The early years: linguistic and developmental psychology models

ECA is a relatively new discipline in which the earliest analyses principally focused on the appearance and evolution of language and symbolism (Bounak Reference Bounak1958; Leroi-Gourhan Reference Leroi-Gourhan1964; Steele et al. Reference Steele, Quinlan and Wenban-Smith1995). The first ECA models were originally based on psychological and linguistic theories, such as those by Chomsky and Lieberman (but see Nowell Reference Nowell2000 and Wynn Reference Wynn1991 for a review). These models were employed to draw analogies between language origin and cognitive evolution, normally through the analysis of archaeological material. Linguistic models are based on the assumption that the operational steps employed in tool manufacture and use reflect the same neural functions as the syntax of language (Atran Reference Atran1982; Leroi-Gourhan Reference Leroi-Gourhan1964; Wynn Reference Wynn, Gibson and Ingold1993a). In terms of lithic technology, such parallels are based on the fact that both processes, language and stone toolmaking, are sequential, hierarchical and goal-directed (Greenfield Reference Greenfield1991; Wynn Reference Wynn, Gibson and Ingold1993a).

One of the earliest linguistic models was presented by Ralph Holloway (Reference Holloway1969). He correlated patterns of stone-tool production with those of syntactical communication, arguing that tool manufacture and language were similar cognitive processes, emphasizing, in particular, the imposition of an arbitrary form (‘symbolisation’) and the hierarchical organization of rules. Holloway (Reference Holloway1969) also suggested the spatial and temporal distance between a stimulus and its consequent action as a way of measuring intelligence, although he clearly established that the lithic or fossil record could not be taken as direct evidence of the existence of language.

Similarly, later on, Glynn Isaac (Reference Isaac1976) proposed the analysis of material culture systems, economic behaviour and adaptive patterns of early hominins to understand the evolution of language. Not only did Isaac propose using artefact complexity as an indicator of cognitive capabilities, but he also suggested a methodology based on different variables: number and variability of artefact classes, number of steps involved in artefact manufacture, use of compound artefacts and appearance of regionalism. He proposed that the increase seen in lithic technological complexity might have affected the enhancement of communication and information exchange systems, resulting in the formation of hominin cultural and communication capabilities (Isaac Reference Isaac1976).

Perhaps the most direct relationship between hominin cognition and lithic remains at this time was made by John Gowlett, who also emphasized the imposition of arbitrary form, variability and standardization, but went further in proposing the analysis of the Acheulean Large Cutting Tool (LCTs, i.e. handaxes and cleavers) chaîne opératoire as a way of inferring forward planning and cognitive mapping (Gowlett Reference Gowlett1979). As most studies at the time relied primarily on the aesthetic properties of tools and their possible use as symbols, Gowlett’s emphasis on the use of the operational sequences involved in tool manufacture is striking, being especially critical of authors who minimized the skill and abilities required for it (Gowlett Reference Gowlett1984).

Similarly, during the 1970s and ’80s, several other researchers first attempted to apply formal cognitive models to the archaeological record using ontogenetic human models. These were used in Developmental Psychology to interpret phylogenetic sequences and implicitly assumed that human cognitive development recapitulates primate and human evolutionary history (Wynn Reference Wynn2017). Initially, many of these ECA models were based on Jean Piaget’s theories on logical and spatial intelligence (Piaget & Inhelder Reference Piaget and Inhelder1956), such as the work of Parker and Gibson (Reference Parker and Gibson1979) and Wynn (Reference Wynn1979; Reference Wynn1981; Reference Wynn1985; Reference Wynn1989). Parker and Gibson (Reference Parker and Gibson1979) applied Piaget’s scheme to non-human primate cognition and to the Homo habilis home base model proposed by Isaac (Reference Isaac1978), concluding that H. habilis possessed some sort of ‘protolanguage’ and rudimentary forms of sensorimotor and early preoperational intelligence similar to those proposed by Piaget for human children (Parker & Gibson Reference Parker and Gibson1979). Simultaneously and independently, Thomas Wynn (Reference Wynn1979; Reference Wynn1981; Reference Wynn1985; Reference Wynn1989) used Piaget’s stages of childhood development to analyse Oldowan and Acheulean artefacts through three main geometric concepts: topological (such as proximity, separation, order, continuity or whole-part competence), projective (perspective) and Euclidean concepts (such as radius, diameter and bilateral symmetry). Both Parker and Gibson’s and Wynn’s models were criticized due to the weakness of the theory underlying them and the lack of consistency in the lithic analyses used and performed. Indeed, by that time, Piaget’s theories were already the subject of criticism (Atran Reference Atran1982; Robson Brown Reference Robson Brown1993): not only do modern children not pass through all the proposed sequential stages of intelligence, but also many of the human cognitive abilities were not indicators of a unique general intelligence, and therefore needed to be treated separately (Wynn Reference Wynn2016; Reference Wynn2017). Moreover, Piaget’s stages were developed from observation of modern human infants, and therefore their application to other species is problematic (Mithen Reference Mithen1995; Nowell Reference Nowell2000). Wynn’s morphological analysis of the archaeological record ultimately failed to confirm the predictions of the Piagetian theory: according to Piaget’s spatial competence development, fully Euclidean concepts appeared after projective concepts, while in Wynn’s analysis, both concepts appeared at the same time, therefore placing the modern adult level of intelligence as far back as 500–300 Kya (Wynn Reference Wynn1979; Reference Wynn1989; Reference Wynn1999; Reference Wynn2016). The Piagetian theory did not match the neuroscience studies on brain functioning, and Wynn eventually abandoned this approach entirely in the 1990s to move into modular mind models and Cognitive Neuroscience (Wynn Reference Wynn2000; Reference Wynn2016; Reference Wynn2017). Wynn (Reference Wynn2002) realized the limits of the Piagetian approach and argued that, while cognition evolved after 300 Kya, spatial perception was probably not an important component of this evolution (Wynn Reference Wynn2000; Reference Wynn2002).

