On Chater and Christiansen’s compelling account, cultural complexity arises from local, directed innovation. Their framework explains why tinkering is directed, as people intelligently solve immediate cognitive challenges. But it misses a key reason why tinkering is local – why innovation remains constrained to close variants of current designs, rather than exploring radical alternatives.

Their answer is cognitive myopia. Cognitive and time constraints prevent humans from considering all relevant variables – or all possible alternatives – and so search well-trod paths to climb local gradients in design space. This is surely an important part of the story – perhaps the main part, in cases of technology and artifacts. In cases of norms and institutions, however, it misses a complementary social mechanism: People often feel normatively committed to well-trod paths, and they get reprimanded when they step out of line.

When working together, social tinkering often produces joint plans that we think should be followed. For instance, siblings with similar income might initially evenly divide the costs of hiring occasional household help for an aging parent. Once implemented, such joint plans set the relevant precedent. And deviation from precedent may be moralized – coming to be seen as a principled matter of fairness or the only natural and just solution. Thus tinkering stays local not only because people fail to think bigger, but also because social tinkering favors solutions that are viewed as normative, with deviation from previously established behavioral patterns inviting sanction. In principle, such conditions could persist beyond the point at which any party would have chosen them ex ante. For instance, suppose one sibling’s income increases considerably, and meanwhile the parent must be transferred to an assisted-living facility. They might nevertheless feel bound by the precedent of evenly sharing costs – constraining the choice of facility – even when both siblings and the parent might have chosen, in the absence of the precedent, greater contributions from the sibling with a higher income (and therefore a nicer facility).

We have begun to explore these dynamics in simple cases of labor division. Two people coordinate on a particular division of tasks and interact repeatedly. The solutions that are first entertained, suggested, and adopted (“what should we do?”) may already be predominantly those that will be viewed as acceptable, appropriate, or fair (Le Pargneux et al., Reference Le Pargneux, Cushman, Zeitoun and Chater2025). Furthermore, once implemented, such solutions set the precedent for future interactions and gradually acquire moral force. But what happens when one party wants to change the arrangement? Perhaps circumstances have shifted and a radically different division would be more efficient. Even if the new arrangement were Pareto-improving, switching to a new equilibrium introduces the risk of miscoordination. The proposer is breaking an implicit agreement and disappointing legitimate expectations (Chater, Zeitoun, & Melkonyan, Reference Chater, Zeitoun and Melkonyan2022; Le Pargneux, Reference Le Pargneux2025; Le Pargneux et al., Reference Le Pargneux, Levine, Zeitoun, Chater, Tenenbaum and Cushman2026).

Recent experimental work supports this picture. In incentivized coordination games, people readily moralize deviations from implicit joint plans – plans that have never been implemented nor explicitly agreed to – in novel contexts (Le Pargneux, Chater, & Zeitoun, Reference Le Pargneux, Chater and Zeitoun2024; Le Pargneux et al., Reference Le Pargneux, Cushman, Zeitoun and Chater2025). Moreover, participants judge deviations from established equilibria as morally blameworthy and punishment-worthy, even when deviation might lead to better outcomes (Calcott et al., Reference Calcott, Braga, Cushman and Le Pargneux2026; Guala & Mittone, Reference Guala and Mittone2010; Przepiorka et al., Reference Przepiorka, Szekely, Andrighetto, Diekmann and Tummolini2022). Such resistance to abandoning conventions may reflect the risk of miscoordination during transitions (Morin, Reference Morin2022). Even clearly superior alternatives may fail to spread because the potential costs of failed coordination outweigh the expected benefits of successful change.

This suggests an extension to the social tinkering framework. People jointly coordinate on solutions they view as best, given the current circumstances. Such solutions set the precedent for future interactions. Precedent gradually becomes moralized, and deviation comes to be seen as not merely suboptimal but wrong. Moralization stabilizes coordination around a given equilibrium, but at the price of constraining innovation to incremental adjustments rather than radical departures.

It also explains something the original framework doesn’t: Why coordination equilibria are so “sticky.” In their account, precedent is cognitively salient but not normatively binding. There’s no particular reason why people wouldn’t switch to better arrangements when they become available. But empirically and historically, people often stick with suboptimal conventions (for instance, tipping). While tinkering with this norm (say, shifting the percentage tipped from 15% to 20%) may be feasible, if a customer refuses to tip or a company abolishes tipping entirely, coordination is at risk. This extension also predicts when cultural change may become possible. When conditions shift such that coordination on alternatives becomes feasible, through public negotiation, mutual observability, or the breakdown of common knowledge sustaining precedent, moralization may be reversed.

Chater and Christiansen argue that tinkering is directed but myopic. We suggest that it is also importantly constrained, not only by cognitive limits but also by social and moral pressures. Innovation that strays too far is thus impermissible, rather than merely unthinkable.