitive systems involved in both the execution and observance of ordinary actions are therefore of primary importance if we are to build a reliable model of the cognitive processes underlying ritual actions. Studies of how humans understand events in the world, in general, and the behaviour of other people in particular are by no means a new endeavour. Philosophers have struggled with the problem for more than two millennia and psychologists for almost two centuries. More recently, however, cognitive science has made significant progress in giving precise descriptions of the mental processes involved in human representations of other people’s actions. Psychological studies have investigated how ordinary adults and children parse actions into chunks defined by an intentional structure (e.g., Newtson 1998; Baldwin et al. 2001). Developmental studies have unravelled how knowledge of causation and intentionality develops (e.g., Leslie 1994, 1995; Wellman et al. 2001; Sommerville & Woodward 2005). Psychopathologists have studied how persons suffering from schizophrenia differ in their understanding of actions (e.g., Zalla et al. 2004) and how autism affects representations of other people’s beliefs and intentions (Baron-Cohen 1995). Computer science has attempted to build reliable simulations of human action representations (e.g., Gobet et al. 2001). And cognitive neuroscience has performed human-ape comparative studies of the neural activity involved in comprehending the actions of other beings (e.g., Gallese 2000a, 2000b, 2001). Taken together, and despite their differences, these studies are supplying significant insight into the mechanisms underly-ing human social cognition. hus, it is now possible to construct a model of the cognitive processes underlying representations of actions, causes, and intentions. he first part of the model must specify the perceptual cues that lead to representations of agency. Ever since Heider and Simmel’s (1944) seminal studies of false ascription of agency, there have been a significant number of studies investigating exactly what perceptual stimuli trigger ascription of agency. At a simple level, the movement of an object through the perceptual field is enough to trigger representations of agency (Mandler 1992; Leslie 1994, 1995; Spelke et al. 1995). However, in order to distinguish movement that indicates the involvement of agency from the movement of inanimate 284 J. Sørensen / Method and heory in the Study of Religion 19 (2007) 281-300 objects, other cues are necessary, most notably the ability of objects to change direction by means of an internal and renewable source of energy or force (Leslie 1995; Gelman et al. 1995). Objects that are able to initiate movements on their own are the prime candidates for agency as this is a necessary condition for being able to interact with the world. he ability to exert force points to the importance of ‘contingent response’ as a cue activating agency detection. Contingent response refers to the ability of agents to react to their surroundings in a non-random manner. Geometrical figures or faceless blobs are conceived as containing agency if they adapt their behaviour to changes in an artificial environment on, for instance, a computer screen (Heider & Simmel 1944; Spelke et al. 1995; Gelman et al. 1995; Johnson et al. 2001; Blakemore et al. 2003). Anyone slightly familiar with the world of video and computer games will recognise the ability of contingent movement to make otherwise non-interesting objects into exciting agents that (some) people will interact with for hours. In a natural environment, animals (including humans) seem to be the prototypical entities fitting these features ( Johnson et al. 2001). Perceptual cues are, however, only a basic level in representations of agency. Humans also employ at least two hierarchically related, non-perceptual systems when ascribing agency. At a fundamental level, actions are generally understood as goal-directed. Agents are not only reacting in a contingent way upon stimuli from the environment. hey instigate movement and react to the environment in order to achieve specific types of goals (e.g., moving to a specific position, avoiding collision, following another agent etc.). his teleological structure of actions entails that perceptible movements are chunked together to form fundamental action gestalts defined by a proximate goal or proximate intention (Searle 1983). Recent studies in primate cognition have exposed ‘mirror-neurons’ that are activated both when the animal performs an action, such as grasping, and when it observes another animal performing the same action (Gallese 2000a, 2000b, 2001). Neuroscientific investigations indicate that humans have a similar ‘mirror-system’ that is activated when performing or observing certain types of simple actions (Rizzolatti & Craighero 2004). Developmental studies show that representations of simple action gestalts arise earlier than more complex action representations and that they are intimately related to representations relating motor action and causal expectations in a teleological schema (Csibra et al. 2003; Sommerville & Woodward 2005). Lesion studies further indicate that patient with specific types of frontal lope damage experience huge problems when asked to parse larger series of behaviour into complex action representations while this does not impede the ability to chunk fine-grained motions into simple action gestalts (Zalla et al. 2003). J. Sørensen / Method and heory in the Study of Religion 19 (2007) 281-300 285 he chunking of perceptual information into basic action gestalt, such as ‘grasping’, ‘walking’ and ‘touching’, organised by a teleological schema, is thus necessary but not sufficient when attempting to explain the cognitive mechanisms underlying representations of agency. Even though the recognition of proximate intention gives some structure to the flow of information coming from the behaviour of fellow humans, the structure is very poor and insufficient to allow the perceiver to navigate in social interaction. he crucial further step is the ability to connect action gestalts into longer strings of intentional actions controlled by an agent