On Neuropsychology and Shamanism in Rock Art

by Robert G. Bednarik, J. D. Lewis-Williams, Thomas A. Dowson
On Neuropsychology and Shamanism in Rock Art
Robert G. Bednarik, J. D. Lewis-Williams, Thomas A. Dowson
Current Anthropology
Start Page: 
End Page: 
Select license: 
Select License

interglacial lasted about 25,000 years (Turner 1975). To-

wards the end of pollen zone Ho IIc there was a sudden

marked increase in Gramineae Dollen concomitant with

a decline in Corylus and axis pollen and a slight in-

crease in Alnus, Betula, and Pinus pollen (Turner 1970).

James (pp. 8-9) interprets the increase in Betula and

Pinus as indicating that deforestation did not take place.

He seems not to have realised, however, that Marks Tey

is a relative-frequency pollen diagram. Therefore, be-

cause the increase in Gramineae pollen (from o to about

10%) does not fully compensate for the decline in Cory

lus pollen (from about 25% to less than IO%), the rela-

tive percentages of all the other species must increase a

little. These increases, of only I-z%, are an artefact of

percentage calculation, a problem long recognised (Ker-

rich and Clarke 1967). Even were these increases real

they would tend to bolster the deforestation hypothesis.

Betula preferentially colonises open ground. Small per-

centages (<5%) of Pinus pollen are well-known to be

the product of long-distance transport (e.g., Lowe and

Walker I 984:I 5 8) and ecologically meaningless. James is

too sceptical. There was indeed a period, which appears

to have lasted some 1,500 years, towards the middle of

the Hoxnian when extensive grassy clearings developed

in the temperate deciduous forests of southern England.

The cause of this is a separate issue. Following West's (1977:357) comment that "it is impossible to say whether the presence of man caused the deforestation, or the deforestation allowed the incoming of man, or whether other environmental factors were responsible," few archaeologists familiar with the Quaternary litera- ture would still argue that this deforestation was of un- doubted anthropogenic origin (e.g., Webb and Baynes n.d.). Its duration would rather suggest some natural cause. One need not deny deforestation in the mid- Hoxnian to argue that the hominids then present in southern Britain were unable to control fire. In this Dar- ticular case, James appears to be in danger of throwing the baby out with the bath water.

References Cited

KERRICH, E., J. AND D. L. 1967. Notes on the possible


mis-use and errors of cumulative frequency graphs for the com-

parison of prehistoric artefact assemblages. Proceedings of the

Prehistoric Society 33:57-69. LOWE,J. J., AND M. J. C. WALKER.

1984. Reconstructing Quater- nary environments. London: Longman.

TURNER,C. 1970. The Middle Pleistocene deposits at Marks Tey,

Essex. Philosophical Transactions of the Royal Society of Lon-

don B 257:373-440.

. 1975. "The correlation and duration of Middle Pleistocene

interglacial periods in northwest Europe," in After the Austral-

opithecine~.Edited by K. W. Butzer and G. L1. Isaac, pp. 259-

308. The Hague: Mouton. WEBB, R. E., AND A. BAYNES. n.d. "Humans as members of in-

terglacial mammal faunas in the British Isles." Proceedings of

the I International Conference on Early Man in Island Environ-

ments, Sardinia, 1988. Edited by P. Sondaar and M. Sanges. In


WEST,R. G. 1956. The Quaternary deposits at Hoxne, Suffolk. Philosophical Transactions of the Royal Society of London B 239:265-356.

Volume 31, Number I, February 1990 1 77

. 1977. 2d edition. Pleistocene geology and biology. Lon-
don: Longman.

J. J. 1974. Clactonian and Acheulian industries in Brit- ain: Their chronology and significance. Proceedings of the Geological Association of London 85:391-421.

On Neuropsychology and
Shamanism in Rock Art


Australian Rock Art Research Association, P. 0.Box 216, Caulfield South, Victoria 3162, Australia. 21 IV 89

Disappointed with Lewis-Williams and Dowson's (CA 29:232-38) response to the valid criticism (and the many valuable points advanced, such as those on the possible involvement of children in rock-art production, by Tur- ner [p. 2281 and Consens [p. 2211) of their paper "The Signs of All Times," I would like to raise several perti- nent issues in the hope of providing this debate with a fresh impetus.

The authors claim Wylie's support of their ethno- graphic analogy when she in fact suggests two strategies for assessing the analogical claims of relevance (p. 232, emphasis added): "One is to press the demonstration of tightness of fit between source and subject (showing that it is unique to this model), and the other is to investigate the persistence and uniqueness of the connection between entoptic-like images in art and their experience in trancelike states in a range of source contexts." It would be relevant, then, to ask whether phosphene forms occur in arts other than those of shamans and whether they are associated primarily with "trance states." Lewis- Williams and Dowson's theory is clearly invalidated by Wylie's test. Phosphene forms are or were most com- monly used by two groups of people: children 3-4 years of age (Kellogg, Knoll, and Kugler 1965) and hominids or humans of the pre-iconic era (Bednarik 1984, 1987, 1988~).The motifs of the art of both groups are largely if not exclusively phosphene forms (the substitution of the term "entoptics" for "phosphenes" has caused such con- fusion that I shall only use the latter term for "normal- state," noniconic entoptics). The evidence concerning the role of phosphenes in the cognitive development of very young children is particularly compelling. Peter van Sommers, a professor of psychology at Macquarie Uni- versity, recently considered the role of graphic univer- sals in the drawings of infants and isolated basic geometric motifs which he called "primitives": they match Knoll's phosphene types. Since van Sommers (1984) never mentions Knoll or afiy of the other writers on phosphenes, his results provide independent corrob- oration, and the notion that very fundamental universals are involved in early art formation becomes even more

