Time, the Middle Stone Age and lithic analyses following the Third Science Revolution

ABSTRACT

This paper is a response to the conference session ‘Generic MSA: fact or fiction?’ held as part of the PanAfrican Association Conference in 2022 in Unguja, Tanzania. It questions the validity of the concept of a ‘generic Middle Stone Age’ and goes beyond the simple debate of terminology that has persisted for nearly 75 years. Instead, it uses this as a starting point to discuss the past, present and future of lithic analyses, including the history of terminology in the African Stone Age, the current topics of lithic enquiry and the role of lithic analysis following the Third Science Revolution. It highlights the effect of tensions between science versus humanities approaches and provides suggestions for future lithic analysts. Lithic studies remain of great importance but are only one source of evidence in modern interdisciplinary human origins research. The paper contends that there is not a single approach suitable for all sites or regions since analyses are contingent upon the questions being asked. Instead, it emphasises opportunities for multivocality and suggests that despite the focus here on the Middle Stone Age the conclusions reached are more widely applicable to other times and places.

RÉSUMÉ

Cet article répond à la session ‘MSA générique: réalité ou fiction?’ tenue dans le cadre de la conférence de l’Association Panafricaine en 2022 à Unguja, en Tanzanie. Il remet en question la validité du concept de ‘Middle Stone Age générique’, et dépasse le simple débat de nomenclature qui dure depuis près de 75 ans. L’article utilise plutôt ce débat comme point de départ pour discuter du passé, du présent et de l’avenir des analyses lithiques, y compris une considération de l’histoire de la terminologie de l’Âge de Pierre africain, des sujets actuels dans les recherches liées au lithique, et du rôle des analyses lithiques après la ‘Troisième Révolution Scientifique’. Il met en évidence l’effet des tensions qui existent entre les approches basées dans les sciences dites ‘dures’ et celles qui touchent plutôt aux sciences humaines, et il fournit des suggestions aux analystes lithiques du futur. Les études lithiques restent d’une grande importance, mais elles ne constituent qu’une source de données parmi d’autres dans la recherche interdisciplinaire moderne portant sur les origines humaines. L’article soutient qu’il n'existe pas d’approche unique qui soit adaptée à tous les sites ou toutes les régions, puisque les analyses dépendent des questions posées. L’article insiste plutôt sur les opportunités de multivocalité et, bien que l’accent ait été mis ici sur le Middle Stone Age, il propose que ses conclusions s’appliquent plus largement à d’autres époques et lieux.

KEYWORDS:

• Eastern Africa
• Middle Stone Age
• lithic artefacts
• Palaeolithic
• human evolution

Introduction

This paper emerged from a conference session held in 2022 at the meeting of the PanAfrican Archaeological Association on Unguja, Tanzania, that posited that where African Middle Stone Age (MSA) assemblages have few or no diagnostic artefacts the term ‘generic Middle Stone Age (MSA)’ is often used. The session organisers and editors of this volume contended that this is unhelpful as it subsumes 250,000 years of spatial and temporal variability into a single category and that greater efforts should therefore be made to distinguish the variability in such assemblages. They sought contributions that would explore the validity of the term ‘generic MSA’ and alternative hypotheses in the wider framework of modern human cultural and biological evolution. This paper has developed considerably beyond that initial premise, culminating in suggestions of how we might accommodate diverse analytical approaches and situating our argument in the context of the so-called Third Science Revolution (Kristiansen 2014).

We begin by establishing that the meaning of any term changes over time and with use. Words accrue additional layers of meaning in time and space and are context dependent. How the term MSA is used now is not the MSA of its first inception. Historic and theoretical contingency alongside the scale at which enquiries are conducted are key. We also note that while the phrase ‘generic MSA’ may be used in general parlance in conferences and academic contexts, it is rarely found in publications. Where it is found the phrase is used in discussing the MSA in northern, southern and central African contexts (Barham et al. 2002; Lombard, 2008; Taylor 2014; Scerri et al. 2021; Timbrell et al. 2022). In the eastern African MSA, where there are no regional technocomplexes such as the Howiesons Poort or Still Bay of southern Africa or the Aterian of North Africa (although see below for a note on the use of the terms ‘Sangoan/Lupemban’) the phrase ‘generic MSA’ is not used.

The second section of our paper moves beyond the conference session theme to consider whether it is helpful to define more variability, how that should be done and whether specific terminologies retain any value. Following Shea (2011a), we argue that lithic artefacts in isolation are not a good way to understand past behavioural variability, diachronic change or mobility etc. For example,

‘Would anyone among us take seriously a proposal to track the migrations, dispersals, and cultural diffusion of ethnohistoric humans from variation in their pens, pencils and personal digital assistants?’ (Shea 2011a: 50).

