Tool evolution as a prerequisite for consciousness

1.1 Examples of the extended mind and externalizations

A comprehensive view of cognition must involve the integration of external objects that are necessary for externalizing memory or other cognitive functions, termed active externalism (Clark 2024; Clark and Chalmers 1998; Menary 2010; Noe 2010; O’Regan and Noe 2001). “External” is defined as being outside one’s body’s space as limited by its physical, discrete borders such as the skin. Active externalism as defined in the original publication by Clark and Chalmers (1998) (see also Menary 2010 for multiple reviews and discussion) is the concept that material objects outside the brain can be essential parts of cognitive processes in a constitutive, symmetrical, and causal, i.e., active way. The idea of active externalism is not that material objects merely somehow assist in cognition but that they are a necessary constituent of a cognitive process, meaning that a particular kind of cognition could not take place without external objects. This, however, does not mean that every cognitive process has an external component. In cognitive processes involving active externalism “the human organism is linked with an external entity in a two-way interaction, creating a coupled system that can be seen as a cognitive system in its own right” (Clark and Chalmers 1998). Consequently, if the external component is removed, cognition will collapse as if a part of the brain had been removed. There is a fundamental continuity between the brain and cognitive extensions into the external world (Clark 2024). This initial radical argument of Chalmers and Clark has subsequently been modified to allow a more malleable internal versus external distinction as “outsourced cognition” (Gerken 2014).

The high degree of how (even transiently) external objects are readily integrated into our perception is, for example, demonstrated by the rubber hand illusion (Botvinick and Cohen 1998) and its subsequent variations, including a whole body illusion (Lenggenhager et al. 2007) where external body (parts) can get “manipulated” into our body image. Along similar lines, external tools such as mirrors have been shown to be able to appease phantom limb pain through feeding back body images into our perception and thereby beneficially complementing aberrant brain-internal functional connectivity (Chan et al. 2007; Ramachandran and Rogers-Ramachandran 1996). Evidence for the deep anchoring of external objects into brain computation also comes from experiments with monkeys where it was demonstrated that training for tool use changes functional network representations (Iriki and Sakura 2008). Furthermore, elaborate forms of memory like autobiographical memory are externalized and dependent on social interactions (Heersmink 2022; Van Bergen and Sutton 2019).

A simple example for everyday’s active externalism is a diary in which appointments are noted (Clark and Chalmers 1998): when looked up, these appointments can be fulfilled, i.e., the diary is causally active. Without the diary, one might not have performed this cognitive process. In this case, memory was externalized, and minimized: instead of prospectively memorizing a series of appointments for upcoming weeks by heart, it was reduced to the instruction: “look in the diary at the appropriate date, then read and short-memorize the time for this or that appointment.”

Another example for progress through externalization is mathematics: beyond the limited individual’s ability to perform mental arithmetics by heart, calculating is mainly done with the help of external tools, be it as simple as using a pen and a paper, in order to externalize memory. Clearly, beyond the number hundred or so, most individuals are not able to perform multiplications or divisions by mental arithmetics alone but have learned to use a pen and a paper to assist in a memorized, abstract algorithmic calculating procedure. A calculator as an advanced tool has now replaced these simple tools used until about 50 years ago. Since advances in the natural sciences necessarily involve the generation of quantitative data and rather complex statistical mathematics, it is fair to state that without cognitive externalism there would be no natural sciences as we know them today (see also Fabry 2019 for an extended perspective of the enculturated origins of mathematics). Mathematics historically emerging from external tool evolution as presented here (detailed in chapter 2d) is thus in opposition to philosophical views assuming mathematics as a given (Tegmark 1998).

While an externalization of memory into “exograms” (Donald 1991) has massively increased memory storage capacity, it also has made memory more vulnerable to manipulation (Risko et al. 2019). Brain-centered (internal) cognition, as opposed to external cognition, is an exception and certainly not as powerful as a cognition that is a hybrid of brain-internal and environmental (outside-brain) tools that are causally coupled and reciprocally active (Menary 2010; Noe 2010). The explicit use of (external) objects for external memory has also been investigated under the term “cognitive off-loading” (Heersmink 2017; Hu et al. 2019; Risko and Gilbert 2016) and has been defined as “the use of physical action to alter the information processing requirements of a task so as to reduce cognitive demand” across several psychological domains (Risko and Gilbert 2016).

1.2 The range of externalizations

The concept of the extended mind is related to the concept of the “extended phenotype” (Dawkins 1982), which states that a phenotype corresponding to a genotype, i.e., the coding genetic information, is not merely constituted in forming a phenotypic body but also equally, via behavioral enactments, in modifying the environment. Examples are beavers building a dam, birds building a nid, and spiders forming a net. Since these environmental modifications executed by individuals require motor and cognitive acts, these animals require primitive “extended minds,” too (Facchin and Leonetti 2024). These primitive externalizations seem rather inborn and not acquired and, therefore, are not subject to social processing or transmission. Thus, while features of externalization themselves are not human-specific, externalizations triggering a cumulative cultural/object-driven evolutionary process are, as will be explained below.

Exactly why the transition of externalizations to tools and foremost a more extensive tool evolution happened only in humans is unclear. Likely, an emergence of rudimentary prosociality setting the frame for enabling social transmission such as the ability for “joint attention” (O’Madagain and Tomasello 2022; Tomasello 2014) and primitive tool use were biological preconditions to spark a coevolution between cultural evolution and culture-conditioned biological fixations (see below chapter 8) facilitating further cultural evolvability, i.e., the facilitation of biological hardware to accelerate cultural uploading and evolution. As will become clearer when following the line of reasoning in this paper, one major claim here is that externalizations by tools promoted prosociality, which in turn further accelerated cultural evolution and this mutual, repetitive reinforcement thus contributed to the evolvability (Kirschner and Gerhart 1998) both of the biological and cultural realm.

Earliest excavated material externalizations are artifacts such as cave paintings or statues that date back to around 40,000 years BCE (Conard 2003; Heersmink 2024). This, however, does not mean that perishable artifacts from wood or bones could not have been generated much earlier. The start of cumulative cultural evolution has been dated to around 600,000 BCE (Paige and Perreault 2024). Language, which in its earliest forms is dated back to 300,000 BCE (Fitch 2010) – 500,000 BCE (Perreault and Mathew 2012), is an externalization too, if we allow a broad definition. Ultimately, through distinct motor acts of the more than 100 different muscles from tongue, lips, jaw, and pharynx, sounds are generated that leave a temporary print on air molecules, which is transmitted between close-by individuals.

