Pleistocene rock art and essentialism

The idea that rock art equals humanity and humanness remained implicit in rock-art studies even after the very early dates obtained in Australia (Noble & Davidson Reference Noble and Davidson1991) shook the Eurocentric narrative of European origins (e.g. Bahn & Vertut Reference Bahn and Vertut1997; Gamble Reference Gamble and Foley1984; Jochim Reference Jochim and Bailey1983; Reference Jochim and Soffer1987; Leroi-Gourhan Reference Leroi-Gourhan1965; Mellars Reference Mellars, Price and Brown1985; Reference Mellars1989; Reference Mellars2006) that was already under fire (Conkey & Williams Reference Conkey, Williams and DiLeonardo1991; Gamble & Gittins Reference Gamble, Gittins, Meskell and Preucel2004; McBrearty & Brooks Reference McBrearty and Brooks2000, 543).

In our view, recent research has reinforced the biological coupling of rock art with a privileged set of cognitive abilities, rather than expanding the array of potential explanations of this cultural practice. We will focus on two examples: the debate around Neanderthal authorship of rock-art depictions in different caves in Spain (Hoffman et al. Reference Hoffmann, Standish and Garcia-Diez2018a; White et al. Reference White, Bosinski and Bourrillon2019), and the recently published ancient rock art found in Southeast Asia (Brumm et al. Reference Brumm, Oktaviana and Burhan2021; Taçon et al. Reference Taçon, Tan and O'Connor2014).

Recent Uranium-series dating of calcite crusts in spatial relationship with rock art has been published for some few specific images in the Spanish caves of Ardales, Maltravieso and La Pasiega (Hoffman et al. Reference Hoffmann, Standish and Pike2018b). Red spots, a hand stencil and a geometric sign were found covered by calcite crust, dated to older than 60,000 years ago. This chronology seems to support the idea of a Neanderthal authorship since there is no Homo sapiens presence attested in Europe for this chronology (Slimak et al. Reference Slimak, Zanolli and Higham2022). The dates, however, have been questioned due to several methodological issues and discrepancy with the previous known archaeological record (Aubert et al. Reference Aubert, Brumm and Huntley2018a; Pearce & Bonneau Reference Pearce and Bonneau2018; Ramos-Muñoz et al. Reference Ramos-Muñoz, Cantalejo and Blumenröther2022; Slimak et al. Reference Slimak, Fietzke, Geneste and Ontañón2018; White et al. Reference White, Bosinski and Bourrillon2019). Beyond potential methodological pitfalls, what is of interest here is the way the supporters of the Neanderthal hypothesis have made their case: specifically, through claims that Neanderthals have been historically considered as inferior to modern humans (Hoffmann et al. Reference Hoffmann, Standish and Garcia-Diez2018a, Reference Hoffmann, Standish and Pikeb; Reference Hoffmann, Standish and Garcia-Diez2019). Therefore, the fact that they painted the walls of caves would demonstrate that they were cognitively as advanced as anatomically modern humans and therefore fully human. Thus, despite the alleged novelty of the findings, the Neanderthal discussion is in fact pushing back into the deep past the cognitivist and origins paradigm for rock-art research and reinforcing the idea of a unique and continuous trajectory of this practice into subsequent time periods, a type of biological determinism that reproduces the traditional tautological arguments about humanity and rock art.

In Southeast Asia, the application of the Uranium series dating method (Brumm et al. Reference Brumm, Oktaviana and Burhan2021) has led to a shift in scientific knowledge. In spite of problematic experiences concerning methodology (Plagnes et al. Reference Plagnes, Causse, Fontugne, Valladas, Chazine and Fage2003), the significant number of published dates (around 50) and their consistency are undisputable (see Figure 1), something not yet achieved in Europe. In Sulawesi, calcite layers covering animals and hand stencils painted in nine different sites have been dated. The oldest dates, related to a pig depiction in Leang Tedongnge cave, offer a date older than 45,000 years. In Leang Balangajia cave the estimate is younger than 73,000 years (Aubert et al. Reference Aubert, Brumm and Ramli2014; Reference Aubert, Lebe and Oktaviana2019; Brumm et al. Reference Brumm, Oktaviana and Burhan2021). Similar results have been obtained at six sites in Borneo, using the U/Th dating system; dates are older than 40,000 years for an animal painting and younger than 51,000 years for a hand stencil in the same panel in Lubang Jeriji Saléh cave in Borneo, always separated from Sulawesi by a significant sea-crossing (Aubert et al. Reference Aubert, Setiawan and Oktaviana2018b).

