Pech Merle, France, Upper Palaeolithic

Relevant articles on this site:
The origins of human modernity, an Open Access article in the new journal Humanities.
Philosophy of science, an article about the application of philosophy of science in rock art research.
Metamorphology: in lieu of uniformitarianism.
Crisis in Palaeolithic art studies, a critical assessment of developments in the study of Pleistocene art.
The seminal paper on taphonomic logic, A taphonomy of palaeoart, PDF file, 347 KB.
Taphonomic logic for dummies is a PDF file, 45 KB, providing a simple explanation of the principles of taphonomic logic – the most powerful theoretical tool in archaeology.
Boundaries, a ‘Semiotic Review of Books Highlight’, discusses epistemological issues in rock art research.
A unified theory for palaeoart studies, an article presenting an epistemology of the study of early art.
Before opening a PDF file you may need to open your Acrobat Reader. If you do not have Adobe Acrobat Reader please click here for a free download of the latest version.

Rock art, taphonomy and epistemology
Robert G. Bednarik
August 2002

Abstract: An epistemological examination of the relationship between traditional archaeological method and the demands of a rigorous study of rock art suggests the existence of significant differences. The effects of archaeological theory and the use of statistics in rock art research are considered, and it is shown that they are likely to generate taphonomically precarious hypotheses. Using taphonomic logic as a means of testing the refutability of archaeological hypotheses it is shown that they are of little consequence in the scientific study of rock art, which is better served by scientific fields, which may be as diverse as semiotics, geochemistry and hominid ethology.

A number of recent papers prompt me to address the subjects of the scientific study of rock art, archaeological perspectives of rock art and other forms of palaeoart, the effects of taphonomic logic and the role of epistemology. I will avoid being directly critical of specific articles or individual scholars (hence the short bibliography), concentrating instead on general issues, epistemological trends in science and current paradigmatic shifts in rock art studies. This field of enquiry, after all, has attained a certain level of independence (Odak 1993).

I will examine some aspects of the decline of certain modes of archaeology; the relationships between its methods and models, and those of ‘prehistoric’ art studies; and the question of how recent developments might affect ‘cognitive archaeology’. I will argue that the latter can survive the epistemological crisis in archaeology: while scientific interpretation of archaeological ‘data’ may often not be feasible, certain hypotheses about pre-Historic (rather than ‘prehistoric’) arts may be susceptible to refutation. Only the briefest discussion is possible of the various topics related here, but they have all been discussed by me elsewhere in greater detail.


Rock art studies and archaeology

In all disciplines, the acceptance of hypotheses is more often based on the biases of scholars than on actual objective merits (Feyerabend 1975; Kuhn 1970; Thomas 1982; Bednarik 1985, 1990a; Cameron 1993). But while in the ‘hard sciences’ hypotheses tend to find themselves subjected to rigorous testing sooner or later, falsification of archaeological interpretations is frequently not possible, especially where they are intentionally couched in non-refutable terms (Bednarik 1992a: 27). In all disciplines, the popularity of an hypothesis may be governed by political factors (Lewis-Williams 1993), but in archaeology it is not necessarily possible to displace unsound models, because falsifiability remains so often elusive. ‘Speculation and the subjective are part of the ‘scientific’ process’ in archaeology (Hodder 1984: 28), its non-scientific nature has been revealed ‘by a scientific style of discourse’ (Hodder 1990: 51), and its role as a sociopolitical tool for the creation of idiosyncratic archaeologies has been recognised (Hodder 1986: 161; see also, for example, Shanks and Tilley 1987; Tangri 1989; Fletcher 1991; Tilley 1993).

Archaeology acquires its ‘accepted fiction’ (Bahn 1990: 75) with
(a) a frequently untestable methodology;
(b) a store of accumulated knowledge, much of which is probably invalid;
(c) vast quantities of ethnocentrically selected materialist ‘data’; and
(d) methods of induction which would not be credible even if the data were valid (see Huchet 1991, 1992).

