R2.1. Overlapping and interacting stages

A few commentators argued against what they perceived as a strict linear progression of the six stages. Instead, they suggest that “[t]he stages form loops” (Menary & Gillett), result in “a multifaceted path of cultural evolution” (Xu & Corriveau), or “proceed along multiple pathways shaped by local conditions and learning arrangements” (Papa et al.). We agree! Indeed, when introducing the six stages in the target article, we explicitly stated that the stages are “overlapping and interacting” and the accompanying Figure 1 contains a backward-pointing arrow labeled “feedback and interaction between stages.” The six stages are not in a strict linear order. For example, in the target article we proposed “a co-evolutionary virtuous cycle between Stages 3 and 4: simple social tinkering provides the basis and impetus for the emergence of simple linguistic communication that, in turn, feeds back to allow for more complex social tinkering, which promotes more complex linguistic abilities, and so on, leading to the enormous cultural and linguistic complexity created by modern humans.” Importantly, though, we maintain that the qualitatively distinct behaviors associated with each stage depend on behaviors at the previous stage (except for Stage 1).

Xu & Corriveau and Papa et al. rightly emphasize that cultural evolution does not follow a single trajectory. Human history and the sheer diversity of extant cultural systems are testament to the multifarious pathways that cultural evolution takes. Again, we agree. Our target article did not suggest that social tinkering is deterministic – far from it. Tinkering depends on whatever is readily available in the local environment. In the case of non-social tinkering, the availability of certain raw materials and prior innovations will constrain what new solutions are easily attainable. The environment for social tinkering includes the community’s socio-cultural context (Xu & Corriveau), which is itself a product of previous social tinkering. As the physical and social environment changes over time (often through tinkering), so does the range of possibilities for non-social and social tinkering, leading to a wide range of possible trajectories for cultural evolution.

R2.2. Cultural niche construction

Several commentators note the importance of cultural niche construction (e.g., Boyd, Richerson & Henrich, Reference Boyd, Richerson and Henrich2011; Laland, Odling-Smee & Feldman, Reference Laland, Odling-Smee and Feldman2000) for cultural evolution and, by extension, social tinkering (Burton et al.; Mollica et al.; Sueur et al). Although we did not mention “niche construction” explicitly, it is implied by our framework as pointed out by Ross: “local social tinkering, without central design or general hierarchy, has produced the cascading complexity of human culture and dynamical niche reconstruction.” Through both non-social and social tinkering, communities reconstruct their local environment – their niche – which in turn both constrains and provides material for further tinkering.

Similarly, Sueur et al. highlight the contribution of social niche construction to cultural evolution, noting the existence of social networks in social animals beyond humans. They further argue that “[c]umulative culture does not begin with the first ad hoc rule; it is preceded by the first self-organized regularity in a social network.” Indeed, we did not suggest that there is no culture before social tinkering. Stages 1 and 2 are cultural – and likely result in some social “rules.” But social tinkering arises when people begin to tinker with the social rules themselves rather than other cultural products (like tools). And we agree that existing social networks will affect subsequent social tinkering and cultural evolution more generally (e.g., intricate innovations are easier to maintain in small, interconnected populations; Reali, Chater & Christiansen, Reference Reali, Chater and Christiansen2018).

Sacco offers the intriguing proposal that the arts constitute a key functional niche, as a safe environment in which “outside-the-box” variation can be generated to assist the development of the social and cognitive skills necessary for advanced tinkering at Stages 3–5. He joins Sperber (Reference Sperber1996) in suggesting that the arts provide a wellspring of ideas, rules, and other kinds of cultural representations to accelerate cultural evolution. Sacco proposes that the arts help create a self-reinforcing loop in which “richer artistic culture enables more creative tinkering, generating coordination challenges that stimulate further exploration, while aesthetic engagement provides each generation with the integrative capacities that the previous generation’s tinkering demands.”

Burton et al. stress the importance of modern “designed environments” which, from our perspective, are themselves a product of tinkering. Most design processes, while being driven by top-down goals (like maximizing profit, increasing efficiency, attracting more eyeballs, etc.), require tinkering to make them work in the real world. Thus, designed environments are part of the cultural niche within which social tinkering takes place. Of course, the non-social environment, too, can influence social tinkering, in particular where the outcomes of prior technological tinkering (e.g., the web, social media, generative AI) directly limit or expand what can be accomplished with future social tinkering.