Figure 1. U/Th dating results in Southeast Asia. Intermittent lines are before and after dates for the same representation (dates from Aubert et al. Reference Aubert, Brumm and Ramli2014; Reference Aubert, Setiawan and Oktaviana2018b; Reference Aubert, Lebe and Oktaviana2019; Brumm et al. Reference Brumm, Oktaviana and Burhan2021).

These findings could show a homogeneous style for animals that also includes hand stencils—and humans in some cases—probably dating between 50,000 and 44,000 bp. These dates are similar to those available for archaeological sites associated with anatomically modern humans in Island Southeast Asia (Hawkins et al. Reference Hawkins, O'Connor and Maloney2017; Kaharudin et al. Reference Kaharudin, Alifah, Ramadhan and Kealy2020). It seems that the presence of modern humans in the wider region can be attested even earlier (Westaway et al. Reference Westaway, Louys and Awe2017), while the expansion of earlier hominin species would have reached Timor and New Guinea (Shipton et al. Reference Shipton, O'Connor and Kealy2021). Recent research suggests similar imagery sharing the same stylistic features in Australia, where images that appear to have been made according to similar formal conventions in Kimberley IIAP are dated to between 13,000 and 17,000 years ago, using wasp nests as an indirect dating method by ¹⁴C-AMS (Finch et al. Reference Finch, Gleadow and Hergt2021); Malaysia, where the Gua Tambun rock-shelter shows animals with a convention of infill lines (Taçon et al. Reference Taçon, Tan and O'Connor2014); and China, where basal rock art in parietal stratigraphy in the Jintsa River is dated older than 5000 years ago (Taçon et al. Reference Taçon, Aubert and Li2012), even around 13,000 in some cases (Wu et al. Reference Wu, Jiao and Ji2022) and, once more, infill lines are used as a formal convention. It could also be possible to track these painting features to the Kosum 1 site, India (Dubey-Pathak & Clottes Reference Dubey-Pathak and Clottes2021), although dates are not available for this region yet.

Even if destabilizing to the Eurocentric ‘master narrative’ of origins, these findings have reinforced the idea that rock art emerged during the Pleistocene in association with the global expansion of anatomically modern humans (Aubert et al. Reference Aubert, Brumm and Taçon2017). Questions such as which were the social contexts of the construction of walls, places and landscapes through rock art; if rock art originated in one particular area or in separate, independent processes; if rock-art making lasted uninterruptedly since the earliest chronologies around 50,000 to 17,000 bp or was produced in independent pulses, are mostly missing. Rather, the Southeast Asian corpus of rock art is used to testify to the dominant narrative that implies the inevitable character of rock art, almost lying in the biological rather than the social realm.

The old chronologies of rock art in Europe and Southeast Asia, however, set us thinking in a different direction. In Europe, Homo sapiens came into contact with Neanderthals. In Southeast Asia, they came into contact with either or both Homo floresiensis or Denisovans. Rock art seems to have emerged in contexts of potential cohabitation between Homo sapiens and other human species. This is just one way to approach a contextual and historical frame of Pleistocene rock art, going beyond the ‘early-centric’ approach to rock art and the biologically based, intellectual abilities of humans.

Holocene traditions

Pleistocene rock art is a very small percentage of the prehistoric rock art preserved across the world. Indeed, Pleistocene rock art is more of an exception than a rule (Renfrew Reference Renfrew, Renfrew and Morley2009). The fact that Palaeolithic rock art does not seem to present continuity into the Neolithic, generally speaking, leads to the conclusion that, in all likelihood, the general lack of Pleistocene rock-art corpora is not solely due to shortcomings in methodologies or research activity. As John Robb (Reference Robb2015, 640) states, ‘there is both a quantum leap in art at the advent of the Neolithic and a steady rise throughout later prehistory’. Robb refers here to all types of prehistoric art, a problematic conflation (see below), but a relevant point as the surprising increase of rock-art making in recent prehistory in many different parts of the world is typically ignored in most discussions (Cruz Berrocal & Millerstrom Reference Cruz Berrocal and Millerstrom2009). In spite of strong historiographical traditions to the contrary, most forager rock art would have been made during the Holocene, largely contemporary with agriculturalist societies which also made rock art (Davidson & Nowell Reference Davidson, Nowell, Davidson and Nowell2021). Two brief examples illustrate this point.

