Agency, art and altered consciousness: a motif in French (Quercy) Upper Paleolithic parietal art.

Lewis-Williams, J.D.

Is the meaning of prehistoric art beyond recovery - especially the meaning of early art in deep caves, a remote and strange location which itself suggests some out-of-the-ordinary purpose? David Lewis-Williams has been extending his explorations of meaning in later southern African rock-art to the famous enigma of the European Palaeolithic, here in the particulars of a single distinctive motif.

Eight years ago Johnson (1989: 190) made a sobering point: 'The individual has been triumphantly reinstated at the centre of the stage in theory, but quietly relegated to the wings, or written out of the script altogether, in practice.' Since then, the situation has not changed radically, at least in the publication of persuasive case studies (but see, for example, De Marrais et al. 1996; Joyce & Winter 1996; Mithen 1996).

In response to the lacuna that Johnson identifies, I lead my argument to a rock-art(1) motif that is known in Upper Palaeolithic art research as 'the wounded man' or 'the vanquished man'. After briefly situating the present study in the development of rock-art research, I explore the notion of agency in five cumulative sections. First, I discuss, in a preliminary way, the role that altered states of consciousness can play in the construction of selfhood. Secondly, I describe some clinically and ethnographically reported somatic hallucinations. I then draw on these reports in a specific case-study, the San. I show that certain somatic hallucinations were construed and manipulated in different ways by shamanic San rock painters. Next, drawing on principles and material adumbrated in my first, second and third sections I examine formal aspects of the 'wounded man' figures. Finally, I essay an explanation of these Upper Palaeolithic images that takes cognizance of human agency and the construction of selfhood.

The role of agency in rock-art research

The study of Franco-Cantabrian Upper Palaeolithic art has, at different times, placed markedly different emphases on the role of human individuality. Soon after the high antiquity of the art was established, the images were seen as the artistic products of artists: art pour l'art, as an explanation for the making of images, emphasized the activities and volitions of talented Upper Palaeolithic individuals who responded to innate compulsions to express themselves. This first phase of explanation did not last. Before long ethnographic analogies led to the proposal of hunting and reproductive magic as a more plausible explanation. In this understanding of the art, influentially advocated by the Abbe Henri Breuil, individuals were believed to have made the images in order to sustain the material basis of life.

Individualism was eclipsed with the advent of Annette Laming-Emperaire's (1962) and Andre Leroi-Gourhan's (1968) structuralism. These writers postulated a 'mythogram', or conceptual template, that persisted throughout the Upper Palaeolithic and that informed the subject matter and the placing of images within the caves. This mythogram derived from the supposed universal binary pattern of human thinking proposed by Claude Levi-Strauss. Levi-Strauss (e.g. 1963) argued that myths think themselves through the minds of people; for Laming-Emperaire and Leroi-Gourhan, it was the (comparably binary) mythogram that thought itself through the minds of people into the art. Although the association has not, to my knowledge, been explored, Laming-Emperaire's and Leroi-Gourhan's structuralism marched well, at least in one respect, with the adaptationist and processual archaeological approaches of the time. On the one hand, human volition and art were seen to be the products of a supra-human mythogram; on the other, human life was similarly reduced to the product of inexorable economic and ecological laws. In both cases, human agency was thought to be of little consequence.

At the beginning of the 1980s, a growing interest in what became known as cognitive archaeology (Renfrew 1982) proved congenial to renewed archaeological concern with agency (e.g. Hodder 1982; Shanks & Tilley 1987). In the new interpretations of the past, adaptationist and 'vulgar Marxist' theories began to give way - at least for some writers - to Pierre Bordieu's theory of practice and Anthony Giddens's related structuration theory (Bordieu 1977; Giddens 1984). Central to the work of both Giddens and Bordieu is the resolution of the dichotomy between the individual and 'society'. Bordieu and Giddens argue that individuals are not pawns moved across an ecological chequerboard by transcendent forces. Rather, at least some individuals know a great deal about the workings of society and are able to manipulate the rules and resources of society; but at the same time individuals are not completely free agents capable of doing exactly as they desire. Their actions are both enabled and constrained by the rules and resources that they manipulate. Despite these theoretical innovations in archaeology, the role of agency and the complex interplay between constraint and enablement in the making and purpose of Upper Palaeolithic art has not been much explored (but see Bender 1989; Hayden 1990; Lewis-Williams & Dowson 1993).

The trajectory of research on southern African San rock-art has in some ways paralleled that of West European Upper Palaeolithic research, but in other ways it has diverged markedly. During the first six decades of the 20th century various forms of art-for-art's sake and sympathetic magic were imported from western Europe, partly as a result of Breuil's visits to southern Africa. Although there was a tentative attempt in the early 1970s, the heady days of structuralism, to challenge the art pour l'art explanation by the detection of binary oppositions in San rock-art (Lewis-Williams 1972: 634), Laming-Emperaire's and Leroi-Gourhan's mythographic interpretation did not take hold in southern Africa.

The major divergence from the course of west European research, however, came in the late 1960s and early 1970s when workers began to heed 19th- and 20th-century San ethnography. The multiple and empirically verifiable fit between this ethnography and specific features of San rock-art opened up hitherto unrecognized avenues of interpretation that, because of an absence of ethnography directly and historically relevant to Upper Palaeolithic art, were not possible in western Europe. Soon a new explanation began to gain ground in southern Africa: it was argued that San rock-art was, in large measure though not exclusively, associated with shamanic rituals, symbols, notions of supernatural power and spiritual experiences (e.g. Lewis-Williams 1980; 1981; 1990a; Lewis-Williams & Dowson 1989; on the use of 'shaman' in the southern African context see Lewis-Williams 1992). Following up this interpretation, Dowson (1994) has explored the role of agency in the production of specific classes of shamanic images. Drawing on historical and ethnographic reports, as well as on the images themselves, he argues that the comparatively egalitarian status of San shamans was challenged by 'shamanic consortia', small groups of shamans who banded together to extend their influence, and also by 'pre-eminent shamans' who exploited their religious functions in order to acquire political power.

Recently, a broadly conceived shamanic explanation has been argued for West European parietal art on the basis of principles developed during the course of the southern African work: on both continents, so the argument runs, the making of rock-art was largely, but not exclusively, associated with institutionalized, ritualized altered states of consciousness, a central feature of shamanism.(2)

It is not my purpose here to rehearse the arguments in favour of this explanation or to explicate the principles on which those arguments are based; they have been set forth elsewhere.(3) Suffice it to say that, although this particular methodological thrust was initially developed in southern Africa, the argument does not consist in a simple analogy between the San and the people of the West European Upper Palaeolithic. Rather, the argument points to the similar (but not identical) ways in which human neurological universals of altered states of consciousness are manifested in different historical circumstances. Drawing on these universals, the shamanic interpretation provides a parsimonious explanation for a constellation of puzzling features of Upper Palaeolithic parietal art; these include: therianthropic figures (a common component of hallucinations); the placing of many images deep underground (sensory deprivation induces altered states); the varied and intimate relationships between images and subterranean surfaces (surfaces were a permeable 'membrane' between those who ventured into the caves and spirit animals in a subterranean realm); the various ways in which the walls of subterranean chambers and passages were touched and otherwise treated (in the establishment of diverse interactions with the spirit world); the ways in which parts of caves were differentially exploited (in the course of various shamanic rituals); and the co-occurrence of representational images and geometric 'signs' (two kinds of hallucination).

Altered states of consciousness and the construction of selfhood

Altered states of consciousness can, of course, be understood only in relation to some sort of 'normal' consciousness. It is this dynamic and complex relationship that opens up discussion of agency in the Upper Palaeolithic. 'Consciousness' is a notoriously difficult concept to define (Edelman 1994: 111ff); everyone thinks that they know what the word means - until they try to articulate its meaning. Michel Foucault (1965) recognized the social nature of definition when he argued that the 'normal' self, the 'normal' consciousness, is constituted by the various ways in which communities define and treat altered consciousness and madness. If there is at least one assumption about the Upper Palaeolithic that we can make confidently, it is that 'altered' consciousness and 'madness' were defined and accommodated differently from the ways in which they are defined and accommodated in late 20th-century Western society. The construction of Upper Palaeolithic 'selves', and hence the foundation of human agency, consequently proceeded along routes that were peculiar to that time and that no doubt changed during the course of the 20,000-year period.

All communities are obliged to formulate definitions of consciousness, various kinds of altered consciousness and madness, whether explicitly or implicitly, for these mental conditions are inescapably part of being human. At the same time, those definitions are always a site of contestation. The resources on which Upper Palaeolithic individuals drew in the construction and transformation of acceptable and powerful social identities therefore doubtless included definitions of various kinds of consciousness. As in many societies and subcultures today, altered, or ecstatic, states of consciousness were a manipulatable resource. Indeed, any account of the past that omits consideration of altered states of consciousness is likely to be incomplete (cf. Sherratt 1991: 52). The hostility of some researchers to discussion of altered states is obscurantist.