persuasive. Extensive controlled phosphene experiments have been conducted with various groups, including Japanese


students, American children, and German air force pilots, but never with shamans. All humans, even some blind people, experience phosphenes, but susceptibility to spontaneous phosphene experiences is by far greatest in infancy. Hallucinogen or trance-induced phosphenes account for only a tiny fraction of such experiences. Hal- lucinations, in contrast, are not physiologically normal phenomena, and Lewis-Williams and Dowson's model of the "three progressive stages of mental imagery" is unconvincing. The iconic images occurring in altered states of consciousness are not conjured up at will (Naranjo 1967); they may even be as "hard-wired" as phosphenes. One can look at original recordings of phos- phenes for hours without ever "seeing" a single object in them. How could Lower Palaeolithic hominids, who probably lacked a concept of iconicity (Davis 1986), have managed to do this?

Lewis-Williams and Dowson "solve" this problem by proposing that "the projection of geometric and iconic imagery was part of humankind's experience throughout the Palaeolithic and in all parts of the world" (p. 216, emphasis added). This is postulated without supporting evidence by researchers who subsequently argue that "a call for proof is inappropriate in rock-art research" (p. 234) and that "observation statements are fallible" and "cannot conclusively falsify a hypothesis" (p. 235). While not conceding that they cannot, upon reflection, sustain their bold claim, they are sufficiently alarmed by my objections to modify their position: "In fact, we claim only that [the] association [of iconic and non- iconic phenomena] is remarkably widespread in rock art" (p. 233).I believe that they owe it to the readers to clarify this key issue by either retracting or reaffirming their claim that Lowtr and Middle Palaeolithic homi- nids experienced iconic imagery.

Lewis-Williams and Dowson not only erroneously equate phosphenes with shamanism and altered states but also equate the trances of the San with those of sha- mans, although the former are communal experiences and not experiences of a shamanistic elite. I am intrigued why-being so interested in shamanism-they concentrate their attention on the San, ignoring the true shamanistic cultures of southern Africa. For instance, what can they tell us about the incidence of phosphene motifs in the art of Zulu shamans (Callaway 1884, Boshier 1974)? More germane than the Coso Range pet- roglyphs (which are ethnographically irrelevant because, according to Whitley, they are often of final Pleistocene1 early Holocene age) would, it seems, be known shaman- istic practices from Siberia, Tibet, the Arctic, West Af- rica (Gorer 1935), the Caribbean (Long 1977), Brazil (Giesler 1983)~ or Peru. The shamanistic status of San art is largely based on Lewis-Williams's own publications (and is not generally accepted even among South Afrkan rock-art specialists), and Reichel-Dolmatoff provides the authors' only ethnographic references for drug-induced hallucinations. As the world's experts in the use of al- kaloids, the South American Indians deserve more than cursory treatment, and the wealth of relevant literature

from South America provides ample challenges for the model of "progressive stages of mental imagery." Lewis- Williams and Dowson conveniently accept the concept of the immutability of phosphene form constants, while postulating that these flickering, ephemeral forms are consciously transformed into iconic motifs during trance. Naranjo's (1967, 1968, 1973) finding that the contents of yage (or yajt) visions are not the result of conscious elaboration of phosphenes squarely con- tradicts their model: specific visions are spontaneously elicited bv harmaline in controlled ex~eriments with subjects lacking the expectations of indigenes, indicat- ing the existence of a collective unconscious of "iconic form constants." Imagery related to death and flying stands out, as do images of felines, snakes, and birds of prey. This, surely, would be more relevant to identifying underlying universals in drug- or trance-induced halluci- nations than the naive explanation that geometric shapes are consciously elaborated into iconic forms (while the subiect's volitional brain functions have suc- cumbed to trance!): a circle becomes an orange, a breast, a cup of water, or a bomb depending on the disposition of the subject. In pondering the possible phylogenetic per- sistence of "iconic form constants" one could enquire, for example, whether the Upper Palaeolithic art has a high incidence of felines, snakes, and eagles. It does not, of course (the total being well below I%)-which does not necessarily preclude shamanism but does render its involvement less likely.

The authors have studied firsthand neither the prehis- toric art of the Upper Palaeolithic nor that of the Ameri- can Southwest or South America, yet they readily reject the advice of those who have studied the occurrence of phosphene forms in all three regions. Similarly, they have made extensive use of Marshack's data but consis- tently misconstrued his illustrations, just as they have misinterpreted Reichel-Dolmatoff (Marshack 1989). Their use of motifs that occur only at a single site to demonstrate a universal mode is, Marshack (1989) ob- serves, "an indication of the subjectivity in the process of selection and construal that Lewis-Williams and Dowson indulged in in order to prove a theory." More- over, they have selected 6 of the 15 phosphene types for consideration (presumably the ones to be found in the rather limited rock-art references cited) but not the most common ones. (The less common types account for only 16% of all electrically and optically induced phosphenes [Eichmeier and Hofer 19741.) They ignore the physiologi- cal causes of phosphenes, and since this is particularly important in understanding the phenomenon it is most unfortunate that they do not consider Meier-Koll's cy- bernetic model of phosphene induction (see Eichmeier and Hofer 1974).