Among the well-known different rock-art traditions in the southern zone of Africa (Smith & Ouzman Reference Smith and Ouzman2004), the San tradition ascribed to hunter-gatherer societies is the most abundant and best researched, normally assumed to be the inheritance of an ancient past, which means different time spans for different scholars (e.g. Le Quellec Reference Le Quellec, Bahn, Franklin, Strecker and Devlet2016 on the ‘chronophobic’ debate), especially because of the strong imbrication of rock-art research in South Africa with ethnographic evidence from the nineteenth and twentieth centuries. In any case, dating is still uncertain.

In Reference Jerardino and Swanepoel1999, Jerardino and Swanepoel published the then oldest minimum age of alleged wall paintings at Steenbokfontein Cave (Western Cape, South Africa) at 3600 bp, based on marine shell contained in the archaeological sediment. The paintings had been made on slabs that fell off the wall, and while the authors assumed that they were made prior to collapse, there is nothing in the evidence to indicate such a relative chronology. Indeed, the preserved paintings on the walls are completely different from the paintings on the slabs. Thackeray (Reference Thackeray1983) published spalls containing traces of paint in deposits dating as far back as 10,000 years bp while Walker (Reference Walker2012) also provided another such example of painted rock spalls potentially dating to 10,000 cal. bp. This kind of indirect dating, and the evidence provided, do not allow for any assessment of the dates. They seem to be discarded from the discussion at this point (see Le Quellec's (Reference Le Quellec, Bahn, Franklin, Strecker and Devlet2016) synthesis of all the dates for South African paintings).

In the Cederberg Mountains of the Western Cape Province, South Africa, Van der Merwe et al. (Reference Van der Merwe, Sealy and Yates1987) obtained one AMS direct radiocarbon date on a painting from Sonia's Cave Upper, Boontjieskloof, Clanwilliam district: 500 ± 140 cal. bp. In the Drakensberg Escarpment of KwaZulu-Natal, South Africa, Mazel and Watchman (Reference Mazel and Watchman1997) obtained a date from a plant fibre embedded within paint at Esikolweni Shelter in the Natal Drakensberg, dating to 330 ± 90 cal. bp. Further samples from carbon-bearing oxalate from Barnes Shelter (1060 ± 65 bp), Main Caves North (spanning 2310 ± 70 bp to 2900 ± 80 bp, with two dates, 2900 and 2760 bp, obtained from crusts overlying the same motif), Highmoor 1 (2310 ± 70 bp), White Elephant Shelter (1930 ± 65 bp) and Maqonqo Cave (3720 ± 100 bp) were published later (Mazel & Watchman Reference Mazel and Watchman2003). The latest date, for Barnes Shelter, derives from oxalates covering the paint. The earliest date, for Maqonqo Cave, derives from the base of the encrustation and the authors assume that the ‘painting overlying the crust date is probably hundreds of years younger than the crust itself, and in all probability postdates 3700 years ago’ (Mazel & Watchman Reference Mazel and Watchman2003, 66). Bonneau et al. (Reference Bonneau, Brock, Higham, Pearce and Pollard2014) published three AMS dates obtained from black pigments from RSA TYN2 (Nomansland, South Africa) spanning from 2072 ± 28 bp and 2100 ± 40 bp. More recently, Bonneau et al. (Reference Bonneau, Pearce and Mitchell2017) have reported direct AMS ¹⁴C dating of rock paintings at three sites in the Thune Dam (southeastern Botswana), five sites in the Metolong Dam (western Lesotho) and six sites in the Maclear District (Drakensberg, South Africa's Eastern Cape Province). Results show 12 dates for the paintings of Lesotho ranging from 300 ± 65 cal. bp to 5700 ± 2000 cal. bp. The authors discard most of the dates and synthesize the chronology of Lesotho rock art as ‘2326–965 cal. bp’ (Bonneau et al. Reference Bonneau, Pearce and Mitchell2017, 331). For the Maclear District of South Africa they synthesize 22 dates ranging from 124 ± 23 cal. bp to 2690 ± 100 cal. bp as dates of ‘2998–2381 cal. bp’. For Botswana, they see dates from the site TD12 as the oldest evidence for extant painting on rock-shelter walls anywhere in southern Africa (Bonneau et al. Reference Bonneau, Pearce and Mitchell2017). They obtained nine dates ranging from 1250 ± 80 bp to 4500 ± 260 cal. bp but synthesize them as ‘5723–4420 cal. bp’. That is, they take the oldest of the dates obtained, which shows a standard deviation of ±260 years, to make their case. The next oldest date they obtained is 3060 ± 30 cal. bp, a much more acceptable date from a methodological point of view, yet still the oldest date for the south of Africa so far. Thus, most of the dates obtained so far are around 2000 years ago or younger.

