Keywords

1 Introduction

The Acheulean is the most persistent of all archaeological cultures with its characteristic artefacts, handaxes and cleavers, produced for over 1.5 million years (Beyene et al. 2013; Haslam et al. 2011). Handaxes and cleavers are elongate stone discs, knapped on both faces of the disc, hence their collective name of bifaces. Handaxes have a typically rounded and globular butt at one end, where they are thought to have been held (Gowlett 2006). At the other, elongate end, handaxes taper in plan view to varying degrees, with their sharp, bifacially worked cutting edge usually extending around much of their circumference (Fig. 1). Cleavers have a straight unworked edge at one end which is thought to have been the principal cutting edge, like the bit of an axe or a chisel (Movius 1948; de la Torre et al. 2014; Shipton 2016) (Fig. 1).

Fig. 1
figure 1

A handaxe (above) and a cleaver (below) from the site of Chirki in India. The handaxe is 118 mm long and the cleaver is 106 mm long. The handaxe is made on a chert cobble with a small patch of remnant rounded cortex visible on the butt. The cleaver is made on a flake of basalt, finished by marginal trimming, particularly of the platform and bulb of percussion which would have been the thickest parts of the flake

Full size image

The Acheulean has long been assumed to be a social tradition like other archaeological cultures, but recently, this idea has been challenged from two directions. On one hand is the proposal that handaxes and cleavers lie in an easy to access ‘zone of latent solutions’ (ZLS) and were repeatedly reinvented (Tennie et al. 2016). On the other hand, it has been suggested that handaxe and cleaver production were in part genetically determined (Corbey et al. 2016). In this chapter, we critique these hypotheses on the grounds that independent invention and genetic control cannot explain the persistence and diversity of Acheulean biface technology, respectively. We then examine evidence for the acquisition of skill among modern knappers using both published studies and our own anecdotal observations. The Acheulean is a vast technocomplex that varies widely across time and space with several subfacies, particularly in its later stages. There is far more to Acheulean stone tools than just handaxes and cleavers, but we focus on these archetypes because both Tennie et al. (2016) and Corbey et al. (2016) make explicit reference to them, and their ubiquity across the Acheulean world makes them the most plausible candidates for all three explanations of genetic control, social transmission and latent solutions.

2 Independent Invention

Tennie et al. (2016) propose ‘the Island Test’ for culture in which they conduct a thought experiment to determine whether or not social transmission is necessary for the repeated appearance of artefact forms. The thought experiment involves imagining a group of hominins raised without socially transmitted knowledge of stone knapping, on an island with stone suitable for flaking and the need to cut things. Under these circumstances, Tennie et al. suggest that the Oldowan stone tool industry (the simpler predecessor of the Acheulean), and even the Acheulean, would be invented spontaneously. However, running this thought experiment in our minds does not produce these results.

Well before the recent discovery of the 3.3 million-year-old Lomekwian industry (Harmand et al. 2015), researchers noted that Oldowan tools already displayed considerable knowledge of knapping, so they were unlikely to be the earliest knapped technology (Panger et al. 2002). Studies of Oldowan artefacts over 2 million years old indicate their knappers understood appropriate platforms and angles to strike at to detach flakes; and how to maintain angles on a core so as to maximise the number of flakes that could be struck from it (de la Torre 2004; Delagnes and Roche 2005; Stout et al. 2010). This latter aspect is particularly important as it sometimes involved forgoing the easiest flake removal in favour of a strike that would allow more future removals, demonstrating an understanding of how individual actions relate to the wider manufacturing sequence. Also noteworthy is the rarity of battering on these early Oldowan artefacts, suggesting hominins knew the appropriate points and angles to strike without extensive trial and error. An experiment comparing early Oldowan flakes with those of modern novice knappers on clasts of the same material showed the Oldowan knappers used a more uniform strategy of unifacial reduction and produced thicker and more standardised flakes (Stout and Semaw 2006). Oldowan knappers were not habitually reinventing the wheel but appear instead to have been part of a knapping tradition that maintained skills.