They cite my work in three places, managing to mis- quote me each time: on p. 205, a table dealing entirely with Australian art is implied to relate to European art, and a typographical error distorts the date of the source (1984); on p. 213, I am listed with several others as hav- ing suggested that shamanism existed in the Upper

Volume 31, Number I, February 1990 I 79

Palaeolithic, when in fact I had never even used the word "shamanism" in print and would not dream of mooting such a notionj and on p. 214 the dates of two papers are again in error. But more constructive than dwelling on errors would be to take up the authors' belief that some misunderstanding on my part is indicated by my relegat- ing of meaning to the trivial aspects of early marking traditions (p. 233). I can only repeat that the semantics of prehistoric art is inaccessible to us, while origins or derivation are not. Some of the scholars sharing my view on meaning are cited by Lewis-Williams and Dowson in their introductory paragraph. The phosphene theory is not about meaning; even the role of phosphenes is rather peripheral to it. It is essentially an epistemological the- ory which, among other things, explores the origins of cognition. I am not aware that this has been preempted, as the authors claim, but perhaps I could be enlightened.

Of course it is highly possible that Upper Palaeolithic people used phosphene forms in their arts-all humans have used them in their image systems since humans evolved from the hominids. We all use them daily, but that does not make us shamans! Nor did such use of recycled motifs make the Aurignacians shamans. Not only is there no unique relationship between shamans and phosphenes, but the latter are eons older than the Aurignacian and phosphene forms must have been used in art and communication long before that culture. The phosphene experiences that Upper Palaeolithic people may have had (irrespective of context) would have had the effect of dramatically reinforcing beliefs in the super- natural qualities of these already enculturated motifs and would thus have validated the metaphysical con- cepts held, whatever these were. It was in fact this vali- dation principle that initially encouraged me to postu- late the phylogenetic antiquity of phosphene forms (Bednarik I 984, I987)-although neurophysiology cer- tainly provides corroborative evidence for it. This is a far cry from the simplistic model of Lewis-Williams and Dowson, a model which can be resolved thus: Everyone who use phosphene forms is a shaman; hence every hu- man is a shaman; hence there are no shamans (since one term becomes superfluous); hence there can be no shamanistic art.

There is a huge corpus of ethnographic evidence from all parts of the world indicating that body painting, cicatrices, tattoos, decorative paraphernalia, and gar- ments all provided significant information about their wearers, and it seems plausible that such enculturated information found its way into rock art. It may well be correct that the geometric markings on anthropomorphs in rock art were ultimately derived from phosphene forms, but at the level at which Lewis-Williams and Dawson proceed this is not relevant: for example, their "meaning" or function may have been emblemic. The form content becomes relevant only at the next level, at which we might consider, for instance, the potential of phosphenes to validate the potency of such recycled motifs; and it becomes especially pertinent when we look at the question of ultimate derivation by examining

the art traditions predating the introduction of two-dimensional iconicity (Bednarik 1988~).

The authors' intransigence in the face of the serious objections raised by nearly all commentators is inexpe- dient. Having long admired Lewis-Williams's dedication in exploring non-positivistic aspects of prehistoric arts and his rejection of naive empiricism or scientism, I re- gret having to conclude that his desire to find a shortcut to a universal model has led him on. Through its omis- sions, misconstruals, and selectivity Lewis-Williams and Dowson's paper actually presents a better case against the involvement of shamans in rock-art produc- tion than for it. No art of true shamanistic traditions is considered, while the arts that are richest in phos- phene forms (e.g., in Australia) are conspicuously non- shamanistic. Phosphene forms constitute less than 5% of the rock arts of the Upper Palaeolithic of Europe, of the Coso Range, and of the San. Prehistoric arts signifi- cantly richer in phosphene types than these occur in all continents, and a reasonably comprehensive table listing their phosphene motif types would need to be about 40- 50 times as large as that provided by the authors. It would include, among others, the archaic petroglyphs of Piaui, Brazil (Bednarik 1989), those of Bolivia (Bednarik 1988b), those of the U.S.A. (Bednarik 1988c), various rock arts in Africa (e.g., dos Santos 1974), the earliest paintings of India and other Asian art, various bodies of European rock art, the largely phosphenic art of New Caledonia (Frimigacci and Monnin 1980)~ and the sev- eral extensive pre-iconic petroglyph traditions of Australia (e.g., Bednarik 1987). In a recent response to the Lewis-Williams and Dowson paper, Bradley (CA 30:68-75) considers the incidence of phosphene types in the megalithic art of Europe. His observations only confirm the ubiquity of these motif types: they are in- deed the "signs of all times" rather than the signs of shamanism. ~ut

Bradley's attempt to extend Lewis- Williams and Dowson's table (fig. 4) also exposes yet another problem with that model: Type VI (filigrees or thin meandering lines) is misunderstood by Bradley as referring to vortices (which are in fact another phos- phene motif, Type 10 of Kellogg, Knoll, and Kugler 1965)~ and IIIE, clearly a radial design in a circle (and thus a combination of phosphene Types 2 and 6), is grouped with dots (Type 7). The confusion is attrib- utable to Lewis-Williams and Dowson's arbitrarv selec- tion of types and selective utilization of different sources to achieve "fits." Future attempts of this kind should always refer to the original sources-the work of Knoll, Kugler, Eichmeier, Hofer, and colleagues.

If shamanism were indicated by the frequency of phos- phene motifs, which is Lewis-Williams and Dowson's central postulate, that frequency would provide an indi- cation of shamanistic influence in an art. According to the antithetical but earlier phosphene theory, in the old- est, pre-iconic art traditions phosphene motifs may dom- inate to the point of exclusiveness (Bednarik 1984), while often being less common in more recent tradi- tions: their frequency thus provides a rough indication of "archaicness." The two theories appear to be mutu- ally exclusive.