The questions of when and how food-producing economies arose in the southern part of Africa are equally unclear so far. Sadr (Reference Sadr2015) contends that livestock appeared in southern Africa over 2000 years ago, and while the exact mechanisms are not clear (a large migration from East Africa, trade down-the-line among hunter-gatherer communities, a long history of diverse small-scale population movements), he proposes two ‘infiltration’ episodes, one along the Atlantic seaboard and a second through the Limpopo River Basin. Overall, he prefers to speak of multiple, small-scale infiltrations of domestic animals (and food producing) into southern Africa. Recent research has confirmed the presence of domesticated sheep at the site of Spoegrivier (Namaqualand, South Africa) dated to 2105 ± 65 bp (Coutu et al. Reference Coutu, Taurozzi, Mackie, Jensen, Collins and Sealy2021). Their sampling focused on the oldest domestic specimens recognized so far, but has not been systematic, and the authors indeed point out the problems for identification of domesticated versus wild fauna. Sadr (Reference Sadr2015) gathers more chronological information of domestic livestock in the Leopard Cave (Namibian coast), with dates between 2270 ± 40 and 2190 ± 40 cal. bp, Toteng 1 (Kalahari Drainage) with dates of 2070 ± 40 cal. bp and 2020 ± 40 cal. bp, and Blombos (southern coast of South Africa), with a date of 1960 ± 50 cal. bp. Other innovations such as particular lithic features appear to have preceded the appearance of livestock in these areas.

Given the evidence, it is legitimate to suppose that large-scale processes such as those pointed out by Sadr may have been at stake here, for a longer time-span than the archaeological appearance of deposits of domestic sheep seems to suggest. This would imply a long familiarity of foragers at least with particular elements of a food-producing system. Indeed, the general contemporaneity of rock art and the evidence of livestock so far is quite striking, and tends to dispute the idea that the makers of the rock art were completely alien to food production.

In South America, extremely old dates from Serra da Capivara (Piauí, Brazil) have been called into question (Steelman et al. Reference Steelman, Rickman, Rowe, Boutton, Guidon and Jakes2002). Although scholarship assumes that rock art dates back to the early Holocene, the available dates are scarce (Troncoso et al. Reference Troncoso, Armstrong, Basile, David and McNiven2017). Neves et al. (Reference Neves, Araujo, Bernardo, Kipnis and Feathers2012) have presented the ‘oldest, indisputable testimony of rock art in the Americas’ (Reference Neves, Araujo, Bernardo, Kipnis and Feathers2012, 3) so far: a petroglyph engraved on the bedrock in Lapa do Santo (Brazil) dated by the ¹⁴C method of a stratigraphic layer to ‘a minimum age of 9,370 ± 40 bp (cal. bp 10,700 to 10,500) … the figure was found approximately 30 millimeters below a hearth dated to 9,470 ± 50 bp (cal. bp 11,060 to 10,580)’ (Reference Neves, Araujo, Bernardo, Kipnis and Feathers2012, 2).

The samples pertaining to the dating of the petroglyph are not accurately identified in the text (Neves et al. Reference Neves, Araujo, Bernardo, Kipnis and Feathers2012, supplementary table S1) and no information is provided for the origin of the date on ‘charred material’. The published dates are consistent, but at least in one case an older date appears higher in the stratification. The petroglyph, on the other hand, was found in an exceptional location within the shelter, potentially at the height of the floor. It stands so far as a unique, isolated and unspecific depiction in the context of the Nordeste Tradition of rock art (Morales Reference Morales2002), whose dates have been limited to c. 3000–2000 cal. bp (Rowe & Steelman Reference Rowe and Steelman2003), based on the best studied area, the Serra da Capivara. That is, there are no parallels for this depiction so far.