The Acheulean must have been invented at least once, albeit very likely out of an Oldowan tradition (Gowlett 1986). Many aspects of Acheulean knapping are however counter-intuitive, so that the propagation of the Acheulean would be more easily achieved by social learning rather than independent invention. For example, even at the most basic level, to remove mass from one surface of a biface, you need to strike it on the opposite surface. In order to produce the thin bifaces characteristic of the later Acheulean, it is necessary to employ techniques such as platform faceting and raising the plane of intersection, which involve small strikes at a steep angle on the surface from which you intend to remove mass, followed by a large strike at a shallow angle on the opposite surface. To master such techniques to the point where they become efficient requires considerable practice even once the method is explained (Stout et al. 2014; Putt et al. 2014). The greater complexity of the Acheulean in comparison to the Oldowan and the later Acheulean in comparison to the earlier Acheulean does not fit with the ZLS model but suggests incipient cumulative cultural evolution at work.

An easy to invent Acheulean would result in similar forms wherever there is basic knapping. A knapping experiment in which the goal was simply to produce the largest flakes from randomised platforms resulted in several bifacial pieces, but none of these resembled a handaxe at any stage of 59 different reduction sequences (Moore and Perston 2016; see Fig. 16 for the closest example). Handaxes and cleavers are not easy to stumble upon without a specific goal in mind.

Perhaps the best evidence in support of the ZLS hypothesis is the appearance of handaxe-like forms at other times in prehistory. Bifacial knapping is the easiest way to shape a piece of stone, and the majority of ancient stone knapping industries contain types of shaped bifaces. Unsurprisingly then, bifaces that could be mistaken for handaxes do occur in later prehistory, for example, in northern Australia, or the Neolithic of north-western Europe (Brumm and Moore 2012; Moore 2003). However, the cleaver is a far more particular tool form (Tixier and Inizian 1983), and it is only known from one other time: the end of the Middle Palaeolithic of north-western Europe, in particular the Cantabrian and western Pyrenees mountains (Utrilla et al. 2015; Thiébaut et al. 2012). This is the best known candidate for the independent invention of the Acheulean, but the Vasco-Cantabrian final Middle Palaeolithic did not appear de novo; it arose out of the existing Middle Palaeolithic industry with deep roots and a piecemeal emergence out of the Acheulean over more than a hundred thousand years (Moncel et al. 2011; de la Torre et al. 2013). Even if the Vasco-Cantabrian Middle Palaeolithic cleavers were independently invented, the industry would not be mistaken for the Acheulean as it includes many other artefact types that distinguish it, such as Levallois points and formal denticulates.

To summarise, there is no evidence for the trial-and-error learning at Oldowan and Acheulean sites that would be necessary to support a scenario of repeated independent invention. Rather Oldowan and Acheulean knappers display sophisticated knowledge of knapping, including counter-intuitive methods that would seemingly be difficult to invent. Only one non-Acheulean stone knapping industry contains both handaxes and cleavers, but this contains other elements that are distinctive from the Acheulean and arises out of a long-standing knapping tradition rather than being invented de novo. We thus dispute the findings of Tennie et al.’s ‘Island Test’ that handaxes and cleavers are easy to invent and that this repeatedly happened during the Acheulean.

3 Genetic Control

The persistence of handaxes, and perhaps also cleavers, over the vast time span of the Acheulean is suggested by Corbey et al. (2016) to be due to these forms being under partial genetic control. We think this hypothesis is inconsistent with the diversity in the Lower Palaeolithic.