The construction of Upper Palaeolithic 'selves' and associated human agency can be examined within the framework of shamanism and the various ways in which altered states of consciousness can be defined because the human nervous system and some of its experiential products are common to all people: all human beings have the potential to experience similarly structured visual, somatic, aural, olfactory and gustatory hallucinations. The neuronal structure of the brain informs

1 the progression of mental imagery from 'light' to 'deep' altered states (though progression through each and every stage is not ineluctable);

2 the forms of pulsating, iridescent geometric imagery (known as phosphenes, form constants or entoptic phenomena) experienced principally in the first stage of altered consciousness and that persist into subsequent stages;

3 entry into a vortex, or tunnel; and

4 the nature of 'deep' state experiences, such as the blending of forms (e.g. human and animal), synaesthesia (confusion of the senses), a sense of dissociation from one's body, and, above all, the sense of being in another realm that has its own rules of order and causality and that is therefore very different from the material world (for a summary see Lewis-Williams & Dowson 1988).

These neurologically generated commonalities account, in large measure, for the striking and often surprising similarities of shamanism worldwide. Further shamanic commonalities are generated by the setting of shamanism in foraging societies that depend on the hunting of animals, the avoidance of dangerous predators, appropriate weather conditions for hunting and the growth of plants (in some regions other factors, such as seasonal salmon runs, may be included here), and various forms of social cooperation and social networks.

By contrast, much of the content of shamanic beliefs, visions and experiences is culture-specific and therefore varies from society to society. Dealing with universals and specifics, Well (1986: 29; see also Siegel 1985: 248) emphasizes the importance of 'set' (individual expectations and personality) and 'setting' (the physical and social environment) in determining how an altered state of consciousness is experienced. Shamanic experience is generated by an interplay between neurological universals and cultural and personal realisations of those givens.

Shamanism is a complex, not monolithic, category. Within any given instance of shamanism, shamans experience different kinds of hallucinations, perform various 'supernatural' tasks, participate in a range of rituals and engage in diverse power struggles. From this uncontroversial proposition there flows an important yet largely overlooked observation. For the human agent, both the universal and the cultural components of altered states of consciousness have the potential to become, in Giddens' terms, a resource capable of manipulation for personal and group ends. In some shamanic societies the making of art that is associated with and that in some ways defines altered states is one of the most important resources available to human agents (Lewis-Williams & Dowson 1989; Dowson 1989; Dowson 1994; Solomon 1994). As is now widely accepted, material culture, taken to include art, does not merely reflect culture and society; in the hands of human agents, it also constitutes culture and society (e.g. Miller 1987; Shanks & Tilley 1987:122-34; 1992; Conkey 1991). Individual image-makers - including rock-artists - actually did things and achieved ends with material culture.

Somatic hallucinations

Up to the present, the neuropsychological contribution to research on West European Upper Palaeolithic parietal art has concerned only visual hallucinations and their possible depiction in the art. By contrast, southern African research has also attended to somatic hallucinations (I take 'somatic' to include haptic, or tactile, and cenesthetic hallucinations), such as the sensations of attenuation and polymelia (having extra limbs or digits) (e.g. Lewis-Williams & Dowson 1989). Somatic hallucinations may occur in varying ways and to varying degrees, whether the subjects' altered state is induced by ingestion of psychotropic substances, sensory deprivation or other extraneous factors, or by pathological conditions such as temporal lobe epilepsy and schizophrenia (Siegel 1977; Brindley 1973: 593).

Somatic hallucinations occur in various parts of the body. For instance, schizophrenic patients report alarming stretching of the scalp; sometimes parts of the scalp seem to be pulled as much as 12 inches beyond the head (Pfeifer 1970: 57). Less dramatic somatic hallucinations are more common and occur in variously induced altered states of consciousness. They include tingling, prickling and burning sensations. Although such sensations may be experienced in various parts of the body, they seem to be concentrated over the scalp, neck, shoulders, sternum, the outsides of the arms, hands and feet, stomach and the front of the upper legs. But there is a problem in determining these locations because, in general, the sensations are ambiguously experienced, and subjects sometimes find it difficult to report exactly where they are located. A schizophrenic invoked this difficulty to explain differences between her reports and the reports of other patients (Pfeifer 1970: 58). This potential variety facilitates individual manipulation and the personalization of experiences.

Subjects, both Western and non-Western, seem to refer to somatic hallucinations less frequently than to visual hallucinations. In their very nature, visual hallucinations are more striking than hallucinations that are experienced in the other senses and that, in any event, often seem to be subsumed under and informed by visual hallucinations, as the examples I give below show. Consequently, neuropsychological research seems to have concentrated on the forms and progression of visual hallucinations at the expense of other kinds.

The ratio between visual hallucinations and hallucinations in other sensory modalities may, in some instances, be affected by the cause of the altered state of consciousness in which they occur. Lysergic acid diethylamide (LSD), for example, is said to induce predominantly visual hallucinations, whereas the pathological condition schizophrenia usually elicits a greater degree of auditory and somatic hallucinations (Winters 1975: 54). In schizophrenia, tactile hallucinations are rarely reported independently of auditory or visual hallucinations (Asaad 1980). Like LSD, psilocybin and peyote induce visual hallucinations minutes after ingestion, but auditory hallucinations are frequently experienced some two hours later; other kinds of hallucinations occur only sporadically (La Barre 1975: 12-13). The issue is further complicated by the observation that hallucinations experienced in one sensory modality can, by the process known as synaesthesia, be perceived in terms of another: for instance, a sensation felt on the skin (somatic) may be perceived as 'blue' (visual) (Kluver 1942: 199; Fischer 1975: 222; La Barre 1975: 10; Emboden 1979: 44; Cytowic 1994).

Attempting to clarify the relationship between visual hallucinations and somatic hallucinations, Siegel (1978: 313) argues that there is 'an orderly progression of hallucinations from simple snow lights through geometric forms to tactile sensations'. Some subjects, however, speak of tactile, or somatic, experiences in earlier stages than those that Siegel specifies. To clarify this apparent contradiction, it is useful to distinguish between somatic sensations and somatic hallucinations. In an early stage of altered consciousness, subjects may experience sensations of itching and tingling in various parts of the body. Because subjects do not 'illusion' these sensations they are known as pseudo-hallucinations. With pseudo-hallucinations, subjects know that the similes they employ to describe their sensations are no more than similes. In a deeper state of altered consciousness subjects lose the ability to discriminate between veridical and hallucinatory impressions. At this point, they no longer compare their physical sensations to something else; the sensations are what the hallucination dictates. Similes become realities.

This progression can be detected in reports by Western subjects. Harner, for instance, gives a detached, somewhat clinical account of a physical sensation that he experienced after one cup of the South American hallucinogen ayahuasca: 'A hyper-excitation is felt in the body, which produces a pleasant agitation in the epiderm' (Harner 1973b: 156). Siegel, who ingested peyote while on a visit to the Central American Huichol, gives a more dramatic, but still 'objective', account, even if he does employ a metaphor: 'Another "Ping!" My skin prickled with electricity' (Siegel 1992: 29). For Westerners, electricity seems to be a ready metaphor; elsewhere, Siegel (1978: 313) records a subject as saying that the sensations were like 'electricity running through the skin'. These subjects are perfectly aware that they are using similes to describe their sensations. In deeper altered states, such as chronic cocaine addiction, subjects are less able to distinguish between reality and somatic hallucinations. One patient tore off his skin and, looking in the bottom of the wound, believed he was pulling out microbes with his fingernails and with the point of a pin (Magnon & Saury, quoted by Siegel 1978: 309-10). The modern Western setting that includes fear of contracting disease as a result of infection by invisible germs here led to the patient's attribution of his sharp sensations to the presence of microbes. In this case, the 'pleasant agitation' that Harner describes became a terrifying hallucination. Other subjects report 'the hallucination of small animals moving in the skin' (Freud, quoted by Siegel 1978: 309). At the same time as these subjects hallucinate the tingling sensation as small animals or insects on or under the skin, a condition known as formication (Asaad 1980), they experience zoopsia (visual hallucinations of animals), thus combining two sensory modalities.