With the increasing importance of the social-cultural niche to human societies, the target article distinguished between two types of learning (section 4.1; see also Chater & Christiansen, Reference Chater and Christiansen2010; Christiansen & Chater, Reference Christiansen and Chater2016a, Reference Christiansen and Chater2022): (i) learning to navigate and manipulate the natural world through non-social tinkering (Stages 1 and 2); and (ii) learning to negotiate the social world, including the accumulated products of social tinkering (Stages 3-6). People must do both, but mostly separately, because the target for each kind of learning is very different. When learning about the natural world, the key is to follow whatever external laws govern what needs to be learned (e.g., in hunting, these external constraints might involve the aspects of aerodynamics and gravity that affect projectile trajectories, and facts about prey-animal speeds and avoidance patterns). But learning socio-cultural rules is a problem of coordination (akin to a Schelling game; Schelling, Reference Schelling1960), where the correct answer is what everyone else is doing (e.g., drive on the right in the US and on the left in the UK). So, contrary to Burton et al., we are not claiming “that people must coordinate with other people instead of aligning with an external reality” (their emphasis). Rather, people need to learn about both the social and the non-social world – but this involves different kinds of learning.

R2.3. Development and cultural evolution

In the target article, we emphasized the relationship between development and cultural evolution, and how learning to negotiate the social world is relatively easy compared to navigating the natural world. Social tinkering is carried out by people with specific cognitive abilities, living in a given socio-cultural niche, which, in turn, shapes the resulting solutions to a substantial degree. So the guesses of each new generation of children about how to align with others in their community can rapidly home in community norms, because these community norms were created by previous generations operating with the very same niche and constraints.

Menary & Gillett worry that we ignore the possibility that such enculturation can feed back and change cognitive abilities. As an example, they note that learning to read recruits parts of the neural architecture, such as the so-called “visual word form area,” previously used for other cognitive skills (Dehaene & Cohen, Reference Dehaene and Cohen2007). We agree that enculturation affects our cognitive abilities and may “recycle” older neural systems. Indeed, we have elsewhere discussed reading, in a very similar way, to highlight how a culturally-evolved system, shaped by cognitive constraints, can result in highly specialized brain areas (Christiansen & Chater, Reference Christiansen and Chater2016a, Reference Christiansen and Chater2022). Moreover, we suggested that language, too, is a cultural product, shaped by the brain (Christiansen & Chater, Reference Christiansen and Chater2008, Reference Christiansen and Chater2016b), and that learning language also results in considerable neural specialization.

Another facet of children’s learning that may support this perspective is mentioned by Xu & Corriveau and Tian & Lindström. Both stress that children are not passive learners but are strategic in choosing whom they learn from. Such strategic learning might speed up the acquisition of both social and non-social aspects of culture, if children choose to copy those they perceive as more competent learners. Xu & Corriveau even suggest that Stage 1 may not be necessary for humans because children tend to be strategic learners, copying from more competent individuals. While we agree that much non-social tinkering in the human case will be at Stage 2, learning at Stage 1 is still needed, for example, when learning to walk, discovering object persistence, and other basic features of the physical environment. Additionally, from a broader evolutionary perspective, Stage 1 is needed to account for learning in non-human animals (indeed, Tennie & Planer claim that only humans have reached Stage 2).

Along with the six stages of tinkering, we also proposed a possible virtuous cycle to explain increased hominin brain size: “a more complex culture (including language) strengthens the biological selection pressures for the large brains required to master and thrive in that culture; and large brains drive the intelligent (if myopic) directed innovations which further complexify the culture.” However, Menary & Gillett challenge us to explain “how the cycle plays out within a single lifetime.” While our account is devised to explain the emergence of cultural complexity in the human lineage, Menary & Gillett do raise interesting questions about whether ontogeny might recapitulate phylogeny. For example, do children go through some approximation of our six stages during development, starting with individual learning (e.g., walking; Stage 1), followed by learning about the natural world from others (Stage 2), then beginning to negotiate social relationships (Stage 3) and acquiring language (Stage 4), and finally using language-mediated means to tinker further with the social world (Stage 5) and the natural world (Stage 6)? And does feedback between developmental stages create a virtuous cycle involving increased cognitive and linguistic abilities that support further social and non-social tinkering, which in turn furnish the child with increasingly complex cognitive, linguistic, and socio-cultural skills? More research is needed to answer these (and related) questions about possible parallels between development and cultural evolution.