with specific beliefs and desires—an ultimate goal or intention (Searle 1983). his ability has been named ‘theory of mind’ (Gopnik & Wellman 1994; Baron-Cohen 1995; Carruthers & Smith 1996; Frith & Frith 1999). Accordingly, the tendency to understand another agent’s actions as determined by a specific set of beliefs and desires are relatively independent of the ability to understand an action as teleological. hus, I can understand an act of walking as goal-directed (‘X wants to cross the street’) without being able to represent the ultimate intention specifying the act (X wants to go to the baker across the street in order to buy a cake). Representing an ultimate goal, however, enriches my understanding of the actions performed and allows me to have certain expectations about the likely actions of a person. Around age four, children know that other agents’ access to information about the world is limited, so they ascribe certain cognitive attitudes, e.g., beliefs, to agents and such cognitive attitudes are represented as informing an agent’s goal-directed behaviour (Wellman et al. 2001; Perner & Ruffman 2005). Further, people suffering from autism seem to lack the ability to form representations of other people’s mental states and therefore their ultimate intentions of performing specific actions (Baron-Cohen 1995). he ability to relate the simple actiongestalt produced by perceptual cues and teleological ascription to the cognitive attitudes of a perceived agent thus facilitates the emergence of more complex representations of intention and, in turn, the ability to envisage the future behaviour of other people. he two systems involved in ascribing goals and intentions to perceived actions will, all things being equal, work in unison in ordinary perception and representation of action. Still, the distinction between the two underlying cognitive systems is crucial. It enables us to distinguish representations of ‘proximate’ intention or teleology of an action (e.g., performing an action with the arm in order to put bread into the mouth), from representations of more complex or ‘ultimate’ intentions (eating in order to satisfy hunger). Humans spontaneously move from perception of ‘proximate’ intentions to ascription of ‘ultimate’ intention. he perception of goal-directed actions, in which an 286 J. Sørensen / Method and heory in the Study of Religion 19 (2007) 281-300 agent performs a series of concerted bodily actions in order to move bread from the plate into the mouth is, automatically, followed by an ascription of an ‘ultimate’ intention of eating in order to satisfy hunger. Further, the two systems are hierarchically related and have different sources. Whereas proximate intentions are constructed automatically based on perceptual features and cognitive simulation, knowledge of ultimate intentions has neither uniform origin nor applicability. Some actions, like eating, are universal human phenomena and as such are cross-culturally recognised. Others, such as moving the queen in the game of chess, involve knowledge of rules of the game delineating possible actions based on cultural convention, and this might lead to complex knowledge of chess that enables the recognition of a move as the reflection of another player’s ultimate intentions. In addition, there is no absolute limit to ascription of ultimate intentions. he player could purposefully be losing a game of chess in order to obtain other goals. An important consequence of this distinction is that agents can use an action defined by its proximate intention (e.g., moving an object) as an instrument in attaining very different ultimate intentions. his points to the final characteristic of ordinary actions: proximate and ultimate intentions are directly and hierarchically related. Ultimate intentions over-determine proximate intentions (I move the Queen in order to . . .), and proximate intentions specify the actions necessary for anything to happen (I move the Queen in order to . . .). As we shall see below, this is not the case in ritual actions. he human cognitive system used in understanding action and agency can thus be summarised as reacting to perceptible features such as movement and contingency, as simulating proximate intentions and agency in perceived actions, and as ascribing ultimate intentions to agents based on represented cognitive attitudes such as beliefs and desires. his action representation is modelled in figure 1. Traces in the Sand Before discussing how particular features of ritual actions tweak the cognitive systems processing perceived actions, I will shortly address the human proclivity to understand events as caused by the actions of non-observable agents. Representation of actions need not be forward directed, i.e. as predicting the relation between perceived actions, their agent and a predictable result. he system is flexible and the entire causal scenario can unfold based on the perception of the consequences of an action. his is found in a simple form in the J. Sørensen / Method and heory in the Study of Religion 19 (2007) 281-300 287 Ultimate intention Action sequence a1 a2 a3 a4 a5 Proximate intention Proximate intention Proximate intention Proximate intention Proximate intention Figure 1: his schema models an ordinary action-sequence: An ultimate intention determines a possible action sequence that in turn specify the causal and temporal relation between a number of individual actions each specified by their own proximate intention. ability to infer the activity of agents from perceptible features, such as the scent of a predator or traces in the sand. he ability to infer the activity of an agent is more developed when it comes to human agents, where perceptions of design, i.e. the deliberate spatial or functional relation between observable elements, is sufficient to trigger representations of agency. Psychological research has for a long time recognised what has more recently been dubbed “promiscuous teleology,” i.e. the tendency in humans to ascribe agency to natural and physical events and to explain natural objects as somehow ‘made’ by an agent for a specific purpose (Kelemen 2004). Pre-school children prefer explanations arguing that