The role of language for evolving social networks and prosociality is eventually more intuitive than that of tools and likely language and tool evolution went hand in hand rather than one preceding the other as newest brain functional magnetic resonance imaging (fMRI) findings in humans suggest that reported activation of similar brain structures by syntactic use of language and tools (Morgan et al. 2015; Thibault et al. 2021). Language is transmitted between individuals in a social network and learning how to speak is usually transmitted vertically from parents to children. Oral language is an externalization process already, even though its mere acoustic dimension makes it a fleeting, transient “object” or transient tool (see chapter 2c), an ultra-fast expiring movement of air molecules requiring temporal resolution abilities both for sending as well as for receiving information. Even though language evolution will not be an object of this paper (for extensive reviews see Dennett 2017; Dunbar 1996; Everett 2017; Fitch 2010), it is important to note that the fleeting character of external “word objects” offers both advantages and disadvantages compared to that of permanently present, modified external objects, i.e., tools, that led to different evolutionary trajectories and cannot be compared here in depth (Figure 1).

Figure 1:

Comparing transient with permanent externalizations. Whereas language consists of transient externalizations, tools constitute relatively permanent externalizations – both with different advantages and disadvantages (the list is selective and incomplete). Initially, transient externalizations were repurposed or transformed to permanent tools at least twice: when motor acts such as gestures were used to execute instructions to make tools and when writing was invented as the permanent correlate to fleeting language (see arrows).

1.3 The ostensive triangle as the germ cell for igniting the evolution of externalizations

Given the likely parallel evolution of language and tool fabrication, it is instructive to analyze the beginnings of the evolution of language with regard to similarities with the beginnings of the evolution of tool use. Rolfe has given an excellent sketch on a plausible gradualistic, phylogenetic evolution of language acquisition, starting with profoundly simple interindividual interactions and ending in epistemic functions (Rolfe 1996) (Figure 2). The application of the ostensive triangle was developed for language but can easily be transposed for explaining tool evolution. The ostensive triangle consists of three parties: a pointer (ostension), an observer, and a pointed-to object (or subject; see Figure 2, see also chapter 3).

Figure 2:

The ostensive triangle. The ostensive triangle as the germ cell of social transmission consists of the pointer, the observer, and the pointed-at subject or object. Whether a pointed-at subject preceded objects in anthropological evolution is discussed in the text.

A precondition of ostension and the first step for establishing social transmission is the solicitation between (at least) two individuals that share “joint attention” (see also chapter 3b), i.e., establish a joint frame of reference. The development of joint attention and the early apprehension of “organizing activity,” i.e., basic dialogical abilities have been the topic of intensive research (O’Madagain and Tomasello 2022; Tomasello 2014) and are developed in humans as early as during breast feeding where, for example, already the sequential nature of chunks of social interaction are trained (Noe 2015).

Once joint attention of two individuals has been established through solicitation, ostension is executed by the active individual aiming to transmit an information: a distinct pointing to designate an object intending to draw the information-receiving individual’s attention toward it so that the object can eventually be associated with a name by either a gesture or a sound (language), which will subsequently represent that object. The pointer sets a relative frame of reference centered on himself (deixis) that the receiving individual can relate to. This frame of reference requires a form of theory of mind since the observer has to adapt to the perspective of the pointer (O’Madagain and Tomasello 2022; Santiesteban et al. 2012). Pointing as a means of demonstrating intentionality seems to be superior to other attention-drawing or intentional gestures such as bodily expressions, gaze following, or attention getting (van Mazijk 2024). Active gestures executed by individuals have language-like syntax and their sequential structures are as fleeting as that of sound within language. Accordingly, it was hypothesized that “the slow-to-emerge precursor [human] from 5 million years ago to 2 million years ago may have built up a gesture language that derived from instrumental actions … it would have been an evolutionary track to pantomime” (Gärdenfors and Hochberg 2024; McNeill 2013). Sequential visuospatial integration of sight and hand movements was clearly an early requirement for pointing, and, later, for the generation of tools (Bruner and Iriki 2016).

The interaction in the above described ostensive triangle (pointer, receiver, object) is used to assign either a representation in the form of a name to an object (vocalization, gesture) or to perform a modification to the object as a first stage of tool evolution. The ostensive triangle is the germ cell of social transmission of externalizations setting in motion through its infinite repetition cumulative cultural evolution. The ostensive triangle tightly interweaves social communication and the transmission of object designations (with representational or instructional content), i.e., externalizations.

The ostensive triangle is similar to what has been termed triadic bodily mimesis, which has been emphasized as a main distinguishing factor in enabling cumulative cultural evolution, not present for example, in the great apes (Zlatev et al. 2005). Dyadic mimesis by contrast, according to this reasoning, just reflects a bipartite relationship that apes are in part capable of such as mirror self-recognition and the ability to copy bodily shapes on command but not fully implement shared intentionality (O’Madagain and Tomasello 2022; Zlatev et al. 2005).

Does externalization have roots in interpersonal recognition and interactions itself? For example, from a first-person perspective other highly mobile individuals could be regarded as (fleeting) external objects. It is conceivable, therefore, that first object externalizations may have been rooted in social individuals of kin or tribe members, as evidenced by the first symbolic artifacts as the “Löwenmensch” from 40,000 BCE (Conard 2003) or findings that go back even further in time (see Tattersall 2024 for a comprehensive review on the history of anthropologic evolution of symbolic use). The aim to externalize another person’s representation, eventually by his/her own ostensive designation, in order to create a memory during that same social peer’s absence as an autograph-like object could be conceived as a primordial motivation creating symbolic, i.e., representative artifacts or tools. Along these lines, representations of persons (dolls), just as in ontogenetic development, may have also preceded tools in anthropological tool evolution (a “doll-theory” of symbolic evolution; leftmost picture in Figure 3).

Figure 3:

The evolution of a third person perspective from the ostensive triangle. Schematic drawing. 1. Joint attention was developed as an interpersonal ability in a primordial ostensive triangle (see Figure 2) where the attention was drawn (depicted by a pointing finger) to another individual. 2. In the ostensive triangle, objects were designated with representational content in the form of sounds (language) or instructions (tools). 3. As the tool-sphere grew, frequent changes in perspective between the pointer and the observer as well as an increasing number of objects led to social synchronization, which resulted in an extrapolation of multiple second person perspectives: the third person perspective, i.e., a joint minimum frame of reference of the material external world.

2.1 Tool definition

What are tools? In this paper, tools are defined in the broadest sense, which includes all objects that have been modified by human individuals for further manipulative use on yet other objects but also those not used for further manipulation such as artwork, signs, or others. There is a spectrum of tool definitions from narrow (Fragaszy and Mangalam 2018) to broad (see also Mangalam et al. 2022; Ruck and Uomini 2024 for discussion). A classic functionalist definition is (Schubotz et al. 2023; Shumaker et al. 2011): 1. A tool is an object that can be manipulated; 2. During tool use, the tool is touched, held, or otherwise directly manipulated by the tool user; 3. The tool is used in an appropriate context, namely purposefully to achieve a goal; 4. The tool is used to alter the state of another object (including other organisms and the tool user); and 5. The tool extends the user’s abilities to achieve certain goals. The latter point is important and clearly relates to the extended mind of the user.