R2.4. The scope of cultural complexity

Papa et al. worry that our account narrows the scope of cultural complexity too much, contending that “no other hominin, and not every modern human group, brought forth the cultural systems described in the later stages of the model.” However, this contention appears to derive from a misunderstanding. Our proposal is that the six stages of cultural complexity arise in all extant human communities, from hunter-gatherers to WEIRD (Western Educated Industrialized Rich Democratic; Henrich, Heine, & Norenzayan, Reference Henrich, Heine and Norenzayan2010) societies, and possibly earlier hominin groups as well. This includes Stages 5 and 6, where language enriches both social and non-social tinkering by making explicit previously unspoken rules, habits, and other ways of doing things, so that they can be more easily debated, rejected, amended, and transmitted. We further suggested that the invention of writing would supercharge both social and non-social tinkering and, in many cases, lead to formal institutions. However, we did not suggest that writing and formal institutions are a necessary part of Stages 5 and 6. After all, it wasn’t until about half a century ago that literacy can be said to be widespread across the world (Buchholz, Reference Buchholz2022). Thus, we agree with Papa et al. that many contemporary sophisticated societies exist “in the absence of writing and formal institutions,” and that such societies do not represent a “failure to progress.”

Denton is also concerned about our notion of cultural complexity, asking us to provide “a clear definition of “cultural complexity” – one that is capable of excluding things like taboos (a norm), tunes (a form of art), and lexical items (such as words), while including examples that the authors invoke elsewhere in the paper such as other moral and social norms, rituals, religious beliefs, art forms, notations, and aspects of language.” In the target article, we suggested that previous epidemiological approaches might offer suitable explanations for “the spread of taboos,Footnote ¹ urban myths, lexical items, tunes” and other potentially “contagious” constructs (such as slang, popular songs, or social media memes like “winter is coming”). These phenomena spread because they become popular among certain groups of people and may even serve some purpose, but not because they provide a solution to a tinkering problem. And although they may change somewhat over time, such modifications are rarely directed toward an improved solution to a challenge, though changes may make them propagate more easily. By contrast, social tinkering innovations are created, given whatever is at hand, to solve specific problems. We suggest that only such directed solutions from social (and non-social) tinkering lead to increased cultural complexity over time as they are amended and recombined with additional innovations to solve new challenges.

R3.1. Tinkering and design

Several commentators focus on the top-down/bottom-up distinction and question the scope of the social tinkering account. Vale & Coughlin note that “the ‘social tinkering’ metaphor effectively captures the incremental nature of cultural growth, yet risks reductionism by casting humans as purely reactive problem-solvers.” Tian & Lindström stress the human ability to engage in goal-oriented planning. Tzafestas stresses the potential importance of leadership and institutional structure. Smaldino emphasizes the impact of coercion, e. g., when one individual or group imposes a plan on others. Xu & Corriveau point out that social tinkering can be shaped by wider cultural factors – so that individual social norms may change as part of economic transformation, such as shifting towards a market economy.

The social tinkering account is entirely compatible with these observations. The account allows that top-down design is entirely possible and indeed is widespread. Constructed languages, such as Esperanto, can be created by deliberate design, as Denton points out. But the overwhelming majority of the world’s roughly 7000 spoken and signed languages are, of course, not designed. Similarly, it is possible to create rules to govern musical composition (for example, Schoenberg’s twelve-tone serialism, Schoenberg, Reference Schoenberg and Stein1984), but the vast majority of musical traditions across the world do not arise through such design. To be sure, once norms and patterns become established, they can be sharpened and codified – just as with the pronouncements of prescriptive grammarians (e.g., in the US, the ubiquitous Strunk & White, Reference Strunk and White2000), the “rules” of harmony and counterpoint (Fux, 1725/Reference Fux1965; Piston, Reference Piston1978), or explanations of the structure of jazz chord progressions (Levine, Reference Levine1995). But these patterns do not originate from abstract design, but emerge after long processes of continual adjustment during linguistic and musical creation. And even designed systems often turn out to be modeled on systems that have originated through social tinkering. Thus, not only is the grammar of Esperanto built primarily on Indo-European elements (Koutny, Reference Koutny2015) but its vocabulary draws on words from Euro-centric sources: 80% Romance, 10% Germanic, 5% Greek, and 5% Slavic (Puškar, Reference Puškar2015).