In the paper’s final section, we reflect on the current state of lithic analysis to suggest how in future it can contribute by considering the wider theoretical landscape beyond stone tools and optimistic approaches that embrace the ambiguity and diversity inherent in artefact studies. It is important that we find ways to integrate the panoply of information we can gather from lithic artefacts with that obtained from other cultural elements (gathered from archaeology) and biological ones. We welcome the proliferation of scientific analyses, but these should not be put on a pedestal above interpretative approaches. At the root of the current discussion is the tension between science versus humanities interpretative paradigms. Lithic artefacts remain the most common archaeological find for significant spans of our evolutionary history. Continuing to develop theoretical and methodological advances in their study remains vital, but this can only happen if we find ways of resolving the paradigmatic tensions arising from multidisciplinary approaches to human evolution.

The past: defining the Middle Stone Age

Here, we briefly consider the history of the term MSA to establish the theoretical context of its first use and acknowledge that from the outset it was considered simplistic.

The term ‘Middle Stone Age’ has had a long and complex history. It was formalised by Goodwin (1928) almost a century ago as an addition to the Earlier and Later Stone Ages (ESA and LSA) and was primarily defined on the basis of the archaeology of southern Africa (Goodwin and van Riet Lowe 1929; Klein 1970; Allsworth-Jones 1993). Importantly, the aim at this time was to differentiate the African and Indian archaeological — specifically lithic — records from that of Europe. Although the tripartite system was widely adopted in southern Africa, interestingly Louis Leakey never used it for the archaeology of eastern Africa, preferring more specific local terms (Masao 1992; Mabulla 1996). The current dichotomy between regional MSA and generic MSA terminologies may be an inadvertent legacy of this history.

The ESA/MSA/LSA divisions sprang from the theoretical context of culture historical approaches in which different stone tools were thought to be associated with specific groups or hominins and were strongly influenced by interpretations of European prehistory. Many mid-twentieth-century researchers thought about lithic artefacts at a global scale and sought to establish the basic sequence of events, culminating in Grahame Clark’s famous ‘modes’ model (J.G.D. Clark 1969). This divided stone tools into a series of Modes (1–5), from the less complex (Oldowan) to the more complex (microliths) with the MSA designated a ‘Mode 3’ technology. These ‘stages’ of technological complexity occurred in approximately chronological order in parallel with human evolution. Some archaeologists continue to use modes (e.g. Foley and Lahr 1997; Barham and Mitchell 2008). Others do not find modes helpful (e.g. Basell 2007, 2008; Shea 2011b) and some have suggested testing whether modes really represent an increase in complexity (Perreault et al. 2013).

Based on notions of unilinear progression stemming from Western Enlightenment thinking, there was a general assumption that increasing technological complexity might be expected from the ESA to the LSA. Attributing artefacts to the ESA, MSA or LSA on typological or technological grounds was thus useful since it conferred a broad chronological association, a point to which we shall return. For the MSA, in particular, a major shift occurred after the 1970s when improvements in chronometric dating methods encouraged a radical revision of Sub-Saharan Stone Age chronologies (Vogel and Beaumont 1972). Initially the ESA/MSA/LSA had been discussed in the framework of pluvial and interpluvial models. With the advent of radiocarbon dating in the 1950s a chronology was introduced for the Sub-Saharan MSA that at first defined its boundaries as ∼40,000–10,000 BP (J.D. Clark 1970), leading to comparisons between the MSA and the European Upper Palaeolithic that resulted in a gradual perception of the MSA as a technologically lacking cultural backwater (Bräuer 1984). As a much longer chronology emerged following Vogel and Beaumont’s (1972) publication, the deep antiquity of the MSA became clear and previous correlations between the MSA and the Upper Palaeolithic were dropped. As more sites have been discovered and dated, it has become possible (with appropriate methodological and taphonomic caveats) to perceive far more temporal and spatial variability in known lithic assemblages. This has resulted in a florescence of further terminologies referring to perceived regional or temporal differences. We do not repeat the debates concerning the complexities and problems of such terminologies here as they have been discussed at length elsewhere (e.g. Kleindienst 1967, 2006; Klein 1970; Cahen 1978; J.D. Clark 1982; Bräuer 1984; McBrearty 1988; Masao 1992; Mabulla 1996; Herries 2011; Shea 2011a, 2011b; Barham 2012; Tryon et al. 2005; Tryon 2015, 2019).

However, it is worth emphasising the importance of Desmond Clark’s seminal paper on the relationship between lithic variability and regional identity (J.D. Clark 1988). This marked a critical turning point in the perception and use of the MSA since it addressed for the first time the relationship between African Stone Age lithic artefacts and identity. It also opened the door to diverse interpretations of lithic assemblages focused on everything from social relations to mobility and skill. Clark’s paper probably remains one of the main reasons for the current discussion concerning regional versus generic uses of the term ‘MSA’ because it relates lithic artefacts and their methods of production to ‘regional identities’. In eastern Africa the terms ‘Ethiopian MSA’, ‘Kenya Rift MSA’ and ‘Mumba Industry MSA’ have not, however, withstood the test of time in the same way that the Aterian or Howiesons Poort. Instead, the MSA of eastern Africa presents a mosaic pattern of variation through space and time. In other words, in eastern Africa there are no regionally specific technocomplexes. It may also be the case that apparently abrupt and horizon-wide transformations in northern and southern Africa (J.D. Clark 1988; Shea 2008) are simply the result of historically coarse-grained analytical approaches and limited chronological control. Until recently the Still Bay of southern Africa, for example, was considered as ‘abrupt’ in its timing, something that recent research suggests may not have been the case (Lombard et al. 2019, 2022).

