Stanford's theory of moral externalisation supposes that agents (1) have self-interest-independent motivations to adhere to enculturated norms, and (2) that agents are motivated to shun those who fail to adhere to these norms, and especially if they take advantage of others’ conformity to free-ride. These motivations operate on (culturally?) selected normative content, and in the context of coordination and cooperation problems this motivational combination supposedly opens up a new space of flexibility and adaptive plasticity, through selection of normative content that automatically (via the secondary motivation) sequesters the agent from conspecifics who would otherwise exploit them.
We suggest that the plausibility of Stanford's picture depends on its idealisation. Stanford largely makes the case for moral externalisation through an idealised model of dyadic interaction, with individuals assessing one another as social partners according to compatibility with respect to some specific norm. We think that this idealisation does not scale to realistic social ecologies, to realistic levels of normative complexity, and to realistic degrees of error and error-correction. Moreover, we think Stanford understates the adaptive plasticity made possible by more limited forms of normative cognition.
Our first scaling problem is heterogeneity: What of normative variation? Our lineage has been obligately social for much longer than any plausible inception of norm-following (certainly complex norm-following and reputation-tracking, mediated by language). And in a small group environment, moral externalisation is potentially dangerous: Start shunning without pre-existing positive assortment, and you will shun yourself out of your group. Likewise, novel coordination problems invite normative disagreements, as agents differ on the appropriate response. In Stanford's analysis, the stakes of disagreement rise: They are not just failures to negotiate a contract, a lost opportunity after which everyone then returns to the status quo. They threaten fracture. Unless disagreement was rare or transient, the cost of disagreement would select against norm externalisation.
This suggests that the normative content to be externalised must diffuse rapidly and be synchronised with high fidelity; conflict costs will be high if a new consensus is established slowly, piecemeal. Perhaps a not-implausibly strong conformity bias might deliver this, though it would have to be fortuitously present. Moreover, any mechanism that generated normative uniformity is in tension with the promised benefits of the model with respect to adaptive plasticity. A moral mutant should be expected to be shunned or out-grouped fairly swiftly, and for a normative innovation to spread within a group (without either fracturing it or being snuffed out), it seems it must do so before anyone really notices; to notice is to object. So, any shunning mechanism efficient enough to weed out free-riders (and mitigate disagreement risk) is also liable to weed out valuable innovations.
The second scaling problem is normative complexity. Externalisation and shunning is a blunt instrument; it motivates agents to want to have nothing to do with norm violators, and this imposes significant opportunity costs in any plausible human behavioural ecology. You don't share the kill with me in the way that I think is right, so externalisation of that norm means I am less likely to hunt with you again. But I am also less likely to coordinate with respect to collective defence or co-parenting, or any number of other fitness-critical activities with a high coordination dividend. Shunning has costly knock-on consequences. So, again, externalisation only has minimal opportunity costs if the agent and their conspecifics are already highly normatively synchronised.
Moreover, normative synchronisation limits costs only if the environment is normatively transparent. Any inaccuracy in identifying normative commitments (such as failure to appreciate mitigating reasons for the violation) risks conflict costs. Perhaps gossip can limit these costs, but it would need to be both honest and accurate.
In short, when we focus on early, ancestral, band-level societies prior to vertical segmentation, we think Stanford has under-counted the costs of externalisation. We also think he has over-counted the relative benefits, because he under-rates the utility of correlated subjective preferences and how they can give rise to quasi-contractual norms. In most circumstances, your preference for ice-cream flavours is of no moment to me because eating ice-cream is an individual activity. But subjective preferences matter a lot if and as they are relevant to collective action. For example, before moving into a joint house, it is important to identify the subjective preferences of the inhabitants with respect to whether they wash up after each meal, once a day, or once a week; if meals are joint, and if so, at what time and who cooks. These are all subjective preferences, but because they need to be coordinated in joint activity, they give rise to quasi-contractual norms; norms neither universal nor externalised, but providing a route through which novel cooperation opportunities can be profitably managed. Many stable and important social forms are based on the alignment of subjective preferences: in complex social worlds, most voluntary clubs and associations. Of course, in the house-share case and many other cases involving these quasi-contractual norms, the agents need a meta-norm or meta-policy: What to do if other agents violate the quasi-contractual norms? In many cases, that policy must be signalled. But the point here is that you do not need anything new as the meta-policy. You just need a standing policy of when and how you cut your losses, and on retaliation against those who fail in their quasi-contractual obligations. Frank's (Reference Frank1988) work on commitment emotions suggests both how the policy works and how it is signalled, without, as Stanford himself notes, requiring externalised norms. On balance, it remains unclear why norms need to be externalised to serve as profitable coordination or signalling devices.
These considerations suggest that normative externalisation is a late-breaking cultural innovation whose adaptive social role (if indeed it is adaptive) post-dates the foundational steps in the evolution of distinctively human cooperation. Rather than being tied to the evolution of gossip and indirect reciprocity in the early or mid-Pleistocene, we see it as more likely linked to the later African and out-of-Africa radiations in the late Pleistocene and early Holocene.