Nevertheless, models based on developmental psychology were (and still are) used, as seen, for example, in the work by Robson Brown (Reference Robson Brown1993) on the Zhoukoudian lithic assemblage and Stephen Mithen (1994; Reference Mithen1996), who placed the appearance of modern cognition around 60–30 Kya due to the increase of cognitive fluidity in the minds of the first Homo sapiens. Instead of using Piaget’s theories, Robson Brown and Mithen’s analyses (Mithen 1994; Reference Mithen1996; Robson Brown Reference Robson Brown1993; Wynn Reference Wynn2000; Reference Wynn2002) were based on the multiple intelligences and the modular mind models (Fodor Reference Fodor1983; Gardner Reference Gardner1983) and concepts of social intelligence. Linguistic models are also still in use nowadays, as in Robert Mahaney’s work based on Acheulean experimental replication (2014; 2015), or Rudolf Botha’s Windows Approach (2022).

Finally, other relevant studies in Cognitive Archaeology at this time were works by Merlin Donald (Reference Donald1993) and Davidson and Noble (Reference Davidson and Noble1993; Davidson Reference Davidson2002). Donald (Reference Donald1993) proposed three major transitional periods in hominin cognitive evolution: 1) the appearance of H. erectus and increase in brain size; 2) a biological change that included an increase in brain size and descent of the larynx, and the emergence of spoken language; and 3) the Late Upper Palaeolithic and the ‘invention of the first permanent visual symbols’ (Donald Reference Donald1993). Similarly, Davidson and Noble developed their own theory of language and perception from ecological psychology (Davidson et al. Reference Davidson, Noble and Armstrong1989; Davidson & Noble Reference Davidson and Noble1993), placing the appearance of modern human cognition around 60 Kya. Both approaches encountered considerable criticism because they were based on general assumptions about prehistory and the Palaeolithic (Wynn Reference Wynn2017) and lacked empirical support (Putt Reference Putt2016).

Evolutionary psychology and social cognition: the social brain hypothesis and the theory of mind

In recent years, hominin social systems have gained importance in archaeological studies (Goren-Inbar & Belfer-Cohen Reference Goren-Inbar and Belfer-Cohen2020; Gowlett et al. Reference Gowlett, Gamble and Dunbar2012; Pappu & Akhilesh Reference Pappu and Akhilesh2019). Through a combination of strategies from Palaeoneurology and Evolutionary and Cognitive Psychology, some Palaeolithic archaeologists (Cole Reference Cole2011; Reference Cole2012; Reference Cole2014; Reference Cole2015a; Reference Cole2017; Gamble et al. Reference Gamble, Gowlett and Dunbar2011; Gowlett et al. Reference Gowlett, Gamble and Dunbar2012; Lycett Reference Lycett2008; Lycett et al. Reference Lycett, Schillinger, Eren, von Cramon-Taubadel and Mesoudi2016; McNabb Reference McNabb2012; McNabb & Cole Reference McNabb and Cole2015; Shipton 2010; Reference Shipton2013; Shipton et al. Reference Shipton, Petraglia and Paddayya2009; Stade Reference Stade2017; Reference Stade2020) have tried to infer cognitive evolution from a Social Cognition perspective. This approach is justified by the fact that shape standardization in lithic implements is perceived as the result of social learning and/or socially agreed cultural practices within hominin groups.

In Evolutionary Psychology, Social Cognition evaluates the cognitive processes responsible for social behaviour and social relationships. One theory deriving from this discipline is the Social Brain Hypothesis, formerly named ‘Machiavellian Intelligence Hypothesis’ (Byrne & Whiten Reference Byrne and Whiten1988), which refers to the direct relationship between primate relative brain size (especially neocortex volume) and social group size (Byrne & Whiten Reference Byrne and Whiten1988; Cole Reference Cole2017; Dunbar Reference Dunbar1998; Dunbar & Shultz Reference Dunbar and Shultz2007; Gamble et al. Reference Gamble, Gowlett and Dunbar2011; Gowlett et al. Reference Gowlett, Gamble and Dunbar2012; Shultz & Dunbar Reference Shultz and Dunbar2007; Shultz et al. Reference Shultz, Nelson and Dunbar2012). The neocortex area is related to cognitive processes associated with reasoning and consciousness (Dunbar Reference Dunbar1998) and some researchers suggest that neocortex expansion was driven by the cognitive demands resulting from more complicated social relationships (Dunbar Reference Dunbar1998). Recent analyses, however, diminish the role of social complexity and argue that, in fact, ecological explanations (DeCasien et al. Reference DeCasien, Williams and Higham2017; González-Forero & Gardner Reference González-Forero and Gardner2018) and the stability of relationships seem to be more important than the number of individuals in the group (Dunbar & Shultz Reference Dunbar and Shultz2007; Shultz & Dunbar Reference Shultz and Dunbar2010).