That the ESA/MSA/LSA system was too simplistic was also recognised from the outset and not everyone used these terms. Many authors have made this point subsequently, particularly as the questions of human evolution have shifted. For example, Mehlman (1991: 194) wrote that ‘our Three Age System is largely illusory; the product of a small sample of occurrences, widely separated in time, giving us the false impression of three plateau stages’. In a similar vein ‘the tripartite categorisation is too simplistic to provide more than an ill-defined shorthand for the archaeological record covering wide temporal and spatial spans’ (Basell 2007, Chapter 2: 2). And such criticism has not been limited to Africa. Some thirty years ago, Geoff Clark was arguing that the ‘Aurignacian’ and the ‘Mousterian’ of Europe are simply modern-day constructs, meaningless over such large temporal and spatial spans (G. Clark 1994, 1997). Over the years we have seen ‘cultures’, ‘industries’, ‘traditions’, ‘technologies’, ‘intermediate’ categories, ‘technocomplexes’ and ‘styles’ all variously adopted, abandoned and resurrected in line with shifting theoretical approaches and the questions that drive our endeavours. Significant effort has been put into specifying what such terms mean (Kleindienst 1967), but this has not prevented authors from continuing to use terms loosely and in interchangeable ways.

The most recent permutation of this can be seen in the heated debate that has arisen around the so-called NASTIES (Named Stone Tools Industries; Shea 2014). NASTIES are linked to the nineteenth-century cultural-historical approach to investigating the past, which identified lithic artefacts with different human groups, most of which were identified in twentieth-century Europe (Shea 2014). There, for example, it famously manifested itself in the dispute between François Bordes and Lewis Binford concerning the correct interpretation of Mousterian stone tool variability (Bordes 1961; Binford and Binford 1966; Bordes and de Sonneville Bordes 1970). In his proposal to abandon NASTIES, Shea (2014) deliberately draws on the Mousterian debate to argue that NASTIES are an obstacle to understanding past processes because of their inductive nature and the lack of systematicity on their definition. The abandonment of NASTIES has also been proposed in southern Africa, with questions raised concerning their utility in interpreting the northern and southern African lithic records (Scerri and Spinapolice 2019; Wilkins 2020).

The present: why should we care about variability, innovation or change?

What is the MSA today?

Today lithic analysts have put significant effort into researching a modern, more satisfactory definition of the MSA and have considered further reasons to drop the term in favour of regional variants or a simple ‘Middle Palaeolithic’ inclusive category (e.g. Tryon and Faith 2013; Cambridge 2014). Although originally defined in relation to lithic assemblages, the term ‘MSA’ explicitly incorporates wider associations. These include chronometric, archaeological, paleoanthropological (possibly different hominin) and geographic factors (e.g. Basell, 2007, 2013; Lombard, 2012; Tryon and Faith 2013; Staurset et al. 2023). Many papers begin by qualifying the use of the term as a generic shorthand for the period during which there appears to be: i) an increase in the variability of lithic types and lithic technology following the Acheulean (Early Stone Age); ii) a progressive abandonment of large cutting tools (LCTs); and iii) a variety of prepared core technologies. The term continues to be used to differentiate the African archaeological record. The MSA is also widely considered as the lithic industry associated with the emergence of the Homo sapiens clade, because of its association with major fossils such as Jebel Irhoud, Omo 1 and 2 and Herto (J.D. Clark et al. 2003; Shea et al. 2007; Richter et al. 2017). Nevertheless, it is possible that the MSA’s variability encompasses the activities of multiple hominin taxa and that for many assemblages we cannot be sure that the knapper involved was Homo sapiens. Currently, its chronological associations for eastern Africa are very broadly >300 kya (and perhaps as early as ∼400 kya) to ∼40 kya, i.e. roughly equivalent to Marine Isotope Stages (MIS) 6 to 3. With the advent of modern dating methods, the date range has changed repeatedly, but while there are significant local variations across the continent at a general level as a temporal indicator the ESA/MSA/LSA framework has retained some integrity, which is why the terms persist.

Compared to the ESA and at a continental scale, the MSA is certainly a period of increased technological and typological diversity. Terminologies include localised or regional sub-stages or industries established via stone tool analyses that may incorporate both typological and/or technological criteria (Lombard 2012). Although the origin of the Levallois method could be older than the widely cited beginning of the MSA (perhaps as early as ∼500 kya in southern Africa and Europe; Moncel et al. 2020; Wilkins 2020), and despite the claim for convergence and no single origin (e.g. Adler et al. 2014), for many the Levallois method remains one of the key features of MSA lithic assemblages (Spinapolice 2020). Equally, however, the MSA is characterised by variability in knapping strategies so for other researchers it is the broader concept of prepared core technologies that is important. These include Levallois, but also the volumetric method employed to produce blades and bladelets, which is present since the Middle Palaeolithic/MSA in both Africa and Europe and traditionally associated with the Upper Palaeolithic (although see Carmingnani and Soressi 2023 for new evidence of blades and bladelets in association with Neanderthals).