Department of Archaeology, University of the

Witwatersrand, Johannesburg, South Africa. 27 VII 89

Rock-art research has never enjoyed much respect or sustained attention in Anglophone archaeology. The reasons for this include its methodological and theoreti- cal poverty, simplistic ethnographic analogies, obsession with trivia, factual errors, ad hominem jibes, failure to attend closely to what others have written, and unsup- ported assertions. A number of these problems are evi- dent in Bednarik's comment.

Many of the points he makes could have been cleared up by attentive reading of our paper, the comments, and our reply (CA zg:zo~-45). This is evident right at the start when he claims that our argument is "invalidated" by procedures proposed by Wylie (pp. 231-32). In fact she writes:

Far from calling into question claims made for [the

neuropsychological model's] plausibility or security,

my aim is to elaborate the suggestion that reliance on

analogical inference should not automatically give

way to "ethnographic despair." . . . the strength of the

model . . . derives not so much from an elimination of

inference as from an effective use of two quite differ-

ent, mutually reinforcing sources which underwrite

strong claims for the relevance of specific similarities

in visual imagery manifested in diverse contexts.

She concludes, "It is a significant strength of the present model . . . that it not only provides an initially compel- ling account of puzzling, seemingly 'anarchic' phenom- ena but so clearly opens up new lines of enquiry." All this is vastly different from the impression Bednarik contrives.

To strengthen the "analogical claims of relevance" Wylie suggests the two strategies Bednarik mentions. Here we must emphasise a point we made in our paper and reply: at the present stage of research and with the model still in what we believe to be an initial stage of development, a tight fit between it (in all its compo- nents) and an art merely suggests that states of altered consciousness played some part. These states may have included, as we allow (p. 213), migraine (Sacks 1970, Richards I 971, Siegel and Jarvik I 975), schizophrenia (Siegel and Jarvik 1975)~ and infantile perception (Kel- logg, Knoll, and Kugler 1965). Bednarik fails to mention that our argument does not end here: it proceeds from this still equivocal point as a "best-fit" hypothesis. For instance, although it is possible that some Upper Palaeolithic visions derived from migraine attacks or schizophrenia, it would be ludicrous to argue that Upper Palaeolithic art can be explained by migraine attacks or

schizophrenia. If some of the imagery did derive from such pathological conditions, those conditions were clearly part of a ritual complex (see, for example, Eliade 1972). Similarly, whatever role children may have played (see below), "children's art" cannot explain Up- per Palaeolithic art as a whole or the geometric forms as a category. If we bear in mind (I)the very widespread occurrence of some form of shamanism in hunter- gatherer societies, (2)the symbolic role of animals in these societies and the prominent role of animals in Up- per Palaeolithic art, and (3) the other points we made in our "Implications" section (pp. 21 3-17), our explanation is clearly a "best fit." As we put it (p. 213)~ "All that the parallels exhibited here allow us to conclude at the mo- ment is an association with altered states and, further, . . .that in the Upper Palaeolithic this association proba- bly had at least some features in common with what we understand by shamanism." This cautious claim is ig- nored by Bednarik; instead he presents a travesty of our argument so that he may the better attack it.

Bednarik wonders why we "concentrate" on San shamanism and ignore the "true shamanistic cultures of southern Africa." Shamanism is, of course, a very di- verse phenomenon, and it is hard to say exactly what constitutes "true" shamanism-if there is such a thing. In a passage Bednarik evidently missed, we discussed our use of "shaman" for the San (p. 204):

We use "shaman" rather than the more usual "medi- cine man," but we do not thereby imply anything about the social position of the person, his or her mental health, or, indeed, many of the other charac- teristics often associated with the very heterogeneous phenomenon called shamanism. Instead we empha- sise what we believe to be the most important and overriding feature of shamanism and the one with which this paper is principally concerned-altered states of consciousness.

We went on to discuss various San words for "shaman" to show that San shamans are closely associated with concepts of supernatural power. In the light of this dis- cussion it is patently absurd to argue that the San do not practise a form of shamanism. (Interesting as their altered-state exveriences are, the Zulus do not make rock art and are-therefore irrelevant to our argument.)

It is also absurd to claim that the shamanistic status of San art "is not generally accepted even among South African rock-art specialists." On the contrary, it is hard to think of any practising professional southern African archaeologist or anthropologist who does not now accept it, and we cited numerous arguments for this position (p. 204). (That list can be extended: see, for example, Huff- man 1983; Maggs and Sealy 1983; Yates, Golson, and Hall 1985; Manhire et al. 1986; Maze1 1989; Parkington et al. 1986; Campbell 1986; Deacon 1986, 1988; S. Hall 19186; M. Hall 1986, 1987; Hammond-Tooke 1983; Morris I 988; Garlake 1987a, b, c; Guenther 1988; Leslie Brooker I 989; Loubser and Dowson 1987; Dowson 1988, 1989; Dowson and Holliday 1989; Sampson 1988~ Par- kington 1989; Vinnicombe 1986; Wadley 1987).