A tool definition that takes an even more active externalist view is from Iriki and Sakura (2008) “external objects in their hands and moving these objects as extensions of their own body … this is the beginning of transitive movement, i.e., tool use.” Iriki sees tool use as the end point of a gradual freeing of the limbs from movement and on a trajectory to support cognitive functions of which tool use is the most advanced one (Iriki and Sakura 2008). Tool use remaps the body scheme (extending boundaries) and thereby affects perception of the world (Noe 2010). Latour defines a tool as “a single element held directly in the hand of a man” and contrasts this with machines where single tools are linked to one another rather than to the hand of an individual (Latour 1987).

2.2 How tools are made and replicated

The more specific and functionalist tool definition given in this paper is as follows:

Tools are made from external materials that are modified in one or several steps and assigned to a particular use. The execution of these modifications requires defined sequences of precise motor acts. The protocol of sequential motor acts necessary to replicate a tool is the instructions that are transmitted from individual to individual (Figure 4).

Figure 4:

The tool replication cycle. Depicted is the flow between an individual’s brain-internal tool instructions and the execution of those instructions during toolmaking. The modified object, the tool, is subsequently evaluated/selected for convenience or fitness during tool use. When the instructions are socially transmitted, they might be modified resulting in a modified, novel tool (or not), and ending one tool replication cycle.

Advanced tools or machines, such as robots, can also fabricate new tools, but this also follows distinct steps (programmed by humans) and the robots themselves as well as their components have been fabricated by humans through yet other motor act-requiring instructions. As such, ultimately, the antecedents of all present tools can be tracked back to a prehistory of components having been generated from raw materials or commodities that have been mined, harvested, or processed by humans, and so forth, extending to an extremely long, but not infinite, chain of linked events. Today’s present tools are end points of branches in a large evolutionary tree where the thicker branches represent the predecessors of today’s tools (thinner branches) similar to a tree of life (i.e., a pedigree of species) in biological evolution.

The important point is that all present or past tools have a position in this pedigree and thus ultimately can all be traced back to the simplest tools, such as handaxes, or similar. The existence of such an unambiguous pedigree proves that there is Darwinian evolution in the tool-sphere, i.e., the “anthropogenic mass” (Elhacham et al. 2020), with its three characteristics replication (by transmitting the instructions tool fabrication), variation (by modifying these instructions), and selection (by deciding whether a modified tool is better suited for a given purpose than its predecessor; see Figure 4). Evolution of the tool-sphere was so successful that, as of today, it exceeds the biosphere by mass (Elhacham et al. 2020).

The three elements of Darwinian evolution: replication, variation, and selection can be abstracted from their exclusive application for modeling biological evolution to other replicating, material systems, provided that those reliably feature these three elements. The overarching concept is that Darwinian evolution is a functionalist model to explain the historical and causal trajectory of phenomena not restricted to genetic evolution (Skinner 1974). Dawkins and others (Lewontin 1970) stated that any system that replicates, induces variation and uses selection necessarily and unconditionally evolves along Darwinian characteristics. Dawkins called these functionalist principles “Universal Darwinism” (Dawkins 1983) (see also chapter 3d). The embodiment and kinetics of the replicating material within Universal Darwinism was never conceived to necessarily resemble genes but was conceived to feature any other plausible form. Darwin himself had proposed that cultural evolution demonstrated similarities to biological evolution in that a large number of gradual changes could amount to large changes over time that, for the uninformed observer who had missed the previous gradual changes, could present itself as inexplicable (Darwin 1871).

Dawkins distinguished the replicator and the vehicle as forming a Darwinian unit, where the replicator is ultimately the material that is replicated and modified during a Darwinian replication cycle of modification and replication, whereas the vehicle is the material into which the replicator and its modifications are translated to and, importantly, subjected to selection (Dawkins 1976/2016). The fitness of the vehicle, i.e., the convenience of how well it interacts with its ecological niche determines selection of the tightly physically connected replicator. In biological evolution, the replicator is the genetic material consisting of deoxy nucleic acid or ribonucleic acid, whereas the vehicles are cells, species, viruses, etc. Dawkins elaborated that the successful, surviving replicator has to display fecundity (a high replication rate), longevity (a long life), and fidelity (a high degree of similarity of the copy to the original) in order to outcompete other replicators. Darwinian evolution is fundamentally gradual rather than saltatory. Seeming gaps in a gradual sequence are explained with a “winner-takes-all” phenomenon: if the gradual development of an innovation is extremely successful (in terms of being adaptive, conveying advantages and being selected), all previous gradual stages are superseded and die out, as, for example, discussed for the evolutionary development of the eye enabling vision.

Could a similar trajectory be conceived for cumulative cultural evolution? Szathmary and Maynard Smith have rated the emergence of culture as one of the major transitions in evolution (Maynard Smith and Szathmary 1995). Clearly, several authors have proposed that cultural evolution follows Darwinian principles, too (Boyd and Richerson 1985; Cavalli-Sforza and Feldman 1981; Mesoudi et al. 2004).

Heyes and others have defined cumulative cultural evolution as being executed by social transmission in such a way that the behavior of a group of individuals is improved over successive generations (Heyes 2023). A “ratcheting the ratchet”-effect has been proposed to occur when previous success in tool achievements is used by successive generations (Tennie et al. 2009; Tomasello et al. 1993; Whiten 2022). In that respect, the term “cumulative” is important because it emphasizes the transgenerational nature of progress: progress that couldn’t be achieved in individual’s lifetime but that critically depends on prior knowledge accumulated by others (Tennie et al. 2017). Nonhuman primates such as chimpanzees have been demonstrated to exhibit primordial evolution of tools (see Arbib et al. 2023; Whiten 2005, 2022 for review). Recent evidence suggests that chimpanzees feature even primordial cumulative culture (Gunasekaram et al. 2024), which would mean that social transmission of tool use in these apes is indeed transgenerational (reviewed in Richerson and Boyd 2024). Intermediate or primordial forms of tool use are proof that cumulative cultural evolution evolved by gradual improvements and that Darwinian mechanisms are at work. Dennett (2017) has provided an illustrative example of cumulative tool evolution: during canoe building in Polynesia, subsequent generations of sailors built improved canoes that were tested for use on the ocean but only those designs were selected for replication that made it back from a life-threatening journey – the unsuccessful designs were selected out by the drowning of their builders, the carriers of the flawed toolmaking instructions.

More similarities exist between biological Darwinian and cumulative cultural evolution: innovation, for example, can also emerge from anastomosis, i.e., when previously separated branches of the evolutionary tree rejoin, such as in biological symbiosis. An analogy in tool evolution would be machines such as a car where several differently evolved tools are assembled to a novel tool with a novel purpose and reward (Dennett 2017; Latour 1987).