Palaeoanthropological studies have used the Social Brain Hypothesis to estimate group size from the cranial volume of extinct hominins, which seems to indicate an increase in group size since the appearance of H. heidelbergensis/Archaic H. sapiens (AHS) around 600 Kya (Aiello & Dunbar Reference Aiello and Dunbar1993; Gowlett et al. Reference Gowlett, Gamble and Dunbar2012). According to Evolutionary Psychology and Primatology studies, these enhanced cognitive skills allow two kinds of social cognition present in primates but not in other species (Dennett Reference Dennett1983; Shultz & Dunbar Reference Shultz and Dunbar2014): self-recognition and the Theory of Mind (or mentalization: recognizing intention and emotion in other individuals). The Theory of Mind corresponds to Dennett’s (Reference Dennett1983) second-order intentionality (Table 1). While mammals and birds can reach first-order intentionality, great apes seem to be able to reach second-order intentionality (and therefore, Australopithecines should have been able to as well), while adult humans can normally attain a fifth order (Shultz & Dunbar Reference Shultz and Dunbar2014). Thus, these higher orders should have been acquired through the evolution of Homo and are linked to abstract or symbolic thinking, perhaps even to ‘visualise the end product of a tool in the raw material of a core’ (Gowlett et al. Reference Gowlett, Gamble and Dunbar2012). As hominin intentionality increased, such ability should be visible in the archaeological record, perhaps when artefacts were used as social representations or icons (McNabb & Cole Reference McNabb and Cole2015). In this sense, cultural continuity throughout the Acheulean technocomplex was related to the existence of stable forms of social transmission involving shared intentionality and imitation (Lycett & Gowlett Reference Lycett and Gowlett2008; Shipton 2010; Reference Shipton2013; Shipton et al. Reference Shipton, Petraglia and Paddayya2009).

Table 1.

Orders of intentionality represent a scale for measuring cognitive complexity. Modern humans can operate at up to four or five orders of intentionality, although most everyday human relationships operate in the second order. (Adapted from McNabb Reference McNabb2012.)

Material Engagement Theory

Bridging anthropology, evolutionary psychology and materiality, archaeologists Colin Renfrew and Lambros Malafouris proposed, in the early twenty-first century, the Material Engagement Theory (MET), a theoretical framework that explores the interaction between the brain, the body and material culture (Malafouris Reference Malafouris2013; Malafouris & Gosden Reference Malafouris, Gosden, Gaskell and Carter2020; Malafouris & Renfrew Reference Malafouris and Renfrew2010; Renfrew Reference Renfrew2012). MET challenges traditional brain-centred models that separate mind and matter, such as the concept of modular mind, advocating instead for an extended-mind hypothesis (Malafouris Reference Malafouris2013; Reference Malafouris2021a; Malafouris & Renfrew Reference Malafouris and Renfrew2010) that addresses both extended and embodied cognition (Wynn et al. Reference Wynn, Overmann and Malafouris2021).

MET argues that cognition is not just in the brain but extends into material interactions, so that objects — or ‘things’ — actively shape human thought and behaviour. While MET serves primarily as an explanatory framework rather than a predictive theory, its applications have expanded significantly in recent years. Although early MET postulates rarely used early stone tool technology or the Acheulean technocomplex as case studies (Coward & Gamble Reference Coward and Gamble2009; Malafouris Reference Malafouris2004; Reference Malafouris2008; but see Malafouris Reference Malafouris2010), more recently this theoretical framework has provided useful in the design and interpretation of analyses on social (Barona Reference Barona2021) and haptic (Bruner et al. Reference Bruner, Spinapolice, Burke and Overmann2018; Cueva-Temprana et al. Reference Cueva-Temprana, Lombao, Morales, Geribàs and Mosquera2019; Wynn Reference Wynn2021) cognition, biomechanics (Baber & Janulis Reference Baber and Janulis2021; Key & Dunmore Reference Key and Dunmore2018), palaeoneurology (Coolidge Reference Coolidge2021; Coward & Gamble Reference Coward and Gamble2009) and neuroarchaeology (Hecht et al. Reference Hecht, Gutman and Kreisheh2015; Malafouris Reference Malafouris2009; Putt et al. Reference Putt, Wijeakumar, Franciscus and Spencer2017; Stout & Chaminade Reference Stout and Chaminade2007; Stout et al. Reference Stout, Hecht, Khreisheh, Bradley and Chaminade2015; see extended discussion below), as it focuses on the interaction of mind, body and objects through the dynamic process of artefact toolmaking (Malafouris Reference Malafouris2010; Reference Malafouris2019; Overmann & Wynn Reference Overmann and Wynn2019). Overall, MET reinforces the concept of metaplasticity, in the sense that human cognition is adaptable and continuously shaped by cultural (i.e. material) engagement, rejecting the idea of a so-called ‘behavioural modernity’ (Malafouris Reference Malafouris2009; Reference Malafouris2013; Reference Malafouris2015; Reference Malafouris2021b; Malafouris & Gosden Reference Malafouris, Gosden, Gaskell and Carter2020; Roberts Reference Roberts2016).

The boost of neuroarchaeology

Embracing the possibilities opened by neuroscience (Putt Reference Putt2016), perhaps the most innovative studies recently made in ECA are based on neuroarchaeological models. Neuroarchaeology applies various neurosciences to solve cognitive archaeological questions, as well as including archaeological data in neuroscience theorizing (Hecht & Stout Reference Hecht, Stout, Wynn, Overmann and Coolidge2023; Laughlin Reference Laughlin2015; Stout & Hecht Reference Stout, Hecht, Wynn, Overmann and Coolidge2023). Neuroarchaeology aims to analyse brain function in relation to hominin behaviour, normally through an experimental approach, so that it allows testing some of the theoretical models earlier discussed. Experimental replication of hominin behaviour, most commonly stone-tool making, is performed while simultaneously (or immediately after) brain activity and patterns of neuroactivation are being mapped through a neuroimaging device (Stout et al. Reference Stout, Toth, Schick, Stout and Hutchins2000; Wynn Reference Wynn2017).