Volume 31, Number I, February 1990 1 81

Never recognising that our model is based on research

conducted by neuropsychologists or addressing those

primary sources, Bednarik gives two simplistic and er-

roneous versions of it in order to lampoon it. First he

asserts that the model "can be resolved thus: Everyone

who uses phosphene forms is a shaman; hence every

human is a shaman; hence there are no shamans (since

one term becomes superfluous); hence there can be no

shamanistic art." He then com~ounds his error: "If

shamanism were indicated by tie frequency of phos-

phene motifs, which is Lewis-Williams and Dowson's

central postulate, that frequency would provide an indi-

cation of shamanistic influence in an art." Even a

superficial reading of our paper shows that the model is

not concerned exclusively with entoptic phenomena. It

comprises three stages, the last of which includes fully

iconic hallucinations, and, furthermore, seven principles

governing the perception of mental imagery. Some of

these principles clarify the relationships between en

toptic and iconic imagery and the ways in which they

are combined. In the face of the neuropsychological evi-

dence we cited, it is hard to see how the three stages of

our model can be challenged. Certainly, in referring to

them as "unconvincing" Bednarik does not attempt to

show that the neuropsychologists are wrong.

We tested the utility of this complex model against San rock art because it is known to be associated with altered states and, quite simply, because we know it best. We also tested the model against Coso rock art. On this subject Bednarik says that we "ignore the solid ob- jections of the American commentators," but when we turn to the comments by North Americans we find that only Vastokas, Faulstich, and Halverson have anything to say about it. Vastokas merely says that we have car- ried our argument "too far" (p. 230). Faulstich says that "the patterns integrated into the Coso anthropomorphs . . . may depict both internal power and ornate gar- ments." Here he seems to be accepting at least part of our case, but later he advocates caution by adding that there is "no conclusive ethnographic evidence" (we be- lieve he is mistaken) and that we should not "prema- turely attribute a shamanistic origin to this art" (p. 225). By contrast, Halverson (p. 225) says that "the heuristic use of ethnography [is] admirable in respect to San and Coso depictions." In any event, the shamanistic status of Coso art is established by the cogency of the argu- ments and the evidence; one writer may be right and all the others wrong.

Bednarik also claims that the "relevant literature from South America provides ample challenges for the model of 'progressive stages of mental imagery1," but, as we have said, according to Reichel-Dolmatoff ( 1978: 12-13) the Tukano themselves speak of three progressive stages in their yaje-induced visions. The evidence, as we re- ported it (p. 204)) is clear enough: for the Tukano there are stages, and those stages are very like the ones we derived from neuropsychology and built into our model. We have not "misinterpreted Reichel-Dolmatoff."

Focussing on the first stage of our model, Bednarik says we claim that a high frequency of entoptic motifs in an art indicates "shamanistic influence." We made it

clear, however, that some arts seem to dwell on Stage I

entoptics while others emphasise iconic hallucinations.

In southern Africa, for example, some rock-art sites have

a high proportion of geometric depictions while others

have very few, yet we know that both kinds of site were

associated with shamanism. Why there are these differ-

ences is a matter for research. Our position on this point

thus makes nonsense of Bednarik's statement that

"phosphene forms constitute less than 5% of the rock

arts of the Upper Palaeolithic of Europe, of the Coso

Range, and of the San." Certainly for southern Africa,

we know of no research that could possibly confirm or

contradict a figure of 5%. Moreover, in Europe, as in

southern Africa, some sites have principally entoptics

while others have very few. There is little point in com-

bining them and assigning them a percentage of the


More interesting is the argument that children experi-

ence Stage I entoptic phenomena and that this in some

way invalidates our position. It may be true that chil-

dren see entoptic phenomena. In replying to Martindale

(pp. 227-28) on this issue, we did not deny that some

geometric rock-art motifs may have been made by chil-

dren, but we challenge the notion that this could have

had much to do with Upper Palaeolithic art (p. 236):

We believe that the long association of entoptic forms with often remarkably "realistic" iconic depictions, various kinds of combinations of entoptic and iconic forms, and combinations of iconic images (therian- thropes), together with evidence for complex Upper Palaeolithic social forms, are better explained by the more extreme varieties of altered consciousness asso- ciated with shamanism than by infantile perception.

Moreover, even though there is some evidence that chil- dren went into Upper Palaeolithic caves, the locations of many of the signs make it highly unlikely that children made them. In any event, are we talking about very young children making entoptic "doodles" deep under- ground or about somewhat older children involved in rituals? Some footprints suggest that children may have been part of whatever rituals were performed in the caves. If Bednarik and others are suggesting the second of these two possibilities, we have no problem with the hypothesis.

Referring to the second stage, Bednarik says, "One can look at original recordings of phosphenes for hours with- out ever 'seeing' a single object in them." We never claimed that people, ancient or modern, try to see things in "original recordings of phosphenes." The point is that, in an altered state of consciousness, people some- times do elaborate entoptic mental imagery, and we cited neuropsychological references to show this. He ar- gues that Stage 3 "iconic images . . . are not conjured up at will." Once again, we made no such claim. He goes on to ask how Lower Palaeolithic hominids, "who probably lacked a concept of iconicity," saw objects in entoptic forms, and he challenges our statement that "the pro- jection of geometric and iconic imagery was part of

humankind's experience throughout the Palaeolithic

and in all Darts of the world." It seems to us clear from

the context of this last quotation that we were referring

to the Upper Palaeolithic, but we nevertheless think it

highly probably that Lower and Middle Palaeolithic peo-

ple (we don't mind the word) experienced both kinds of

hallucination. Bednarik is confused about the concept

of iconicity. We were clearly writing about hallucina-

tions of things such as animals, which are surely

possible independently of a capacity for understanding

two-dimensional pictures. Neuropsychological research

strongly suggests that animals such as cats and apes can

hallucinate (Siege1 and Jarvik 1975:87-104). This im-

plies that the capacity to hallucinate is built into the

mammalian, not just the human, nervous system. If this

reasonable assumption is valid, then it seems highly

probable that Lower and Middle Palaeolithic people did

hallucinate. But, as Davis (1986) points out, they proba-

bly lacked a capacity for iconicity and did not make pic-

tures of their hallucinations.