Just as dating techniques have become more refined over the decades, lithic studies now incorporate new and improved scientific analytical techniques, many of which are being used in MSA contexts. For example, developments in geochemical provenancing methods and tools allow new perspectives on raw material distributions and hominin mobility patterns (e.g. Brooks et al. 2018; Nash et al. 2022); residue and microwear analyses allow us new insights into how lithics were used or combined with other materials (mastic and hafting) (e.g. Donahue et al. 2004; Werner and Willoughby 2018); experimental approaches continue to offer exciting new insights into both production processes and tool function (e.g. Brown et al. 2009; Schoville et al. 2016); and new theoretical approaches offer novel possibilities for understanding aspects of past lives such as skill, cognition, social transmission, mobility, function, the life histories of artefacts, their interactions with other materials and neural function (e.g. Dobres 2000; Wynn 2002; Shea 2011b; Stout and Chaminade 2012; Tostevin 2012, 2019; Grove and Blinkhorn 2020; Hussain and Will 2021, Liu et al. 2023). Such improvements in both chronological control and interpretations of lithic artefacts permit a better alignment between the available non-lithic archaeological data and the questions that we seek to answer about human evolution.

In most MSA lithic analyses the typological approaches of the past have been replaced by primarily quantitative methods as the primary frame of reference both for analysing variability within assemblages and for comparing artefacts at a wider scale (regional or even continental) (e.g. Blinkhorn and Grove 2018, 2021; Perreault et al. 2013; Will et al. 2019; Fusco et al. 2021). These continue to be undertaken via manual human analyses, but with increasing use of laser scanning, photogrammetry, automated shape capture technologies, and associated analytical software including artificial intelligence (Elliot et al. 2021; Gellis et al. 2022; Grosman et al. 2022). With increased computing power, it has become possible to measure, record and visualise variability in more dimensions, better and faster than ever before. The proliferation of quantitative methods of analysis in turn permits large-scale statistical modelling (e.g. Blinkhorn and Grove 2018, 2021).

Why and how are researchers looking at lithic variability?

Characterising lithic variability relates directly to the key questions that researchers have sought to answer about human evolution over the decades. Beyond establishing chronologies, identifying increased variability in methods of production and types of stone tools has been considered important in understanding stone tool functions, adaptive variability within and between hominin species across time and space, cognition or ‘modernity’ and, in a related vein, the propensity of hominins to innovate. There has also been an interest in defining cultural ‘packages’ to understand how hominins were moving regionally and globally (migration) and the degree to which groups and/or species might have acculturated with one another. Today, the dominant questions in human evolution are closely aligned with the five categories into which Kintigh et al. (2014) group 25 ‘grand challenges’ of ‘the most important scientific challenges’ in archaeology, namely:

(1)	Emergence, communities and complexity;
(2)	Resilience, persistence and collapse;
(3)	Movement, mobility and migration;
(4)	Cognition, behaviour and identity; and
(5)	Human-environment interactions.

To address such questions requires the identification of continuity or rupture in the archaeological record. Lithic artefacts remain one of the most prolific sources of behavioural evidence that we have for understanding continuity or rupture, tradition and/or innovation. Identifying variability in form, production methods and function, as well as social and spiritual meanings, is an issue that is strictly related to the intrinsic capacity of stone tools to store cultural information, in other words, whether we can infer inter-generational cultural transmission from lithic data and how we can measure and analyse it (Liu and Stout 2023). Cultural transmission theory is a useful tool for understanding the processes of transmission, modification, preservation and loss of learned behaviours, including the technical choices of artefact makers, in an evolutionary perspective (Premo and Hublin 2009; Premo and Kuhn 2010). It may be that lithic analyses of this kind, provided they are considered in combination with other data sources, could play a role in identifying past population movements.

It is crucial, then, to identify which significative elements in the lithic record are susceptible to transmission through learned behaviours and how to successfully measure and analyse them (Spinapolice 2020). This is why it is important to have strong geochronological and geoarchaeological controls and a sound understanding of the scale of analysis. Equally, however, it is necessary for us to understand mechanisms of social transmission, chaînes opératoires, learning and skill and to consider what the drivers of innovation or change might have been. With the proliferation of analytical tools, it is now possible to consider far more types of variation. In every lithic assemblage there are artefacts that reflect the explicit and/or implicit choices of the knapper. Some are related to a spatiotemporal context, while others may reflect a common, socio-cultural background or general understanding of ‘how things are done’ that might have a relationship to identity. Blinkhorn and Grove (2018) have argued that some so-called constellations of stone tools span most of the MSA and could represent what used to be called a ‘substratum’. It might be helpful to consider such substrata as forming a ‘generic MSA’, but only if they are closely defined. This cannot be merely by the presence or absence of specific tool types, but could perhaps involve closely defined technological approaches, specific kinds of ‘know-how’ shared in time and space, and potentially across species, and specific tools where it is possible to be sure they were used as a specific adaptive strategy in a particular situation.