You have
Access
Stanford's theory of moral externalisation supposes that agents (1) have self-interest-independent motivations to adhere to enculturated norms, and (2) that agents are motivated to shun those who fail to adhere to these norms, and especially if they take advantage of others’ conformity to free-ride. These motivations operate on (culturally?) selected normative content, and in the context of coordination and cooperation problems this motivational combination supposedly opens up a new space of flexibility and adaptive plasticity, through selection of normative content that automatically (via the secondary motivation) sequesters the agent from conspecifics who would otherwise exploit them.
We suggest that the plausibility of Stanford's picture depends on its idealisation. Stanford largely makes the case for moral externalisation through an idealised model of dyadic interaction, with individuals assessing one another as social partners according to compatibility with respect to some specific norm. We think that this idealisation does not scale to realistic social ecologies, to realistic levels of normative complexity, and to realistic degrees of error and error-correction. Moreover, we think Stanford understates the adaptive plasticity made possible by more limited forms of normative cognition.
Our first scaling problem is heterogeneity: What of normative variation? Our lineage has been obligately social for much longer than any plausible inception of norm-following (certainly complex norm-following and reputation-tracking, mediated by language). And in a small group environment, moral externalisation is potentially dangerous: Start shunning without pre-existing positive assortment, and you will shun yourself out of your group. Likewise, novel coordination problems invite normative disagreements, as agents differ on the appropriate response. In Stanford's analysis, the stakes of disagreement rise: They are not just failures to negotiate a contract, a lost opportunity after which everyone then returns to the status quo. They threaten fracture. Unless disagreement was rare or transient, the cost of disagreement would select against norm externalisation.
This suggests that the normative content to be externalised must diffuse rapidly and be synchronised with high fidelity; conflict costs will be high if a new consensus is established slowly, piecemeal. Perhaps a not-implausibly strong conformity bias might deliver this, though it would have to be fortuitously present. Moreover, any mechanism that generated normative uniformity is in tension with the promised benefits of the model with respect to adaptive plasticity. A moral mutant should be expected to be shunned or out-grouped fairly swiftly, and for a normative innovation to spread within a group (without either fracturing it or being snuffed out), it seems it must do so before anyone really notices; to notice is to object. So, any shunning mechanism efficient enough to weed out free-riders (and mitigate disagreement risk) is also liable to weed out valuable innovations.
The second scaling problem is normative complexity. Externalisation and shunning is a blunt instrument; it motivates agents to want to have nothing to do with norm violators, and this imposes significant opportunity costs in any plausible human behavioural ecology. You don't share the kill with me in the way that I think is right, so externalisation of that norm means I am less likely to hunt with you again. But I am also less likely to coordinate with respect to collective defence or co-parenting, or any number of other fitness-critical activities with a high coordination dividend. Shunning has costly knock-on consequences. So, again, externalisation only has minimal opportunity costs if the agent and their conspecifics are already highly normatively synchronised.
Moreover, normative synchronisation limits costs only if the environment is normatively transparent. Any inaccuracy in identifying normative commitments (such as failure to appreciate mitigating reasons for the violation) risks conflict costs. Perhaps gossip can limit these costs, but it would need to be both honest and accurate.
In short, when we focus on early, ancestral, band-level societies prior to vertical segmentation, we think Stanford has under-counted the costs of externalisation. We also think he has over-counted the relative benefits, because he under-rates the utility of correlated subjective preferences and how they can give rise to quasi-contractual norms. In most circumstances, your preference for ice-cream flavours is of no moment to me because eating ice-cream is an individual activity. But subjective preferences matter a lot if and as they are relevant to collective action. For example, before moving into a joint house, it is important to identify the subjective preferences of the inhabitants with respect to whether they wash up after each meal, once a day, or once a week; if meals are joint, and if so, at what time and who cooks. These are all subjective preferences, but because they need to be coordinated in joint activity, they give rise to quasi-contractual norms; norms neither universal nor externalised, but providing a route through which novel cooperation opportunities can be profitably managed. Many stable and important social forms are based on the alignment of subjective preferences: in complex social worlds, most voluntary clubs and associations. Of course, in the house-share case and many other cases involving these quasi-contractual norms, the agents need a meta-norm or meta-policy: What to do if other agents violate the quasi-contractual norms? In many cases, that policy must be signalled. But the point here is that you do not need anything new as the meta-policy. You just need a standing policy of when and how you cut your losses, and on retaliation against those who fail in their quasi-contractual obligations. Frank's (Reference Frank1988) work on commitment emotions suggests both how the policy works and how it is signalled, without, as Stanford himself notes, requiring externalised norms. On balance, it remains unclear why norms need to be externalised to serve as profitable coordination or signalling devices.
These considerations suggest that normative externalisation is a late-breaking cultural innovation whose adaptive social role (if indeed it is adaptive) post-dates the foundational steps in the evolution of distinctively human cooperation. Rather than being tied to the evolution of gossip and indirect reciprocity in the early or mid-Pleistocene, we see it as more likely linked to the later African and out-of-Africa radiations in the late Pleistocene and early Holocene.