The earliest examples of neuroarchaeological studies appeared in the 2000s, with several analyses using different neuroimaging techniques such as Positron Emission Tomography (PET), functional magnetic resonance imaging (fMRI) or functional near-infrared spectroscopy (fNIRS), among others, to elucidate different aspects of stone-knapping behaviour. Images collected during experimental tasks are normally compared with those collected under controlled conditions so that isolated changes in brain-activation patterns during toolmaking can be identified (Stout Reference Stout2005; Stout et al. Reference Stout, Toth, Schick, Stout and Hutchins2000).

Some neuroarchaeological studies have been able to correlate brain functioning with Oldowan (Stout et al. Reference Stout, Toth, Schick, Stout and Hutchins2000; Reference Stout, Passingham, Frith, Apel and Chaminade2011; Reference Stout, Hecht, Khreisheh, Bradley and Chaminade2015) and Acheulean (Stout et al. Reference Stout, Toth and Schick2006; Reference Stout2011; Reference Stout, Hecht, Khreisheh, Bradley and Chaminade2015) knapping techniques. These studies have revealed, for example, that when replicating Oldowan knapping strategies, the activated brain structures are related to complex spatial perception, grip perception and sensorimotor coordination (Stout Reference Stout2005), while in Acheulean replication, other cortex areas implicated in hierarchical organization are also activated (Hecht et al. Reference Hecht, Gutman and Kreisheh2015; Stout et al. Reference Stout, Passingham, Frith, Apel and Chaminade2011). Moreover, other experiments have looked at brain activation in novice and expert knappers, intending to understand brain functioning during learning (Stout & Chaminade Reference Stout and Chaminade2007; Stout et al. Reference Stout, Toth, Schick and Chaminade2008; Reference Stout, Passingham, Frith, Apel and Chaminade2011; Reference Stout, Hecht, Khreisheh, Bradley and Chaminade2015; Putt Reference Putt2016) with and without spoken language (Putt Reference Putt2016). When comparing Oldowan and Acheulean tool manufacture, researchers (Stout et al. Reference Stout, Toth, Schick and Chaminade2008; Reference Stout, Hecht, Khreisheh, Bradley and Chaminade2015; Putt et al. Reference Putt, Wijeakumar, Franciscus and Spencer2017; Reference Putt, Wijeakumar and Spencer2019) suggest that Oldowan technology activates areas of the brain related to visual attention and sensorimotor control, but not those required for executive functions (Stout & Chaminade Reference Stout and Chaminade2007). According to these studies, Oldowan is an ‘ape-like’ technology, as previously proposed (Wynn Reference Wynn1989; Wynn & McGrew Reference Wynn and McGrew1989). In contrast, Acheulean technology involved areas of the brain related to higher-order motor planning, the central executive of working memory and auditory feedback mechanisms, which suggests that Acheulean technology may have helped the development of neural connections involved in speech perception (Putt et al. Reference Putt, Wijeakumar, Franciscus and Spencer2017; Stout et al. Reference Stout, Hecht, Khreisheh, Bradley and Chaminade2015).

Of course, as with any other scientific field, neuroarchaeology has limitations. It primarily faces two major drawbacks: first, the use of modern human brains as analogues of those of extinct hominins (Pargeter et al. Reference Pargeter, Khreisheh and Stout2019; Putt Reference Putt2016; Wynn Reference Wynn2017); second, experimental neuroarchaeology does not offer direct resources for inferring hominin behaviour from archaeological remains, which creates ambiguity in inferences about human evolution (Pargeter et al. Reference Pargeter, Khreisheh and Stout2019). Nevertheless, it has proved to be an extraordinary tool for interpreting hominin behaviour and the evolution of cognition. Neuroarchaeology demonstrated that earlier studies were at least partially right in assuming that manual and perceptual motor adaptations were important in the earliest hominin technologies and enhanced cognitive control was more important in later ones, reflecting the increasing complexity of hierarchical organization involved in Acheulean operational sequences (Stout et al. Reference Stout, Passingham, Frith, Apel and Chaminade2011; Reference Stout, Hecht, Khreisheh, Bradley and Chaminade2015).

Tennie’s Zone of Latent Solutions

Yet another significant theoretical framework related to ECA and, especially, studies in the origins of stone-tool production is the Zone of Latent Solutions (ZLS) (Corbey et al. Reference Corbey, Jagich, Vaesen and Collard2016; Tennie et al. Reference Tennie, Call and Tomasello2009; Reference Tennie, Braun, Premo and McPherron2016; Reference Tennie, Premo, Braun and McPherron2017). This refers to novel behaviours (i.e. ‘solutions’) that lie dormant in an individual until triggered by social and/or environmental cues and sufficient motivation on the part of the learner (Tennie et al. Reference Tennie, Braun, Premo and McPherron2016). That is, behaviours that are genetically driven (Corbey et al. Reference Corbey, Jagich, Vaesen and Collard2016) and emerge within a species without any prior cultural transmission or cumulative culture (Tennie et al. Reference Tennie, Call and Tomasello2009). ZLS behaviours are, therefore, those that arise solely through individual innovation but always within an organism’s latent cognitive ‘repertoire’ (and when the appropriate learning conditions are met). Tennie’s ZLS hypothesis highlights the significance of understanding triple (i.e. genes-culture-environment) inheritance theory (Tennie et al. Reference Tennie, Braun, Premo and McPherron2016), while also emphasizing the need to distinguish between behaviours that rely on high-fidelity transmission from those that can be spontaneously discovered or rediscovered by individuals. Tennie et al. (Reference Tennie, Call and Tomasello2009) propose that ZLS low-fidelity behaviours arise from the interaction between an individual’s genetic predisposition and their environment and that, additionally, the emergence of these behaviours can be shaped by the influence of low-fidelity social learning mechanisms.