R3.2. Tinkering is forward-looking

Denton argues that social tinkering is too “near-sighted” and that much cultural change is “far-sighted” and planned. We agree that tinkering is often forward-looking – indeed, it is central to Adam Ferguson’s vision of human social complexity as generated by human action, not human design, that individual actions are goal-directed. Thus, legal arguments about the lawful basis for deciding a case will look backwards to past precedents but also forward to possible precedents that may be set for the future. Ferguson’s point is not that people are oblivious to future consequences, but their primary concern is local adjustment in the present (rather than redesigning the entire social system from scratch), and that their ability to foresee consequences is limited. In language, law, science, technology, and elsewhere, the long-term consequences of local changes can be substantial and totally unexpected. So, even in Denton’s example of setting up seed banks (presumed to be the antithesis of tinkering), which may have a long-term foreseeable objective (e.g., preserving biodiversity), the specific route by which preserved biodiversity may translate into new pharmaceuticals, disease- or drought-resistant crops, or might be grist for future genetic engineering whether for good or ill, is not foreseeable.

We agree, too, with Hong that social tinkering may draw not just on the forward-looking coordination that we stress (e.g., in planning a joint activity), but in retrospective coordination – e. g., in Hong’s example, jointly deciding why a child has fallen ill, and hence what action we should take. Such retrospective coordination may be important in shaping systems of belief (e.g., modifying folk beliefs about illness in the light of the new case) and guiding future actions (taking steps, whether effective or not, aimed at reducing the likelihood of future illness).

Undoubtedly, much human activity is the product of design rather than tinkering: architecture, product design, mass manufacturing, building large IT systems, playing from sheet music, or following instructions for self-assembly furniture. Unlike tinkering, these activities distinguish between a design stage, in which the process is envisaged, formalized, and codified, and an implementation stage, with the goal of carrying out the process as exactly as possible rather than to make ongoing adjustments.

Still, even where this separation is clear, the design process itself is likely to involve tinkering by the designer or designers. Over longer timescales, any given design will typically be a variant on prior designs, incorporating small adjustments. This design, too, is often embedded within a broader process of tinkering. Indeed, Stages 5 and 6 in our account of social tinkering focus on how, once communicative conventions (languages, notations, diagrams) have been established (at Stage 4), tinkering can operate over designs, theories, and plans. Thus, we agree with Burton et al. that contemporary society is a complex interaction of both tinkering and design. To paraphrase Newton’s famous quote, engineered design stands on the shoulders of a giant number of tinkerers.

R3.3. Tinkering and group selection

Some commentators ask how social tinkering might unfold in light of possible cultural group selection. Mollica et al. highlight the importance of competition and selection within and between groups, asking how groups decide to cooperate in the first place, which they see as prior to social tinkering (construed as coordination). They also note that cooperation can be costly for individuals and thus that institutions are needed to make sure that everyone complies with agreements. We broadly agree but stress that coordination problems arise without (obvious) common goals, where coordination is mutually beneficial (e.g., following the same linguistic rules, traffic norms, etc.); and even in social dilemmas or competitive contexts, it is still possible to agree on common goals that may differ from the personal goals of any individual (e.g., virtual bargaining would tell us that two people can implicitly agree, e. g., to share, even if both are selfish; Chater, Zeitoun, & Melkonyan Reference Chater, Zeitoun and Melkonyan2022; Misyak & Chater, Reference Misyak and Chater2014). Moreover, we concur that cultural group selection can play a role in cultural evolution – but this isn’t just a matter of “selection.” After all, groups continually borrow institutions from each other, because they work well (at least from the point of view of those proposing the borrowing); e. g., legal frameworks, constitutions, notations, organizational structures, languages, etc., are continually subject to borrowing and subsequent tinkering within and across groups.