Identifying variability is also useful for those who seek to understand the distinctiveness of human cumulative culture. For many this remains a big question to which lithic analyses could contribute. Definitions of cumulative cultural evolution have been much debated over the last decade, in part due to its application in multiple fields of enquiry. A useful broad definition is that it refers to processes of cultural change involving modifications in cultural traits occurring over multiple episodes of social transmission (Caldwell 2023). With time, traits (behaviours, artefacts, tools etc.) become more effective and beneficial to their users (Caldwell 2018; Dunstone and Caldwell 2018). At a global scale, and across the entirety of human evolution, increasing technological complexity is undeniable. However, a general global trend in complexity should not be confused with notions of efficacy (i.e. ‘more complex’ does not necessarily equate to ‘better’) or ‘progress’. Importantly, local adaptations frequently buck global trends since they are context- and task-specific. They are definitely not linear and skill and knowledge may be lost as well as passed on or developed. Comparisons at local and regional scales can also be problematic where different researchers use different analytical tools or terms that can say more about researcher choice in the present than variation relating to hominin behavioural changes in the past. Nonetheless, identifying and explaining technological variability at the scale of the site, as well as regionally, remains a key objective for many researchers and certainly contributes to understandings of larger scale trends.

The ‘generic MSA’

So where does this leave the phrase ‘generic MSA’? Historically in eastern Africa there was no specific definition of the MSA. Thus, unless an assemblage fits — or is compared to — a regional definition the only option is to refer to it as “generic MSA’ in the absence of enough information to define a new term. Regional terms used in eastern Africa have not generally withstood the test of time (e.g. Hargesian, Magosian (J.D. Clark 1954, 1988)). The most prominent exceptions are the Sangoan and Lupemban, terms with a complex history but generally associated with sites in central and western Africa (see Barham and Mitchell 2008; Basell 2010; Taylor 2014, 2022; Ssemulende 2022 for wider discussions). Although there are sites in Uganda, Tanzania, Kenya and Sudan where these terms have been applied (McBrearty 1987, 1988, 1992a, 1992b; Van Peer et al. 2003, 2004; Basell 2007, 2008, 2010; Basell and Brown 2012; Basell and Posnansky 2019; Ssemulende et al. 2022) it is currently more appropriate to consider them as eastern outliers of that distribution rather than a well-defined and chronologically constrained East African technocomplex, although we acknowledge that this situation may change in the future.

More generally, we cannot find any published examples of ‘generic MSA’ being used in an eastern African context. The term is only ambiguous when it is not specified whether it is being used to: 1) provide a rough indication of how old a surface scatter, or poorly dated or undated a sequence might be (because the lithic artefacts present look similar to other MSA stone tools in a region); or 2) to define something as MSA that does not have specific characteristics that allow it to be ‘regionally connotated’ (e.g. as Aterian or Howiesons Poort etc.). This resonates with the idea of using names to reflect observations (age, provenience) rather than metaphors (e.g. generalised, specialised, curated, expedient, specific/generic, etc.) that mean different things to different people (John Shea, pers. comm., 2022). Naming things is part of an iterative and interpretive process. As such, it can actually be helpful (if the criteria for naming are clearly defined) as well as problematic (i.e. NASTIES). What words we use in describing lithic artefacts, and how we understand those words, will ultimately have an influence on the narratives that we produce.

In certain circumstances it may indeed be helpful to have a generic catch-all phrase, especially when looking at a surface scatter that has no stratigraphic context and no chronological controls, something all too common in many contexts across the African continent. Many archaeologists examining a lithic assemblage will, as part of an initial visual assessment, say that it looks ‘Middle Stone Age’. In the absence of multiple diagnostic stone tools that allow insights into technological strategies or specific adaptations, the phrase ‘generic MSA’ might prove useful in such circumstances. All the analyst is then saying is that it resembles other lithic artefacts of a certain date and type based on their wider knowledge at the time of publication. As discussed above, this could relate closely defined technological approaches, specific forms of ‘know-how’ shared in time and space and the presence of particular tools where it is possible to be sure that they were used as a specific adaptive strategy to a particular situation. In our view this is an entirely acceptable position to take and may well be useful as a starting point to test hypotheses and initiate data collection. Of course, further analyses, dating and consideration of the wider context, including more detailed or comparative lithic analyses, may ultimately lead to a conclusion that the assemblage in question is a mixed one or does not fulfil the criteria for categorisation as MSA at all.

The future: considering what we want to know from our lithic assemblages and future analytical approaches

Historically, lithic analyses were used to build chronologies and define ethnic groups (e.g. Bordes and de Sonneville Bordes 1970). Later, following more discoveries and with the advent of chronometric dating, they were used to try and establish dispersal routes or increasing complexity that might reflect ‘modernity’, often in conjunction with other artefact types (e.g. Mellars 1990; McBrearty and Brooks 2000; Henshilwood et al. 2002). As one of the most prolific sources of behavioural evidence available to archaeologists, stone tools were dominant in tackling big questions of evolution. So, why has doubt arisen over their role in helping to answer those questions? One reason is that lithic analysis lies at the very interface of scientific methods and archaeological interpretation. Human evolution is currently dominated by scientific approaches deeply embedded in evolutionary biology and the methodologies employed by such approaches are often at odds with interpretative humanities-based approaches (e.g. Sørensen 2016, 2017).