In the context of lithic technology, the ZLS hypothesis serves as a framework for interpreting appearance and variations of Early Pleistocene cultural industries (Corbey Reference Corbey2020; Corbey et al. Reference Corbey, Jagich, Vaesen and Collard2016; Tennie et al. Reference Tennie, Braun, Premo and McPherron2016; Reference Tennie, Premo, Braun and McPherron2017). Since primate archaeology and primatology studies suggest that behaviours such as chimpanzees’ nut cracking or simple flaking by capuchins fall within their respective ZLS (Tennie et al. Reference Tennie, Call and Tomasello2009), this raises the question of when and why high-fidelity social learning and cumulative culture appeared and whether the first lithic technologies, such as the Lomekwian and the Oldowan, were in fact dependent on cultural transmission as opposed to being rediscoverable innovations within the ZLS of early hominins (Corbey Reference Corbey2020; Corbey et al. Reference Corbey, Jagich, Vaesen and Collard2016; Tennie et al. Reference Tennie, Braun, Premo and McPherron2016; Reference Tennie, Premo, Braun and McPherron2017).

Applying the ZLS framework to Acheulean technology, however, presents a more complex picture. Tennie argues that shape variation in Acheulean LCTs is due to differences in raw material variability and reduction intensity (Tennie et al. Reference Tennie, Braun, Premo and McPherron2016, after White Reference White1998; McPherron Reference McPherron2007), and this seems to be the case at least for the oldest Acheulean (Diez-Martín et al. Reference Diez-Martín, Wynn and Sánchez-Yustos2019; Martín-Ramos Reference Martín-Ramos2022, but see Sharon Reference Sharon2008; Reference Sharon2010; McNabb et al. Reference McNabb, Binyon and Hazelwood2004). This overall idea of cultural conservatism is due, however, to a research bias that tends to look at the lithic record from a pure aesthetic (i.e. morphological) perspective (Tennie et al. Reference Tennie, Braun, Premo and McPherron2016: 127). Even from a morphometrical approach alone, this view might still result in an oversimplification of a technocomplex that spans over 1.5 million years (McNabb Reference McNabb2019). While low-fidelity core-and-flake technologies might lie within the ZLS of early Homo (Morgan et al. Reference Morgan, Uomini and Rendell2015), the hierarchical organization and technical variability observed in Acheulean LCT and large-flake predetermination (Martín-Ramos Reference Martín-Ramos2022; Sharon 2008; Reference Sharon2010) necessarily implied the existence of more or less complex or enhanced cognitive processes (such as shared intentionality and cumulative cultural learning) and socially agreed and transmitted design imperatives and procedural knowledge (Herzlinger et al. Reference Herzlinger, Wynn and Goren-Inbar2017; Shipton Reference Shipton2010; Shipton & Nielsen Reference Shipton and Nielsen2015). Overall, this binary distinction between so-called individual learning and social learning represents a too-simplistic view, perhaps as much oversimplification of social and cultural processes as it is to reduce the vast spectrum of past human behaviours to the mere visual shape of lithic tools (Martín-Ramos Reference Martín-Ramos2022: 37). Still, while the ZLS theory may not fully explain the cognitive and cultural underpinnings of all lithic technologies, its emphasis on the role of individual learning and environmental affordances provides a valuable counterpoint to models emphasizing solely cultural transmission.

Working memory and the Expert Cognition model

An archaeologist, Thomas Wynn, and a neuropsychologist, Frederick Coolidge, together developed the ‘enhanced working memory’ approach based on Baddeley and Hitch’s working memory system (1974). Working memory is a cognitive process with a long evolutionary history (Coolidge & Wynn Reference Coolidge and Wynn2009a; Putt Reference Putt2016; Wynn & Coolidge Reference Wynn and Coolidge2011). It is inheritable, related to general fluid intelligence and short and long-term memory, and has been supported by cognitive psychology experiments (Baddeley Reference Baddeley2010). The most widely accepted working memory model was proposed by Alan Baddeley and Graham Hitch (Baddeley 1993; Reference Baddeley2010; Baddeley & Hitch Reference Baddeley, Hitch and Bower1974) (Fig. 1) and explains the operations of short-term memory, how memory is instructed and directed, and how it relates to long-term memory.

Figure 1.

(A) Schematic representation of the multicomponent Working Memory Model by Alan Baddeley (Reference Baddeley2010); (B) extended version of Baddeley’s (Reference Baddeley2001) model by Coolidge and Wynn (Reference Coolidge and Wynn2005).

Through the application of the ‘enhanced working memory’ hypothesis, Coolidge and Wynn (Reference Coolidge and Wynn2005; Reference Coolidge and Wynn2009a) attempted to adapt Baddeley’s model to the archaeological record. The technical evidence they use to correlate with enhanced working memory includes traps and snares, reliable weapons, and hafting technology (Wynn & Coolidge Reference Wynn and Coolidge2011). They argue for a late genetic mutation occurring at around 100 Kya that increased working memory capacity in H. sapiens, although preceded by other components of modern cognition that had evolved long before (Coolidge & Wynn Reference Coolidge and Wynn2005; Coolidge et al. Reference Coolidge, Haidle, Lombard and Wynn2016; Wynn & Coolidge Reference Wynn and Coolidge2004). Still, Coolidge and Wynn do not see any signals for a ‘modern’ working memory in the minds of H. erectus and other pre-sapiens hominins. They argue that there is no evidence for long-range planning and innovation (Coolidge & Wynn Reference Coolidge and Wynn2009b), although they acknowledge some cognitive development in H. heidelbergensis/AHS due to the existence of prepared core technique and the fact that Acheulean LCTs could have been seen as icons or ‘something else than just tools’ (Coolidge & Wynn Reference Coolidge and Wynn2009b).