According to Corbey et al. (2016), variation in handaxe form is too little to accommodate the copying error inherent in cultural transmission. They draw support for this argument from the lower than expected size variation in archaeological handaxes (Kempe et al. 2012) when compared to outline shapes drawn on a tablet computer in an experimental transmission chain. However, this experiment does not account for the fact that handaxe size is a function not only of transmission but intrinsic constraints of hand size and strength among other factors (Gowlett and Crompton 1994). Therefore, the size of real stone handaxes should not be expected to vary as much as two-dimensional shapes drawn on a tablet. A follow-up experiment found that in reproducing three-dimensional handaxe shapes from foam blocks, there was less copying error when the parameters of social transmission were enhanced to include observation of the act of reproduction, rather than just the end product—i.e. in an imitation condition (Schillinger et al. 2015).

The lack of variation in handaxes across different environmental conditions is touted as evidence of genetic transmission (Corbey et al. 2016). However, the interregional homogeneity is overstated: handaxes are not monolithic, and there are significant regional and temporal differences in handaxe morphology (Wynn and Tierson 1990; Shipton and Clarkson 2015; Shipton and Petraglia 2010). Moreover, the relevance of a lack of variation across environments is dependent on knowing the function of handaxes and demonstrating different functional requirements for different environments. The most widely demonstrated function for handaxes is as butchery tools for large animals (e.g. Mitchell 1994), and large animal butchery seems to have been a feature of hominin behaviour across the Acheulean world (Piperno and Tagliacozzo 2001; Goren-Inbar et al. 1994; Roberts and Parfitt 1999).

It is suggested that birds’ nests and bowers, widely agreed to have a strong genetic component (Frith and Frith 2004), are similar in technological complexity to handaxes (Corbey et al. 2016). But building a nest from sticks and fluffy things or a bower from sticks and shiny things is conceptually very different from handaxe knapping. The sticks are rigid to begin with, the fluffy things are fluffy, and the shiny things are shiny, and though they may be combined in intricate and impressive ways, their fundamental properties are not changed. But a lump of stone prior to knapping is not sharp, nor are the properties which make it suitable for knapping easily determinable. Corbey et al. (2016: 14) state that it was ‘probably not just a simple target form, but rather a predisposition toward the basic behavioural routines involved’ that was genetically determined in handaxe manufacture. But there is no one basic behavioural routine applicable to all handaxe manufacture. Depending on whether a biface is made on a large flake or a slab/cobble of stone, the process of manufacture is conceptually entirely different (Shipton et al. 2009). Even within the paradigm of flake handaxe production, a review of methods for creating large flake blanks for handaxes identifies ten distinct methods found in different regions of the Acheulean world (Sharon 2007).

The dearth of bifaces in East Asia is explained by Corbey et al. through the lack of high-quality rock for their manufacture over large swathes of the region. When such rock does become available, the genes are then expressed. However, it has been demonstrated experimentally that handaxes of the same shape can be made on a broad variety of rock types (Eren et al. 2014), and in any case, most East Asian biface assemblages do not contain genuine handaxes and cleavers (Shipton and Petraglia 2010). The rock quality explanation offered by Corbey et al. for assemblages lacking handaxes also does not explain assemblages with high-quality material lacking handaxes. At the northern fringes of Acheulean occupation in Europe, there are large assemblages without handaxes that are contemporary with handaxe-bearing assemblages further south. These ‘Clactonian’ assemblages date to the beginning of interglacial periods and are thought to be a pioneer phase of Acheulean (re)colonisation by small populations unable to sustain more complex handaxe technology (White 2000; Ashton 2016). Under the genetic model, there is no reason why regions and even sites with high-quality flint, which later have handaxes, should lack them for some periods of the Lower Palaeolithic.

In a study of handaxe shape, Lycett (2008) found that, consistent with the founder effect, handaxes get less variable with distance from their East African homeland, and Corbey et al. state that this pattern fits their model of genetic inheritance. The founder effect in genetics works because inheritance is particulate, and once you have lost a particular allele from a population, it is unlikely to reappear: but there is no evidence that the quantitative variation in biface shape measured by Lycett is any way particulate or discontinuous. The correlation Lycett identifies between variation and distance from East Africa is weak (<0.3), and it is noteworthy that of the assemblages with reliable age estimates, those from within Africa are older than those from outside Africa, so that this could just as easily be a chronological pattern as a geographical one.