In a similar vein, Singh & Toutée argue that while social tinkering between individuals might be useful for capturing the emergence of social order in small-scale societies, such as early American colonial settlements (Ellickson, Reference Ellickson1993), it may be less relevant for larger, more complex systems where between-group competition might play a more central role. We agree that in these cases, tinkering is not always between individuals but can involve representatives of groups of individuals (elected or otherwise). Yet, even in such situations, there are few “great leaps” – rather, tinkering remains key at the group level (e.g., trade agreements, peace treaties, labor contracts) – though the various parties engaging in such tinkering may not have equal power and influence over the tinkering process. Thus, cultural evolution through between-group competition is still largely myopic but forward-looking; for example, different organizations, countries, and cultures may actively attempt to learn from each other or be more passively influenced, but this is very different from group selection in the standard sense where “bad” cultural structures just “go extinct” (e.g., Wilson & Sober, Reference Wilson and Sober1994).

More broadly, social tinkering is just one of many interacting processes that shape cultural and social change, which may include processes of conflict, lobbying, oppression (Hong; Smaldino; Nowak et al.; Ross; Stibbard-Hawkes; Tzafestas), between-group competition (Mollica et al.; Singh & Toutée), and socio-cultural niche construction Burton et al.; Mollica et al.; Sueur et al.). But we believe that social tinkering is central to understanding how complex human societies and their institutions of language, social norms, laws, money, and religions can emerge.

R3.4. Tinkering and the dark side of human nature

Does the social tinkering account adequately accommodate the darker sides of human social life – conflict, self-interest, domination, exploitation, and oppression? Human societies are not merely systems of peaceful negotiation, cooperation, and mutual accommodation. They are also marked by persistent struggles over status, resources, prestige, ideology, and power. Social institutions may emerge that entrench hierarchy, legitimize inequality, or enable some individuals and groups to dominate others in ways that are coercive, exploitative, and profoundly illiberal (Hong; Smaldino; Mollica et al.; Nowak et al.; Ross; Stibbard-Hawkes; Tzafestas).

Social tinkering is entirely compatible with these darker outcomes: tinkering is not inherently benign, progressive, or cooperative. It is simply the process through which social practices, norms, and institutions evolve through local adjustments, improvisation, imitation, negotiation, and contestation. Those processes may generate systems that are humane and egalitarian; but they may equally produce systems that are exclusionary, violent, unstable, or oppressive.

Ross contrasts an idealized vision in which social order emerges harmoniously from individuals acting within a shared system with a more conflictual metaphor: rather than thinking of Neurath’s boat continually repairing itself at sea by ingenious improvisation, we should see culture as a flotilla of competing boats engaged in continual collision and exchange of fire. This image may better capture much of cultural and political evolution. Human groups are not only coordinating with one another; they are also competing, bargaining, resisting, coercing, deceiving, and struggling for advantage against other groups; and group membership and identity is itself continually being adjusted. Cultural evolution therefore unfolds not only through cooperation but through conflict among individuals, factions, classes, institutions, and societies.

Indeed, many stable social arrangements may persist precisely because they work to the advantage of powerful groups. Norms, conventions, and institutions are not selected solely because they are collectively beneficial, efficient, or morally defensible. They may survive because they are enforceable, because they coordinate expectations, because alternatives are difficult to organize, or because dominant groups possess disproportionate influence over cultural transmission and institutional design. We agree with Stibbard-Hawkes that active top-down lobbying is a powerful force in many areas of culture, rather than mere bottom-up diffusion.

The social tinkering account is a descriptive framework – there is no automatic tendency for cultural evolution through social tinkering to yield benign outcomes. There is no inevitable pressure for human values to become increasingly inclusive and universal, toward the growth rather than constriction of human rights, or to favor open debate and inquiry over the entrenchment of dogma. Similarly, there is no inherent tendency for belief systems arising from social tinkering to become closer to the truth or more instrumentally useful. Indeed, social tinkering can lead both to the growth of modern mathematics and science, but also to any number of elaborate and incompatible religious systems, cults, and conspiracy theories (where the power of the imaginary can disconnect us from reality, Ross), and potentially maladaptive practices of magic, divination, and taboo systems, as stressed by Hong.