2.3 Language as a transient, fleeting “tool”

Language evolved in the acoustic modality where externalizations/signals are transient and limited to a moment, contrary to the visual modality where externalizations are usually permanent (to variable degrees; Figure 1). Language, as a highly specialized motor memory (of the phonetic muscles), is entirely stored in internal memory, which has to grow along with increasing linguistic skills and vocabulary. Pronounced single words are single externalizations just as the simplest tools, but these motor acts that are – like gestures – fleeting and do not lead to permanent modifications of an object, i.e., a tool or artifact. Complex arrays of words connected by syntactic structures are thus similar to more elaborated tools as shaped by multiple motor acts (instructions). A breakthrough discovery was the similarity of activated brain areas both during tool-related acts and language (Morgan et al. 2015; Thibault et al. 2021) suggesting that their biological correlates have a common origin. These findings suggest that language and tool evolution cannot be conceptualized separately, and it will be interesting to figure out the mutual communicative interactions between transient and permanent externalizations, i.e., language and tools, respectively, and the possible effects on their evolvability.

In language, the evolution of semiotic progression from indexes via icons as the earliest forms of intentional linkage to symbols required a normative convention (ostensive triangle, Figure 2, see also Everett (2017). Language allows to variably direct attention (Clark 2011) and heavily relies on the ability to internally memorize and process sequences of a more general nature. The domain general nature of language is supported by evidence that in dyslexics not only reading but also other “sequential functions” are impaired (Heyes 2018). Even though the cultural evolutionary origin of language was questioned for a long time (Chomsky 1965), increasingly, this view is replaced by a cultural, evolutionary view of language supported by a wealth of empirical data (for discussion see Dennett 2017; Everett 2017; Heyes 2018). In that context, for example, it has been claimed that grammatical construction is a unit of Darwinian-like selection itself (Steels and Szathmary 2018). Likely, alternating cycles of cultural evolution and genetic fixations during language evolution are at the origin of today’s language (Henrich 2016). Language was subsequently “materialized” upon the invention of writing and thereby became a more permanent tool losing its fleeting character.

2.4 The language-writing transition: transforming externalizations from the transient to the material stage

Writing/reading recreates the transience of language by transposing phonetic movements to eye movements when tracking sequential letter strings. For simplicity, writing/reading will subsequently be referred to as “writing” only. The analysis of the evolution from language to writing, or in the context of this paper, the evolution of language micro-instructions, i.e., motoric acts of the phonetic system with representational content to writing micro-tools, i.e., tools with representational content eliciting defined affordances, is an excellent and instructive example of how the materialization of previously fleeting internal instructions leads to a higher fidelity, higher longevity, and, eventually, higher fecundity of the replicated instructions.

Archeological evidence for the origins of writing starts to emerge from around 8,000 BC on in mesopotamia with the appearance of clay tokens in the wake of agriculture and the need to document transfer and property of commodities (Schmandt-Besserat 1996). Clearly, the initial purpose of writing was both economic and arithmetic in nature (for a detailed view on the enculturated practice of arithmetics see Fabry 2019; Menary 2015). Each clay token corresponded to one kind of good, had a unique geometric form, and was an entirely man-made artifact. By representing a kind of good, the token system allowed to simultaneously process data of different categories, allowed handling of a theoretically infinite number of goods without the risk of depending on human memory, and, importantly, in the absence of the real goods (Schmandt-Besserat 1996). In this initial first step, two events are remarkable: 1. There is an intention to externalize memory, which as a side effect allows to perform functional operations with those external, materialized memories, i.e., to recombine them in special or in random ways (which, “mutation-like” may lead to novel, originally unintended insights), and 2. The appearance of tokens as symbolic representations for goods including a numerical representation for documenting debts also intricately links the appearance of writing to the emergence of mathematics. The materialization of language certainly led to a massive expansion of external memory. The further evolution of the clay tokens allows a gradual reconstruction of steps leading from real goods and commodities to their abstract two-dimensional representations (Schmandt-Besserat 1996): 1. The original clay tokens representing each one a specific amount of commodities were 2. assembled in so-called envelopes, i.e., hollow vessels used for storing and grouping tokens, where the contents was marked upon, to 3. tablets, i.e., the former covers of envelope (vessels) where the contents was marked but without the prior content, to 4. the final step, the mere marking of signs with a pencil-like instrument on a two-dimensional, plain surface representing commodities achieved the full abstraction of representational content. Phonological representations, for example, for designating the names of the proprietor, came later (Malafouris 2013). Thus, gradual, “Darwinian” steps in the language-writing (tool) evolution can be precisely tracked.

Language which when orally transmitted has relatively low high variability thus got transformed into an external, permanent, and potentially lasting tool upon writing. The permanence of the material, i.e., written word increased longevity of the replicator, its fidelity (compare Chinese whispers with copy-writing of the same sentence), and its fecundity. Once paper was invented, subsequently, the book press, or ultimately the internet, the replicator fecundity of writing language rose exponentially. At the same time, the spontaneous drift in change of spoken languages particularly in small communities, due to the relatively low fidelity of the spoken word, was decelerated and the high fidelity of writing fed back to stabilize language variability (Noe 2015). Individual knowledge of writing also changes the attitude of speaking such as when the speaker is influenced by written texts to adopt an attitude of speaking as one would write (Noe 2015). What drives the replication of tools and how are tool-replicating instructions (Figure 4) transferred between humans?

3.1 Copying motor acts

Social transmission of language is intuitive and a prototype of social transmission (Dennett 2017) because the fleeting sound is literally transmitted, i.e., a motor pattern of the phonetic muscles generates a sound, the sound travels instantly through oscillations of air molecules before it dissipates, which is long enough to activate a pattern in the brain of the receiver associating the sound with a designated representation. This kind of social transmission similarly applies to toolmaking and use, except that motor patterns in the latter involve a wider range of body muscles rather than only those of the phonetic system, and the product, the tool, is more permanent than the fleeting sound. Ultimately, all transmitted instructions can be reduced to performing a distinct sequence of motoric acts (Tamariz 2019). The specific transition of a simple nonintentional motor act to one with intentional content for a specific achievement has been termed praxis (Ruck and Uomini 2024). Praxis is “an action which is semantically rich, and a physical imposition of higher order (functional and social) internal states upon the external world” (Ruck and Uomini 2024). The instructions for motor acts could be transferred by language or text, which would be an indirect (but efficient) way for social transmission but copying is also possible without language through imitation of movement, similar to what is done during dancing (Hannay et al. 2024; Laland et al. 2016). The evolutionary accumulation of behaviors through an interaction with external memories has been termed a “non-genetically transmitted behaviourome” (Müller 2020).