The Movius–Schick hypothesis, which states that biface knapping was a necessary precursor to the development of Middle Palaeolithic Levallois technology, has been put forward to explain the dearth of Levallois in East Asia (Lycett 2007). Under the genetic model, there should not be a relationship between handaxe knapping and Levallois as they are propagated by different transmission mechanisms, but cumulative cultural evolution under the social transmission model explains the geographical correlation between the Acheulean and Levallois.

Cognitive change at the end of the Acheulean is key to explaining the more rapid rate of cultural change after the Acheulean in the social transmission model of handaxe conservatism. Corbey et al. attack the idea of cognitive change on the grounds that it would require convergent evolution between Neanderthals and Homo sapiens, but endocranial evidence indicates increases in brain size and changes in brain shape do occur in the emergence of both species (Bruner 2010; Bruner and Holloway 2010). Corbey et al. also use cultural features from Africa in the last 90,000 years (well after the end of the Acheulean) as their benchmark for evidence of cognitive change, rather than evidence from the Acheulean to Middle Palaeolithic transition, which is the period of relevance. Elsewhere we have argued that the Acheulean to Middle Palaeolithic transition is characterised by increases in hierarchical organisation, recursion and, most importantly for cultural change, generativity (Shipton et al. 2013). Throughout Corbey et al.’s article, the argument is that Acheulean handaxe variation is not consistent with modern human models of cultural transmission. However, cultural transmission in Acheulean species with less capacity for innovation should not be expected to generate the same predictions.

The reappearance of handaxe-like forms in later prehistory presents a problem for the genetic hypothesis, as later examples, including those made by our own species, should also be interpreted as genetically determined. It is argued that handaxes are fundamentally different from Middle Palaeolithic bifaces in that the former were platforms on which different working edges were made, while the latter were complete tools with integrated working edges that were maintained during resharpening (Corbey et al. 2016). However, there is no evidence to suggest that handaxes were platforms for multiple working edges rather than integrated tools, and indeed it has been shown that distinctive handaxe shapes are conserved through episodes of reduction, suggesting they were conceptualised as holistic tools (Shipton and Clarkson 2015). The relatively short-lived nature of the Middle Palaeolithic biface industries is used by Corbey et al. (2016) to suggest that it is unlikely they were under genetic control. However, some of the distinctive handaxe forms of the British Acheulean, such as twisted ovates, are similarly short-lived (Ashton 2016; Wenban-Smith 2004; Bridgland and White 2014).

A final criticism of the social transmission model offered by Corbey et al. is that in non-hominin species, genes not culture are invoked as the first resort of explanation for complex behaviour, so to invoke culture in hominins is an anthropocentric double standard. We agree with Corbey et al. that it is indeed anthropocentric, but we contend that in our immediate ancestors within the last million years, anthropocentrism is justified and more parsimonious than proposing a different transmission mechanism for knapped stone to the one which exists in the recent descendants of these hominins.

In sum, there is far too much variation in the morphology of handaxe assemblages in different times and places; between the different stones on which handaxes are made; between the manufacturing sequences through which handaxes were made; and between handaxe and non-handaxe assemblages made by closely related populations, for handaxes to have been genetically controlled. The reappearance of handaxe-like forms in several later industries requires a double standard of explanation in the genetic model for Acheulean and later hominins. It seems wholly unparsimonious to suggest that in the most recent ancestor of Homo sapiens and Neanderthals, the two most social species this planet has seen, a fundamental part of their artefact repertoire was genetic and not cultural.