All other old dates for South American rock art are also based on different forms of indirect dating. In the best-known case, Hornillos 2 (Argentina), ¹⁴C dates of c. 11,000–10,000 cal. bp were obtained from an archaeological layer where a wood carving of a camelid and grinding stones with pigments were recovered (Yacobaccio et al. Reference Yacobaccio, Solá, Alonso and Clottes2012). This would imply that the found mortars were directly related with the making of the paintings, and that pigments were not used for any other activity. This is also the case in the recent example of Motumachay (Pérez Maestro & Bueno Ramírez Reference Pérez Maestro and Bueno Ramírez2022). In another well-known case, the site of Inca Cueva (Argentina), the chronological assessment (10,600–8900 cal. bp) was proposed based on the assumption that the dated stratum (9900 ± 200, 9230 ± 70, 10,620 ± 140, 9650 ± 110 cal. bp; no identification of the charcoal samples is provided), was the floor when the paintings were executed (Podestá & Aschero Reference Podestá, Aschero and Clottes2012). The remaining evidence is based on painted flakes within the stratum. Since these shelters were used recurrently for most of the Early and Middle Holocene and are the scenario of periodic water overflow, this kind of indirect association may remain underdetermined. Recently, rock art from Huenul 1 Cave in Patagonia (Argentina) has been directly dated starting at 8200 cal. bp (Romero-Villanueva et al. Reference Romero-Villanueva, Sepúlveda, Cárcamo-Vega, Cherkinsky and de Porras2024). This is a more accurate date, yet reinforcing a mid-Holocene chronology.

Other lines of evidence for the late Pleistocene/early Holocene chronology of rock-art paintings in the region are based on the alleged depiction of extinct fauna (Morcote-Ríos et al. Reference Morcote-Ríos, Aceituno, Iriarte, Robinson and Chaparro-Cárdenas2021; Paunero Reference Paunero and Clottes2012). The timing of the extinctions of megafauna, as well as the landscape implications involved, are still uncertain. Interestingly enough, Morcote-Ríos et al. (Reference Morcote-Ríos, Aceituno, Iriarte, Robinson and Chaparro-Cárdenas2021) relate this early human presence (and rock-art making) with potential early management of forest resources, which would have had an enduring impact on establishing the tropical forest. In this context, the authors suggest that rock-art making would have been a strategy to create a cultural landscape related to territorial control, ritual practice and the development of social networks. If early Holocene dates for rock art are finally confirmed, it could be the case that the artists did not belong to hunter-gatherer societies (in the more traditional and limited sense), but were plant managers and even pre-domesticators. For the South Andes, widely accepted chronologies for widespread rock-art making have been assigned to a chronological horizon of c. 3000 and 1000 bce, associated with increasing social complexity, demographic growth, the appearance of specialized technologies, the intensification of practices of exploitation of plant resources and a reduction in the mobility of these populations (see Troncoso et al. Reference Troncoso, Armstrong, Basile, David and McNiven2017). Direct dates on rock art for South America so far go back only to c. 1600 bce in La Placa 5 and Covacha Pintada (North Central Chile) (Troncoso et al. Reference Troncoso, Moya, Sepúlveda and Carcamo2015).

Despite the long-standing epistemological significance of early, well-preserved and aesthetically striking (to Western eyes) cave art of the European Upper Palaeolithic, it turns out that a significant number of rock-art corpora and traditions in the world (Fig. 2) have relatively recent dates, mostly well after the early Holocene, and were made by food-producing societies (Cruz Berrocal & Millerstrom Reference Cruz Berrocal and Millerstrom2009). It could be possible to argue against this that this situation is simply due to the much more widespread distribution of farmers. But we argue that this prevalence of farmers’ rock art is not only a function of demography, but has to do with particular social and historical conjunctions. In this context, the known interruptions of the traditions of rock-art making are extremely significant.

Figure 2. Pleistocene hunter-gatherer rock art versus Holocene rock art from hunter-gatherer and food-producing societies (authors’ own elaboration, based on Anati Reference Anati1984; Bednarik Reference Bednarik and Clottes2012; Fritz et al. Reference Fritz, Tosello, Barbaza and Pinçon2017).