How do sequences of motoric acts transition from more random events to less random ones, i.e., more significant intentional ones, such as praxis (Ruck and Uomini 2024) or mimesis (Donald 1991; Zlatev et al. 2005) (Figure 5)? Going back to active externalism, for an individual to access external cognition, motor acts have to be executed in order to access the external object, which usually involves the manipulation of this object. For example, this can consist in mere walking to somewhere for finding information or accessing external memory, browsing a book, or it can be more complex tool using. The important thing here is that the individual’s interaction with an external material can be unambigously defined by a motor sequence, a sequential “organized act,” executed by this individual with the intention of a particular achievement (Noe 2015; Ruck and Uomini 2024; Tamariz 2019).

Figure 5:

The continuum of externalizations from gestures to modified objects to tools and machines. Schematic drawing showing how simple externalizations such as motor movements evolve to defined sequences of movements such as gestures and, ultimately, lead to object modifications and tools.

The execution of a motor act with an external object by the first individual may draw the attention of a second individual. Any attended external material and the first individual’s actions will then also become part of the second individual’s passive cognition, initially merely drawing attentional resources. At the beginning, the aim or intentions of any actions of the first individual with the external material may not be fully “understood” by the second, observing individual (see chapter 6), but a “meaning” of the first individual’s motor sequence, i.e., a change in the evaluation of the observed motor sequence may eventually unfold as further specific interactions of the first individual with the external material are recognized. This process is catalyzed by prior knowledge or experience of the second individual about the specific external material used by the first individual, or prior knowledge about specific motor sequences related to the external material. Prior knowledge will 1. prime specific motor sequence associations elicited by the specific external material memorized in similar situations in the past (by simple associative learning Birch et al. 2020) and thereby 2. greatly restrict the degrees of freedom for motor sequence associations performed in the context of that external object.

During the developmental ontogenesis of an individual, i.e., one’s lifelong learning history starting in early childhood, specific external objects become associated to specific uses by mental representations of motor sequences or may become integrated into the child’s body scheme (Bruner and Iriki 2016; Iriki 2006). While these initial associations occur in a social context through observations, imitation, or pedagogy, lateron, the specific external object will elicit those associations, or affordances (see chapter 4) also spontaneously in the absence of other individuals, or aid to predict potential actions of other individuals related to the use of this specific external object.

The passing on of an instruction leading to the making or use of a tool is thus transmitted within the ostensive triangle (Figure 2) where a pointer (or sender in semiotic terms) is in contact (by joint attention) with the observer (receiver in semiotic terms) to point at an object (signifier in semiotic terms) and eventually associates this object with a representative gesture or sound (sign in semiotic terms). When an observing, second individual reproduces the motor act observed from the pointer/sender, the motor act or instruction of the first individual has been replicated after passing through the cognition of the observing individual.

3.2 Joint attention

Joint attention in humans can be established in several domains: eye contact with gaze following to direct attention and perspective, language and sound in the acoustic realm, and gestures, or combinations thereof: all of these domains ultimately involve motor acts as outputs requiring perception as input between the sender and the recipient and, eventually, reverse interaction when sender and receiver are switched. “Our visible eyes provide information about eye direction and cues of mutual gaze that seem crucial to our unique forms of learning, cooperation, and communication” (Hare 2017). “Joint attention, which allows children to learn the association between labels and objects, relies on a child’s ability to follow an adult’s line of gaze” (Tomasello 2019; Tomasello et al. 2007). Experimental data report that when eye blinking is synchronized between individuals, they improve the coordination of their subsequent motor activities (Koike et al. 2016).

The question arises how the ostensive triangle and joint attention emerged in the first place. It has been pointed out that selection pressure has to act both on the sender and the receiver for a general benefit of social learning as compared to asocial learning (Turner et al. 2023). In that context, it is noteworthy that the achievement of joint attention itself may have evolved as a cognitive gadget (Heyes 2018), i.e., a culturally acquired ability that only later on got biologically fixed to ascertain its reliable availability. Ontogenetic development of pointing is seen as being rooted in grasping objects within the process of social shaping and has been hypothesized to develop from the simple transposing of objects towards the active referring to objects during joint activity (Liszkowski and Rüther 2024).

Along these lines are anthropological insights that during fire domestication in anthropological evolution (800–400,000 years ago), a cognitive prerequisite was not only the ability to ignite fire itself as a distinct sequence of a motor acts involving several objects, but also required prospective organizational abilities to maintain the fire (such as wood piling, fire supervision to prevent its premature extinction, etc.) (Twomey 2019). All these executions required complex social skills including joint attention and are evidence for an early evolution of social transmission of task and attention sharing itself.

An early externalized object might have been a peer or prey, i.e., another living individual, and pointing may have evolved for joint foraging or for other reproduction-related purposes. In this context, joint attention would have evolved due to a higher selection pressure on getting and staying in contact to align (“share”) intentions, for elementary needs. The transition within the ostensive triangle from a mobile living subject to a permanent, externalized object (Figure 3) may then have gradually occurred through, for example, associating the third individual in the ostensive triangle with doll-like objects as personal memory-artifacts representing real persons, and such practices may have occurred with the beginning of person-related memories when burial practices developed (Chapman 2013). Alternatively, the transition person-to-object may be related to the idea of “social tools” (Hare 2017), i.e., the reification of social peers and their assignment of roles in social interactions where under some circumstances individuals are handled as interchangeable tools.

3.3 Imitation and mimesis

There is a variety of mechanisms of social transmission and learning (see overview in Whiten 2022). A particularly important mechanism is imitation (Heyes 2016, 2023; Heyes and Frith 2014). One narrow definition of imitation states “more specifically, in imitation an observer learns and reproduces something about the topography of body movements” – which part of the body to arrange (e.g., head or foot), or how to move body parts relative to one another (e.g., to bring appendages closer to the trunk or to move them further away) (Heyes 2023). Imitation, i.e., the copying of another’s body movements as precisely as possible, has invariably been considered key in social transmission. This is supported by the phenomenon of “overimitation,” defined as including copying of causally irrelevant and perceivably unnecessary actions in relation to the given goal (Hoehl et al. 2019). Overimitation thus seems to have evolved to ascertain copying redundancy: motor patterns are first reproduced in overabundance without recognizing context-given termination points or borders, or any “understanding.” This could mean that imitation is really a process that does not involve any (conscious) insight in line with that Dennett has called “competence without comprehension” (Dennett 2017). Other kinds of social learning include stimulus enhancement, observational conditioning, and many others (reviewed in Heyes 2018, 2023).

Mirroring movements or postures of other individuals clearly is at the center of imitation. Following that line, mirror self-recognition is considered to constitute a primordial form of self-representation (Anderson and Gallup 2015). A mirror-independent form of imitation is “movement mirroring” that is, for example, performed already in the early mother–child interaction where the mother imitates the child’s spontaneous facial movements resulting in the association of self-movements and facial perceptions of others in “matching vertical associations” (Heyes 2019).