Not only is it quite impossible to infer the substance of a complex archaeological phenomenon from its surviving ‘observable phenomena’ alone (Bednarik 1985), the naive logic archaeologists sometimes apply results in misinterpretations. To quantify data, subjective and entirely etic taxonomies are imposed on the finds. The arbitrary classes so derived are then treated as if they were real and emic entities, even though they reflect the cognitive categories of archaeologists, not of the producers or consumers of the artefacts. ‘Crucial’ indices are perceived, and changes in them are interpreted as indicating some demographic, cultural or ecological changes. Yet there are pronounced sedimentary and various other biases in the preservation of archaeological material, which are likely to be reflected in perceived distributional and compositional patterns of the evidence. These taphonomic biases are not adequately accounted for in the interpretation of data, or in translating the empirical evidence into archaeological interpretation.

The ultimate (long term) purpose of pre-Historic art studies is to explore the processes that have in some way contributed to the formation of human concepts of reality. All human knowledge claims are in terms of these concepts, and if we were to find means of illuminating the origins of anthropocentrism (the interpretation of reality in terms of the material stimuli experienced by humans) we would be likely also to acquire a new understanding  of the limitations it imposes on the human intellect. Such insights may free that intellect from the restrictions imposed by its epistemological limits, in the distant future. The deficiencies of a conceptual model of reality cannot be perceived from within such a model, by uncritical recourse to the biological intelligence that is its own product (Bednarik 1990b). In an anthropocentric system of reality, ideas or mental constructs must adhere to its inherent order not only to be acceptable, but even to be able to be conceived. (An analogue is provided by my present paper, which will have to be rejected by those whose thinking is dominated by epistemological models that grant the human intellect access to objectivity.) In short, palaeoart studies are intricately connected with epistemology; indeed, I see them as providing the raw data on which all epistemology should ultimately be based.

The processes that led to human models of reality are attributable to the frames of reference created by the early cognitive evolution of hominids (Bednarik 1992b). The earliest cosmologies that are ‘accessible’ to us are those that are detectable in the oldest known externalisations suggesting the existence of human consciousness. We may still lack viable methods of dealing systematically with this tenuous link with the origins of our anthropocentrism, but to disparage palaeoart studies by relegating them to a subsidiary role in the humanistic branch of hominid ethology, as some archaeologists would tend to do, implies a grave misapprehension about the purpose of the scientific endeavour.


About the scientific study of rock art

Many archaeologists believe that studies of early art are either of no relevance to ‘proper archaeological practice’, or that they can only be of very limited use in archaeological interpretation. Indeed, the ideas some archaeologists have about the purposes of palaeoart studies seem absurd when one considers the sort of models they have in mind. These include the iconographic identification (i.e. using one’s own cortical processes of matching neural patterns of objects with visual graphic input) of objects supposedly depicted in the arts (such as tools, weapons, animal species, tracks etc.), which is scientifically untenable (cf. Clegg 1983; Tangri 1989; Cameron 1993); or the determination of ‘stylistic’ and other perceived distribution patterns which are seen as encoding various types of demographic, emblemic and cultural information: for instance, tribal boundaries, cultural diffusion, even population densities. Some archaeologists have adopted a distinctly ahistorical approach to rock art, they regard the assemblage of a site as belonging to a single tradition unless otherwise demonstrated (Bednarik 1988). The culturally crucial function of rock art, that of a cultural determinant (Bednarik 1991-92), remains ignored, as does its study as a record of cultural dynamics (rock art re-use, retouch, cultural responses). Statistical approaches are favoured, but they provide a false sense of security, producing unreliable results of generally limited relevance (see below). Thus archaeologists have not only neglected the true scientific potential of palaeoarts, they have utilised them for comparatively trivial purposes, and in schemes for which they are poorly suited.

To begin with, distribution and changes in rock art ‘styles’ over time are not necessarily functions of economic, environmental, cultural, social or even religious factors (Tangri 1989). Neither is apparent stylistic continuity proof for cultural (or any other) continuity. Even if it were it would be of no help, in view of archaeologists’ ambivalence as to where style resides in an artefact (Conkey & Hastorf 1990). Thus direct correlation between ‘quantifiable’ archaeological ‘data’ and rock art poses serious problems, and the lack of reliable dating for nearly all rock art in the world only aggravates these. To make matters worse, the principal tool in the archaeological analysis of rock art, statistics, is scientifically invalid when applied to quantitative rock art data. Taphonomic logic decrees that such data can only be relevant in describing a present state of a corpus of rock art, they are not directly related to the archaeological significance of the art.