4 How Modern Knappers Learn

Percussive technology is known among non-human primates such as chimpanzees, macaques and capuchins. Some West African chimpanzee populations crack oil palm nuts in what has been described as a tradition (McGrew 1992), but Corbey et al. (2016) suggest that such behaviour may also be genetically controlled or alternatively it could be independently invented under conducive ecological conditions. An experiment with East African chimpanzees which do not crack nuts in the wild showed that they will learn to do so by observing a model, demonstrating genetic differences between populations do not explain the acquisition of nut-cracking behaviour (Marshall-Pescini and Whiten 2008). In addition, this study also showed that a control group provided with nuts and the materials to crack them but no model to observe did not spontaneously invent nut cracking, indicating percussive technology will not easily arise even given the right ecological circumstances.

Elsewhere we have argued that high-fidelity social learning was key to the longevity of handaxe and cleaver production (Shipton 2010; Shipton and Nielsen 2015; Nielsen 2012). Support for this position can be found among modern people learning how to knap. For example, in one of the rare ethnographic examples documented of people learning to knap large bifaces, Stout (2002) writes that stone adze makers in New Guinea work in a line commenting on the work of less experienced neighbours, reminding them where on the adze to strike and even taking over to overcome particular difficulties.

In the second half of the twentieth century, knapping was revived in many places by pioneers such as Don Crabtree, Errett Callahan and Bruce Bradley in North America, and Mark Newcomer and Francois Bordes in Europe. Having no access to other individuals who could knap, these pioneers were necessarily self-taught, trying to replicate forms that they saw in the archaeological record, in what may be described as emulation learning. It should be noted that these pioneers had the luxury of time that hominins with a spear to sharpen or a carcass to butcher would presumably not have had. Callahan describes the process (http://www.webring.org/l/rd?ring=stonetool;id=3;url=http%3A%2F%2Fwww%2Eerrettcallahan%2Ecom%2F): ‘Those first 10 years were a real struggle. I had to work it all out by trial and error. I didn’t even know what the questions were, much less the answers. Sometimes I’d find myself banging away for years, making one mistake after the other, trying to isolate what causes what. As slowly as evolution itself, I eventually sorted most of it out’.

Similarly Bruce Bradley (pers. comm.) states: ‘I began in the 1960s on my own having never seen or heard of the process. It was entirely trial and error (a lot of error)’.

The Australian knapper Chris Clarkson (pers. comm.) has a similar story of the initial steps: ‘For me, knapping began as a first year student in 1989 trying to work out how a beautiful handaxe of Bergerac flint my lecturer had brought to class had been made. I messed around with hard stones, poor quality cherts with incipient fracture planes from the creek, and inappropriately sized hammerstones of different types, including steel claw hammers. Needless to say, I almost gave up’.

As more knappers began to learn the art, they began to teach others. Callahan writes that, after the first 10 years, reading and ‘hands-on instruction from the Master-level knappers’ were important factors in the development of his knapping skill. Bradley states: ‘In 1968 I witnessed Don Crabtree giving a knapping demonstration for about 5 min and this really started me in serious knapping… Over the years I have been fortunate to learn from and also teach many of the best knappers… I have learned from all of them but also contributed to their learning’.

In line with the above, John Lord (pers. comm.) recalls that his knapping developed by watching Newcomer, going away and practising himself for a while and then asking Newcomer for advice on overcoming specific problems. Chris Clarkson says that he ‘began reading up on flintknapping, getting into Crabtree … and Whittaker. An old grainy movie of Crabtree and Bordes making pressure blades, fluted points, Levallois and punch blades was shown in second year, and after that I was really hooked. …In 2004 … a student of Bruce Bradley … showed me some valuable stuff about isolating platforms and turning edges that I knew in theory, but didn’t really know too well in practice. …The age of YouTube began around 2007, and for a while I was glued to channels … until I had reached a point of satisfaction in my own knapping. …I found by this point that I could obtain a great deal of information by watching even once a video of someone … knapping, and trying it for myself’. Similarly, Whittaker (2004) writes that an average ‘how I learned to knap story’, including his own, would go: ‘I found an arrowhead, wondered how it was made, and started experimenting. I figured out a little bit, but then got stuck. Then I met X, and he taught me enough to go on from there’.