Mirror neurons, i.e., neurons originally discovered as being activated both on perceiving and executing hand mouth movements in monkeys (Rizzolatti and Craighero 2004) have been hypothesized to be a neural correlate for imitation. To this end, mirror neurons have been detected in nonhuman primates and humans (Bonini et al. 2022; Heyes and Catmur 2022). They are likely the result of extensive ontogenetic training rather than an innate endowment (Cook 2012; Heyes and Catmur 2022). A gradual evolution of imitation has been proposed, from grasp copying to complex imitation, gestural copying and protolanguage corresponding to distinct stages of early hominin toolmaking and use, and mediated by the mirror neuron system (Arbib 2011).

The report of tool-specific mirror neurons, i.e., cells that display highest activity when tool-manipulation is performed or observed suggests deep biological rooting of tool use (Ferrari et al. 2005; Rochat et al. 2010), conceptualized either as tools becoming part of the body scheme or, alternatively, hands or other body parts becoming objectified (Iriki 2006). What is certain today is that mirror neurons play a causal role in copying of body movement topography and contribute to processing of observed actions (such as grip) but they do not seem to play a role in inferring observed actors’ intentions (Catmur et al. 2014; Cook et al. 2014; Heyes and Catmur 2022).

A precondition for multistep tools is the ability of memorizing a complex (motor) sequence. This requires episodic memory (Bevandic et al. 2024; Huston and Chao 2023; Tulving 1983), which is present in humans, apes, but also in many other species. The efficient process of social transmission of instructions was powered by a strong individual reward to replicate tool-related instructions. The benefit of tool use as increasing the convenience to achieve a goal was that without that tool, the benefit would have been costlier – thus, improving social transmission of tool-related instructions was rewarded and provided a strong reinforcing motivation to replicate tools (see Figure 4). Supporting the notion of tool-related rewards is the observation that nonhuman primates make profitable decisions relative to the reward quality in the context of preferred tool use choices (Laumer et al. 2019).

3.4 Universal Darwinism

On the assumption of Universal Darwinism, Dawkins coined the term “meme” to designate a replicator that, in analogy to a gene, could be the unit of cumulative cultural evolution. In the almost 50 years after its conception, however, and for reasons that become apparent below, the definition of a meme has undergone transformations that today make it too ambiguous to be used for more specific scientific definitions, such as the ones mentioned in this paper.

Dawkins originally defined memes as: “a unit of cultural transmission, or a unit of imitation” … “Examples of memes are tunes, ideas, catch-phrases, clothes fashions, ways of making pots or of building arches … Popular songs and stiletto heels are examples. Others, such as the Jewish religious laws …” (Dawkins 1976/2016). In these unfortunately fuzzy, attempted definitions, it is problematic that things of different complexity are mixed together: tunes (acoustic melodies), ideas (a thought most likely articulated in (representational) language), clothes (something material), and religious laws (written and nonwritten norms for group behavior). Dawkins attempted a more precise definition later (Dawkins 1982) mainly trying to clarify the replicator/phenotype distinction: “A meme should be regarded as a unit of information residing in a brain…” (Dawkins 1982). A comprehensive history of memes can be found in Daniel Dennett’s recent book on memes (Dennett 2017). Blackmore has defined memes as “instructions for carrying out behavior, stored in brains (or other objects) and passed on by imitation” (Blackmore 1999), which introduces the term “instruction” but still limits meme transmission to imitation and locates units of selection in brains only. The idea of memes being related to instructions was also brought up by Delius “Memes are capable of instructing … behavior” (Delius 1991).

The concept of the meme as an algorithm or a software-like subroutine was extensively elaborated by Dennett (1995). In his latest publication, Dennett defined memes (Dennett 2017): “what are memes a kind of? They are a kind of way of behaving (roughly) that can be copied, brandished, ridiculed, parodied, censored, hallowed.” Dennett (2017): “Memes are informational things. They are ‘prescriptions’ for ways of doing things that can be transmitted, stored, and mutated without being executed or expressed.” Dennett tackles the problem by defining the units of memes as “the smallest elements that replicate themselves with reliability and fecundity” (Dennett 1995). The inability of previous authors to come to a crisp and clear definition of a meme ultimately limited its usefulness and left it mainly for colloquial use, for example, to denominate rapidly shared messages in the advertisement industry or in social media.

The here presented idea that tools and corresponding instructions are two different occurrences of the same underlying phenomenon has previous been termed “petrified memes” (Korth 2020) to emphasize that for tools – in contrast to the meme concept outlined above – a distinct material stage can be described (petrified = in stone). Both, the instructions transferred between social individuals on one side and the corresponding material tools on the other side were proposed to represent the same particular meme, existing in a duality, or dual states, similar to light that can be described both as a wave and as a particle. These dual states, embodied tools, and corresponding instructions are principally equal in significance and necessitate each other: the tool could not be built without the associated instructional mental representation, but the tool itself instructs the apprehending human individual about its making and use within a social context. Tool-related instructions, by general mechanisms of associative learning (Ginsburg and Jablonka 2019), can recombine with other instructions to yield novel instructions and thus fulfill the Darwinian imperative of variation (see Figure 4).

Powerful Darwinian evolutionary systems seem to have in common, that the replicator and the vehicle, i.e., the form that the corresponding gene is translated into, operate in different dimensions without or with rare functional contact. This strict division of labor between the coding replicator and its translation into the vehicle for selection has also been highlighted as a striking feature of evolution by Maynard Smith and Szathmary (Maynard Smith and Szathmary 1995, page 12f). Thus, a “clear cut” between genotype and phenotype in biology, where the genotype-encoded cells or bodies are under selection pressure rather than the physical form of the genotype, i.e., DNA or RNA themselves seem so lead to a particularly robust evolutionary system with high evolvability. This “clear cut” or separation between the encoding and the selected-for material offers several advantages for biological evolution: 1. one gene can code for several phenotypes without being selected out when one phenotype does not fit anymore (environmental flexibility), 2. redundancy in that several genes can code for a similar phenotype allowing the loss of one pathway without leading to a system-relevant loss, 3. evolvability in that a pool of genotypes can silently “evolve” (genetic drift) without being selected against but eventually becoming relevant in future changes in the environment.

The Darwinian evolutionary system for cumulative cultural evolution presented here as an alternating shuttle between the material stage of tools that are subject to selection on one side and on the other side the socially transmitted instructions for their making and use as sequences of motor acts (Figure 4) also constitutes such a “clear cut” between the encoding and selection dimensions indicating a stable evolutionary system. The analogy to biological Darwinian evolution is the genotype corresponding to the instructions/motor acts that are transmitted/replicated and the phenotype corresponding to material tools that are selected for their convenience or advantages for facilitating survival in the widest sense.