rocks are pointy in order to avoid having elephants sitting on them to more naturalistic explanations of erosion by rain and wind (Kelemen 1999). hus, the human cognitive system is not only able to ascribe agency to actions observed, but is also able to infer the actions of an absent agent based on observations of states in the world (e.g., the bread on the plate is no longer there). his tendency is especially strong in children—an observation that led Piaget to argue that children are natural animists (Piaget 1992). Recent developmental research, however, indicates that this feature remains a default strategy also evoked by adults when lacking adequate information (Kelemen 2004). his line of research is further backed by other studies pointing to the human proclivity to ascribe agency to all sorts of events and states (Guthrie 1996; Boyer 2001; Atran 2002). he branch scratching the window and the uncanny 288 J. Sørensen / Method and heory in the Study of Religion 19 (2007) 281-300 sound in the dark house are by default interpreted as the result of (often harmful) agents, and are only subsequently reclassified as natural events either by means of physical investigation or by a conscious symbolic effort. Anthropologist Stewart Guthrie (1996, 2002) has argued that the proclivity to see agents behind natural events and design behind natural objects can explain the origin of religious representations. Religion would be a by-product of an evolved ability to scan an ambiguous perceptual environment for agents. his makes good evolutionary sense, since false positives are relatively cheap (seeing a stone as a bear) in contrast to a false negative. Not perceiving a potentially dangerous agent (seeing a bear as a stone) is most likely to be fatal. Because of a selective pressure humans are oversensitive to cues of potential agents in the environment (Barrett 2000). Ancestors, spirits and gods become relevant concepts as they explain events (why did this happen) and objects (why are they designed this way) that human cognition has a proclivity to process as resulting from the activity intentional agents. herefore, the popularity of culturally transmitted concepts of religious agents might be conceived as the result of human tendency to infer the actions of intentional agent from scarce an ambivalent perceptual cues. As we shall see, something similar takes place when perceiving the performance of ritual actions. Agency in Ritual Actions Classical anthropological theories have addressed the roles of agency in ritual action from different angles. Traditionally, it has been argued that the performance of ritual action is motivated by beliefs in specific superhuman agents. Ritual actions are accordingly special because of the special character of the beliefs involved. Further, gods, spirits and ancestors are understood as fulfilling specific ritual roles because of the beliefs entertained. In its purest form this intellectualist approach sees ritual as a method of interacting with the world based on flawed explanatory theories of the world (e.g., Horton 1970). More symbolically inclined anthropologists have argued that this rational motivation is only a (minor) part of the story and that the real function of a ritual, and therefore its raison d’étre, must be found elsewhere, whether in social cohesion (e.g., Durkheim 1965), Freudian repression (e.g., Roheim 1962) or as the expression of basic social values (e.g., Beattie 1964). None of these approaches, however, have adequately addressed why representations of superhuman agents such as gods, spirits and ancestors, are so closely related to the performance of ritual actions. he relation between J. Sørensen / Method and heory in the Study of Religion 19 (2007) 281-300 289 alleged religious beliefs and the presence of such special agents in rituals are merely stated rather than explained. Further, few of these studies have paid any close attention to how peculiar features of ritual actions affect representations of agency (or the opposite). When we turn our attention to representations of agency, ritual actions initially seem to be rather normal. We find actions performed by an agent and sometimes involving different types of patients, instruments and objects. A Christian priest baptizes a child, a Jain sacrifices rice to the image of a Jina (Humphrey & Laidlaw 1994), or an Ndembu priest heals a childless couple (Turner 1969). All are actions or, rather, series of actions that can be analysed as observable agents performing specific actions with proximate and purported ultimate intentions on patients, possibly employing different types of instruments. In short, ritual actions exploit cognitive systems that are used to process ordinary types of actions (Lawson & McCauley 1990; McCauley & Lawson 2002). Further, ritual agents generally seem to be special by having privileged positions that enable them to perform specific ritual actions. Religious guilds usually impose a strong monopoly on ritual services (Boyer 2001), and even if this is not the case, individuals performing rituals often need to be initiated into a religious group, or at least perform certain actions in order to evoke the right conditions for performing the ritual. However, not all rituals are performed by agents with special qualities, and the efficacy of the ritual action can be ensured by other means, e.g., by the correct pronunciation of a spell, or by possessing the correct ritual object. his is particularly pronounced in the co-called ‘magical practices’, where agents perform ritual actions in order to achieve some specific goal. So, even though ritual agents are often endowed with special qualities and there seems to be a spontaneous tendency to ascribe ritual agents with such qualities, rituals can attain their efficacy through other means than a special agent. A theory of ritual agency must address how different ritual elements can become represented as wielders of the agency supplying the force necessary for a ritual to have any efficacy. An obvious solution to this problem lies in the fact that many rituals contain more or less direct reference to non-observable superhuman agents. he structural presence of gods, spirits, ancestors, and