With rapid technological advances and increased abilities to manipulate vast quantities of data (the so-called ‘Third Science Revolution’ or ‘new empiricism’) we have an incredible array of high-specification techniques informing our research questions in human evolution (Kristiansen 2014; 2017; Sørensen 2016). These include genetics, improved chronologies, scanning, isotopes, sedimentary DNA and palaeoproteomics, to name just a few. We are able to analyse vast quantities of data (‘Big Data’), which in lithic studies are often geospatial in nature or include a geospatial component, at unprecedented speeds and scales of analysis. However, it is critically important to acknowledge that large quantities of data alone and correlation of those data are not enough (Anderson 2008: 17; McCoy 2017; Huggett 2020; VanValkenburgh and Dufton 2020). Just as adding more data will not overcome problems of bias and effectively reduce/remove the influence of data errors (Beer 2019: 6–7; Huggett 2020), defining greater variability in lithic analyses will not provide a better understanding of hominin behaviour. The archaeological data generated from lithic analyses are (and remain) undoubtedly ‘shadowy’ (cf. Wylie 2017: 204–205), i.e. there are many gaps and absences in the primary data and considerable instability in the inferential resources on which they rely. While novel discoveries capture the public imagination and make headlines, those that are really significant are often the result of painstaking re-examination of exiting bodies of data, including re-analyses that question previous understandings. Nevertheless, data remain powerful and can ‘bite back’ in unanticipated ways (Wylie 2017).

Despite the spectacularly interdisciplinary nature of human evolutionary studies, and the ‘biomolecular turn’ in archaeology more generally, some have argued that lithic studies still need to fully affirm their role in the so-called ‘grand narratives’ within research on human evolution (Hussain and Soressi 2021). The interdisciplinary nature of the topic alongside the variety of new scientific tools for analysing stone tools means that we now have new ways to address old questions, but also requires us to rearticulate the role of lithic artefacts in our interpretations. One of the reasons for perceived difficulties in incorporating lithic analyses in the burgeoning body of scientific data is that their subject matter provides one of the most valuable tools available to us for addressing the social dimensions of the past (identity, mobility, economy, exchange, consumption and production). Immaterial qualities, processes of disappearance, or indeed emergence are particularly challenging for archaeological scientists seeking indisputable observations or unambiguous rhetoric.

These ideas chime with wider discussions in archaeology concerning the (re)turn to science as the main provider of knowledge of the past, and how ‘legacy’ data can be put to work in new ways (Sørensen 2016; Wylie 2017). We agree with Shea (2014: 169) that:

‘properly analyzed, the lithic record can shed light on change and variability in hominin behavior with sample sizes and statistical power that the sparse hominin fossil record simply cannot equal. And yet, there is often a mismatch between how Paleolithic archaeologists analyze and interpret the lithic record and the “big questions” in human evolution.’

This mismatch is fundamentally an epistemological clash between big science questions and methods and the need for archaeological interpretation to embrace paradigms that are underlain by humanities-based analyses as much as by science. Lithic datasets are often far from ideal when it comes to mathematical or hard science approaches. Even the most carefully excavated and curated lithic assemblage will not include all the fragments of the reduction process. Many older assemblages were selectively sampled in the field, thereby introducing significant biases, and such biases may be compounded in a museum setting through loss of labels, artefacts etc., especially in places where staff and funding are limited. Where primary access to assemblages is not possible, analysis can only be undertaken using published literature. Again, lithic data are variably presented and often only partially published in summary tables. This means that many datasets cannot be analysed statistically because of inherent biases or very low sample sizes or because archaeological researchers are not trained in — or do not understand — appropriate statistical mitigation strategies.

More positively the concept of ‘research agenda’ can be very useful (Love 2008). Here, investigation can be co-ordinated around sets of problems, called ‘problem agendas’, that have associated criteria of explanatory adequacy (Love 2008). A problem agenda is a list of questions unified not by a theoretical perspective or methods but by a phenomenon of interest (Menganzin and Currie 2022). A research agenda focuses on phenomena more than on paradigms and, in this sense, is a useful tool to deal with Palaeolithic/Stone Age archaeology. In particular, it helps to solve the dichotomy between nomothetic and idiographic approaches (e.g. Schiffer 1975), where nomothetic ones are based on a tendency to generalise, as is typical for the natural sciences, while idiographic ones are based on a tendency to specify, as is typical for the humanities, which focus on more contingent, unique and often cultural or subjective phenomena. Using a research agenda as a tool, we can reconcile the desire to answer ‘big questions’ about human evolution with the necessity of analysing thousands of artefacts, context by context and site by site.