The local character of agreements – typically forged between particular agents in specific social interactions – extends naturally beyond abstract conventions to encompass roles, responsibilities, rights, and obligations (Bicchieri, Reference Bicchieri2006). Through this process, increasingly complex political and organizational structures can emerge, through gradual processes of spontaneous order, in which arrangements are continually adjusted to close loopholes, extend to new cases, and to be generalized and modified in a myriad of ways (Ostrom, Reference Ostrom1990). These processes often unfold under highly adversarial conditions, as different groups within and between societies attempt to reshape the rules of the game in ways that advance their own interests, often with threats or violence (North, Wallis, & Weingast, Reference North, Wallis and Weingast2009). The resulting changes may be incremental, but they may also be revolutionary, as when one political order is displaced by another. As ever, there is no inherent drive for political orders to develop in benign directions. Cultures can evolve into liberal democracies, brutal dictatorships, and anything in between (Fukuyama, Reference Fukuyama2011). Thus, the same processes of social tinkering that form oppressive institutions may also generate movements for emancipation, equality, and social change (Acemoglu & Robinson, Reference Acemoglu and Robinson2012).

Baran et al. are skeptical that institutions and organizations can arise in this way. By contrast, while we agree that there is sometimes a crucial role for top-down design, we argue that social change typically occurs through processes of modification and adaptation of existing arrangements. Moreover, even in cases where new institutions are ostensibly the product of central planning, their survival and functioning depend on subsequent processes of social tinkering. We suggest that durable social arrangements, whether liberal or oppressive, typically emerge only through extended processes of adjustment, negotiation, and adaptation, in which norms, institutions, and expectations gradually become aligned.

R4.1. Functional pressures on social tinkering

Baumard and Dubourg stress that social tinkering is guided by motivational systems that may themselves be highly stable. Dubourg lists motives such as harm avoidance, hunger, fairness, cooperation, mate selection, offspring protection, status, and exploration. Baumard, drawing on Jacob’s (Reference Jacob1977) metaphor that biological evolution is a tinkerer, further argues that these motivations are ultimately organized around the underlying criterion of inclusive fitness. Without such a criterion, he suggests, biological evolution would resemble a random walk rather than tending toward the emergence of increasingly complex and functional systems.

Baumard suggests that the social tinkering account lacks an analog of inclusive fitness for cultural evolution. Yet we suggest that this absence is not a weakness but a defining feature of culture itself. In cultural evolution, there is no single criterion of success. Rather, goals, standards of evaluation, and even the problems to be solved are themselves open to continual revision and tinkering.

Cultural evolution is not a random walk, of course. Indeed, it is a distinctive feature of human sociality that enormous amounts of thought, argument, and collective deliberation are devoted not merely to following rules, but to determining which rules we ought to live by. Indeed, human societies continually debate what counts as success, what constitutes an appropriate social norm, rule of etiquette, or dress code, which conventions and standards are defensible and which should be abandoned, and which objectives individuals, groups, and entire societies should strive to achieve.

At the most abstract level, these debates take the form of ethics, political philosophy, and aesthetics – domains in which disagreement persists even over the very possibility of reducing human judgment to any single organizing principle. In ethics, for example, there are powerful yet incompatible traditions of utilitarianism, deontology, contractualism, virtue ethics, and many others (although these varied traditions may have deep connections, e. g., Levine et al., Reference Levine, Chater, Tenenbaum and Cushman2026). Cultural evolution therefore permits political, social, moral, and artistic traditions to diverge along multiple trajectories rather than converging upon a single criterion of optimality.