holy figures is exactly the defining characteristic of magical and religious rituals. Such agents are often referred to as ‘really’ performing the actions, the logical agent merely working on their behalf. However, superhuman agents are by no means always directly evoked, and they can be related to the ritual both directly as either agents or patients, or indirectly as lending credence to the efficacy of specific practices or objects. In some cases gods and spirits themselves act through intermediaries 290 J. Sørensen / Method and heory in the Study of Religion 19 (2007) 281-300 (e.g., baptizing) or are acted upon by humans (e.g., sacrifice). In other cases the superhuman agents have instituted ritual actions or objects thereby ensuring their efficacy. In Bringing Ritual to Mind, McCauley and Lawson treat these aspects of religious rituals in depth (McCauley & Lawson 2002; see also Lawson & McCauley 1990). McCauley and Lawson argue that a theory of ritual form can explain why rituals are represented as efficacious actions. Further, by focussing on the structural role of superhuman agents, a formal analysis can elicit which rituals are considered to be central in a given religious tradition, and whether any given ritual is repeatable or non-repeatable (among other explanatory facts). In short, the more immediately gods or spirits are represented to act in the ritual, the more important and central the ritual will be for the participant (the Principle of Superhuman Immediacy). Further, whereas ritual acts performed with superhuman agents as patients (e.g., sacrifice) can be frequently repeated in order to renew the effect, rituals having these gods or spirits as more or less direct agents (e.g., baptizing) are, in principle, non-repeatable (the Principle of Superhuman Agency). hus, cultural rituals can be divided into non-repeatable, ‘odd-numbered’ rituals in which superhuman agency is directly related to the agent slot in the ritual (e.g., a consecrated priest baptizing), and repeatable, ‘even-numbered’ rituals in which a superhuman agency is directly related to the patient or instrument slot in the ritual (e.g., sanctified bread in Holy Communion). According to McCauley and Lawson, participants’ intuitions about ritual performance are informed by knowledge about how superhuman agents are related to a particular ritual. his entails tacit representations of how previously performed rituals enable particular ritual elements to function as agents in a ritual (the priest can baptize because he is ritually initiated; the bread is active because it is consecrated by an initiated priest). he efficacy and ‘wellformed-ness’ of every ritual is thus judged by its relation to a vast network of embedded rituals that must be taken into account in order to give a full structural description of any given ritual (McCauley & Lawson 2002: 19). It is an empirical question, however, whether it is plausible that participants have access to vast implicit networks of structurally embedded rituals when judging the performance of a single ritual. Here I want to address another but related matter. McCauley and Lawson argues that religious rituals are defined by the role of Culturally Postulated Superhuman agents (e.g., gods) and as such, this is the most important aspect distinguishing them from ordinary actions. McCauley and Lawson therefore assert that “ . . . participants presume that CPS-agents act definitively in ritual, because they are already inclined to credit them the power to direct matters in life generally” (McCauley & Lawson J. Sørensen / Method and heory in the Study of Religion 19 (2007) 281-300 291 2002: 22, original emphasis). hus, beliefs in acting gods, spirits and ancestors are prior to their participation in ritual actions, whether as agents or patients. I claim that it might very well be the other way around: hat ritual actions are the primary context of acting superhuman agents and that this subsequently leads to representations of gods acting in life generally (Sørensen 2007). his can be expressed in the hypothesis that in religious traditions involving gods, spirits, and/or ancestors but having no rituals in which these are really believed to act, people will be very unlikely to ascribe real world events to the intervention of such agents. Contrary, in religious traditions involving both beliefs and strong convictions that gods, spirits or ancestors act in rituals, people will be much more susceptible to interpreting real world events as caused by such agents. Ritual actions entailing representations of the involvement of superhuman agents facilitates more general beliefs. his hypothesis is backed by psychological experiments where children (age 3-5) are told a story about a magical object capable of changing anything it touches into what it was two years before. his narrative has no effect in the judgement of the children about the possibility of such an object’s real existence, which is flatly denied. In contrast, an action sequence seemingly changing an old postal stamp into a new one does not make the children doubt the credibility of the action (they do not suspect cheating) but rather makes them believe that such magical objects are real and in fact has been responsible for the observed transformation (Subbotsky 1994). hus, actions prevail where concepts fail to substantiate beliefs in superhuman agency. Now, it might reasonably be argued that, in contrast to the psychological experiment, most ritual actions contain no direct perceptible evidence of the workings of a superhuman agent. Even though many rituals claim to effect direct, if nonperceptible changes, we therefore need to direct our attention to the distinguishing features of ritual actions in order to understand how they can more or less automatically produce representations of active superhuman agents. he first important feature concerns representations of intentionality in ritual actions. In he Archetypal Actions of Ritual (1994), Humphrey and Laidlaw claim that we find “consistent displacement of intentional meaning” (1994: 260) in ritualized actions. Rituals are characterised as actions that must be performed in the correct manner, no matter their instrumental effect or the often diverse intentions participants can entertain as their personal reason for performing a particular ritual. he identity of particular ritual actions does not depend on participants’ intentions but, rather, on the adherence to the stipulated actions defining a particular ritual sequence. he intentions of ritual participants are thus directed to performing the ‘whole’ ritual sequence. 