In new analyses, lithic artefacts must be considered in their wider geoarchaeological context using the full suite of evidence that we have for understanding hominin behaviours to answer questions about progressive knowledge transmission and populations and more contingent questions about past behaviour at specific time scales and in specific geographic areas. To use lithic analyses effectively to say anything meaningful about past behaviours is not dependent on common terminologies, although this can help with wider comparisons in answering some questions. However, it does require careful and detailed analysis of lithic artefacts in context before building regional comparisons, something that this is significantly facilitated when there is a degree of standardisation between the quantitative analyses applied, especially in the context of statistical modelling. Analyses need to focus on understanding the specific objects/process, look at the chaîne opératoire and (ideally) involve a clear understanding of the geoarchaeological context of the finds. Understanding site formation processes and the sedimentary context of an assemblage has significant implications for understanding what has caused artefact associations and variability. It is only with this detailed analysis in combination with effective chronological controls that it becomes possible to understand technological continuity and diversity and to establish if this is behaviourally meaningful over wider spans of time and space.

Through careful, detailed and contextualised lithic analyses at local and regional scales, lithic interpretations can yield fascinating information on past behaviours. It is important to stop thinking of lithic artefacts exclusively in progressive terms. Stone tools do not evolve, but they can indicate cultural change and adaptation, potentially at evolutionary scales (see Režek et al. 2018). Even though few organic remains have been preserved, with which many stone tools would have been combined, analytical techniques now exist that allow us to access this information, e.g. regarding questions of hafting and tool use. With such knowledge the interpretation of a particular suite of artefacts can completely change. A tool that is relatively simple in appearance may have been a highly effective adaptation to a specific task. As an example, consider the Hoabinhian of Southeast Asia, in which river cobbles were once interpreted as indications that their makers were ‘backward’ compared to previous stone tool producers in the region until it was understood how they had been used to process a wide variety of bamboos, providing fantastic adaptations to an ecologically specific situation (White 2011; Blench 2013).

Since the terms we use and the methods of analysis we employ will always be specific to the questions that we seek to answer it is important that researchers are explicit about the questions that their analyses are addressing and the scale of on which those analyses are conducted. Indeed, the nature of a site and the interests of a researcher can mean that questions vary considerably even between researchers working contemporaneously in the same region. Methods may be underlain with significantly different theoretical paradigms and especially those relying on microscopic or computer-generated analyses/pattern matching, for example. Where researchers are analysing assemblages with the same questions in mind, the thorny issue of inter-analyst measurement comparability arises. However, recent comparative performance testing suggests a high replicability in many attributes, as well as providing useful recommendations for future comparative studies (Pargeter et al. 2023). Continental scale comparisons need to be undertaken with caution and anchored to well-dated sites and potentially specific environmental characteristics. In certain circumstances there may be questions of migration/population expansion or adaptations to specific environmental conditions that can be answered at a regional, or even a continental, scale. However, this requires compatible theoretical paradigms, excellent geochronological control, the integration of multiple sources of evidence and a recognition that the larger the geospatial and temporal scale of analysis the greater the interpretative possibilities. Continental comparisons of lithics should not focus on fossiles directeurs or individual technological approaches. Identifying the reasons why variability within and between assemblages exists at different scales is important and numerous interesting new approaches to this exist that link behavioural ecology and the formation of the archaeological record (e.g. Brown et al. 2013; de la Torre and Hirata 2015; Luncz et al. 2016; Harmand and Arroyo 2023; Reeves et al. 2023).

Conclusion

Although our focus here has necessarily been on the MSA we hope that our concluding points are more widely applicable to other times and places. We suggest that researchers should acknowledge historical contingency in the use of all lithic terminologies and analytical approaches. Previous interpretations cannot just be ignored, and many remain valuable provided that the historical context of interpretation is adequately acknowledged. Following Wylie (2017), this might be through secondary retrieval of primary data, the recontextualisation of such data or by using old data in experimental simulations (modelling).

Generic terms can be useful and will not disappear. Their use is dependent on the context of the assemblages being discussed, the scale of analysis being undertaken and the time at which that analysis was done. The MSA always has a host of additional meanings. Even though Goodwin’s (1929) initial definition of it was based on lithic artefacts alone, he considered those stone tools to be associated with specific populations and was trying to define the MSA in the context of ‘race’ and ‘invasion’. This is what Wylie (2017) refers to as ‘scaffolding’, an element of all archaeological analyses that is consistently challenged and rebuilt. Provided that we accept that material culture can house ‘data imprints’ of past lives, then we have a rich evidential resource at our fingertips able to provide us with fascinating insights about those past lives.