To illustrate effects of applied taphonomic logic, I give some well-known examples. The distribution of the Upper Palaeolithic female figurines found in Eurasia is often described as reflecting the distribution of the hypothetical tradition that produced them (Bednarik 1990b). This is most probably a fallacy, because it would be an incredible coincidence if the distribution of that hypothetical tradition coincided with the distribution of sediments suitable for the preservation of these artefacts. Nearly all of them consist of calcium carbonate (dentine or limestone), and all were found in loesses or limestone caves, i.e. in calcite or dolomite-rich, high-pH soils (Bednarik 1992c). Similarly, the apparently exclusive occurrence of Upper Palaeolithic rock art in caves indicates not that it was endemic to caves, but in fact that it was not so (Bednarik 1986: 41). Archaeological misinterpretations, then, are attributable to erroneous determination of the common denominator of phenomenon categories (see below). For instance, a geomorphological common denominator is interpreted as an archaeological one, and it should be self-evident that this can result only in misinterpretation. Therefore taphonomy demands that most interpretation in archaeology must be expected to be false, and this is particularly evident in palaeoart studies (for discussions of taphonomic logic as applicable to palaeoart, see Bednarik 1993, 1994a).

The vastly differing longevities of pigment types, or the changes of colour many pigments experience over time (e.g. all iron oxides and hydroxides used in rock painting pigments are metastable; Bednarik 1979, 1987, 1992d; I have also described white pigments that become completely black, and McNickle 1991 reports black pigments that change to light-green) lead to similar conclusions, as do the effects of petroglyph depth on repatination processes (Bednarik 1979) or on petroglyph longevity, and a variety of similar aspects. It is no coincidence that the oldest rock arts outside of caves regularly consist of deep-red pigment of a particular hue (ochres age towards the haematite phase, which is also the most stable iron pigment, and the one best suited for interstitial retention in sandstones; Bednarik 1992e) or of deeply carved line motifs. For example, most Australian petroglyphs are not engravings, but sgraffiti (not to be confused with graffiti), in which natural cutaneous rock laminae of differing colours rather than groove relief were utilised for contrast. Much of this art cannot be expected to survive for longer than total repatination takes, and what have been seen as ‘chronological’ trends in rock art are far more likely to be attributable to selective survival: the quantitative characteristics of extant rock art are a result primarily of taphonomic processes, and secondarily of art production. Most rock art can be assumed to have been lost over time, which means that the cultural significance of extant statistics is subordinate to their taphonomic significance. The fact that the latter has not been considered so far at more than the most rudimentary level does not bode well for archaeological speculations based on statistics.

The extent of archaeology’s deficiencies can be fathomed by considering that perceived trends are often presented as evolutionary, chronological (by circular argument) or empiricist evidence. In addition to geomorphological biases, many other factors can also greatly distort the statistical characteristics of rock art. Among them are location, recorder’s bias, historical responses to alien iconographies (e.g. iconoclasm), or indeed any process that contributes to the degradation of the art.


The statistical analysis of rock art

Generally speaking, only a remnant sample of the art can have survived at most rock art sites (Bednarik 1992e). We cannot know which quantifiable characteristics of the surviving sample are culturally determined, and which ones are determined by other factors, such as location, type of rock support, or other environment-related circumstances. To illustrate this point one could imagine a cultural convention restricting all motifs of one type to granite surfaces, those of a second type to limestone: after a few thousand years, none of the motifs of the second type may have survived, because exposed limestone surfaces are dissolved by weathering. The statistical characteristics of the surviving sample would be so much influenced by geomorphological selection processes that it would be futile to speculate about their archaeological significance. They would be significantly different from the statistical variables of another site of the same tradition that had been subjected to very different selection processes.

The most serious limitation of statistical analyses in palaeoart research is that posed by the inherent subjectivity of the data. Irrespective of the actual method used, statistics that address the content of rock arts always involve a taxonomy of motif elements, because the grouping of motifs perceived to be similar is a prerequisite for such treatment. Yet any such taxonomising process is entirely based on the iconographic perceptions or graphic and depictive conventions defining the researcher’s own (etic) system of reality, and does not necessarily reflect the artists’ graphic (emic) cognition.