By contrast, in highly charged religious contexts Western subjects sometimes seem to be ambivalent about whether they are speaking in similes or not. Their statements are nevertheless clearly informed by their setting, that is, their supportive emotional circumstances and the sort of imagery in use in those circumstances. Goodman (1972: 58) reports that charismatic Christians, delighting in the ecstasy of heightened religious experience, tell of a 'gentle rain coming down on neck and shoulders and penetrating the chest'; the sensation extends down into the legs and the middle of the back. This sort of language reflects the notion of 'showers of blessing', the outpourings of divine grace for which they yearn. Although Goodman does not mention the point, references to divine blessing in the form of rain are common in the Bible, written as it was in an arid and semi-arid environment. For instance, Ezekiel (34: 26), with characteristic eloquence, represents the Lord as saying, 'And I will make them and the places round about my hill a blessing; and I will cause the shower to come down in his season; there shall be showers of blessing.' Clearly, this sort of imagery informs the somatic sensations triggered by charismatic Christians' altered states of consciousness. But whether they believe that the 'rain' is real or metaphorical is not entirely clear; a lack of language sophistication can lead to the suppression of 'as if' in descriptions of hallucinations (Sarbin 1967:371). Either way, these somatic experiences are a resource which individuals manipulate to achieve religious and social prominence.

The Western experiences that I have so far described range from variously successful attempts to observe one's own reactions objectively to the overwhelming experiences of pathological conditions. Non-Western shamans who experience the same universally neurologically structured somatic hallucinations do not attempt to assess their experiences objectively. Instead, they construe them in culturally contingent ways that are often different from Western reports of 'microbes', 'electricity' or 'divine rain'. For instance, in the 1920s Marius Barbeau recorded the experience of Isaac Tens, a Gitksan Native American shaman. Tens spoke of frequently falling spontaneously into a trance and of how, after he had undergone training, powerful chants forced themselves out of him, the phenomenon known as glossolalia. Speaking of one of these experiences, he said, 'The bee-hive's spirit stings my body ... In my vision, I went round in a strange land which cannot be described. There I saw huge bee-hives, out of which the bees darted and stung me all over my body' (Barbeau, in Halifax 1980: 189). The hallucination that developed out of Isaac Tens' somatic sensations in trance was thus similar to the hallucinations of Westerners who speak of insects crawling over them. The sharpness of Tens's sensations, however, together with his zoopsia produced a complex multi-sensory hallucination of bees stinging him. Tens was able to use this hallucination to entrench his shamanic status.

For the Jivaro of the Amazon Basin, sharp, pricking sensations lead to a different hallucination. Jivaro shamans believe that they can keep magical darts in their stomachs indefinitely and regurgitate them at will. These darts bear a supernatural potency that can also take the form of spirit helpers, or tsentsaks, that enable shamans to perform their tasks (Harner 1973a: 17). The Jivaro notion of mystical darts is developed in beliefs about pasuks, another kind of spirit helpers who aid malevolent bewitching-shamans by shooting their own objects into victims (Harner 1973a: 21-2). While in an altered state of consciousness, curing-shamans can shoot a tsentsak into an eye of a threatening pasuk, the eyes being the pasuks' only vulnerable area. Apart from tsentsaks, there are other supernatural darts that can kill or injure people; they are called anamuk. They are invisible to people who are not under the influence of the hallucinogen natema. In complex hallucinations, a shaman's darts sometimes take the form of animals that protrude from his skin. The darts also cover a trancing shaman's body as a protective shield (Harner 1973a: 24). The Miwok of the Central Sierra Nevada, California, held very similar beliefs. 'Poison doctors', tu yu ku, rubbed various kinds of poison on a pin-like stick or porcupine quill and then, by magical means, shot or threw the poisoned dart at a person who may have been as far as 50 miles away (Bates 1992: 101-2). Amongst the Jivaro and the Miwok, then, the sharpness of the prickling experienced universally during altered states of consciousness contributes to beliefs about and hallucinations of pointed missiles and, recalling the Western subjects' reports, creatures under and coming out of the skin (formication). Jivaro and Miwok shamans weave these experiences into accounts of their prowess and, like Isaac Tens, thus enhance their social positions.

As numerous writers have pointed out, the attainment of extra-human shamanic power and concomitant social status is frequently situated in the midst of an ordeal and an encounter with death. Often, it seems, the more horrific and painful the ordeal, the greater the prestige and power that accrue to the initiate. Death, suffering, excarnation, dismemberment, transformation and rebirth are indeed common elements of shamanic initiation (Eliade 1972). Piercing is frequently a part of that experiential sequence. A Siberian Tungus shaman, for instance, told how his shaman ancestors initiated him: 'They pierced him with arrows until he lost consciousness and fell to the ground; they cut off his flesh, tore out his bones and counted them; if one had been missing, he could not have become a shaman' (Eliade 1972: 43). In a comparable account, Eliade (1972: 43) notes that the souls of a Siberian Buryat initiate's ancestors 'surround him, torture him, strike him, cut his body with knives, and so on'. During this initiatory torture, the neophyte 'remains for seven days and nights as if dead' (Eliade 1972: 44). A Kazak Kirgiz initiate of southern Siberia spoke of five spirits in heaven who cut him with 40 knives and pricked him with 40 nails (Eliade 1972: 44).

Halifax (1982) sums up the suffering of the shaman in the title of her book, Shaman: the wounded healer. For the frontispiece she chose a photograph of an Inuit greystone, ivory and bone carving of a shaman harpooning not a seal but himself: he holds a harpoon that goes right through his body [ILLUSTRATION FOR FIGURE 1 OMITTED]. According to Halifax (1982: 5), this carving of a pierced figure 'captures the essence of the shaman's submission to a higher order of knowing'. The carving, I argue, is also an expression of the widely reported hallucination of piercing and so a visible, tangible guarantee of the 'higher order of knowing' that sets the shaman apart from others.

Somatic hallucinations and San rock-art

I now show that the somatic hallucinations that I have described constitute a resource that San shamans manipulated, and still manipulate, in a number of ways. I argue that certain San rock paintings represent some of those manipulations and point to a possible explanation for the West European 'wounded man' figures.

Today San shamans do not resort to hallucinogens (but see Schultes 1976; Dobkin de Rios 1986); they achieve an altered state of consciousness through intense rhythmic dancing, audiodriving, concentration and hyperventilation. They speak in a variety of ways about the kinds of somatic sensations and hallucinations that are thereby induced. For instance, Old K"au, a !Kung shaman (n/um k"a u, owner of supernatural potency), spoke of flies clustering all over his sides when he was in god's presence. There were, he said, also mumbas, pythons, bees and locusts: 'When you go there, they bite you. Yes, they bite you (gestures to his legs) ... Yes, they bite your legs and bite your body' (Biesele 1980: 57). I argue that Old K"au was construing his tingling, prickling somatic sensations and his visual hallucinations of extracorporeal travel and animals and insects. Only the greatest !Kung shamans actually reach god's house, and Old K"au was manipulating and developing his hallucinatory experiences to achieve that status.

Bees, one of the creatures that Old K"au mentioned, are closely associated with god. Both the 20th-century !Kung and the 19th-century/Xam are recorded as saying that god's wife is 'the mother of the bees' (Marshall 1962: 226, 245; Bleek 1923: 47). God himself likes bees and, if people burn bees when they are trying to smoke them out of a hive, he will send sickness to them; Marshall (1962: 245) heard of two men who were said to have died because they burned bees. God, moreover, can change himself into honey and place himself in a tree like a honeycomb. If he wants to kill someone, he attracts that person to the honey; if the person eats the honey, he or she dies (Marshall 1962: 245). Not surprisingly, 'Honey' is the name of a medicine, or trance dance, song (Marshall 1962: 249). Such songs are believed to be imbued with the supernatural potency that shamans are said to 'own' and that they activate in order to enter trance. The !Kung of the northern Kalahari like to dance when the bees are swarming, though not, of course, actually in a swarm (Wilmsen pers. comm.); they believe that bees are vehicles for potency (Katz 1982: 94).

An association between bees and trance dancing is evident in an Eastern Cape Province rock painting, part of which is shown in FIGURE 2. Here bees are depicted by means of small white crosses; in other paintings they are depicted with red bodies and tiny white wings [ILLUSTRATION FOR FIGURE 2 OMITTED]. Some of the bees in FIGURE 2 seem to be issuing from a roughly quadrilateral form that may represent a hive, or, if !Kung beliefs are taken into account, perhaps god himself. In addition, a swarm is depicted over the heads of the dancing figures to the right. Some of these dancers have red lines on their faces that probably represent the flow of nasal blood that San shamans sometimes experience in trance (Lewis-Williams 1981: 95-7, 98-9); at the far right, blood falls from a figure that bends forward and wears the eared cap that/Xam San informants associated with shamans who were believed to have the ability to control the movements of game (Bleek 1935: 46; 1936: 144). This dancer reaches out to what is probably a hallucinatory creature that holds dancing-sticks, bleeds from the nose, has red lines on its face, erect hair on its back and two trailing 'streamers' (for explanations of these features see Lewis-Williams 1981; Lewis-Williams & Dowson 1989). None of the bees is depicted actually on any of the human figures.