292 J. Sørensen / Method and heory in the Study of Religion 19 (2007) 281-300 According to Humphrey and Laidlaw the very ontology of rituals is based on the stipulation of the right sequential performance, or as philosopher John Searle would have phrased it, in their constitutive rules (Searle 1969). A basic problem with this approach is that it only covers stipulated ritual actions (liturgical rituals) and not what Humphrey and Laidlaw call ‘performative’ rituals (Humphrey & Laidlaw 1994: 12-18). he problem can be solved if not only actions, but also agents and objects can be ritualized and thereby facilitate representations of ‘magical agency’ in the ritual action. In such cases, it is the special quality of the object or agent that ensures the establishment of a connection between action and result, and the actions surrounding such ritualised objects or agents are non-intentional (and not unintentional) due to their relation to these ritualized elements. he ‘strange’ behaviour of a shaman is not interpreted by participants as actions with ultimate intentions, but rather as an index of the underlying special quality of the agent and his or her ability to produce the desired end. Further, such cases usually also contain some of the qualities that distinguish rituals from other types of actions, such as condensation, redundancy and iteration, strange prosody, grammar and vocabulary, or unusual communicative behaviour (Sørensen 2005a). hus, rituals are not only actions in which superhuman agents are active, but are actions distinguished from ordinary actions by the ritualization of one or several elements, whether agents, action sequences, or objects. herefore, we can refine the description by relating the findings of Humphrey and Laidlaw to the defining characteristics of ordinary human action representations discussed above. First, at a basic level, the perceptual features of actions performed by participants will immediately entail the recognition of these as agents, no matter their role in the larger action sequence. Individuals are moving inside the perceptual field; they are reacting to each other in a contingent manner; and they are manipulating objects and instruments. he activation of the cognitive systems dealing with such teleological actions will lead to representations of proximate intentions: e.g., an agent hands out bread to participants and participants open their mouths in order to receive the bread in Communion. As such, rituals will be regarded as ordinary actions involving agents, actions, patients and objects. However, in ritual actions, representations of ultimate intentions, ordinarily related to the action sequence perceived, are disconnected from the automatically produced representations of proximate connections. he amount of bread received during Communion, the special quality of the bread, the fact that the bread is being fed to you even as an adult etc. immediately disconnects all ordinary representations of ultimate intentions (e.g., eating in order to satisfy hunger). he very format of the ritual underscores J. Sørensen / Method and heory in the Study of Religion 19 (2007) 281-300 293 Ultimate intention Ritual action sequence a1 a2 a3 a4 a5 Proximate intention Proximate intention Proximate intention Proximate intention Proximate intention Figure 2: his schema models a ritual action representation: he performance of a specific ritual action sequence is determined by an ultimate intention. his, however, does not specify the actions performed or the causal relation between them. Instead the sub-actions are frozen into a stipulated series of proximate intentions, whose performance only signals that the ritual action sequence has been performed. the fact that this is not to be understood as an ordinary act of eating. As such it facilitates interpretations of the agency responsible for this connection and the necessary force used to obtain the desired result. he transformation of representations between intention, action and causality is modelled in figure 2. he disconnection of proximate and ultimate intention has a curious result, namely that rituals attain an ‘event-like’ character. In contrast to action, events are changes in the world that are not specified by the ultimate intention of an agent. At the same time, however, the actions performed will necessarily produce representations of proximate intentions related to the specific actions performed, and as such are represented as actions involving agents, action, patients and objects. So, what are we to make of the fact that rituals are both like events and like actions? he solution to this problem lies in the human proclivity to ascribe ultimate intentions to actions whenever presented with proximate intentions. As participants in ritual actions cannot use their own ultimate intentions to explain the proximate intentions found in the actions performed, agents able to connect these two aspects of the actions must be found elsewhere. Superhuman agents are highly relevant in this respect as their projection into the ritual representation explain whose ultimate intentions specify the actions performed. Humans have a proclivity to infer the activity 294 J. Sørensen / Method and heory in the Study of Religion 19 (2007) 281-300 of non-observable or absent agents in understanding real-world events and objects. his tendency is even stronger in the relation to performance of ritual actions as these are obviously actions. Further, they make possible the stabilising of which agents are evoked due to the repeated performance of the same ritual in the same social context. here is an additional aspect of ritual action that makes the evocation of superhuman agency highly relevant. Rituals are not only distinct from ordinary actions because of the transformation