Ultimately, according to Coolidge and Wynn’s hypothesis, ‘pre-sapiens’ technologies required procedural and long-term memory but not enhanced working memory. Several researchers, particularly within neuroarchaeology, disagree with this hypothesis and support the emergence of different aspects of working memory over time (Putt Reference Putt2016), suggesting, for example, that in comparison with Oldowan technology, Acheulean toolmaking required increased working memory and cognitive control (Putt et al. Reference Putt, Wijeakumar, Franciscus and Spencer2017; Stout et al. Reference Stout, Hecht, Khreisheh, Bradley and Chaminade2015). In addition, neuroarchaeological studies have recently shown that both Oldowan and Acheulean industries require visual working memory, and that Acheulean toolmaking requires complex motor planning (Putt et al. Reference Putt, Wijeakumar, Franciscus and Spencer2017; Stout Reference Stout2008). Enhanced working memory would then have been a necessary component of Acheulean toolmaking, where the knapper needs to keep in mind the goal and sub-goals related to handaxe flaking. Technological studies, such as that performed by Herzlinger and colleagues (Herzlinger et al. Reference Herzlinger, Wynn and Goren-Inbar2017) on Large Flake Acheulean and cleaver manufacture at the Gesher Benot Ya’aqov site (Israel) suggest that Acheulean toolmaking required more and longer procedural sequences held in long-term memory and also an increase in working memory capacity. Furthermore, recent neuroarchaeological work by Shelby Putt (Reference Putt2016; Putt et al. Reference Putt, Wijeakumar and Spencer2019) proposes the ‘working memory hypothesis for hominin brain expansion’, suggesting that LCT production was a trigger for the evolution of larger working memory capacities because of the reproductive benefits this enhanced working memory would have had. Hominins who were better knappers and those whose tools were most functional would have been the healthiest in the population, something that would have reproductive benefits and advantages for the early years of their offspring (Putt Reference Putt2016). Ultimately, the Enhanced Working Memory approach is related to expert skill. The Expert Cognition or Long-term Working Memory model is based on the fact that expert performance is stored in long-term memory and can be assessed through several characteristics (Table 2) evident in the stone tool record or through experimental replication (Coolidge et al. Reference Coolidge, Haidle, Lombard and Wynn2016; Fajardo et al. Reference Fajardo, Kozowyk and Langejans2023; Wynn & Coolidge Reference Wynn and Coolidge2004; Reference Wynn and Coolidge2011; Wynn et al. Reference Wynn, Haidle, Lombard and Coolidge2017).

Table 2.

Characteristics of expert performance, according to the Expert Cognition model (modified from Wynn et al. Reference Wynn, Haidle, Lombard and Coolidge2017)

Essentially, working memory is the ability to keep a goal in mind when performing secondary duties (Baddeley Reference Baddeley2010; Coolidge & Wynn Reference Coolidge and Wynn2005; Reference Coolidge and Wynn2009a; Hallos Reference Hallos2005; Wynn & Coolidge Reference Wynn and Coolidge2011), which is necessary when performing complex tasks (Fajardo et al. Reference Fajardo, Kozowyk and Langejans2023; Putt Reference Putt2016) and relates to inhibition and self-control (Green & Spikins Reference Green and Spikins2020). Haidle (Reference Haidle2010) expands this definition by adding that working memory serves to focus attention by maintaining memory representation in a conscious state despite interference or response competition (Coolidge & Wynn Reference Coolidge and Wynn2005; Haidle Reference Haidle2010). While neuroarchaeology has proved its validity in hominin cognitive evolution, there is still the problem of its application to the archaeological material record. Nevertheless, combined with procedural analysis of tool manufacture, so far, the Working Memory model has proved to be useful and reliable in the analysis of the cognitive implications of archaeological remains.

Lithic technology, the Acheulean technocomplex and the cognitive abilities of extinct hominins

Lithic technology, the Acheulean technocomplex and the cognitive abilities of extinct hominins are closely interlinked subjects that examine early human tool-making skills and mental capabilities. As shown throughout this paper, research on hominin cognition has often relied on stone tools to investigate the origins of language, abstract thought, symbolism and ‘modern thinking’ (Nowell et al. Reference Nowell2003). This focus is partly due to the excellent preservation of lithic artefacts in archaeological contexts as direct evidence of hominin behaviour. Additionally, stone tools offer insights into technological and subsistence strategies, which may have been shaped by cognitive differences among our ancestors.

In the case of the Acheulean technocomplex, it has been hypothesized, and so far also tested experimentally, that it required more cognitive control and working memory capacity than the preceding Oldowan technocomplex (Gowlett Reference Gowlett1986; Reference Gowlett1996; Isaac Reference Isaac1986; Stout et al. Reference Stout, Toth and Schick2006; Wynn Reference Wynn1989; Reference Wynn1993b). This is based on the premise that LCT manufacture requires the toolmaker to proceed through a series of complex sequences of actions that involve long-term planning and hierarchical goals (Belfer-Cohen & Goren-Inbar Reference Belfer-Cohen and Goren-Inbar1994; Putt et al. Reference Putt, Wijeakumar, Franciscus and Spencer2017; Stout Reference Stout2011; Wynn & Coolidge Reference Wynn and Coolidge2010). Likewise, Acheulean LCTs are also the first evidence of the deliberate imposition of shape and form. Such intentional façonnage, produced through more or less complex knapping strategies, reflects the existence of procedural and mental templates in the minds of knappers, as well as degrees of forward planning (Ashton & McNabb Reference Ashton and McNabb1994; Gowlett Reference Gowlett1986; Reference Gowlett2006; Pope et al. Reference Pope, Wells and Watson2006; Sharon et al. Reference Sharon, Alperson-Afil and Goren-Inbar2011).