Archaeological empiricists seem to realise part of this problem, but their belief in their own objectivity still allows them to perceive meaningfulness in the arbitrary selection of criteria to create taxonomies. They believe that they can detect and correctly recognise the visual clues hidden in an alien art without recourse to the cognitive or epistemological framework of the culture concerned; that they can perceive ‘meaning’ and identify objects reliably (Bednarik 1990b). It has been demonstrated that the reliable identification of the iconic content of an art is restricted to participants of the culture in question (e.g. Macintosh 1977), yet the most common preoccupation of students of rock art is to muse about the nature of the objects depicted in the art, and to base chronologies and interpretations of meaning on their subjective notions. Not only are animal species confidently identified, many other aspects of the pictures are interpreted in a seemingly ethnocentric fashion. For instance, motifs of the Franco-Cantabrian ‘cave art’ are often interpreted as pregnant, falling, dead or dying, and so forth, just as depictions in other arts are prone to be pronounced as dancing, praying, adoring, flying, worshipping etc. In general, these interpretations are based on the assumption that depictive conventions in the arts of pre-Historic or ethnographic peoples are identical to those determining the enculturated perceptions as well as the cognitive abilities and limitations of the contemporary observer. It may well be true that the modern beholder can in certain cases correctly interpret iconographic aspects of palaeoart, but this is not the issue. It would need to be demonstrated that the cortically established patterns of detecting iconicity are identical in the mute artist and the contemporary observer, i.e. that the past and present strategies for detecting and interpreting visual clues in a marking are similar. (Iconicity refers here to the visual quality of a motif which conveys to most contemporary observers, especially Westerners, that a specific object is depicted. It is merely a subjective definitional tool.) Otherwise such interpretations would be no more susceptible to falsification than other archaeological interpretations.

It would initially appear that non-figurative art, consisting of ‘geometric’ arrangements, is more accessible to statistical analysis than figurative art. In fact the opposite is true, because while there is some merit in assuming that the observer shares iconic perceptions of the artist, that cannot be verified in non-iconic art. No two rock art motifs are identical in every respect, yet in order to prepare a site’s assemblage for statistical treatment, apparently similar motifs must be grouped and considered together. Without this process no statistical treatment of any corpus of rock art is possible. Each and every motif possesses thousands of characteristics, and the analyst must decide which ones are to be considered. Otherwise there will inevitably be as many motif types as there are motifs at any one site. Among the characteristics available for selection are metrical indices, qualitative or formal aspects, aspects of the motif’s relationship to the site, to other motifs at the site, to the motifs at other sites, to the topography of the site, to petrological or past vegetational or hydrographic aspects of the site or of the surrounding landscape, spatial or syntactic context, the identity or status of the artist (is a circle engraved by a man the same motif type as one engraved by a woman, and if not, how would an archaeologist detect the difference?). Not only can single elements have wide ranges of ‘meaning’, definable only in terms of pre-Historic context (consider Munn 1973), the number of characteristics of each motif is practically unlimited. Many of those characteristics that defy archaeological definition may be crucial in identifying motif types (e.g. the sex of the artist). Thus the parameters the analyst chooses will inevitably reflect her/his personal, cultural, historical, ethnocentric, academic and cognitive biases and limitations. Consequently the information so derived is useful only in studying the analyst’s own culture and cognition, and in studying the way in which s/he applies these in examining the surviving graphic traces of alien cognitive systems to which s/he has no cognitive access. However, that information is not scientifically relevant to the study of the pre-Historic culture concerned, except for limited descriptive purposes. It cannot be expected to provide valid data for statistical analysis, or valid interpretations of the palaeoart.


About science and epistemology

Everything said so far in this paper implies a rejection of interpretational approaches to palaeoart. One can selectively gather data about rock art and portable art, one can compare these, speculate about them, generate an endless variety of hypotheses from them (cf. Marshack 1989: 17), discuss the individual merits and probability ratings of these, but no amount of iconographic or statistical information provides falsifiable (Popper 1934) or refutable (Tangri 1989) interpretations.