of the link between proximate and ultimate intentions. he causal link between action and purported result is also radically altered. Even if some rituals seem to be almost totally devoid of representations of any immediate result (e.g., the Jain puja described by Humphrey and Laidlaw), many rituals are much more intimately connected to representations of an effect of the actions, and in the domain of ‘magical rituals’ this instrumentality seems to be a defining characteristic. Still, even if ritual actions are represented as having a particular effect, this representation does not arise in a manner similar to causal expectations guiding non-ritual actions. he causal unfolding of ordinary actions is guided by a significant number of intuitive expectations. If I throw a ball towards you, you can attempt to catch the ball by processing its likely trajectory, you’ll expect it to fall down eventually, but you will not expect it to alter its direction by its own will. By contrast, there are no intuitive expectations linking the action of making a baby’s head wet and being baptized, casting spells over an axe used to cut down the bush and getting an abundant harvest (Malinowski 1935), or burying magical medicine in the ground and getting revenge over a thief (Evans-Pritchard 1937). his lack of causal transparency, however, does not entail that the fundamental characteristic of ritual action is its meaninglessness (as proposed by Staal 1979). A meaningful relation can be established by means of two hermeneutical strategies—one based on available perceptual clues, another based on symbolic interpretations. Since an obvious causal outcome of the action performed is lacking, the participants can direct their attention to directly perceptible cues. he first hermeneutical strategy is thus characterised by the cognitive processes underlying so-called ‘magical actions,’ namely perceived relations of contagion and similarity (originally proposed by Frazer 1975). hese impose severe formal constraints on the actions performed: the water must touch the child’s head, the spell must be uttered in the vicinity of the axes, and the magical medicine must be directed towards its goal. hese perceptual features produce representations of what ethologist Hans Kummer describes as “weak causal connections” relating actions and possible result (Kummer 1995). J. Sørensen / Method and heory in the Study of Religion 19 (2007) 281-300 295 Ritual actions are seldom performed in a contextual or semantic void. Besides the weak causal connections, a second hermeneutic strategy is available to ritual participants: they can relate ritual actions to more or less established symbolic interpretations that connect actions to pre-specified conditions and to pre-specified results. Instead of focusing on perceptual features of the actions performed, participants can interpret the effect of actions by means of habitual and conventional symbolic relations. hese are to a large extent underdetermined by the actions performed, even though the actions do place certain constraints on meaningful symbolic interpretations. hus, the act of touching a child’s head with water is exactly an act of baptizing by means of the symbolic interpretation chosen to relate the actions performed to mythical narratives of Christ being baptized, to notions of essential change made through contact with sanctified water etc. All such interpretations are only loosely constrained by the basic schematic properties involved in the actions and, because of this under-determination, rituals are readily connected to larger conceptual and institutional networks, and thus to both myth and dogma. he two cognitive systems involved in representations of actions might help us explain why interpretations of ritual action apparently cluster around these two attractors. he perceptual interpretation is based on associative relations between elements in the perceived action and a locally defined pragmatic ‘purpose.’ Instead of attempting to integrate the whole ritual sequence into an encompassing intentional structure, the simple action gestalts found the ritual sequence can each be related directly to a specific goal. hus, ingesting sanctified bread will transfer sacred essence by means of contact, but the same bread can be used in other action gestalts such as preventing theft by burying it under the doorstep (homas 1991). In short, ‘magical interpretations’ of ritual actions entail that perceived schematic properties of individual action gestalts are directly related to locally defined pragmatic purposes. his leads to the formation of a context-near goal structure where the only thing that constrains the linking of a ritual act to a specific pragmatic purpose is its inherent schematic structure. By contrast, the symbolic interpretation is based more or less solely on a top-down structure in which the purported ‘meaning’ of the ritual can be constructed relatively independent of the actions taking place. Understanding baptism as a symbolic entrance to a new fellowship does not entail any great focus on the actions performed—these need only have a marked temporal structure dividing ‘before’ and ‘after.’ he focus on the ‘meaning’ of a ritual leads to the formation of a context-distant intentional structure in which the performance of ritual actions is related to symbolic interpretations instead of local pragmatic situations. All instances of correctly 296 J. Sørensen / Method and heory in the Study of Religion 19 (2007) 281-300 performed rituals are thus understood as instances of the same symbolic procedure that has either generalizable and rather abstract results (the performance of ritual X leads to salvation) or cease to be ascribed any efficacy what so ever. In both of these cases, however, ascription of agency plays a key role in representing rituals as actions with efficacy. In case of magical interpretations, agency is relegated to individual ritual elements that are really doing the work and that are represented as indispensable if the ritual is to have any efficacy (no transfer of sacred essence or protection against theft without sanctified bread). hus, ritual actions understood as instrumental for specific