By contrast, the complex somatic and visual experience of which Old K"au spoke - bees stinging (biting) him - is, I argue, depicted in a small number of rock paintings. One such painting is in the KwaZulu-Natal Drakensberg. The large panel comprises many depictions, among which are two swarms of bees that, like those in FIGURE 2, are represented by small crosses, though in this case the crosses are painted in red. One of the swarms is associated with dancing figures and a hallucinatory 'elephant' with toes, a bifurcating trunk and possibly human hind legs (Lewis-Williams 1983: figure 74). 160 centimetres to the right there are seven elaborately painted human figures with white crosses on their red legs [ILLUSTRATION FOR FIGURE 3 OMITTED]; they seem to be running in the direction of the swarms. The crosses on these figures, like those of similar size that are painted in swarms, probably represent bees, but here the bees are crawling on and possibly stinging a number of shamans.

In the Eastern Cape Province, a painting shows a similar association of bees with the human body, but it places this association in a highly unusual congeries of motifs. At the left there is a standing figure with 18 red crosses painted on its legs; there is also one cross on its body [ILLUSTRATION FOR FIGURE 4 OMITTED]. the San ethnography and painted examples that I have cited, these crosses almost certainly again represent bees crawling on and possibly stinging a shaman's body. Nearby is a markedly attenuated figure that, in a typical trance dancing posture, bends forward and holds a stick. Attenuation is a somatic hallucination that is frequently depicted in San rock-art (Lewis-Williams & Dewson 1989: 76-7). This figure and the figure with the crosses are associated with other images that include five highly unusual depictions of human heads. Four of these heads have white faces with red lines at the nose; the lines on the head farthest to the right are the most numerous and most clearly suggest the nasal haemorrhage of San trance experience. The trance dance itself is explicitly suggested by the four white figures at the extreme right, one of which holds two dancing sticks and bends forward in the frequently painted and ethnographically reported dancing posture. Two antelope are also depicted. The recumbent one is probably a rhebuck; the standing or walking one is an eland, the creature most closely associated with supernatural potency (Lewis-Williams 1981). There is a great deal more of interest in this remarkable panel. Here I note only the numerous indicators of San shamanism and hence altered states of consciousness. The somatic component of those states is represented by the attenuated figure and by the crosses on the large standing figure.

These paintings of human figures with crosses on their legs and bodies strongly suggest that a culturally situated understanding of universally, neurologically structured somatic and visual hallucinations was expressed in San rock-art. But the differences between these paintings and between both of them and the dance shown in FIGURE 2 are highly significant for an understanding of the ways in which somatic hallucinations can become a resource which shamans can manipulate. In FIGURE 2, the bees are associated with a large number of dancing figures, not with an individual or small group of individuals. Moreover, no one figure stands out in any striking way. The power of the bees seems to be open to all the dancers. This group is probably what Dowson (1994) calls a 'communal group'. By contrast, the running figures in FIGURE 3 suggest that a small group of shamans chose to identify itself with a particular kind of shamanic hallucination. I know of no similar paintings in this area, and the unusualness of this kind of painting suggests that some shamans sought a novel group identity. Dowson (1994) calls such cooperative groups 'shamanic consortia'. In yet greater contrast, the panel shown in FIGURE 4 carries the notion of individuality further, first by the depiction of individual heads and, secondly, by the placement of bees on the legs and body of a single, large figure. The singularity and prominence of this figure suggest that an individual shaman appropriated one component of somatic experience in order to establish his special status (cf. Dowson 1988; 1994).

I now turn to another way in which San shamans understood sharp, pricking somatic sensations of altered states and in which San artists exploited those understandings. A !Kung shaman, a man, said (Katz 1982: 46):

In !kia, around your neck and your belly you feel tiny needles and thorns which prick you. Then your front spine and your back spine are prickled by these thorns.

!Kung women who enter trance describe a similar experience: they say that their 'middle feels as if it is full of thorns' (Katz 1982: 165). The pricking of needles and thorns are, I argue, two San construals of the somatic sensation experienced universally in certain altered states of consciousness.

As the Jivaro beliefs about supernatural darts show, a hallucination cognate with pricking thorns readily suggests itself in the setting of hunting and gathering societies that hunt with bows and arrows, darts and spears. San shamans describe the shooting of what they call arrows-of-potency that are small and invisible to ordinary people. Some of these mystical arrows-of-potency are sent by god or the spirits of the dead; they cause sickness and death (Bleek 1935: 5, 7). If people allow the central fire at a trance dance to die down, they say, 'God's arrows will strike us and make our skin painful' (Katz 1982: 120). It is the shamans' task to remove these arrows during a medicine dance and to hurl them back at the spirits and nameless malevolent shamans who lurk beyond the fire light.

Other, benign, arrows-of-potency are shot by an experienced shaman into a novice's//gebesi ('stomach'; Katz 1982: 168). A !Kung shaman said, 'You fire them in and fire them in until these arrows of n/um, which are a lot like long thorns, are sticking out of your //gebesi.' The informant went on to give a vivid and highly significant description of how this appears: 'Your abdomen is like a pin cushion, with arrows sticking out in all directions' (Katz 1982: 214). Even if the peculiarly Western pincushion metaphor was contributed by Katz, the statement gives a vivid impression of a St Sebastian-like figure and recalls the other ethnographic reports that I have cited.

This visual construal of a physical sensation is probably represented in some San rock paintings. One such painting [ILLUSTRATION FOR FIGURE 5 OMITTED] shows an isolated recumbent person wearing the sort of eared cap that was associated with shamans who were said to have power over game (cf. FIGURE 2; Bleek 1935: 45; 1936: 144). The figure is impaled and surrounded by many short lines that are clearly not 'realistic'. The painting certainly recalls the remark about feeling like a pincushion. The short lines may represent arrows-of-sickness or, perhaps, mystical 'thorns', but, because the figure lacks a painted context, it is difficult to say whether the 'arrows', if so they be, are carrying sickness or beneficial potency. (For further examples of 'impaled' figures in southern African rock-art see Garlake 1987a: figure 67; 1987b; figure 6; 1995; figure 185; Bond 1948.)

Although it is tempting to interpret such paintings as historical narrative (perhaps a murder), the large number of arrows piercing the figure, the eared cap, and the nature of southern African rock-art in general suggest they have to do with spiritual experiences. I argue that those experiences were associated with the tingling and pricking sensations that were experienced in certain altered states of consciousness and that the San themselves describe.

In sum, I argue that San paintings of figures with bees on their bodies and those apparently pierced by arrows negotiated desired status by, first, their novelty and, secondly, their factuality. Their novelty was a challenge to the ordinary run of images; although they drew on widespread San beliefs about bees, arrows and supernatural potency, they constituted a contestation of the San iconographic canon. At the same time, the factuality of the images (Lewis-Williams & Dowson 1990; Lewis-Williams 1995) consisted in a reification, not just a 'picture' or 'representation', of an unusual ineffable experience, cultivated, very probably, for social and political ends. As powerful things-in-themselves (Lewis-Williams 1990a), these and many other southern African rock-art images were more than mere statements about personal experiences. As palpable reifications of those experiences they aspired to incontrovertibility. One cannot argue with such a reified spiritual experience without challenging the whole belief and ritual system and thereby undermining one's own status. We can therefore speak of the experiences of altered states of consciousness and certain items of material culture - here rockart images - as a resource. The images had the potential to be instruments of human agency, not merely reflecting beliefs and society in a vague and general way but being used by individuals and interest groups to act upon social and political structures. Ethnographic and historical studies have shown that shamans do indeed play an active role in social and political change (e.g. Guenter 1975; Hitchcock 1982; Aldenderfer 1993; Thomas & Humphreys 1996). Here I have identified one of the ways in which they can achieve their ends.

The 'wounded men' of Upper Palaeolithic art

The empirical neuropsychological, ethnographic and painted evidence that I have so far adduced lays a firm foundation for a consideration of the Upper Palaeolithic 'wounded man' figures [ILLUSTRATION FOR FIGURE 6 OMITTED].

Anthropomorphic images are far less numerous in West European Upper Palaeolithic parietal art than they are in San art. Leroi-Gourhan (1982: 50) estimated that there are about only 75 anthropomorphic figures in all Upper Palaeolithic parietal art, a number that would constitute a very small percentage indeed of the (unknown) total number of images. Recent discoveries have not materially affected that percentage. By contrast, quantitative surveys of rock paintings in defined geographical areas in southern Africa suggest that human figures constitute 50% to 75 % of all representational images in those areas (Maggs 1967; Vinnicombe 1976; Pager 1971; Lewis-Williams 1972; 1974). This striking difference may seem to count against any argument based on the universality of the human nervous system; I therefore point out that the apparent contradition derives from a misleading generalization.