A discontinuous history

The continuity/discontinuity of rock-art making during the European Palaeolithic is not well understood (Davidson Reference Davidson2023); at this point, it is not possible to confirm if it was constantly made throughout the entire period or periodically abandoned for millennia. As Conkey and Fisher (Reference Conkey and Fisher2020) rightly point out, there is something fundamentally wrong in looking at it as an uninterrupted practice over millennia and continent-wide. However, this is the way in which many archaeologists, anthropologists, and art historians have approached rock art since the end of the nineteenth century. This explains, for instance, the persistence (until recently) of monolithic theories about the meaning of rock images, such as shamanism.

To counteract the acritical universalist view of rock art as a mere product of enlarged cognitive abilities, one valuable research approach is to study the many discontinuities in rock-art making. They make visible those features that are potentially unsustainable or dysfunctional in a process of social change. The examination, rather than avoidance, of the discontinuities (Davidson Reference Davidson2023) may highlight the great social significance (to the artists, to the observers, to those who could know but not observe) of rock art in particular contingent historical contexts; traditional teleological narratives built on linearity, uniformitarian principles and ahistorical implicit assumptions, can be weakened. This strategy puts the emphasis on ‘negative evidence’ or ‘absence’, a very particular kind of data that usually go under the radar. A focus on the temporal and geographical discontinuities of rock art, what we might call the ‘presence of absence’, is a null hypothesis that challenges the existence of rock-art making as a given. The Pacific provides an example.

In Taiwan, the alleged cradle of Austronesian expansion (e.g. Blust Reference Blust2019), only four rock-art sites are known—in Wanshan, in the south Central Mountains. Moreover, they feature only weak and unclear connections to the large rock-art corpus to the east (O'Connor et al. Reference O'Connor, Louys, Kealy and Mahirta2015). Jalandoni et al. (Reference Jalandoni, Taçon and Haubt2019) have shown that only sporadic rock art is to be found in Micronesia. In the Central Pacific, a striking scarcity of rock-art sites exists in Fiji, Tonga and Samoa (Cruz Berrocal & Millerstrom Reference Cruz Berrocal and Millerstrom2013). The pattern of rock-art location on these islands (specifically Fiji) is unusual, showing in most cases just one rock-art site per island, which highly contrasts with the impressive number of sites in the east and western Pacific. Cruz Berrocal and Millerstrom have argued that the existence of a unique rock-art site does not diminish its potential social significance, but enlarges it; also, that ‘the existence of a shared iconography throughout Oceania is undeniable. But the “Fijian rock art gap” argues for a chronologically independent origin of rock art making in Near and Remote Oceania’ (Cruz Berrocal & Millerstrom Reference Cruz Berrocal and Millerstrom2013, 164). This implies that ‘rock art making would be the product of particular, independent historical conjunctures, used as part of different cultural and social strategies in every archipelago’ (Cruz Berrocal & Millerstrom Reference Cruz Berrocal and Millerstrom2013, 164).

Wilson and Ballard (Reference Wilson, Ballard, David and McNiven2018, 26–7) contend that there are deeper connections between rock-art assemblages in Near and Remote Oceania based on their formal similarities, and propose a ‘widespread availability of a regional suite of motifs and their accompanying social grammars, which are sustained, communicated and transmitted between and across multiple symbolic media, and then applied in particular locations, as geological circumstances permit, to durable rock surfaces’. In our view, there is no contradiction between the idea of a gap or discontinuity in rock-art making, the independent ‘invention’ of rock art in different chronologies and archipelagos, and the social availability of symbols and grammars (Cruz Berrocal & Millerstrom Reference Cruz Berrocal and Millerstrom2013; see also Jalandoni et al. Reference Jalandoni, Taçon and Haubt2019). Iconography and meanings may have been passed down through different material means such as carving in wood, bone or stone, tattoo, pottery and barkcloth, as Wilson and Ballard propose (Reference Wilson, Ballard, David and McNiven2018, 18). Rock art just seems to belong in a much more specific social and historical context and should not necessarily be reduced to a lineal practice. It was made only under certain circumstances, that we might well try to understand. The availability of an iconographic repertoire does not by itself explain the making of the rock art.