pragmatic endeavours will automatically generate representations of ‘magical agency’ responsible for the effect. hese might be more or less strongly identified with cultural representation of superhuman agents but the spread of magical objects and instruments, without such connections (e.g., magical medicine among the Azande, Evans-Pritchard 1937), indicates that this is a secondary process. Symbolic interpretations that downplay both perceptible elements and local pragmatic context makes representations of superhuman agents even more relevant as they can guarantee that otherwise opaque actions are actually connected to their purported results. Superhuman agents thus not only solve the problem of relating proximate and ultimate intentions in ritual actions—specifying why particular actions are performed. he performance of ritual actions also makes their evocation highly relevant as they effectively construct a ‘causal’ link between the actions and their purported effect. It is important to emphasise that the two hermeneutic strategies do not characterise specific religious traditions or even particular rituals. hey are strategies that can be employed by ritual participants/observers to extract a purpose or meaning of ritual actions. he difference between the two strategies, however, has potentially important ramifications for understanding the development of religious institutions. Representatives of religious institutions are generally sceptical or at least ambivalent towards magical interpretations as it is difficult to control locally entrenched interpretations and these might eventually produce competing symbolic interpretations. A method to control this is to enforce strict symbolic and context-distant interpretations of the ritual performed. he problem for religious institutions is that this is not only unlikely to be met with great success, as individuals spontaneously invest context-near magical efficacy to ritual elements. Rituals specified by symbolic interpretations alone also tend to become irrelevant to participants due to a J. Sørensen / Method and heory in the Study of Religion 19 (2007) 281-300 297 lack of connection to locally defined, pragmatic concerns (see Boyer 2001 and Sørensen 2005b, 2007). his, in turn, might prompt movements of ritual ‘revitalisation’ or a search for sources of context-near ritual efficacy outside the religious monopoly. Conclusion Attempting to explain the role of agency in human ritual actions makes it necessary to pay close attention to the cognitive systems responsible for understanding ordinary actions. Based on convergent evidence I claim that two, hierarchically related, systems can be discerned when constructing an operational model of human action representation. One chunks perceptual motions into simple action gestalts defined by a teleological structure or proximate intention. he second system unites different action gestalts into a sequence defined by an ultimate intention. Together the two systems create an action representation system in which actions with proximate intentions are instrumental to ultimate intentions. Ritual actions differ from this model of ordinary actions on two accounts. First, by means of ritualizing specific elements, proximate intentions related to specific actions inside a ritual are disconnected from ultimate intentions that might be entertained by participants. Even if participants do entertain ultimate intentions when performing the ritual, these cannot motivate the particular actions performed. Second, ritual actions are causally disconnected from their purported result. Strong causal relations based on intuitive expectations are replaced either by weak, perceptually motivated relations or by more or less culturally approved symbolic interpretations. Together, these two aspects of ritual actions make the evocation of superhuman agency highly relevant: (a) by relating the proximate and ultimate intentions of ritual actions (specifying why those actions are performed); and (b) by guaranteeing the relation between the actions performed and their purported result (specifying why the actions work). hus, theories that argue that beliefs in superhuman agents motivate the performance of religious rituals might be on the wrong track. An alternative explanation focuses on the role of ritualized behaviour in producing cues that makes the evocation of existing culturally transmitted superhuman agents highly relevant or, alternatively, lead to the production of new superhuman agents. 298 J. Sørensen / Method and heory in the Study of Religion 19 (2007) 281-300 References Atran, S. (2002). In Gods We Trust: he Evolutionary Landscape of Religion. Oxford: Oxford University Press. Baldwin, D. A., J. A. Baird, M. M. Saylor & M. A. Clark (2001). “Infants Parse Dynamic Action,” Child Development 72 (3): 708-717. Baron-Cohen, S. (1995). Mindblindness: An Essay on Autism and heory of Mind. Cambridge, MA: MIT Press. Barrett, J. (2000). “Exploring the Natural Foundations of Religion,” Trends in Cognitive Sciences 4: 29-34. Beattie, J. (1964). Other Cultures. London: Cohen & West. Blakemore, S.-J., P. Boyer, M. Pachot-Clouard, A. Meltzoff, C. Segebarth & J. Decety (2003). “he Detection of Contingency and Animacy from Simple Animations in the Human Brain,” Cerebral Cortex 13: 837-844. Boyer, P. (2001). Religion Explained: he Human Instincts that Fashion Gods, Spirits and Ancestors. London: Vintage. Carruthers, P. & P. Smith (eds). (1996). heories of heories of Mind. Cambridge: Cambridge University Press. Csibra, G., S. Biro, O. Koos & G. Gergely (2003). “One-year-old infants use teleological representations of actions productively,” Cognitive Science 27: 111-133. Durkheim, E. (1965 [1915]). he Elementary Forms of the Religious Life. New York: Free Press. Evans-Pritchard, E. E. (1937). Witchcraft, Oracles and Magic among the Azande. Oxford: Clarendon. Frazer, J. G. (1975). he Golden Bough: A Study in Magic and Religion: Part 1: he Magic Art and the Evolution of Kings, Volume One. London: MacMillan Press Ltd. Frith, C. D. & U. Frith (1999). “Interacting Minds—A Biological Basis,” Science 286: 1692-1695. Gallese, V. (2000a). “he Inner Sense of Action. Agency and Motor Representations,” Journal of Consciousness Studies 7 (10): 23-40. Gallese, V. (2000b). “he Acting Subject. Towards the Neural Basis of Causal Cognition,” In homas Metzinger (ed.), Neural Correlates of Consciousness. Cambridge, MA: MIT Press. Gallese, V. (2001). “he ‘Shared Manifold’ Hypothesis. From Mirror Neurons To Empathy,” Journal of Consciousness Studies 8 (5-7): 33-50. Gelman, R., F. Durgin & L. Kaufman (1995). “Distinguishing between animates and inanimates: not by motion alone,” In D. Sperber, D. Premack & A. J. Premack (eds.), Causal Cognition: A Multidisciplinary Debate. Oxford: Clarendon Press. Gobet, F., P. C. R. Lane, S. Broker, P. C.