Whilst it is true that there are many depictions of human beings in southern African rock painting sites (located largely, but not exclusively, in the more mountainous regions of the subcontinent), most rock engraving sites (located on the plains of the interior plateau) have hardly any. In one section of the Vaal and Gariep River basins Fock & Fock (1989: 143) found that only 5% of the representational engraved images were anthropomorphic. Yet it has been persuasively argued that both the engravings and the paintings were associated with shamanic practices (Deacon 1988; Dowson 1992; Lewis-Williams 1988; Lewis-Williams & Dowson 1989). A stipulated proportion of human to animal depictions is not, in fact, a crucial component of the hypothesis that an art is largely associated with ritualized altered states of consciousness. The proportion is historically contingent: for whatever reasons, some communities seem to be principally interested in the geometric mental imagery of the early stages of altered states of consciousness, while other groups are more concerned with the iconic imagery of deep trance - the animals, monsters and spirits of the 'other world'. Then again, some societies focus on animals as sources of power or as 'helpers', while others emphasize anthropomorphic depictions of people or spirits.

The Upper Palaeolithic human figures known as 'wounded men' occur at Cougnac and Pech Merle, two sites in the Quercy district of France. The images are of human figures with three or more lines radiating from their bodies (FIGURE 6). Leroi-Gourhan (1982: 54, chart xxviii) places these 'wounded' figures in a larger category that he calls 'vanquished man' and in which he includes depictions of men who appear to have been knocked down by a bear or bison. Clottes & Courtin (1996: 155-61) call the category 'L'Homme Tue', 'the killed man'. They too extend the category beyond the Quercy figures and include the recently discovered figure in Le Grotte Cosquer on the Mediterranean coast (1996: figure 158), a figure in Sous-Grand-Lac (1996: figure 166; Bahn & Vertut 1988: 152), and three figures in Le Grotte de Gabillou (1996: figures 167, 168). All these figures are certainly associated with lines, one in the case of the supine Cosquer figure, more in the other depictions, but the lines do not seem to emanate from the bodies of the figures as clearly as they do in the Quercy examples; nor are the lines in most cases as numerous. I therefore concentrate on the Quercy figures; they constitute a clear, unambiguous group.

The sex of the Quercy figures is difficult to establish unequivocally: with the possible exception of the image shown in FIGURE 6c, which may have a penis, primary sexual characteristics are not depicted. Yet, if these images are compared with those Upper Palaeolithic parietal images that are clearly female, such as the figures on the ceiling at Pech Merle, the 'reclining women' at La Magdeleine (Leroi-Gourhan 1968: figures 501, 502) or the profile figure at Les Combarelles (Leroi-Gourhan 1968: figure 514), their general form seems to be male rather than female. I therefore retain the current phrase 'wounded man' but add that the sex of the figures is, in any event, not crucial to my argument.

Meroc & Mazet (1977: 35-7, 70) describe the three examples at Cougnac. The first, painted in black, is part of a large painted panel. It appears to be running towards the right and has three lines emerging from its lower back and buttocks [TABULAR DATA FOR FIGURE 6a OMITTED]. The torso of the figure appears to lean forward slightly, and this posture gives, at any rate to modern viewers, a sense of movement, perhaps of fleeing. The head and upper parts of the body are not depicted. It is placed near to depictions of large-horned deer (megaceri) and ibexes (Leroi-Gourhan 1968: figure 383). It is, in fact, placed on the lower chest of a megaceros, the throat and chest-line of which follows a natural contour of the rock face. The second 'wounded man' figure is part of another panel. It has seven or eight lines emerging from various parts of its body and has short arms without hands; no feet are depicted [TABULAR DATA FOR FIGURE 6b OMITTED]. It bends towards the left and its head seems to be fitted into the head of a mammoth that, like the magaceros, is partially defined by the natural contours of the rock. The figure is surrounded by double applications of paint perhaps made with two fingers (Leroi-Gourhan 1968: figure 384). Below it, there is another but smaller mammoth that is partly superimposed on the larger one; to the right is an ibex. The group of images is painted near the entrance to a low side chamber that contains painted dots. A third figure, described by Meroc & Mazet (1977) as the oldest and as being painted in a dark redish colour, is said to have three lines, two in the breast and one in the back. This figure is, however, very difficult to make out; certainly, I cannot decipher it.

A further example is at Pech Merle, some 30 km away [ILLUSTRATION FOR FIGURE 6c OMITTED]. Painted on a sloping ceiling, it is more erect than the others, and, like one of the Cougnac figures, it has vestigial arms. It has nine lines protruding from it, but one of these may represent a penis. A significant feature of this figure is that one of the so-called brace signs is painted just above it so that the right-hand 'arm' of the brace touches the back of its head. Given the ample space available, a connection between the figure and the sign seems to be intended. At Cougnac there is a panel of such signs. There may be eight or more of them, but some are now very fragmentary (Lorblanchet 1984: [ILLUSTRATION FOR FIGURE 8 OMITTED]). This Cougnac panel is some 35 m from the 'wounded men'. Nevertheless, the brace signs, like the 'wounded men', clearly suggest some sort of link between the two sites, and the way that the sign at Pech Merle touches the back of the 'wounded man's' head in turn suggests a link between these human figures and brace signs.

The bending-forward posture of one of the Cougnac figures and the slightly less bent position of another require comment. Leroi-Gourhan (1982: 53) notes that 'an important number' of all human figures have the body bent forward 'to around 30-45[degrees]'. This sort of posture may be occasioned by one of the physiological effects of some altered states of consciousness. When a San informant was describing the pricking of 'needles and thorns', he went on to say, 'Your //gebesi tightens into a balled fist' (Katz 1982: 46). This painful experience causes San trance dancers to bend forward until their bodies are almost at right angles to their legs (Marshall 1969: 363-4). Southern African rock paintings often depict trancers in this position (e.g. Lewis-Williams 1981: figures 19, 20, 23, 28, 32; Lewis-Williams & Dowson 1989: figures 15, 16c, 17, 20, 28, 32a). Similarly, in the 'wounded man' figures, as well as in other Upper Palaeolithic human figures (e.g. the female figures from Lalinde (Marshack 1972: figures 181, 182) and the Les Trois Freres 'sorcerer' (Leroi-Gourhan 1968: figure 57)), this posture may represent a physical response to the prickings and contractions induced by some altered states of consciousness. This interpretation does not, of course, derive from a simple ethnographic parallel with the San (Lewis-Williams 1991a). Rather, it is based on what is a common physiological response that is governed, at least in part, by the human nervous system.

Indeed, as is often remarked, there is no directly relevant ethnography to guide the interpretation of Upper Palaeolithic art, as there is for San rock-art. Writing of all four examples, Meroc & Mazet (1977: 36) categorize the depictions with paintings of 'wounded animals' and argue for 'la magie de la destruction'. One context for this kind of magic and image may, they go on to argue, have been conflict over territory, the images having been made and ritually 'pierced' prior to a real fight between competing communities. Meroc & Mazet, however, doubt this explanation, arguing that, if it were correct, the makers of the images would have depicted the heads of the figures as faithfully as possible. Pointing to the non-human or animal-like faces of three of the figures and the absence of the fourth's head, they suggest that the putative magic was more probably related to the exorcism of malicious demons rather than to the defeat of adversaries. Similarly, Clottes & Courtin (1996: 160-61) argue that the imprecision and incomplete nature of the figures was deliberate and may have constituted a safeguard of some sort. In a comparable explanation, Graziosi (1960: 182) argues for 'magic death practices'. Meroc & Mazet, Graziosi and, to a lesser extent, Clottes & Courtin, thus argue that the images were made by individuals to secure economic or personal survival.

On the other hand, Breuil (1979: 272), in a literal 'reading', suggested that FIGURE 6c depicts a masked man pierced by 'many arrows, unless he is carrying them'. Similarly, Leroi-Gourhan (1982: 54) sees the Quercy figures, together with the other images in his 'vanquished man' group, as the most 'pictographic' (narrative) in Upper Palaeolithic art, but he allows that the different aggression themes supposedly expressed by the images in this category - killed by animals or by human-made artefacts - could have had different 'mythographic' (symbolic) messages. He cites Laming-Emperaire's observation that male human beings most often occur in what appear to be scenes of a tragic character (Leroi-Gourhan 1968: 130). This is, of course, a Western 'reading' of the images. In his descriptions of all the 'impaled' figures in his 'vanquished man' category, Leroi-Gourhan takes as unproblematic the interpretation that they are, literally, pierced by spears. Clottes & Courtin (1996: 160-61), too, argue that the most plausible explanation is that the lines represent spears of some sort. They conclude that the supine figure in La Grotte Cosquer that is traversed by a single horizontal line indisputably depicts a slain man (1996: figure 158). For Clottes & Courtin, this conclusion lends support to the view that the Quercy figures also represent people struck by projectiles of some sort.