Defining ever increasing diversity in the lithic record will not get us closer to an unambiguous ‘answer’ or ‘truth’ about our past. To jettison the terms MSA/LSA is not possible. They are embedded in the literature of the last 75 years and unless we choose not to reference previous work then we are tied to them. Although they are typological terms, they remain a useful means of communication (when properly qualified) and operate as a framework for systematising data and reducing complexity (Boozer 2015). Issues do arise with typologies when they become entrenched and canonised (Boozer 2015; Wylie 2017), but this does not appear to be the case in the African Middle Stone Age. If ambiguity is honoured, then the term ‘generic’ need not be seen as a source of despair, but rather understood as being valuable because it acknowledges the uncertainty of its descriptive and interpretative claims (cf. Sørensen 2016; Wylie 2017). If researchers acknowledge that what are recognised as relevant data, and data of evidentiary significance, relate to the ‘pre-understandings’ and assumptions that they bring to bear in their interpretations, then incorporating words like ‘generic MSA’ becomes unproblematic since it forms part of the mutability of conceptual and technical ‘scaffolding’ that we employ (Bell 2015; Wylie 2017). Instead of endless discussions on their value and precise meaning, we should embrace the diversity of approaches to possible interpretations of the past. Instead of dwelling on the inadequacies of past approaches and worrying about trying to reconstruct knowledge about the past from trace evidence, a more productive and positive approach would be to acknowledge that scaffolding is constantly built and rebuilt because this allows the horizon of enquiry to be shifted such that a graded set of possibilities becomes possible instead of generating more dichotomies (Wylie 2017 after Currie 2014).

The record of human evolution is highly fragmentary even with today’s rich multidisciplinary approaches. But the challenges of working with trace evidence are not unique to archaeology (Wylie 2017). There are many interpretative possibilities (although constraints on interpretations do exist) such that there will never be a single narrative or ‘truth’ about human evolution. Such interpretative and methodological diversity and innovation are what makes archaeology a dynamic and fascinating discipline. This is not the same as accepting people’s analyses uncritically, but vagueness or ambiguity are not a research deficiency (Sørensen 2016). Here, we have tried to suggest ways in which accommodating vagueness can be both useful and beneficial. Archaeologists have a strong drive to shelve old ideas, methods, observations and concerns in a relentless quest for novelty, or at least its illusion (Sørensen 2018). Instead of continuing to worry about terms (a recurrent theme since the 1950s) perhaps it is now time to get on with the job of identifying what it is that we want to understand about human evolution. If we are explicit about identifying what we can understand from stone tools, (both as materials and as sources of information about the human decisions and structures that surround them) then we will be better placed to evaluate how they are useful in integrated interpretations of multiple sources of evidence. Approaching lithic artefacts like this should allow researchers to develop interesting (though not necessarily new) ways of extracting information, given that stone tools are the most prolific pieces of evidence from our prehistoric past.

The main reason researchers think that it would be helpful to have closer definitions of variability is because they are often seeking to create a single narrative for the human past, a grand narrative at a global scale. If we accept that there is not a single truth about the past that we are trying to reconstruct, but that there are multiple interpretations through time and space, then far more exciting possibilities emerge. We can explore the wide array of different hominins and adaptations and, since we only have a fragmentary record of their time on this earth and their material remains, there is a significant amount of flexibility in the number of stories that we can tell about these past populations. Importantly, the range of interpretations that we can come up with is limited only by the materiality of those things we discover and by the affordances of the analytical tools that we have at our disposal.

The narrative of Palaeolithic and Stone Age lithic artefacts is historically grounded and its history is rooted in the Western academic tradition. It is also worth emphasising that access to analytical tools is unequal. Many archaeological researchers based in Africa may not have access to the suite of technologies available to those working in the Global North. That should not mean that the lithic interpretations of those working in Africa are any less valid or valuable, but it is hard to imagine a future where this continues.

Instead, we are hopeful that, with the next generation of African and Africanist archaeologists, new ways of thinking about stone tools will emerge. Breaking out of specific academic and historical traditions, considering views that are not solely rooted in science and decolonising approaches to the past should open up different narratives. As scholars of human evolution, we recognise the remarkable plasticity of hominins. If we can work with colleagues from different cultural backgrounds, who knows what exciting new possibilities may arise? There is huge potential in our increasingly globalised world for a future where we can embrace and come to terms with alternative viewpoints or, through collaboration, arrive at new ones.

Acknowledgements

We should like to thank the conference organisers, the organisers of the session in which this paper was delivered and all the presenters and session discussants who took part in it for such a stimulating conference. Funding for the authors’ attendance was provided by the Quaternary Research Association (Basell) and Grandi Scavi Sapienza, MAE (Italian Ministry of Foreign Affairs) and PRIN 2022 (Spinapolice). We are grateful for the informed and helpful feedback from three anonymous reviewers and for some helpful comments on an early draft from Marlize Lombard. The ideas presented here evolved via discussions about Spinapolice’s conference presentation in Unguja in 2022 with both authors conceiving and writing the paper in equal part.

Additional information

Notes on contributors

Laura S. Basell

Laura Sophie Basell is a landscape geoarchaeologist who directs field and research projects in East Africa and Europe. Her interdisciplinary research integrates interpretations of behavioural landscapes with Quaternary climatic change to understand cultural and technological innovation. She is interested in population migration and how cultural heritage informs identity creation. Her practical skills are in excavation, geomatics and the methodological development of digital technologies in heritage contexts.

Enza E. Spinapolice

Enza Elena Spinapolice is a Palaeolithic archaeologist working in the field of Evolutionary Anthropology who directs field and research projects in East Africa and Europe. Her research focuses on the adaptations and extinctions of Neanderthals and the African record of Homo sapiens. She is fascinated by the cognitive aspects of tool construction from primates to the present, variability within the MSA and the pan-African connection of people and technologies.