6.1 Understanding an object or a tool

To understand a tool is the ability to experience the same or very similar affordances as other individuals relating to the tool’s making or use. It is the ability to deconstruct a tool into its instructions, to reverse-engineer the tool in efficient ways, i.e., to select a parsimonious best fitting deconstruction among many other possible ones. Teaching is “defined as the sender (or pointer) modifying their usual behavioral repertoire to actively facilitate social learning in others,” which occurs when the benefit is larger than an action that involves only individual learning (Turner et al. 2023). The simultaneous perception of the tool by the teacher, the student (=receiver), and, eventually, a demonstration of its use will align the affordances of the teacher to those arising in the student and it is the gradual convergence of affordances related to a tool that is the equivalent of “understanding” the tool. The teacher optimizes relevance for the student regarding the tool in question such that inference of the most parsimonious tool-related instructions sufficiently reconstructing the affordances is possible (Goodman and Frank 2016). Since synchronization of affordances by a tool is also an alignment of predictive processing between teacher and student, in order to control apprehension of the receiver, the teacher checks arbitrarily the similarity of their affordances by observing cues, or evaluating answers of the receiver to the teacher’s questions (verbal or nonverbal) that the receiver can only answer correctly when he/she has achieved a similar deconstruction.

This dyadic (or triadic if the tool is added) process with rapid bilateral communicative exchanges is similar to the process of the search for “hidden causes” representations within continuous prediction error minimization in the active inference paradigm (Metzinger and Wiese 2017). By recombining possible elementary instructions for best fitting a tool, i.e. the subject of understanding, teaching is thus like engineering a (virtual) tool itself. It is like building a fleeting novel cognitive gadget (Heyes 2018) of instructive steps that can be exchanged via motor representations (language) or a visualized motor execution with the tool. The importance of an organized teaching environment, i.e., a kind of permanent object or architecture, that includes tools for instructed pedagogy has been highlighted (Sterelny 2012). In this iterative process, both the teacher and the student are gradually improving to think through another mind (Veissiere et al. 2019). At the end, tool-inspired deconstruction or a reverse-engineering of affordances ultimately synchronizes teacher and receiver that ends with the teacher saying “the other individual understood.” Having understood generates an experience of “familiarity” of the student relating to the novel object or tool. This process of iterative deconstruction has also been described as being “able to transform the new object into, so to speak, older objects” (Latour 1987).

6.2 Understanding an unfamiliar tool is similar to the process of cognitive archeology

The process of learning about novel tools and, ultimately, understanding them is similar to the method employed by cognitive archeology where the material remnants of ancient times are the only substrates left and the ancient world and human cognitions around it are reconstructed from these remains. The here-proposed tool-based theory of psychological understanding is similar to the method of cognitive archeology. In fact, cognitive archeology is the “study of past ways of thought as inferred from material remains” (Renfrew and Zubrow 1994), where starting from the external, material objects, cognition is inferred. Archeologists have to infer microscale processes, e.g., the making of a tool solely from the materials remnants (Overmann and Wynn 2019; Shennan 2011). For example, archeologists have investigated how handaxe knapping of different symmetries requires different levels of advanced cognitive and even neural brain functions (Hodgson 2019).

In cognitive archeology, this has led to a “the material engagement theory, a theoretical framework to set the sole substrate of archeology, material objects, as starting points for reconstructing past society’s or individuals’ cognitions” (Malafouris 2013; Renfrew and Zubrow 1994), where thinking with the help of tools (i.e., active externalism) or “things” has been termed “thinging” (Malafouris 2020). At its extreme, cognitive archeology even departs from a subject-centered view of cognition to adopt a completely decentralized view of cognition where materials are “cognitive scaffolds for thinking, material agency: materials have an effect on us, they make us do things” thus even reversing common agency dogmas (Malafouris 2013).

Obviously, there is a difference in time scale between psychological understanding and archeological deconstruction: whereas the process of deconstructing the very instructions that possibly led to the construction of an archeological object can take years, in interpersonal communication a deconstruction from a tool in the process of understanding eventually happens in minutes or even seconds – but both deconstructions still have a similar formal structure.

6.3 Understanding someone else

How do I understand someone else? The question is better rephrased as: How do I understand what someone else has in mind related to a specific topic? When detached from the context of tools, the explanation of such an understanding is more speculative but it follows a similar pattern: the questioner tries to achieve a deconstruction of the topic named by the person he/she wants to understand. Starting from a first guess that considers all possible contexts of the specific topic in question, an iterative process is started between questioner and respondent where an affordance alignment is the goal and its achievement is constantly confirmed or rejected by control questions (verbally or nonverbally). This process of alignment of mental states can be assisted efficiently by the presence of representations of tools (language), images, metaphors, or material evidence such as external objects to accelerate synchronization. Thus, when one’s goal is to understand someone else by simple observation, searching for the most parsimonious deconstruction that explains this person’s behavior or a situation is continuously ongoing and part of regulating behavior to establish common grounds (Vasil et al. 2019). It is similar to an engineering process similar to what has been described by Braitenberg as “synthetic psychology” (Braitenberg 1984): a behavior is most easily understood by reconstructing it from the most simple elements rather than by complicated speculations. In a sense, understanding is reverse-engineering of predictive processing. Affordance comparisons and mutual perspective taking are thus basic elements for inferring each other’s goals (Creem-Regehr et al. 2013). Synchronization of brain activation is reported upon the mutual understanding of conceptual understanding (Stolk et al. 2014). The question “why” is a request for joint deconstruction in the act of understanding. A joint deconstruction is a first step to a theory of mind.

6.4 Competence without comprehension

Is understanding a process that depends on consciousness and insight? Not when a wide sense of the word is used and motor learning as well as the ability to replicate instructions are also comprised in the term “understanding,” - not the deliberate exchange of representations by language alone. Understanding without rational insight has also been called “competence without comprehension” (Dennett 2017). An exchange of (tool-related) instructions can best explain understanding when defined as an interindividual alignment of affordances as proposed in this paper. It was recently shown in a reductionist experiment that technological advances can be made without any causal understanding of the phenomenon (Derex et al. 2019) referring to broader research that many cultural optimizations can evolve without any individual insight such as, for example, cultural instructions transmitted across generations in prescientific societies (Henrich 2016, 2021).

8.1 Evolvability

Evolvability, in the original biological definition, is an organism’s capacity to generate heritable phenotypic variation (Kirschner and Gerhart 1998). Translated into cumulative cultural evolution it means an individual’s capacity to generate variations in tools that can be socially transmitted. In a narrower sense, evolvability may also be defined as a mechanism that increases heritable phenotypic variation, i.e., that speeds up evolutionary solutions to faster achieve adaptation. Based on the concept of evolvability, eight major transitions in Darwinian evolution were claimed from the chemical evolution of replicating molecules to the evolution of societies (Maynard Smith and Szathmary 1995). This concept was subsequently extended to the stepwise transitions in cognitive domains (Barron et al. 2023; Dennett 2017; Ginsburg and Jablonka 2019; Tomasello 2014). In each of these steps, novel mechanisms of selection were established opening novel phenotypic spaces.