Smith (1992) has formulated an alternative explanation. He argues that the multiple lines represent 'life forces' that link shamans to animals. 'Life forces', a generalization that Smith draws from his survey of shamanism worldwide, is an explanation that has some merit, but, by itself, it seems too vague a concept to provide a firm basis for an explanation of the 'wounded man' figures. The empirically established experiences generated by the human nervous system in altered states that I have described hold out the hope of more precision and may provide, though in a different way, support for Smith's suggestion that the lines do not represent spears.

An ethnographic example illustrates this point. When San trancers experience a tingling sensation in the top of the head, they ascribe it to the soul, or spirit, leaving on extracorporeal travel (they explain trance by spirit loss, not by spirit possession). Some San shaman-artists depicted this experience by drawing long lines emanating from the heads of trancing figures (Vinnicombe 1976: figures 109, 152, 247; Lewis-Williams 1981: figures 31, 38; Lewis-Williams & Dowson 1989: figures 22, 32a, 32b, 33b, 65b). Similarly, Upper Palaeolithic shamans may have construed the pricking sensations felt in various parts of the body not as being caused by sharp lances but by some sort of potency or spirit entering or leaving the body; they may then have depicted such a concept by the lines emanating from the Cougnac and Pech Merle figures. But, as Clottes & Courtin (1996: 160) point out in their argument against Smith's explanation, the 'wounded man' figures irresistibly call to (our Western) mind a person struck by projectiles, especially in the case of the apparently fleeing figure in Cougnac [ILLUSTRATION FOR FIGURE 6 OMITTED].

There is at any rate one point on which writers agree: the 'wounded man' figures are not easily interpreted. Clottes & Courtin (1996: 161) believe that, without some extraordinary discovery, it will be impossible to achieve certainty, and Leroi-Gourhan (1982: 54) concludes that the figures pose 'serious problems that we are not yet capable of resolving except through hypotheses which have little foundation.' I leave aside inappropriate notions of proof or certainty; as with most archaeological explanations we shall have to relinquish a desire for absolute proof (whatever that may consist in) and content ourselves with degrees of confidence as constituted by evidence, logic, internal consistency, analogies with strong relations of relevance, and so forth. I argue that a hypothesis derived from the ideas and data that I have developed in previous sections of this paper moves from a well-attested foundation towards a persuasive resolution of at least some of the problems posed by the 'wounded man' figures.

First, the 'pictographic', or 'realistic', nature of the images must be questioned. The partial representation of the human body in all the examples, the possibly zoomorphic aspect of the heads of two of the examples and, especially, the large number of 'lances' in the figures are features that suggest that the Quercy paintings are not literal depictions of real, violent events. At the same time, it should be noted that concepts of 'realism' and 'non-realism' are culturally situated and not universal. They may be handy labels if loosely applied, but they probably obscure other and more complex notions of what may constitute 'reality'. A rigid distinction along the lines of Western concepts of what is real and what is not seems unlikely to have obtained in the Upper Palaeolithic.

Secondly, as the clinical and ethnographic examples that I have given show, Upper Palaeolithic people's understanding of somatic sensations and hallucinations would have been as much controlled by their set and setting as are Westerners' or the San's. Yet, the fact that Upper Palaeolithic people were hunter-gatherers suggests that there may have been similarities between the ways in which shamans of that time and San, Gitksan, Jivaro and Miwok shamans interpret their somatic sensations and hallucinations.

Indeed, the two informing factors that I have emphasized - the universality of the human nervous system and the shamanic hunter-gatherer setting - suggest that the artists who painted the Cougnac and Pech Merle figures may have experienced the pricking sensations of trance and have hallucinated them as multiple stabbings with sharp pointed lances. So whilst the radiating lines may represent spears, they are not 'literal' spears, and the images do not depict violent incidents. Rather, they represent spiritual experiences. The possible ritual setting of those experiences is the next question to be addressed.

As I have shown, beliefs in missiles of some sort that carry potency and pierce a shaman are a widespread concomitant of one of the physical experiences of some trance states. Moreover, a context in which piercing is commonly reported is shamanic initiation. Shamans must suffer before they can heal, die before they can bring life to their people. The ethnographic examples of initiatory piercing that I cited and indeed the title of Halifax's survey of shamanism worldwide, Shamanism: the wounded healer (1982), inevitably recall the Upper Palneolithic images of what seem to be people pierced by spears. The 'wounded man' figures may, I argue, represent a form of shamanic suffering and initiation that was closely associated with somatic hallucinations.

This interpretation can be extended by a consideration of the locations sometimes chosen for shamanic initiations. In central Australia, an aspirant Aranda 'medicine man' goes to the mouth of a cave where he 'fails asleep'; spirits then throw invisible lances at him, piercing his neck and cutting off his head (Eliade 1972: 46). Similarly, a Smith Sound Inuit initiate must go to a cliff containing caves: 'If he is predestined to become a shaman, he will enter a cave ... As soon as he has entered the cave, it closes behind him and does not open again until some time later' (Eliade 1972: 51). A particularly vivid North American account of subterranean shamanic initiation combines visual and auditory hallucinations. A 50-year-old Paviotso man wishing to become a shaman entered a cave and prayed. He tried to sleep but was prevented by strange noises - the grunts and howls of bears, mountain lions and deer. When he finally did fall asleep, he saw a healing ceremony. Then the rock split open and a 'man appeared in the crack. He was tall and thin. He had the tail feather of an eagle in his hand.' He taught the initiate how to cure (Eliade 1972: 101). The wall of the cave was but a thin membrane between the initiate and the spirits, both animal and human, of the underworld (cf. Lewis-Williams 1995; Clottes & Lewis-Williams 1996). In citing these and numerous other ethnographic instances of shamanic initiation, Eliade (1972: 50-53) emphasizes the importance of caves and the fact that the 'death' of an initiate often involves a descent to the lowest realm of the shamanic cosmos.

These and other instances of piercing and entry into caves add a further dimension to the explanation that I propose: the 'wounded men' of Upper Palneolithic Quercy art may, together with other images, have been associated with chthonic shamanic initiation (cf. Pfeiffer 1982; Hayden 1990). The neuropsychological and ethnographic evidence that I have adduced strongly suggests that, in these subterranean images, we have an ancient and unusually explicit expression of a complex shamanic experience that is informed by altered states of consciousness: that experience comprised isolation and sensory deprivation by entrance into an underground realm, 'death' by a painful ordeal of hallucinatory multiple piercing, and emergence from those dark regions of an inspired, re-born shaman. Ritualized altered states of consciousness and these images were thus intimately associated with the construction of social personoe. We are now in a position to consider the role of human agency in the making of the 'wounded man' Upper Palneolithic images.

Art and agency

First, we must note that the group of figures that I have defined was made in a restricted geographical area and, probably, within a comparatively restricted period (Clottes & Courtin 1996: 159). Their existence therefore points to a historically situated event or complex of events.

Something of the nature of those events is suggested by the placing of the novel 'wounded man' images in the 'traditional' shamanic Upper Palneolithic contexts of subterranean chambers and painted panels comprising depictions of animals and 'signs'. These two contexts, topographic and iconographic, suggest that the 'wounded man' images were associated with a social and cognitive movement situated within and accepting (at least in part) of existing cosmology. As I have argued, all novelty is a form of contestation. But, like the making of the southern African paintings that I described, the making of the 'wounded man' figures was probably not, despite their originality, an attempt to challenge the entire cosmology. Rather, it was probably an attempt to manipulate the resources of existing cosmological, social, religious and iconographic frameworks so that the novel images and what they represented would be acceptable to at least some people. Cosmological, social, religious and iconographic frameworks are not immutable givens. They are reproduced through complex processes and forms of representations and are therefore always open to negotiation.

It is not clear if the contestation implied by the 'wounded man' figures came from emergent individuals or from interest groups in some ways similar to Dowson's (1994) 'shamanic consortia'. The notion of an individual is complex and always historically and culturally situated (for a range of ideas on this point see Carrithers et al. 1985). Without exploring the philosophical implications of the notion of personhood, we can draw a distinction between large, complex paintings that must have been communally produced (such as the well-known ones in the Rotunda and elsewhere in Lascaux) and those simpler images that were probably made by individuals (such as the images in the Cabinet de Felines, Lascaux: Leroi-Gourhan & Allain 1979: figures 314-50). Communally produced art in large chambers was probably associated with group rituals, while smaller and more remote images may have been made by individual initiates or vision questers (cf. Bender 1989; Hayden 1987; Lewis-Williams & Dowson 1993). The making of personal images was probably associated with rituals that were different from those that attended the making of communal images (Lewis-Williams 1994; 1995). Although all Upper Palaeolithic individuals probably had to have some following in order to get to the point of making images in the subterranean chambers, it seems likely that self-differentiating individuals played a key role in the making of the novel 'wounded man' images. Indeed, entry into the caves, as well as the preparation of paint and the making of images, was probably part of a series of interrelated, socially differentiating ritualized contexts (Lewis-Williams 1995).