Additionally, it has also been argued that LCT form is sometimes over-determined (i.e. excessively worked, beyond what would be functionally necessary), with Acheulean knappers applying special effort in producing specific shapes not necessary for functional reasons (Wynn & Gowlett Reference Wynn and Gowlett2018). This relates to perhaps the most debated feature of (some) Acheulean LCTs: symmetry. Whether bilateral, volumetric or cross-section, because it does not have a substantial effect on LCT performance (Machin et al. Reference Machin, Hosfield and Mithen2005; Reference Machin, Hosfield and Mithen2007) and it is not a by-product of the flaking process (Shipton et al. Reference Shipton, Clarkson and Cobden2018), LCT symmetry has been taken as a consequence of brain (Hodgson Reference Hodgson2009) and spatial cognition development (Wynn Reference Wynn1979; Reference Wynn1989; Reference Wynn2000; Reference Wynn2002).

Finally, cognitive implications have not solely relied on the visual aspect of LCTs. Discussions on cognitive evolution have also looked at the concept of predetermination of form and the increase of technological complexity when compared to Lomekwian and Oldowan artefacts (Petraglia et al. Reference Petraglia, Shipton and Paddayya1999; Ranov Reference Ranov2001). Thus, the Acheulean chaîne opératoire has been taken as a foundation for the study of H. erectus and Middle Pleistocene hominins’ mental development, arguing that longer artefact transportation and more complex and hierarchized knapping sequences must have required enhanced working memory and forward planning (Cole Reference Cole2011; Herzlinger et al. Reference Herzlinger, Wynn and Goren-Inbar2017; Hodgson Reference Hodgson2015; Shipton Reference Shipton2013; Stout et al. Reference Stout, Hecht, Khreisheh, Bradley and Chaminade2015). All these are indicative of important cognitive developments occurring during the long time span of the Acheulean technocomplex.

Methodological approaches for assessing hominin cognition from a technological perspective

Hopefully, this review has highlighted the usefulness of experimental approaches (including neuroarchaeology) in the assessment of hominin cognitive capacity. Additionally, archaeologists have also employed the longstanding concept of the chaîne opératoire in the assessment of hominin cognitive capabilities through direct analyses of stone tools. This framework was proposed in the early years of cognitive archaeology as a way of understanding the evolution of hominin cognition and language (Leroi-Gourhan Reference Leroi-Gourhan1964; Schlanger Reference Schlanger1994), with the idea of developing conceptual operative schemas already introduced in the 1980s and the early 1990s (Karlin & Julien Reference Karlin and Julien1994; Pelegrin Reference Pelegrin1990; Perlès Reference Perlès1992; Pigeot Reference Pigeot1991; Schlanger Reference Schlanger1994). Around this time, John Gowlett (1982; Reference Gowlett1984) also mentioned ‘operational chains’ and ‘procedural diagrams’ as a way of mapping technological complexity and individual planned actions in Early Stone Age industries. In this sense, while some of the strictest applications of chaînes opératoires lack explicit cognitive justification and their conclusions are relatively vague (Stout Reference Stout2011; Wynn et al. Reference Wynn, Haidle, Lombard and Coolidge2017), several authors recently developed different ways of addressing and representing conceptual operational sequences or ‘mental chaînes opératoires’ (Fairlie & Barham Reference Fairlie and Barham2016; Haidle 2010; Reference Haidle2012; Reference Haidle, Wynn, Overmann and Coolidge2023; Muller et al. Reference Muller, Clarkson and Shipton2017; Stout Reference Stout2011).

Dietrich Stout, for example, employed tree diagrams that represent hierarchical and subordinate actions and goals involved in Acheulean tool manufacture, providing a standard format for a technological comparison, which can be useful in assessing cognitive constraints. By using action hierarchies (Fig. 2A) he was able to represent goals, sub-goals and temporally prolonged processes, from the overall objective (i.e. Early Acheulean flake production) to specific motor acts (i.e. rotate core). Drawing action hierarchies of Oldowan and Acheulean technologies permitted him to infer cumulative cultural change in Lower Palaeolithic industries, with the appearance of more varied end-products being driven by an increase in technical and hierarchical complexity (Stout Reference Stout2011).

Figure 2.

Hierarchical diagrams showing Early and Late Acheulean handaxe manufacture by (A) Stout (Reference Stout2011) and (B) Muller et al. (Reference Muller, Clarkson and Shipton2017). Stout’s models focus more on representing the chaîne opératoire, while Muller and colleagues underline mental goals and active foci through the knapping process.

Fairlie and Barham (Reference Fairlie and Barham2016) presented a method not based on direct stone technology analysis but on observational analysis derived from psychological science, the chaîne opératoire approach and the perception-action theory (which infers cognition processes from motor activities). They addressed several behavioural variables observable through experimental replication of stone-tool manufacture (handling and rotation of the object, flow and pace during the knapping sequence, etc.), which can be used to outline cognitive changes in Stone Age manufacture. As with neuroarchaeology, its drawback lies in the difficulties of applying their results to direct artefact analysis.