It seems almost inevitable that palaeoart studies will share the fate of ecological/materialist or empiricist archaeology (cf. Lewis-Williams 1993), but I will argue here that the opposite applies. Some archaeologists have traditionally treated rock art research in a condescending manner, because its cultural and idiosyncratic dimensions, i.e. its essentially human qualities, make it so difficult to extract what is considered to be ‘valid archaeological information’ from it. Yet it is precisely these much derided human aspects that permit truly scientific access to early art.

Philosophically, this should have been self-evident. All phenomena of the physical world are made up of large numbers of variables, of which humans can only detect those their sensory faculties and intellect allow them to perceive (Bednarik 1984: 29). From them they select what I call ‘crucial common denominators of phenomenon categories‘, which are the basis of all cosmological taxonomies. Their selection is not determined by objective criteria in terms of how things really are in the world, but by the ‘anthropocentrising’ dynamics of hominid reality building processes: by how phenomena were interpreted and integrated into a system of understanding based entirely on human faculties. Since the latter were derived from human evolution, which was never in terms of defining cosmic reality but in terms of survival value, they must be assumed to provide only a narrow spectrum of objective reality. Consequently, scientific constructs of reality cannot be expected to reflect real and objective reality.

There are, however, exceptions to this rule: a phenomenon that is created by humans specifically for the purpose of relating to a human sensory faculty can only consist of crucial variables determining its phenomenological externalisations which are accessible to human perception. Art is such a phenomenon; there can be no crucial common denominator for phenomenon categories in art that is entirely inaccessible to humans. Therefore art is the only phenomenon in the real world that can be shown to provide human access to its crucial variables. Indeed, one can invert this postulate by defining art as the collective phenomena in human experience about which we can argue objectively. This is a philosophically sound definition of art (cf. Bednarik 1991).

This truism explains readily how humans attained their unique neural structures of relating to the world, i.e. their humanness, through art (‘art’ as defined here). The introduction of phenomena consisting only of humanly perceptible variables (earliest mark production) would have prompted several developments: ‘intentional’ marks (consider, for example, those resulting from rhythmic manipulation of tools) would have rendered perceived reality ‘conceptually manageable’, by providing complete sets of percepts rather than fragmentary ones, promoting visual and mental taxonomising processes and the inclusion of the new neural structures in cybernetic feedback systems. ‘Conscious experience’, or rather, what we understand by it, became possible because the neural structures prompted by the early art production became available for the processing of stimuli of the non-artificial material world, in a taxonomising format. This explains why the ultimate result, humanly perceived reality, is in the final analysis determined by art, and why it is at the same time valid and inadequate. As I have noted elsewhere (Bednarik 1992b), a false, cultural cosmology or epistemological model can be formed and maintained indefinitely by a biologically intelligent (Jerison 1973) species.



If the artificiality of art permits us an empirical access, at least theoretically, that is not available for natural phenomena, ‘direct’ scientific access to palaeoarts may be similarly feasible. Assuming that the human faculties that are involved in present-day art production (especially the visual system) have remained physiologically similar during recent human evolution (Deregowski 1988), some heuristic access should be similarly possible to pre-Historic arts. (This is the ‘neurological bridge’ I proposed in the early 1980s; it has never been discussed, and archaeologists have been more fascinated by the pseudo-scientific confirmationist arguments about shamans, a most unfortunate offshoot of my work; Bednarik 1990c.) Other physiological factors affecting art production would be the tactile ability of the fingers, the competence of performing the so-called precision grip, and that of precise and well-controlled arm movement. A precondition for producing art (other than finger marks and stencils) must have been to procure, hold and direct marking instruments with confidence and precision, involving not only motor skills, but also the co-ordination of tool-mediated skills with the operation of the visual system (Marshack 1984). A proficiency in tool making provided two prerequisites for art production: precision of vision-oriented motor skills, and the experience of observing the effects of altering crucial aspects of the perceptible environment through participation. These factors would have contributed to establishing neural structures that would prompt the behavioural patterns producing early ‘art’ (Bednarik 1994b).