In suggesting that some iconic hallucinations may be as "hard-wired" as entoptic phenomena, Bednarik raises an intriguing issue that we intend to address in a future publication. For the present we simply note that he over- looks the fact that the Westerners who took part in labo- ratory experiments had well-formed concepts of felines, snakes, and birds of prey and that South American sha- mans and Westerners can be expected to share at least some ideas about these creatures. The ex~eriments show not, as he supposes, that images of these creatures are "wired" into the human brain but that even Westerners experience the stresses of certain altered states of con- sciousness in terms of animal imagery: in all societies animals are probably the most powerful and pregnant symbols. The threatening nature of some of these altered states is embodied in frightening animals that are asso- ciated with death (a very widespread way of speaking about entry into altered states). At the same time it must be said that people who have no knowledge at all of, say, polar bears will clearly not hallucinate them. Bednarik goes on to make the extraordinary suggestion that a low incidence of felines, snakes, and birds of prey in an art renders shamanism as a factor in its ~roduction


less likely. He is here assuming that shamans automat- ically depict the images they "see" in the same propor- tions as they see them, as if depiction were some sort of mechanical photocopying process operating on mental imagery. On the contrary, it is abundantly clear that depiction, entoptic and iconic, is culturally controlled. That is why it is unnecessary for our hypothesis to show a high incidence of any particular species or the occur- rence of distinctive entoptic forms in a shamanistic art in the same proportions as in laboratory experiments.

Cultural control must be taken into account when ad- dressing another point as well. Whether certain geomet- ric forms, especially those associated with human figures, are "directly fixed" entoptic phenomena or whether they derived "ultimately" from entoptic shapes is, contrary to what Bednarik implies, a question we did address (pp. 215-16):

This is not to say that all Upper Palaeolithic depic- tions are images fixed by people in altered states or experiencing afterimages. Once that initial step had been taken, the development of Upper Palaeolithic art probably followed three courses. One stream re- mained mental imagery fixed while it was being ex- perienced. A second stream derived from recollected mental imagery processed in the ways we have sug- gested. A third stream derived from contemplation of the graphic products of the first two streams and the realization that they could be duplicated even by someone who had never experienced an altered state of consciousness.

The images that will be "fixed," recollected, processed, contemplated, and duplicated will be culturally con- trolled, but there is more to it than that. In southern Africa entoptic forms are sometimes integrated into clearly hallucinatory "monsters" and therianthropes (Lewis-Williams and Dowson 1989, Dowson 1989). This is in fact what neuropsychology predicts, and we cited references to support'this. We believe that such San, Coso, and upper-Palaeolithic depictions represent vi- sions in which the two kinds of hallucination (entoptic and iconic) were combined. Whether some depictions remesent "third-stream." derived forms is another aues- tion; clearly, we allowed that this may well be the case.

Although Bednarik says that he has "long admired Lewis-Williams's . . .reiection of naive em~iricism," he fails to see that he himself follows &at falladious methodology and mind-set. Echoing a point made by Marshack (1989)~ he claims that we "have studied firsthand neither the prehistoric art of the Upper Palaeo- lithic nor that of the American Southwest" (Coso is ac- tually in the Great Basin). As it happens, one of us (DLW) has studied Upper Palaeolithic art through the kindness of the late Andre Leroi-Gourhan and the staff of the Les Eyzies Museum and North American rock art through the good offices of David Whitley, Larry Loen- dorf. Stuart Conner, and others. More recentlv both of us have been able to $tudy a large number of cives in the company of Paul Bahn and many French researchers. Borrowing again from Marshack, Bednarik exhibits the empiricist mind-set when he refers to the "selection" we engaged in "in order to prove a theory." Elsewhere he sarcastically (but correctly) quotes us as claiming that a call for "proof is inappropriate in rock-art research" and that "observation statements are fallible" and "cannot conclusively falsify a hypothesis." Clearly, Bednarik puts his faith in "proof" and inductive arguments from theory-free data (Marshack's "analytical data"). As Con- key (1987, 1988) has emphasised, this kind of empiri- cism is one of the most debilitating features of rock-art research. The empiricist programme envisages four stages of research: the unbiased collection of theory-free data, "objective" categorization and analysis of these data, induction of an explanation from this analysis, and, finally, a test of the explanation against "new" data. Phi- losophers of science have amply demonstrated the im- possibility of this alluring approach. It is impossible to

collect data without making discriminations according to preconceived ideas; it is impossible to categorize and analyse these data without selecting discriminating fea- tures according to a preconceived classification; it is im- possible to induce inescapable conclusions; and it is im- possible to test those conclusions without an element of circularity. One of us has discussed empiricism in rock- art research (Lewis-Williams 1983, Lewis-Williams and Loubser 1986)~ and more recently we have collaborated in a critique of Marshack's empiricism (Lewis-Williams and Dowson 1989); we recommend the references we have cited in our publications (e.g., Chalmers 1978, Hempel 1966). At the same time, we emphasise that careful empirical work is as important in rock-art re- search as in any other field. Unfortunately, the distinc- tion between an empiricist research programme and the empirical work that links theory and data is not fully appreciated, and criticisms of empiricism are often taken to mean that accuracy does not matter.

Bednarik considers us to have been "sufficiently alarmed" by his earlier objections to modify our posi- tion. Alarm is not the emotion we experience when we read his observations, and we have shown that no modification of our position was necessary. He says that he hopes to provide the debate "with a fresh impetus," but his misreadings and errors do not accomplish that end. For fresh impetus we must turn to Bradley (CA 30:68-7s; see also Lewis-Williams and Dowson ad.).