Miriam Haidle (Reference Haidle2009; Reference Haidle2010; Reference Haidle2012; Reference Haidle, Wynn, Overmann and Coolidge2023; Lombard & Haidle Reference Lombard and Haidle2012) developed an interesting way of addressing working memory and executive function development through archaeological remains: the cognigrams. The theoretical basis for her method is the problem-solution distance approach, which recognizes tool behaviour as a process of indirect thinking (Lombard & Haidle Reference Lombard and Haidle2012). The model is related, in fact, to Holloway’s (Reference Holloway1969) approach, one of the earliest examples of ECA studies mentioned at the beginning of this paper, which suggested measuring the spatial and temporal distance between a stimulus and its subsequent action as a way of determining intelligence development. Haidle’s methodology enables the measurement of cognitive complexity, flexibility and decision-making by reconstructing the thought-and-action sequence involved in tool manufacture and use. The model allowed her to compare the manufacture and use of tools between animals and hominins, including Oldowan and Acheulean technologies, and composite, projectile and complementary tools. Her studies (Haidle Reference Wynn2010; Reference Haidle2012; Lombard & Haidle Reference Lombard and Haidle2012) indicated that problem-solving in animals is restricted to problems for which a solution can be found in the spatial and temporal vicinity. During human evolution, however, the complexity of tool behaviour increases with a higher number of active foci and operational steps and a larger spatial and temporal frame. Haidle’s cognigram methodology (Haidle Reference Haidle2010; Reference Haidle2012) enables the representation of a subject’s perception of a need and the different attention foci that are present throughout the following process until that need is satisfied, including goals, sub-goals, problems and sub-problems.

A second and more recent study also addressed cognitive complexity through problem-solution distance modelling and the reconstruction of thought-and-action chains. Through an experimental replication of bipolar, discoidal, Acheulean, Levallois and blade knapping techniques, Muller and colleagues (Muller et al. Reference Muller, Clarkson and Shipton2017) created, as Stout (Reference Stout2011) had done, hierarchical diagrams that allow quantitative assessment of the degree of hierarchical organization within knapping sequences (Fig. 2B). The resulting diagrams illustrate a ‘primary focus’ that is divided into a series of bifurcating and hierarchically structured sub-foci (similar to Haidle’s cognigrams but with more emphasis on technological reconstructions of knapping sequences and less on mental perception).

All these studies represent discrete actions in modular structures that show a hierarchical organization of goals. Such methods can provide quantitative and qualitative results to argue for the presence/absence of certain cognitive traits and cumulative cultural change, and can also be applied to the study of the archaeological record. Combined with the insights provided by neuroarchaeological studies, these analyses constitute promising methodological frameworks that contribute to a more nuanced understanding of the cognitive processes underlying stone tool manufacture and use.

Towards holistic approaches in ECA and lithic studies. A reflection

Beyond Acheulean technology and the study of stone tools themselves, it is now widely emphasiszed that ECA studies should account for the broader environmental and behavioural contexts surrounding tool production and use. For example, raw material procurement and adaptability can reflect changes in spatial cognition and working memory capabilities. Similarly, the evidence for curation and transportation of lithic tools indicates forward planning and an understanding of resource value over time. These behaviours, often inferred indirectly through lithic technological analyses, provide crucial insights into the selective pressures shaping hominin cognition. Likewise, the social dimensions of learning and teaching require greater emphasis. Cultural transmission, imitation and the role of active teaching and learning in knapping skill acquisition are vital for understanding how lithic technological traditions persisted and evolved. While this review has primarily focused on insights derivable directly from the analysis of lithic artefacts, future research should integrate these broader biological, behavioural and ecological dimensions to provide a fuller picture of hominin cognitive evolution (Fig. 3).

Figure 3.

Evolutionary cognitive studies should prioritize empirically replicable methodological frameworks that integrate genetic paradigms, materialist perspectives and extended cognition theories, while also considering the social and ecological contexts that shaped the cognitive development of extinct hominins.

In some way, ECA must perhaps undergo a ‘paradigm’ shift comparable to that of the New Archaeology revolution, one that actively integrates qualitative scientific methods into its analytical framework. This is because while this field has produced extensive epistemological and theoretical discussions, these remain largely disconnected from empirical methodologies capable of supporting or challenging their claims. Exceptions to this, fortunately, do exist, such as experimental neuroarchaeological analysis or problem-solution distance approaches. The challenge lies still in linking these studies to the physiological, genetic, social and environmental contexts of extinct hominins.

A fundamental challenge within ECA is its reliance on evolutionary theory without a reliable and replicable methodological bridge to qualitative insights from cognitive science, anthropology and psychology. Similarly, while quantitative approaches—such as computational modelling and statistical inference—have significantly advanced our understanding of hominin cognition, they often overlook the interpretative and context-sensitive dimensions that qualitative methods can offer. To address these limitations, the ECA must embrace an interdisciplinary synthesis that integrates ethnographic analogies, experimental archaeology, phenomenology, materiality and neuroarchaeological approaches (Fig. 3). This shift would strengthen the empirical foundation of theoretical claims and provide a more nuanced understanding of how material culture and cognition co-evolved.

By fostering a methodological pluralism that values qualitative insights alongside quantitative approaches, ECA can move beyond theoretical speculation towards a more holistic empirically grounded discipline. Such a transformation would mirror the impact of the New Archaeology movement, which revolutionized archaeological practice by insisting on explicit methodologies and hypothesis-driven research. In doing so, ECA can better bridge the gap between theoretical discourse and scientific practice, ensuring a more comprehensive exploration of the development of human cognition.