Individuality that was opposed to a dominant social and religious order is suggested by the fact that the images are not of 'otherness' - animals, animals as sources of power, visionary animal-helpers and so forth. On the contrary, I argue that the 'wounded man' images are probably highly manipulated representations of the subjects themselves. In these self-images, the subject is foregrounded. The distinctive 'wounded man' images were therefore probably an answer or challenge to the general paucity of human figures in Upper Palaeolithic parietal art. What that paucity meant is a difficult question. The hegemony of animals in the (representational) symbolling systems of the time may have tended to place the source of supernatural and, in all probability, political power outside of the human sphere. In these depictions of (albeit transformed) human beings, we may have evidence for a particular, historically situated challenge to the locus and control of supernatural and political power: some people were wishing to become a visible part of the interrelated symbolling and power systems but in a new way, and this desire meant that new representations had to be fashioned out of the existing and potentially divisive religious experiences.

Conclusion

In mounting challenges to the status quo individuals adopt roles that are different from their familial and economic roles. In seeking initiation as a shaman, an initiate hopes to don a new persona that has specific social advantages. Drawing on the dramaturgical origins of 'persona', Hollis (1985: 222) discusses the difficulty of separating out the 'fusion of man and mask': self and role become intertwined. I argue that this fusion is particularly closely achieved when the act of adopting a new persona involves altered states of consciousness, because those states are situated within an individual and inevitably transform an individual's conception of him- or herself: role becomes reality in the same way that simile can become metaphor, and metaphor can become 'reality'.

Fusion and transformation are unusually clear in the 'wounded man' figures. In one powerful image they represent person and role, a new kind of shamanic role made possible by a conscious, deliberate decision to highlight a distinctive and, at that time, unrepresented component of altered states - piercing somatic hallucinations. The individuals who made the images were, I argue, manipulating a resource (altered states of consciousness) to advance their own religious, social and political positions. They were presenting their religious experiences as related to, yet different from, the experiences of others; probably, they were emphasizing their personal suffering as a superior road to advancement. The contestation of the Upper Palaeolithic iconographic canon that the images constituted is therefore an example of how material culture can, in the hands of human agents, play an active, formative role.

Acknowledgements. I am especially grateful to numerous people who made it possible for me to visit Upper Palaeolithic caves in France in 1972, 1989, 1990 and 1995 and with whom I had useful discussions: they include Norbert Aujoulat, Paul Bahn, Robert and Eric Begouen, Brigette Delluc, Jean Gaussen, Yanik Le Guillou, Andre LeroiGourhan, Michel Lorblanchet, Jaques Omnes, Aleth Plenier, Jean-Philippe Rigaud, Yoan Runeau, Dominique Sacchi, Georges Simonnet, Denis Vialou, Luc Wahl and, especially, Jean Clottes. Parts of this paper were presented at a seminar in the McDonald Institute, University of Cambridge, in April 1996. The illustrations were prepared by Anne Holliday [ILLUSTRATION FOR FIGURES 3 & 5 OMITTED] and Rory McLean; FIGURES 2, 3 & 5 are based on Thomas Dawson's and Paul den Hoed's fieldwork. I am grateful to colleagues who commented on drafts: Geoff Blundell, Anne Holliday, Ben Smith and Rory McLean. The Rock Art Centre is funded by the Centre for Science Development (Human Sciences Research Council) and the University of the Witwatersrand. Neither institution is responsible for the views herein expressed.

1 Some writers (e.g. Conkey 1991) try to avoid the word 'art'; with good reason, they prefer phrases such as 'image-making'. I retain the handy monosyllable 'art', but I do not imply that Upper Palaeolithic image-making resembled 20th-century Western art in its functions and social roles.

References

ASAAD, G. 1980. Hallucinations in clinical psychiatry: a guide for mental health professionals. New York (NY): Brunner/ Mazel.

BAHN, P. & J. VERTUT. 1988. Images of the Ice Age. London: Windward.

BATES, C.D. 1992. Sierra Miwok shamans, 1900-1990, in L.J. Bean (ed.), California Indian shamanism: 97-115. Menlo Park (CA): Ballena Press.

BENDER, B. 1989. The roots of inequality, in D. Miller et al. (ed.), Domination and resistance: 83-93. London: Unwin and Hyman.

BIESELE, M. 1980. 'Old K"au', in Halifax (ed.): 54-62.

BLEEK, D.F. 1923. The Mantis and his friends. Cape Town: Maskew Miller.

1935. Beliefs and customs of the/Xam Bushmen. Part VII: Sorcerers, Bantu Studies 9: 1-47.

1936. Beliefs and customs of the/Xam Bushmen. Part VIII: More about sorcerers and charms, Bantu Studies 10: 131-62.

BOND, G. 1948. [Cover design], South African Archaeological Bulletin 3(11).

BOURDIEU. P. 1977. Outline of a theory of practice. Cambridge: Cambridge University Press.

BREUIL, A. 1979. Four hundred centuries of cave art. Montignac: Centre d'Etude et de Documentation Prehistoriques.

BRINDLEY, G.S. 1973. Sensory effects of electrical stimulation of the visual and paravisual cortex in man, in R. Jung (ed.), Handbook of sensory physiology VII/3B: 583-94. New York (NY): Springer-Verlag.

CARRITHERS, M.. S. COLLINS & S. LUKES (ed.). 1985. The category of the person: anthropology, philosophy, history. Cambridge: Cambridge University Press.

CLOTTES, J. & J. COURTIN. 1996. The cave beneath the sea: Paleolithic in, ages at Cosquer. New York (NY): Abrams.

CLOTTES, J. & J.D. LEWIS-WILLIAMS. 1996. Les chamanes de la prehistoire: transe et magie dans les grottes ornees. Paris: Le Seuil.

CONKEY, M.W. 1991. Contexts of action, contexts for power: material culture and gender in the Magdelenian, in J.M. Gero & M.W. Conkey (ed.), Engendering archaeology: women and paintings: 57-92. Oxford: Blackwell.

CYTOWIC, R.E. 1994. The man who tasted shapes. London: Abacus.

DE MARAIS, E., L.J. CASTILLO & T. EARLE. 1996. Ideology, materialization, and power strategies. Current Anthropology 37: 15-86.

DEACON, J. 1988. The power era place in understanding southern San rock engravings, World Archaeology 20: 129-40.

DICKSON, B. 1990. The dawn of belief: religion in the Palaeolithic of southwestern Europe. Tucson (AZ): University of Arizona Press.

DOBKIN DE RIOS, M. 1986. Enigma of drug-induced altered states of consciousness among the !Kung Bushmen of the Kalahari Desert, Journal of Ethnopharmacology 15: 297-304.

DOWSON, T.A. 1988. Revelations of religious reality: the individual in San rock art, World Archaeology 20: 116-28.

1992. Rock engravings of southern Africa. Johannesburg: University of the Witwatersrand Press.

1994. Reading art, writing history: rock art and social change in southern Africa, World Archaeology 25: 332-45.

EDELMAN, G. 1994. Bright air, brilliant fire: on the matter of the mind. London: Penguin.

EICHMEIER, J. & O. HOFER. 1974. Endogene Bildmuster. Munich: Urban & Schwerzenberg.

ELIADE, M. 1972. Shamanism: archaic techniques of ecstasy. London: Routledge & Kegan Paul.

EMBODEN, W. 1979. Narcotic plants. New York (NY): Macmillan.

FISCHER, R. 1975. Cartography of inner space, in Siegel & West (ed.): 197-239.

FOCK, G.J. & D. FOCK. 1989. Felsbilder in Sudafrika Teil III: Die Felsbilder im Vaal-Oranje Becken. Koln: Bohlau Verlag.

FOUCAULT, M. 1965. Madness and civilization. New York (NY): Random House.

FURST, P.T. 1976. Hallucinogens and culture. Novato: Chandler & Sharp.

GARLAKE, P. 1987a. The painted caves: an introduction to the prehistoric art of Zimbabwe. Harare: Modus.

1987b. Themes in the prehistoric art of Zimbabwe, World Archaeology 19: 178-93.

1995. The hunter's vision: the prehistoric art of Zimbabwe. London: British Museum Press.

GIDDENS, A. 1984. The constitution of society: outline of the theory of structuration. Cambridge: Polity Press.

GOODMAN, F.D. 1972. Speaking in tongues: a cross-cultural study of glossolalia. Chicagu (IL): University of Chicago Press.

1988. Ecstasy, ritual, and alternate reality: religion in a pluralistic world. Bloomington (IL): Indiana University Press.