References Cited

ARING,C. D. 1972. The migrainous scintillating scotoma. Journal

of the American Medical Association 220:5 19-22. BEDNARIK, G. 1984. On the nature of psychograms. TheAr-

tefact 8:27-32.

Volume 3I, Number I, February 1990 1 83

tory of Bleek and Lloyd's /Xan San informants. African Studies 45:135-55. -. 1988. The power of a place in understanding southern San rock engravings. World Archaeology 20: 129-40. DOS SANTOS,J. R. 1974. AS gravuras rupestres do Tchitundo hulo. Trabalhos do Instituto de Antropologia "Dr. Mendes Cor- r&a" 26. D OWSON,T. A. 1988. Revelations of religious reality: The indi- vidual in San rock art. World Archaeology 20:I 16-28.

-. 1989. Dots and dashes: Cracking the entoptic code in Bushman rock paintings. South African Archaeological Society Goodwin Series 6:84-94. DOWSON,T. A., AND A. L. HOLLIDAY. 1989. Eland andzigzags: An interpretation of an idiosyncratic combination. South Afri- can Archaeological Bulletin 48:46-48. DRIVER,H. E. 1937. Cultural element distributions 6: Southern Sierra Nevada. Anthropological Records I :35-I 54. EICHMEIER,J., AND O. HOFER. 1974. Endogene Bildmuster. Munich: Urban and Schwarzenberg. ELIADE,M. 1972. Shamanism: Archaic techniques of ecstasy. New York: Routledge and Kegan Paul. FRIMIGACCI, MONNIN. 1980. Un inventaire des pet-

D., AND J.
roglyphes de Nouvelle-CalCdonie. Journal de la Societe des
OcBanistes 36:17-59.


P. S. 1987a. Structure and meaning in the prehistoric art of Zimbabwe. Hans Wolff Memorial Lecture, African Stud- ies Program, Indiana University.

-. 1987b. The painted caves: An introduction to the prehis- toric art of Zimbabwe. Harare: Modus. . 1987c. Themes in the prehistoric art of Zimbabwe. World Archaeology 19: 178-93. GIESLER,

P. V. 1983. "Parapsychological anthropology: Multi- method approaches to the study of psi in the field setting," in Research in parapsychology 1982. Edited by W. G. Roll, J. Be- loff, and R. A. White, pp. 241-45. Metuchen, N.J.: Scarecrow Press.

GORER,G. 1935.African dances: A book about West African Ne- groes. New York: Knopf. GUENTHER,

M. 1988. "Animals in Bushman thought, myth, and art," in Hunters and gatherers: Property, power, and ideology. Edited by T. Ingold, D. Riches, and J. Woodburn, pp. 192-202. Oxford: Berg.

-. -.

1987. Engramme und Phosphene. Zeitschrift fur Ethnol- ogie 112:223-35. . 1988a. Art origins. Paper presented in Symposium K of the First AURA Congress, Darwin, September 2. 1988b. El arte rupestre Boliviano visto desde el exterior.

Sociedad de Investigacibn del Arte Rupestre de Bolivia Boletin,

no. 2, pp. 22-28.

HALL,M. 1986. The role of cattle in southern African agropastoral societies: More than bones alone can tell. South African Ar- chaeological Society Goodwin Series 5:83-87.

-. 1987. The changingpast: Farmers, kings, and traders in southern Africa. Cape Town: David Philip. HALL.S. L. 1986. Pastoral adaptations and forager reactions in the eastern Cape. South African Archaeological Society Goodwin

-. 1988c. Archaeological potentials of the Parowan Gap site, Utah. Paper presented in Symposium B of the First AURA Con- gress, Darwin, August 29.

-. 1989. On the Pleistocene settlement of South America. Antiquity 63:1o1-11. B OSHIER,A. K. 1974. "African apprenticeship," in Parapsychology and anthropology. Edited by A. Angoff and D. Barth, pp. 273-84 New York: Parapsychology Foundation. CALLAWAY,

H. 1884. The religious system of the Amazulu. Lon

don: Folk-Lore Society. CAMPBELL,

C. 1986. Images of war: A problem in San rock art

research. World Archaeology 1925 5-68. CHALMERS,

A. F. 1978. What is this thing called science! Milton Keynes: Open University Press.

CONKEY,M. 1987. New approaches in the search for meaning? A review of research in "Paleolithic art." Journal of Field Ar- chaeology 14:413-30.

-. 1988. "The structural analvsis of Paleolithic art." in Archaeological thought in ~merica. Edited by C. C. amber^-Karlovsky, pp. I 35-54. Cambridge: Cambridge University

Series 5:42-49. HAMMOND-TOOKE,

W. D. 1983. Reply to A. R. Willcox. South African Archaeological Bulletin 38:5-6. HEMPEL,C. G. 1966. Philosophy of natural science. Englewood

Cliffs: Prentice-Hall. HUFFMAN,

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

KELLOGG, R. M., M. KNOLL, AND J. KUGLER. 1965Formsimilarity between phosphenes of adults and preschool chil- dren's scribblings. Nature 208:I 129-30.


M. 1989. The Holocene sequence from
Uniondale Rock Shelter in the eastern Cape. South African
Archaeological Society Goodwin Series 6:I7-32.


J. D. 1983. Introductory essay: Science and rock art. South African Archaeological Society Goodwin Series 4:3-13.

LEWIS-WILLIAMS, J. D., AND T. A. DOWSON. 1989. Images of power: Understanding Bushman rock art. Johannesburg: South- ern Book Publishers.



n.d. Neuropsychology and megalithic art. MS.