Two factors seem to be of paramount importance in learning to knap: practice and contact with an expert. Once the pioneers had discovered an array of knapping techniques, the next generation of knappers who could learn from these early masters had a much easier time of it. In Callahan’s words, ‘My students are now learning in one week what it took me the first 10 years to learn on my own’.

In 1994, prior to the proliferation of instructional videos on knapping, Whittaker conducted a survey of knappers to establish how they learned (Whittaker 2004). Thirty-two respondents regarded artefacts or self-teaching as the most important factors, while 68 regarded social learning (friends, lessons, watching experts, books on knapping or attending knapping groups) as the most important. When respondents were broken down into two groups of more or less than 10 years of knapping experience, self-teaching was less important to the less experienced knappers than it was for the more experienced group (12% vs. 35% regarded it as the most important factor) (Whittaker 2004). This indicates that from its beginnings as individual learning, the acquisition of modern knapping is becoming increasingly social.

Young children engaging with adults in stone working were observed among modern quarry workers in the Hunsgi–Baichbal Valley in India, working with the same stone as Acheulean knappers, and it was hypothesised that knapping could begin at a very young age (Petraglia et al. 2005). An anecdotal observation from CS’s experience also suggests knapping could be transmitted early in life:

My 5 year-old nephew asked me to make him a stone arrowhead (his mother had told him that I could do this). In front of him, I took the small lump of flint that I had and reduced it in a discoidal fashion, telling him to look for any pointed flakes (needless to say he cut himself while doing this). I offered no explanation as to what I was doing because I believed it would not be something he could do at his age. About 6 weeks later he was visiting my farm again, and, without any prompting, in the exact same spot where I had been knapping I saw my nephew attempting to knap! Interestingly, he was not trying to use the remnant lump of flint or the stone hammer that I had used, as he would not have been strong enough. Instead, he was using freehand percussion to hit a slab of slate 1 cm thick, with a stick about 2 cm thick and 20 cm long. In this way he was able to remove at least nine flakes over 2 cm in maximum dimension, the largest being 6.4 cm, and he was even flaking the core bifacially. I gathered up the flakes and core he left behind after he had finished (Fig. 2).

Fig. 2
figure 2

Slate core, 10 mm thick, flaked bifacially using a stick as a hammer by CS’s 5-year-old nephew

Full size image

This example indicates that very young individuals can begin learning to knap through social observation.

Several experimental studies have looked at the acquisition of knapping skill. A transmission chain experiment with naïve subjects learning Oldowan technology showed that increased social learning significantly improved the uptake of knapping ability, with imitation providing an improvement over emulation, gestural teaching providing an improvement over imitation and verbal teaching providing an improvement over gestural (Morgan et al. 2015). A handaxe knapping experiment compared novices and expert knappers in a learning condition where they were given the necessary materials and the finished artefact to copy, but were not able to view the knapping process (Geribàs et al. 2010). Participants were given as many attempts as they wished in a single session to achieve a result they were satisfied with. Novices failed to achieve the combination of acute edge angles, tilted cores and soft support of the cores that are critical to freehand percussion. This accords with the suggestion above that even the basics of freehand knapping evident in the Oldowan are difficult to acquire spontaneously. Experts rotated the core more than novices including more bifacial rotations between the faces. Novices meanwhile tried to strike the sides of blank rather than the two main faces. These deficiencies meant that the novices did not produce anything that would be archaeologically recognisable as a handaxe. This experiment supports the observation above that bifacial knapping is counter-intuitive and difficult to master.