GRAZIOSI, P. 1960. Palaeolithic art. London: Faber & Faber.

HALIFAX, J. 1980. Shamanic voices: a survey of visionary narratives. Harmondsworth: Penguin.

1982. Shaman: the wounded healer. London: Thames & Hudson.

HARNER, M.J. (ed.). 1973a. Hallucinogens and shamanism. New York (NY): Oxford University Press.

1973b. Common themes in South American Indian Yage experiences, in M.J. Harner (ed.), Hallucinogens and shamanism: 155-75. New York: Oxford University Press.

HAYDEN, B. 1987. Alliances and ritual ecstasy: human responses to resource stress, Journal for the Scientific Study of Religion 26: 81-91.

1990. The cultural capacities of Neanderthals: a review and re-evaluation, Journal of Human Evolution 24: 11346.

HEDGES, K. 1983. The shamanit: origins of rock art, in J.A. van Tilburg (ed.), Ancient images on stone: rock art of the Californias: 46-58. Los Angeles (CA): Institute of Archaeology, University of California Los Angeles.

HODDER, I. 1982. Sequences of structural change in the Dutch Neolithic, in I. Hodder (ed.), Symbolic and structural archaeology. Cambridge: Cambridge University Press.

HOLLIS, M. 1985. Of masks and men, in Carrithers et al. (ed.): 217-33.

HUFFMAN, T.N. 1983. The trance hypothesis and the rock art of Zimbabwe, South African Archaeological Society, Goodwin Series 4: 49-53.

JOHNSON, M,H. 1989. Conceptions of agency in archaeological interpretation, Journal of Anthropological Archaeology 8: 189-211.

JOYCE, A.A. & M. WINTER. 1996. Ideology, power, and urban society in pre-Hispanic Oaxaca, Current Anthropology 37: 33-86.

KATZ, R. 1982. Boiling energy: community-healing among the Kalahari !Kung. Cambridge (MA): Harvard University Press.

KLUVER, H. 1942. Mechanisms of hallucinations, in Q. McNemar & M.A. Merrill (ed.), Studies in personality: 175-207. New York (NY): McGraw Hill.

LA BARRE, W. 1972. Hallucinogens and the shamanic origins of religion, in P.T. Furst (ed.), Flesh of the gods: the ritual use of hallucinogens: 261-278. London: George Allen & Unwin.

1975. Anthropological perspectives on hallucinations and hallucinogens, in Siegel & West (ed.): 9-52.

LAMING-EMPERAIRE, A. 1962. La signification de l'art rupestre Paleolithique: methodes et applications. Paris: Picard.

LEROI-GOURHAN, A. 1968. The art of prehistoric man in western Europe. London: Thames & Hudson.

1982. The dawn of European art: an introduction to Palaeolithic cave painting. Cambridge: Cambridge University Press.

LEROI-GOURHAN, A. & J. ALLAIN. 1979. Lascaux inconnu. Paris: CNRS.

LEVI-STRAUSS, C. 1963. The structural study of myth, in C. LeviStrauss, Structural anthropology. 206-31. Harmondsworth: Penguin.

LEWIS-WILLIAMS, J.D. 1972. The syntax and function of the Giant's Castle rock paintings, South African Archaeological Bulletin 27: 49-65.

1974. Superpositioning in a sample of rock-paintings in the Barkly East district, South African Archaeological Bulletin 29: 93-103.

1980. Ethnography and iconography: aspects of southern San thought and art, Man 15: 467-82.

1981. Believing and seeing: symbolic meanings in southern San rock paintings. London: Academic Press.

1983. The rock art of southern Africa. Cambridge: Cambridge University Press.

1988. Reality and non-reality in San rock art. Johannesburg: Institute for the Study of Man in Africa.

1990a. Discovering southern African rock art. Cape Town: David Philip.

1990b. On Palaeolithic art and the neuropsychological model, Current Anthropology 31: 407-8.

1991a. Wrestling with analogy: a methodological dilemma in Upper Palaeolithic art research, Proceedings of the Prehistoric Society 57(1): 149-62.

1991b. Upper Palaeolithic art in the 1990s: a southern African perspective, South African Journal of Science 87: 422-9.

1992. Ethnographic evidence relating to 'trancing' and 'shamans' among northern and southern Bushman groups, South African Archaeological Bulletin 47: 56-60.

1994. Rock art and ritual: southern Africa and beyond, Complutum 5: 277-289.

1995. Modelling the production and consumption of rock art, South African Archaeological Bulletin 50: 143-54.

LEWIS-WILLIAMS, J.D. & T.A. DAWSON. 1988. Signs of all times: entoptic phenomena in Upper Palaeolithic art, Current Anthropology 29: 201-45.

1989. Images of power: understanding Bushman rock art. Johannesburg: Southern Book Publishers.

1990. Through the veil: San rock paintings and the rock face, South African Archaeological Bulletin 45: 5-16.

1992. Art rupestre San et Paleolithique superieur: le lien analogique, L'Anthropologie 96 (4): 760-90.

1993. On vision and power in the Neolithic: evidence from the decorated monuments, Current Anthropology 34: 55-65.

LOMMEL, A. 1967. The world of early hunters. London: Evelyn, Adams & Mackay.

LORBLANCHET, M. 1984a. Grotte de Cougnac, in L'Art des cavernes: atlas des grottes ornees Paleolithiques Francaises: 4837. Paris: Ministere de la Culture.

1984b. Grotte du Pech-Merle, in L'art des cavernes: atlas des grottes ornees Paleolithiques Francaises: 467-74. Paris: Ministere de la Culture.

MAGGS, T.M.O'C. 1967. A quantitative analysis of the rock art from a sample area in the western Cape, South African Journal of Science 63: 100-104.

MARSHACK, A. 1972. The roots of civilization. London: Weidenfeld & Nicolson.

MARSHALL, L. 1962. !Kung Bushman religious beliefs, Africa 32: 221-51.

MEROC, L. & J. MAZET 1977. Cougnac. Gourdon: Edition des Grottes de Cougnac.

MILLER, D. 1987. Material culture and mass consumption. Oxford: Blackwell.

PAGER, H. 1971. Ndedema. Graz: Akademische Druck.

1973. Rock paintings in southern Africa showing bees and honey hunting, Bee World 54: 61-8.

PFEIFER. L. 1970. A subjective report of tactile hallucinations in schizophrenia, Journal of Clinical Psychology 26:5760.

PFEIFFER, J.E. 1982. The creative explosion: an enquiry into the origins of art and religion. New York (NY): Harper & Row.

RENFREW, C. 1982. Towards an archaeology of mind. Cambridge: Cambridge University Press.

SARBIN, T.R. 1967. The concepts of hallucination, Journal of Personality 35: 359-60.

SCHULTES, R.E. 1976. Hallucinogenic plants. New York (NY): Golden Press.

SHANKS, M & C. TILLEY. 1987. Social theory and archaeology. Cambridge: Polity Press.

SHERRATT, A. 1991. Sacred and profane substances: the ritual of narcotics in later Neolithic Europe, in P. Garwood et al. (ed.), Sacred and profane: 50-64. Oxford: Oxford Committee for Archaeology.

SIEGEL, R.K. 1977. Hallucinations, Scientific American 237: 132-40.

1978. Cocaine hallucinations, American Journal of Psychiatry 135: 309-14.

1985. LSD hallucinations: from ergot to Electric Keel-Aid, Journal of Psychoactive Drugs 17: 247-56.

1992. Fire in the brain: clinical tales of hallucination. New York (NY): Dutton.

SIEGEL, R.K. & L.J. WEST (ed.). 1975. Hallucinations: behavior, experience, and theory. New York (NY): John Wiley.

SMITH, N.W. 1992. An analysis of Ice Age art: its psychology and belief system. New York (NY): Peter Lang.

SOLOMON, A. 1994. 'Mythic women': a study in variability in San rock art and narrative, in T.A. Dawson & J.D. LewisWilliams (ed.), Contested images: diversity in southern African rock art research: 331-71. Johannesburg: Witwatersrand University Press.

SUMMERS, R. (ed.). 1959. Prehistoric rock art of the Federation of Rhodesia and Nyasaland. Salisbury: Rhodesia and Nyasaland National Publications Trust.

VINNICOMBE, P. 1976. People of the eland: rock paintings of the Drakensberg Bushmen as a reflection of their life and thought. Pietermaritzburg: Natal University Press.

WEIL, A. 1986. The natural mind: an investigation of drugs and the higher consciousness. Boston (MA): Houghton Mifflin.

WINTERS, W.D 1975. The continuum of CNS excitatory states and hallucinosis, in Siegel & West (ed.): 53-70.

WOODHOUSE. H.C. 1987. Bees and honey in the prehistoric rock art of southern Africa, South African Bee Journal 59 (2): 36-41.