This divergence is, of course, not unconstrained. The huge variety of human languages, for example, is nevertheless constrained by common functional pressures: languages must support communicative richness, flexibility, and open-endedness, while also fitting the perceptual, motor, memory, and learning capacities shared by human beings (Christiansen & Chater, Reference Christiansen and Chater2008, Reference Christiansen and Chater2016b) – leading, for example, to linguistic systems with a roughly compositional structure, inventories of sound-based units, common maxims of conversation, and so on, but not converging on any rigid systems of “language universals” (Evans and Levinson, Reference Evans and Levinson2009). More generally, cultural systems evolve within broad cognitive, social, ecological, and physical constraints. But within those constraints, both the goals and the standards by which cultural practices are evaluated remain open to continual revision. Culture therefore differs fundamentally from biological evolution in lacking any fixed overarching objective analogous to inclusive fitness; instead, the criteria of success themselves can continually be renegotiated through social tinkering.

R4.2. Stability and inertia in social tinkering

Cultural evolution does not have a uniform trajectory across history, populations, and environments. While the target article focused on how social tinkering can lead to increased cultural complexity, several commentators rightly stressed the importance of stability during cultural evolution (Liu et al.; Nowak et al.; Papa et al.; Ross). For example, Papa et al. discuss how some cultures have evolved long-lasting stable solutions to sea voyage navigation and others for mushroom foraging. Liu et al. suggest that social network properties, such as a high degree of clustering, may help stabilize such local traditions. And Nowak et al. argue that cultural evolution can involve relatively long periods of stability followed by abrupt transitions, with institutions serving as “attractors” of stability. Our social tinkering framework is compatible with such stability (we discussed both “reuse” and “entrenchment” of past innovations in the target article). When a society works out a useful solution to either a non-social tinkering problem (like navigation or foraging) or a social tinkering problem (via systems of social norms), then it makes sense to stick with it (following the adage “if it ain’t broke, don’t fix it”). But even in such cases, some tinkering likely takes place “at the edges.”

Of course, a downside to stability can be excessive inertia. Calcott et al. raise an important caveat concerning the challenge of renegotiating implicit agreements on the fly. In coordination problems, there are often many possible equilibria. If the environment changes, and the current equilibrium becomes clearly inferior to some alternative equilibrium, it may still be difficult for a group to shift collectively to the better option. This may be true even when it is common knowledge that the alternative equilibrium is superior, because unilaterally switching to a new action while others remain with the old action can lead to poor outcomes for everyone, and potentially to social opprobrium for the individual attempting the change.

This phenomenon, which Calcott et al. demonstrate experimentally, is likely to arise very widely: from the difficulty of shifting from imperial units to the scientifically superior metric system, to large-scale political and economic transitions, such as the uneasy movement from communism to a mixed economy discussed by Novak et al. Exploring such effects, and in particular the mechanisms that allow people to shift more easily toward better agreements, is an important topic for future research. Sacco suggests that the arts might serve as an accelerator for new ideas that may help break out of inferior equilibria (though Ross cautions that our powerful imagination may also conjure up notions so disconnected from reality that they may threaten ecological stability).

R4.3. The emergence of normativity from social tinkering

Suriano and Baran et al. raised the question of whether the social tinkering account can explain the origin of normativity: that linguistic and social constraints and norms do not merely describe how people actually behave, but provide criteria concerning how people ought to behave, so that their violation is met with disapproval, punishment, or even ostracism.

Rather than seeing this as a challenge for the social tinkering account, we view normativity as emerging naturally within our framework. Social tinkering involves creating implicit, and sometimes codified, agreements concerning behavior. And agreements, by their very nature, have a normative element. To agree to a set of norms or conventions is to accept that these norms or conventions should guide one’s own and other people’s behavior, and to expect disapproval when such norms are violated.

Indeed, the appropriateness of forms of disapproval and/or punishments is often built into the implicit agreement itself. Thus, it is normatively appropriate to criticize or sanction a person who violates an agreement without good reason. Moreover, sanctions can themselves become conventionalized, whether in informal contexts – where a norm might emerge that a person should compensate someone whom they have injured, or engage in an act of apology or penance – or in formal systems of law, regulation, and policing.

The accumulation of case law in the British and US common law traditions provides a particularly interesting illustration of social tinkering in a normative domain. Judicial decisions in particular cases, together with the reasoning used to justify them, generate precedents for future cases. Over time, these precedents may become generalized, codified in legislation, and further organized and systematized through legal analysis and scholarship.