-H. Cheng, G. Jones, I. Oliver & J. M. Pine (2001). “Chunking mechanisms in human learning,” Trends in Cognitive Sciences 5 (6): 236-243. Gopnik, A. & H. M. Wellman (1994). “he heory heory,” In L. Hirschfeld & S. Gelman (eds.), Mapping the Mind: Domain Specificity in Cognition and Culture. Cambridge: Cambridge University Press. Guthrie, S. E. (1996). Faces in the Clouds: A New heory of Religion. Oxford: Oxford University Press. ——— (2002). “Animal Animism: Evolutionary Roots of Religious Cognition,” In I. Pyysiäinen & V. Anttonen (eds.), Current Approaches in the Cognitive Science of Religion. London: Continuum. Heider, F. & M. Simmel (1944). “An experimental study of apparent behaviour,” American Journal of Psychology 57: 243-259. Houseman, M. (1993). “he interactive basis of ritual effectiveness in male initiation rite,” In P. Boyer (ed.), Cognitive Aspects of Religious Symbolism. Cambridge: Cambridge University Press. J. Sørensen / Method and heory in the Study of Religion 19 (2007) 281-300 299 Horton, R. (1970). “African traditional thought and Western science,” In B. Wilson (ed.), Rationality. Oxford: Blackwell. Humphrey, C. & J. Laidlaw (1994). he Archetypal Actions of Ritual: A heory of Ritual Illustrated by the Jain Rite of Worship. Oxford: Clarendon Press. Jacob, P. & M. Jeannerod (2005). “he motor theory of social cognition: a critique,” Trends in Cognitive Sciences 9 (1): 21-25. Johnson, S., A. Both & K. O’Hearn. (2001). “Inferring goals of a non-human agent,” Cognitive Development 16: 637-656. Kelemen, D. (1999). “Why are rocks pointy? Children’s preference for teleological explanations of the natural world,” Developmental Psychology 35: 1440-53. ——— (2004). “Are Children ‘Intuitive heists’? Reasoning about Purpose and Design in Nature,” Psychological Science 15 (5): 295-301. Kummer, H. (1995). “Causal knowledge in animals,” In D. Sperber, D. Premack & A. J. Premack (eds.), Causal Cognition: A Multidisciplinary Debate. Oxford: Clarendon Press. Lawson, E. T. & R. N. McCauley (1990). Rethinking Religion: Connecting Cognition and Culture. Cambridge: Cambridge University Press. Leslie, A. M. (1995). “A heory of Agency,” In D. Sperber, D. Premack & A. J. Premack (eds.), Causal Cognition: A Multidisciplinary Debate. Oxford: Clarendon Press. Leslie, A. M. (1994). “ToMM, ToBy, and Agency: Core architecture and domain specificity,” In L. Hirschfeld & S. Gelman (eds.), Mapping the Mind: Domain Specificity in Cognition and Culture. Cambridge: Cambridge University Press Malinowski, B. (1935). Coral Gardens and heir Magic, Two Volumes. London: Gorge Allen & Unwin Ltd. Mandler, J (1992). “How to build a baby II: Conceptual primitives,” Psychological Review 99: 587-604. McCauley, R. N. & Lawson, E. T. (2002). Bringing Ritual to Mind: Psychological Foundations of Cultural Forms. Cambridge: Cambridge University Press. Newtson, D. (1998). “Dynamical Systems and the Structure of Behavior,” In K. M. Newell & C. M. Peter (eds.), Applications of Nonlinear dynamics to developmental process modelling. Mahvah, N. J.: Lawrence Erlbaum Associates. Perner, J. & T. Ruffman (2005). “Infants’ Insight into the Mind: How Deep?,” Science 308: 214-216. Piaget, J. (1992). Barnets psykiske udvikling. København: Hans Reitzels Forlag. Rizzolatti, G. & L. Craighero (2004). “he mirror-neuron system,” Annual Review of Neuroscience 27 (1): 169-92. Roheim, G. (1962). Magic and Schizophrenia. Bloomington: Indiana. Searle, J. R. (1969). Speech Acts: An Essay in the Philosophy of Language. Cambridge: Cambridge University Press. ——— (1983). Intentionality: An Essay in the Philosophy of Mind. Cambridge: Cambridge University Press. Sommerville, J. A. & A. L. Woodward (2005). “Pulling out the intentional structure of action: the relation between action processing and action production in children,” Cognition 95: 1-30. Spelke, E. S., A Phillips & A. L. Woodward (1995). “Infant’s knowledge of object motion and human action,” In D. Sperber, D. Premack & A. J. Premack (eds.), Causal Cognition: A Multidisciplinary Debate. Oxford: Clarendon Press. Staal, F. (1979). “he meaninglessness of ritual action,” Numen 26: 2-22. Subbotsky, E. V. (1994). “Early rationality and magical thinking in preschoolers: Space and time,” British Journal of Developmental Psychology 12: 97-108. 300 J. Sørensen / Method and heory in the Study of Religion 19 (2007) 281-300 Sørensen, J. (2000) Essence, Schema and Ritual Action: Towards a Cognitive heory of Religion, Ph.D. thesis, University of Aarhus. ——— (2005a). “he problem of magic—or how gibberish becomes efficacious action,” Recherches sémiotiques / Semiotic Inquiry 25 (1-2): 93-117. ——— (2005b). “Charisma, Tradition, and Ritual: A Cognitive Approach to Magical Agency,” In H. Whitehouse and R. N. McCauley (eds.), Mind and Religion: Psychological and Cognitive Foundations of Religiosity. Walnut Creek: AltaMira Press. ——— (2007). A Cognitive heory of Magic. Walnut Creek: AltaMira Press. homas, K. (1991). Religion and the Decline of Magic. London: Penguin Books. Turner, V. (1969). he Ritual Process: Structure and Anti-structure. Ithaca, Cornell University Press. Wellman, H. M., D. Cross & J. Watson (2001). “Meta-analysis of heory-of-Mind Development: he Truth about False Belief,” Child Development 72 (3): 655-84. Zalla, T., I. Verlut, N. Franck, D. Puzenat, A. Sirigu (2004). “Perception of dynamic action in patients with schizophrenia,” Psychiatry Reasearch 128: 39-51. Zalla, T., P. Pradat-Diehl & A. Sirigu (2003). “Perception of action boundaries in patients with frontal lobe damage,” Neuropsychologia 41: 1619-27.