The scientifically most reliable access to palaeoart is therefore likely to be via non-conscious universals. The above epistemological model postulates the existence of such universals, and aspects of them are likely to be detectable by humans. In any comparison of universals in art we will first need to establish whether the proposed common denominator can reasonably be assumed to have applied at the time the palaeoart in question appears to have been produced. Subject to this proviso, I propose that it should be possible to detect in archaic art universals that can be identified, studied and quantified in contemporary art production (e.g. Arnheim 1964; Berger 1972; Getzels and Csizszentmihalyi 1976; Levy and Reid 1976) under conditions of full refutability. To be eligible for consideration, universals must be unrelated to such factors as cultural experience, cognitive systems or individual preferences. Universals that may be susceptible to conscious variation, or that may themselves be the product of long-established traditions of art production (such as the canons of aesthetics), are of no value to us here.

I emphasise that, because of the special position of art in the human construct of reality, it is far more accessible to scientific examination than any other external stimulus in the human experience, and any universals in it are fundamentally more reliable than the known variables of any other external stimulus known to us: the phenomenon they define exists wholly in the realm of human experience, it has no dimensions of reality other than the ones we as humans can know. Hence the universals I refer to (I emphasise that this concept of universals is far more sophisticated than, and not connected to, notions of shamanic or epigraphic ‘universal’ connotations) may offer an epistemologically sound approach to rock art. Together with the greatly effective analytical tool of taphonomic logic, such approaches promise a scientific rigour in the palaeoart studies of the twenty-first century that is probably unattainable by conventional archaeology.