From a social tinkering perspective, normativity is therefore present from the beginning. Norms, their enforcement, and the institutional structures through which they are organized arise largely through bottom-up debate and negotiation, typically focused on the problem at hand. The approach thus provides a bottom-up account of how the complex systems of laws, regulations, and enforcement mechanisms characteristic of governments and societies can gradually accumulate.

We agree with Suriano that the mere reproduction of behavior is not sufficient to establish normativity. As Suriano notes, large language models (LLMs) capture normative regularities embedded in conventional language remarkably accurately. But LLMs are not participants in agreements concerning communicative conventions, nor are they committed to those conventions in the way that people are. According to our account, language arises from a history of communicative interactions, in which momentary communicative “conventions” are implicitly agreed upon (and are hence treated normatively) to achieve an immediate communicative goal, but which provide precedents for future communication. Over time, cross-cutting and only partially consistent patterns become entrenched, and the systematic patterns studied by linguistics emerge through processes of spontaneous order (Chater & Christiansen, Reference Chater, Christiansen, Lappin and Bernardy2023; Christiansen & Chater, Reference Christiansen and Chater2022). But LLMs skip this process of social tinkering entirely, jumping directly to Stages 4 (language), Stage 5 (language-mediated manipulations of social norms, rules, etc.), and Stage 6 (language-mediated manipulations of science, technology, etc.). An interesting open question is whether skipping cultural evolutionary stages in this way leads to LLMs having only a relatively shallow mimicry of linguistic behavior (Bender et al, Reference Bender, Gebru, McMillan-Major and Shmitchell2021), or whether human-level linguistic competence can be achieved “directly” from copying human linguistic behavior (see Braga et al., Reference Braga, Hayes, Myers, Turrini and Chater2026; Chater, Reference Chater2023).

R4.4. Cognitive prerequisites for social tinkering

Finally, several commentators consider the cognitive prerequisites for social tinkering. Tennie & Planer focus on the traditional notion of “social learning” as copying from others, which is involved in Stage 2 of our account. Specifically, they argue that complex cultural evolution requires what they call “know-how” copying – being able to understand the causal components of copying actions as opposed to simply mimicking the action sequence. On their reading of the current evidence on non-human primate cognition, know-how copying is not present in apes but is uniquely human. The social tinkering account is agnostic regarding when sophisticated copying might have appeared, but sophisticated copying is needed for subsequent social tinkering stages to take off.

A partial mechanistic answer to Tennie & Planer can be found in the commentary by Jon-And et al. They propose that sequence-processing abilities may be a key prerequisite for representing and passing on complex behaviors faithfully. We agree that complex sequence learning is important for cultural evolution (we have elsewhere discussed at length how cognitive constraints on sequence processing have shaped language evolution, see e. g., Christiansen & Chater, Reference Christiansen and Chater2008, Christiansen et al., Reference Christiansen, Dale, Ellefson, Conway, Cangelosi and Parisi2002; Cornish et al., Reference Cornish, Dale, Kirby and Christiansen2017). However, the need for precision may differ across domains. It may be important for the correct assembly of complex artifacts (where memory constraints may be alleviated to some degree by offloading to the external world). By contrast, capturing long sequences of spoken language in real time is not possible because of memory and attentional constraints (the Now-or-Never bottleneck, Christiansen & Chater, Reference Christiansen and Chater2016b), and sequence processing need only be good enough for current communicative purposes (Goldberg & Ferreira, Reference Goldberg and Ferreira2022).

The commentary by Tian & Lindström can be interpreted as adding a more causally-oriented component to the faithful sequence copying of Jon-And et al., which in combination may support Tennie & Planer’s know-how copying. Specifically, Tian & Lindström suggest that humans have evolved planning skills that allow for goal-directed explorations using internal models, given environmental structure (when relevant structure is not present, people revert to random solutions instead; Yaman et al., Reference Yaman, Tian and Lindström2026). With the advent of language, networks with associations between words and objects/events (and their properties and functions) represented in semantic memory can then help create directed innovations. Such goal-directed planning may help provide the kind of explanation of social tinkering that Baumard and Dubourg ask for – though more research is needed to uncover what cognitive prerequisites are needed for social tinkering, which of these are present in other animals, and when they may have appeared in the human lineage.