8.2 Prosociality and consequences

Clearly, genes steering brain development must first have developed primordial brain hardware to enable the ostensive triangle and primordial social transmission of sequences of motor acts (instructions), but once ignited, abilities for social transmission were rapidly accelerated by evolvability. Social tolerance as early prosociality is a necessity for stabilizing populations of large overall size and high connectedness, which in turn are preconditions for generating, retaining, and diffusing cultural innovations (Cieri et al. 2014; Muthukrishna et al. 2014). Increased social tolerance promotes cultural transmission itself through the evolution of teaching and conformity (Tennie et al. 2009) as well as through increasing memory (Urban Levy et al. 2023).

Anthropological findings support the view that humans self-domesticated, i.e., selected themselves for prosociality (Cieri et al. 2014), as have done other species, e.g., possibly bonobos in contrast to chimpanzees (Hare et al. 2012). Biological features of the “domestication syndrome” such as craniofacial feminization and increased serotonergic and dopaminergic innervation of the amygdala coincide in the humanoid fossil record with the transition to settlements and increase in parallel to population density (Hare 2017). In a famous experiment recapitulating the domestication history of dogs, wild foxes were domesticated in only 45 years demonstrating the degree by which selective mating can accelerate the emergence of prosociality (Hare et al. 2005). Domesticated foxes shared many of the morphological and behavioral features (feminized skull, sensitivity to human social cues such as gestures, prolonged juvenile development, etc.) of dogs versus their wolf ancestors (Hare et al. 2005). It has been suggested that (self-) domestication genes are primarily found in neural crest development (Sanchez-Villagra et al. 2016), but others have argued that shared reproductive disruption would be sufficient to explain specific differentially expressed genes in domesticated animals (Gleeson and Wilson 2023).

Increasing prosociality led to the formation of very large groups and sedentariness, and together with the co-emergence of domesticated animals, also guided coevolution of zoonotic pathogens and adaptive immune responses to these. In history, social groups with efficient immune responses to zoonotic pathogens came to dominate other groups that had not developed such an immunity as, for example, happened in the selection of prosocial life forms over hunter/gatherers (Diamond 1997; Kennedy 2024), thus further selecting and thereby accelerating prosociality as a human individual trait.

8.3 Assortative mating

The progression in cognitive and toolusing abilities in individuals likely increased the preference to company with individuals of similar cognitive abilities, leading to assortative mating, i.e., the nonrandom preference for mating partners. The increased synchronization of mates with improved communicative and social abilities consequently greatly accelerated fixation of genes favoring social transmission (Creanza et al. 2017; Zeng and Henrich 2022). As a consequence of assortative mating, social transmission-relevant but less frequent genes would have had a higher and faster chance of becoming penetrant and thereby guide biological evolution through the fixation of these traits. Socially biased learning and communication has been shown to select for intelligence (van Schaik and Pradhan 2003) and would thus accelerate assortative mating in a loop-like, self-reinforcing manner.

8.4 Baldwin effect

Acquired behaviors can be genetically fixed between generations by the “Baldwin effect” (Baldwin 1896). The Baldwin effect is a phenotypic change occurring in an individual as a successful adaptation to its environment that becomes gradually fixed in its developmental genetic or epigenetic repertoire (Baldwin 1896; Dennett 2003; Maynard Smith 1987). This means that a continued advantageous behavior of the same kind for which flexibility was not an advantage anymore, put biological selection pressure on genes that fixed this behavior to overcome its inherent instability and delayed reaction for the sake of a rapid and stable response. Such genetic fixations are gradual and concern domain-general abilities such as, for example, specific processing speed, episodic memory, social recognition, social transmission, and others, rather than whole complex behaviors. Alternatively, under a similar selection pressure, a rare random mutation is also conceivable to rather suddenly enable a useful adaptation.

8.5 Niche construction

Yet another mechanism how tool use shaped biological evolution is niche construction (Clark 2011; Laland et al. 2000; Odling Smee et al. 2003). Niche construction refers to “activities, choices, and metabolic processes of organisms, through which they define, choose modify, and partly create their own niches” and thereby actively changing selection pressure on the niche creator (Laland et al. 2000). Human cultural activities, including those involving tools, have fed back on genetic evolution by modifying natural selection and thereby altering genotype frequencies (Laland et al. 2000; Odling Smee et al. 2003). This has been abundantly reviewed and includes, for example, the persistence of lactose genes and light skin color to counter vitamin D deficiency when northern hemisphere habitats were colonized after the control of fire and the domestication of animals were achieved (Henrich 2016). In the process of niche construction, a cycle of mutual fitting between individual and niche is started that gets optimized and that includes both the natural and the social environment (Veissiere et al. 2019). Another example for a cumulative cultural solution replacing a hard-wired biological solution is the relative atrophy in digestive organs in response to the acquisition of controlling fire and cooking: the relative ease to obtain essential nutrients by the achievement of cooking led to smaller jaws and intestines and the saved energy needs of these organs could be expended on a growing brain (Wrangham 2017).

8.6 Neural reuse model

The neural reuse model assumes that existing brain structures are used for novel functional behaviors (Anderson 2014, 2016) and could explain how massive behavioral – ultimately sequential motor acts – could be stored and processed with increasing social transmission of language and tool instructions in a relatively short amount of time. According to the neural reuse model, “language works because it has developed to take advantage of and is fitted to our interactive sociality (and not because we evolved specialized, dedicated, modular neural machinery to support it)” (Anderson 2016). According to this concept, learning-initiated brain plasticity scaffolded by cultural practices drives the acquisition cognitive skills such as arithmetics, language, or writing (Fabry 2019).

The brain is extremely plastic, i.e., neural processing can be taken over by other regions. For example, it is a frequent clinical observation that when lesions occur at very young age, other parts of the brain can take over with little functional disability (de Schotten and Forkel 2022). In a landmark experiment, young ferrets were operated and rewired postnatally such that they could “see” with the auditory cortex (von Melchner et al. 2000). These observations clearly indicate that there are no inborn, i.e. predetermined specialized areas per se in the brain responsible for one function but rather that specialized areas are assigned during development and interaction with the environment. Along these lines, it has been shown that learning to read and write changes the structure of the embodied brain (Dehaene 2009). Significant rewiring of the brain upon externalizing large parts of computation through the extended mind is also believed to explain the factual shrinking of brain mass in Homo sapiens since the last ice age (10,000 years ago) by around 12 % (Tattersall 2024).