DAVIS, W. 1986. The origins of image making. CURRENT ANTHRO-LEWIS-WILLIAMS, J. D., AND J. H. N. LOUBSER. 1986. "Decep- POLOGY 27:193-215. tive appearances: A critique of southern African rock art stud- DEACON,J. 1986. "My place is the Bitterpits": The home terri- ies," in Advances in world archaeology, vol. 5. Edited by F.


Wendorf and A. E. Close, pp. 253-89. New York: Academic


LONG,J. 1977. "Extrasensory ecology: A summary of evidence," in Extrasensory ecology: Parapsychology and anthropology. Edited by J. Long, pp. 371-96. Metuchen, N.J.: Scarecrow Press.


J. H. N., AND T. A. DOWSON. 1987. Tombo-La-Ndou: The Venda perception of San rock art. South African Archaeo- logical Bulletin 42:5 1-54.

MAGGS,T. M. o'c., AND J. SEALY.1983. Elephants in boxes. South African Archaeological Society Goodwin Series 4:44-48. MANHIRE, A. H., J. E. PARKINGTON, A. D. MAZEL, AND

T. M. O'C. MAGGS. 1986. Cattle, sheep, and horses: A review of domestic animals in the rock art of southern Africa. South Afri- can Archaeological Society Goodwin Series 5:zz-30.


A. 1989. Methodology in the analysis and interpre- tation of Upper Palaeolithic image: Theory versus contextual analysis. Rock Art Research 6:17-38.

MAZEL,A. D. 1989. People making history: The last ten thousand years of hunter-gatherer communities in the Thukela Basin. Natal Museum Iournal of Humanities I:I-168.

MORRIS,D. 1988. Engraved in place and time: A review of vari- ability in the rock art of the northern Cape and Karoo. South African Archaeological Bulletin 43: 109-21.


C. 1967. "Psychotropic properties of the harmala al- kaloids," in The ethnopharmacologic search for psychoactive drugs. Edited by D. Efron. U.S. Public Health Service Publica- tion 1645.

. 1968. Psychotherapeutic applications of new fantasy-
enhancers. Clinical Toxicology I :66-68.

. 1973. "Psychological aspects of the yage experience in an experimental setting," in Hallucinogens and shamanism. Edited by M. S. Harner, pp. 176-90. New York: Oxford Univer- sity Press.

PARKIN GTO N, J. E. 1989. Interpreting paintings without a com- mentary: Meaning and motive, content and composition in the rock art of the western Cape, South Africa. Antiquity 63:13-26.

PARKINGTON, J. E., R. YATES, A. MANHIRE, AND D. HAL- KETT. 1986. The social impact of pastoralism in the south-wes- tern Cape. Iournal of Anthropological Archaeology 5: 3 I 3


G. 1969. El context0 cultural de un alucinogeno aborigen, Banisteriopsis caapi. Revista de la Academia Colombiana de Ciencias Exactas, Fisicas y Naturales 3:327-45.

-. 1978. Beyond the Milky Way: Hallucinatory imagery of the Tukano Indians. Los Angeles: UCLA Latin American Cen- ter.


W. 1971. The fortification illusions of migraines. Scientific American 225:89-96. SACKS,O. W. 1970 Migraine: The evolution of a common disor-

der. London: Faber. SAMPSON,

G. 1988. Stylistic boundaries amongmobile hunter- foragers. Washington: Smithsonian Institution Press.

SIEGEL,R. K.,AND M. E. JARVIK. 1975. "Drug-inducedhallucina- tions in animals and man," in Hallucinations: Behaviour, expe- rience, and theory. Edited by R. K. Siege1 and L.J. West, pp. 81-161. New York: Wiley.


P. 1984. Drawing and cognition. Cambridge:

Cambridge University Press. VINNICOMBE,

P. 1986. "Rock art, territory, and land rights," in

The past and future of !Kung ethnography: Critical reflections

and symbolic perspectives, essays in honour of Lorna Marshall.

Edited by M. Biesele, R. Gordon, and R. Lee, pp. 275-301 Ham- burg: Helmut Buske Verlag.


E. W. 1938. Tubatulabal ethnography. Anthropolog1-90.

ical Records ~(1): WADLEY,

L. 1987. Later Stone Age hunters andgatherers of the southern Transvaal: Social and ecological interpretation. British Archaeological Reports International Series.

YATES,R., J. GOLSON,AND M. HALL. 1985Tranceperformance: The rock art of Boontjieskloof and Sevilla. South African Ar- chaeological Bulletin 40:70-80.

All anthropologists and librarians know:


The Journal of the Royal Anthropological Institute

but do you know

foranthropology with ZIP?

During 1989, ANTHROPOLOGY TODAY has run articles on: the 'crack' economy in Spanish Harlem (Philippe Bourgois); China's June 1989, Dramas of revolution on street and screen (Stephan Feuchtwang); books on psychoanalysis (Roland Littlewood); Azaria Chamberlain and the dingo (J~llie Marcus); the murder of Dr David Webster in Johannesburg (Andrew Spiegel and Rhoda Kadalie), etc.

Individual Fellows receive both jo~urnals automatically

(US$59or £36 overseas residents, £42 UK residents, US$34 or £21 if under 28).
Or you may subscribe just to A.T. as a 'member' for £9 or US$15.
(Rates are for 1990.Library rates are Man US$91/£58,A.T. US$28/£17.)

Write now for a free sample copy, or send order to: RAI,50 Fitzroy Street, London WlP 5HS.

  • Recommend Us