The experience of modern knappers indicates that given enough time, it is possible to teach oneself the basics of knapping using purely emulation learning. However, even the basics of percussive technology are not easy to acquire, and the process becomes vastly easier with direct observation and active teaching. Social transmission of knapping appears to happen spontaneously when the opportunities are there. Whiten (2015) presents a ‘helical curriculum’ model of the social learning of complex skills: an observational episode in which the knowledge of a skill is acquired, followed by a period of practice in which the ability to enact that skill is assimilated; a new observational episode, even of an identical task, then affords new insights; and a subsequent period of practice allows the new skills to be assimilated. This process continues in a cyclical manner with the learner acquiring an increasingly complex skill repertoire. The importance of both social learning and practice in the examples presented here suggests the helical curriculum model may be apposite for the acquisition of knapping skill.

5 Conclusion

Three mechanisms have been proposed to explain the vast scale of the Acheulean: handaxes and cleavers are particularly easy to invent from scratch; they are to some extent genetically determined; and they were replicated through high-fidelity social transmission. We can find no evidence to support the genetic hypothesis, which cannot account for the variability of the Acheulean. The repeated invention hypothesis is a more serious contender as both handaxes and cleavers may have been independently invented in the final Middle Palaeolithic of north-western Europe, and the first generation of modern knappers had to reinvent the Acheulean without other knappers to copy. It should be remembered though that the north-western European final Middle Palaeolithic is currently the only known place and time that cleavers reappear and they arise in the context of a long-standing western European knapping tradition stretching back to the Acheulean. The reproduction of the Acheulean by the first modern knappers was dependent on copying artefacts in a form of emulation learning, rather than completely independent reinvention.

It is possible to recreate the Acheulean without imitation; indeed the first invention of the Acheulean cannot by definition have been learned through imitation. However, the experience of modern knappers indicates it is very difficult to learn to knap without some observation of the process. Learning through observation will also occur spontaneously if the opportunity arises. That chimpanzees acquire simple percussive technology through social learning suggests that this would also have been the default position for early Homo. Archaeological evidence for separate canalised knapping methods in the production of handaxes and cleavers at the Acheulean site of Isampur Quarry in India indicates both the method and the goal were socially transmitted—as in true imitation (Shipton et al. 2009; Shipton 2010). The transmission of some of the more complex Acheulean knapping sequences, including those used to make cleavers at Isampur Quarry, may, due to preparatory steps that were causally opaque, have required overimitation (Shipton and Nielsen 2015)—that is, the tendency to copy others with such high fidelity that they will incorporate visible but causally irrelevant actions (Nielsen et al. 2014; Nielsen and Tomaselli 2010). The existence of regional clusters of sites using complex techniques for biface production, such as Victoria West in southern Africa or Tabelbala–Tachenghit in north-western Africa (Sharon 2007), indicates that such techniques were not independently invented but socially transmitted across these regions.

We invoke Whiten’s (2015) model of a ‘helical curriculum’ to explain the acquisition of knapping skill both among modern knappers and in the Acheulean: observation of a particular knapping sequence, followed by practice, recurring in cycles whereby new insights are made with each observation and skills are assimilated through practice until the learner achieves the skill level of whoever they are observing. It has been suggested that the abundance of bifaces at many Acheulean sites might be evidence of deliberate practice (Rossano 2003; Suddendorf et al. 2016). Practice has been shown experimentally to be important for mastering the skill of platform preparation required to produce the refined bifaces characteristic of the Late Acheulean (Stout et al. 2014).

The Acheulean is perhaps the most mysterious of all archaeological periods because it is unlike anything from recent prehistory and has no modern analogues, either from ethnography or primatology. Its most distinctive feature is the vast temporal and geographical scale over which it was produced. We suggest that the hominins who produced the Acheulean shared our propensity for imitation and overimitation, but did not have our capacity for innovation through generativity, recursion and hierarchical organisation (Shipton et al. 2013); and it was this combination of traits that gave rise to the unparalleled longevity and ubiquity of the Acheulean.