Robert G. Bednarik
IFRAO Convener
P.O. Box 216
Caulfield South, Vic. 3162



ARNHEIM, R. 1964. Art and visual perception. University of California Press, Berkeley.
BAHN, P. G. 1990. Motes and beams: a further response to White on the Upper Paleolithic. Current Anthropology 32: 71-76.
BEDNARIK, R. G. 1979. The potential of rock patination analysis in Australian archaeology – part 1. The Artefact 4: 14-38.
BEDNARIK, R. G. 1984. The nature of psychograms. The Artefact 8: 27-32.
BEDNARIK, R. G. 1985. Editor’s response to letter by B. J. Wright. Rock Art Research 2: 90-91.
BEDNARIK, R. G. 1986. Parietal finger markings in Europe and Australia. Rock Art Research 3: 30-61.
BEDNARIK, R. G. 1987. No pictographs at end of Rochester Creek rainbow. La Pintura 15(2+3): 14-18.
BEDNARIK, R. G. 1988. Comment on: ‘Rock art sequences, a matter for clarification’, by F. D. McCarthy. Rock Art Research 5: 35-38.
BEDNARIK, R. G. 1990a. Comment on: ‘!Science, hypothesis testing and prehistoric pictures’, by D. Tangri. Rock Art Research 7: 67-69.
BEDNARIK, R. G. 1990b. More to Palaeolithic females than meets the eye. Rock Art Research 7: 133-37.
BEDNARIK, R. G. 1990c. On neuropsychology and shamanism in rock art. Current Anthropology 31: 77-80.
BEDNARIK, R. G. 1991. Standardisation in rock art terminology. In Rock art and posterity, C. Pearson and B. K. Swartz (eds.), pp. 116-18. Occasional AURA Publication 4. Australian Rock Art Research Association, Melbourne.
BEDNARIK, R. G. 1991-92. Rock art as cultural determinant. Survey 5/6: 11-20.
BEDNARIK, R. G. 1992a. Palaeoart and archaeological myths. Cambridge Archaeological Journal 2(1): 27-43.
BEDNARIK, R. G. 1992b. On Lower Paleolithic cognitive development. In S. Goldsmith, S. Garvie, D. Selin and J. Smith (eds), Ancient images, ancient thought: the archaeology of ideology, pp. 427-35. Proceedings of the 23rd Annual Chacmool Conference, University of Calgary.
BEDNARIK, R. G. 1992c. Natural line markings on Palaeolithic objects. Anthropologie 30(3): 1-8.
BEDNARIK, R. G. 1992d. Mehr ueber die rote Farbe in Vorgeschichte. Almogaren 23: 179-89.
BEDNARIK, R. G. 1992e. Prioritetnye napravleniya v rabotakh po konservatsii pamyatnikov naskal’nogo iskusstva. In Naskal’nye risunki evrazii, Pervobytnoe iskusstvo, pp. 8-13. Novosibirsk ‘Nauka’, Sibirskoe otdelenie, Novosibirsk.
BEDNARIK, R. G. 1993. Who’re we gonna call? The bias busters! In M. Lorblanchet and P. Bahn (eds), Rock art studies: the post-stylistic era, or where do we go from here? Oxbow Books, Oxford.
BEDNARIK, R. G. 1994a. A taphonomy of palaeoart. Antiquity 68: 68-74.
BEDNARIK, R. G. 1994b. Art origins. Anthropos 89: 169-80.
BERGER, J. 1972. Ways of seeing. BBC and Penguin Books, London.
CAMERON, D. W. 1993. The archaeology of Upper Palaeolithic art: aspects of uniformitarianism. Rock Art Research 10: 3-17.
CLEGG, J. 1983. Recording prehistoric art. In G. Connah (ed.), Australian field archaeology: A guide to techniques, pp. 87-108. Australian Institute of Aboriginal Studies, Canberra.
CONKEY, M. W. and C. A. HASTORF (eds) 1990. The uses of style in archaeology. Cambridge University Press, Cambridge.
DEREGOWSKI, J. B. 1988. The eye, pictures and cognition. Paper presented to Symposium L, First AURA Congress, Darwin.
FEYERABEND, P. K. 1975. Against method. Verso Editions, London.
FLETCHER, R. 1991. Refutation and tradition: an uneasy relationship. Australian Archaeology 33: 59-62.
GETZELS, J. W. and CSIZSZENTMIHALYI, M. 1976. The creative vision: a longitudinal study of problem solving in art. Wiley, New York.
HODDER, I. 1984. Archaeology in 1984. Antiquity 58: 25-32.
HODDER, I. 1986. Reading the past. Cambridge University Press, Cambridge.
HODDER, I. 1990. Style as historical quality. In M. W. Conkey & C. A. Hastorf (eds), The uses of style in archaeology, pp. 44-51. Cambridge University Press, Cambridge.
HUCHET, B. M. J. 1991. The nature of analogies in Australian archaeology. The Artefact 14: 3-12.
HUCHET, B. M. J. 1992. Processual archaeology: discrepancies between norms and performances. Norwegian Archaeological Review 25: 111-19.
JERISON, H. J. 1973. Evolution of brain and intelligence. Academic Press, New York.
KUHN, T. S. 1970. The structure of scientific revolutions (2nd edition). University of Chicago Press, Chicago.
LEVY, J. and M. REID 1976. Variations in writing posture and cerebral organization. Science 194: 37-39.
LEWIS-WILLIAMS, J. D. 1993. Southern African archaeology in the 1990s. South African Archaeological Bulletin 48: 45-50.
MACINTOSH, N. W. G. 1977. Beswick Creek Cave two decades later: a reappraisal. In P.J. Ucko (ed.), Form in indigenous art, pp. 191-97. Australian Institute of Aboriginal Studies, Canberra.
McNICKLE, H. P. 1991. A survey of rock art in the Victoria River District, Northern Territory. Rock Art Research 8: 36-46.
MARSHACK, A. 1984. The ecology and brain of two-handed bipedalism: An analytic, cognitive and evolutionary assessment. In H. L. Rotitblat, T. G. Bever and H. S. Terrace (eds), Animal cognition, pp. 491-511. Lawrence Erlbaum, Hillsdale, N.J..
MARSHACK, A. 1989. Methodology in the analysis and interpretation of Upper Palaeolithic image: theory versus contextual analysis. Rock Art Research 6: 17-38.
MUNN, N. D. 1973. Walbiri iconography. Cornell University Press, Ithaca.
ODAK, O. 1993. The birth of a new scientific discipline. Rock Art Research 10: 114-25.
POPPER, K. 1934. Logik der Forschung. Julius Springer Verlag, Vienna.
SHANKS, M. and C. Tilley 1987. Social theory and archaeology. Plenum Press, London.
TANGRI, D. 1989. !Science, hypothesis testing and prehistoric pictures. Rock Art Research 6: 83-95.
THOMAS, N. 1982. In lieu of a critical self-consciousness. Australian Archaeology 14: 1-5.
TILLEY, C. 1993. Material culture and text: the art of ambiguity. Cambridge Archaeological Journal 3: