The target article argues that the distinctive psychology that regards moral properties as objective, external features of the world is adaptive because it allows us to engage in beneficial cooperation through correlated interactions. Implicit in the account are two theses: (1) the correlation thesis that moral commitment serves as a correlating device, allowing fellow norm-followers to associate with each other and to ostracize defectors; and (2) the indispensability thesis that externalized moral norms of this sort are necessary to achieve pro-social cooperation, at least at the high rate seen in humans.
We argue that the first thesis is plausible, but that it undermines the second thesis. If the correlation thesis is true, there is good reason to think that externalized moral attitudes are not indispensable for achieving cooperation, but are merely one possible solution among many.
There is already a well-understood set of mechanisms by which agents may ensure correlated interaction with fellow biological altruists. It is not clear how Stanford's account is supposed to relate to this menu of options. Is it a type of reciprocity, costly signaling, group selection, or something else? We suggest there are three possibilities (not mutually exclusive) that are especially promising.
Costly signaling models require a diversity of types in the population (e.g., cooperators and defectors) who – at least in simple cases – face differential signaling costs, or who stand to make differential gains from being believed. Some types can afford to send signals that are uneconomic for other types, and, hence, any sufficiently costly signal is credible. Costly signaling mechanisms have been proposed to explain phenomena like unconditional sharing, costly punishment, apology, and guilt (Gintis et al. Reference Gintis, Smith and Bowles2001; Jordan et al. Reference Jordan, Hoffman, Bloom and Rand2016; Ohtsubo & Watanabe Reference Ohtsubo and Watanabe2009). Could externalizing moral psychology be a form of costly signal also?
It is conceivable. By rigid commitment to moral attitudes, an agent might incur a cost that a defector would not be willing to pay. But for this hypothesis to be plausible, we need an explanation for why externalized moral attitudes are especially burdensome or costly. Prima facie, it is hard to see that they are. Compared to better established examples of costly signals, including rituals such as fasting, bodily mutilation, and animal sacrifice, it does not seem as if having the belief that moral requirements are objective features of the world is especially costly at all.
The green beard hypothesis is that altruistic traits are genetically linked to a distinctive phenotypic trait. If only altruists can develop green beards, then a strategy of cooperating with fellow green beards is a plausible evolutionary outcome. A genetic barrier to green-bearded defectors keeps the world safe for cooperation. Recent work (cf. Gardner & West Reference Gardner and West2010) has shown that green beard mechanisms can sustain cooperation even if the link between beard and altruism is not strict but permits some plasticity. The result is an unstable dynamic, in which the population cycles through different beard colors, with bursts of cooperation in the beginning of each beard cycle, followed by invasion by defectors, followed by development of a new beard color (Jansen & van Baalen Reference Jansen and van Baalen2006; Traulsen & Nowak Reference Traulsen and Nowak2007). Perhaps the best elaborated account of how these sorts of dynamics might explain actual human cooperation is the case of accents, which are certainly hard to fake and are also not tightly linked to any particular genes (Cohen Reference Cohen2012).
Both costly signaling and green beard accounts undermine the indispensability thesis, however, because both imply that the cooperative correlating mechanism is arbitrary. Stanford emphasizes that the moral psychology, which he seeks to explain, appears to be cross-culturally robust. This makes it quite unlike rituals or accents, which vary dramatically across time and space.
Finally, the most promising mechanism to underlie Stanford's account is social selection. Suppose that a competitive mating environment exists in which fitness is enhanced by finding reliable long-term cooperative partners. In such a market, it is adaptive to have a reputation for being reliable. So, we predict adaptations that make one sensitive to reputation (cf. Haley & Fessler Reference Haley and Fessler2005). Cooperative, altruistic behaviors may then be adaptive because they enhance one's reputation – even if those behaviors are done cynically, for reputation-enhancing reasons. The competitiveness of the mating market may then drive this process so that ever better demonstrations of reliability are required in order to obtain a mate. In this setting, it may be more cost-effective to be reliable than to merely appear reliable (Sperber & Baumard Reference Sperber and Baumard2012). One particular way to be reliable is to take moral facts as external, objective demands. Notably, this process explains the development of “high-quality” types as emerging from a competition among lower-quality types who are more self-serving in their pursuit of reputation.
This better explains the adaptive function of externalizing psychology in particular, but without more explicit modeling, it remains open that a population in equilibrium may have only a small minority who display this phenotype. The main mechanism – reputation-sensitive coordination with the prevailing norms – may explain most observed cooperation. Indeed, this is plausible. Consider the evidence of social norms prevailing over moral commitments, revealed both in history and in experimental settings (e.g., Haney et al. Reference Haney, Banks and Zimbardo1972). Externalizing psychology is apt to fascinate philosophers, who are in the grip of related meta-ethical puzzles dating back to Plato's Euthyphro, but this is not yet evidence that it plays a significant role in achieving actual cooperation.
Debates about the genesis of human cooperation are unlikely to make significant advances without comparison of models “in the field” – using disciplines such as archaeology, ethnography, genetics, and experimental economics. This work, however, requires models that can deliver testable predictions. We wait with interest to see whether Stanford's proposal constitutes a novel mechanism, with novel predictions, or if it can be assimilated to existing mechanisms. We have suggested here that if it is assimilated to existing mechanisms, there is little hope for the indispensability thesis. The “categorical imperative” nature of moral commitments may be a contingent artefact of our idiosyncratic evolutionary history.
The target article argues that the distinctive psychology that regards moral properties as objective, external features of the world is adaptive because it allows us to engage in beneficial cooperation through correlated interactions. Implicit in the account are two theses: (1) the correlation thesis that moral commitment serves as a correlating device, allowing fellow norm-followers to associate with each other and to ostracize defectors; and (2) the indispensability thesis that externalized moral norms of this sort are necessary to achieve pro-social cooperation, at least at the high rate seen in humans.
We argue that the first thesis is plausible, but that it undermines the second thesis. If the correlation thesis is true, there is good reason to think that externalized moral attitudes are not indispensable for achieving cooperation, but are merely one possible solution among many.
There is already a well-understood set of mechanisms by which agents may ensure correlated interaction with fellow biological altruists. It is not clear how Stanford's account is supposed to relate to this menu of options. Is it a type of reciprocity, costly signaling, group selection, or something else? We suggest there are three possibilities (not mutually exclusive) that are especially promising.
Costly signaling models require a diversity of types in the population (e.g., cooperators and defectors) who – at least in simple cases – face differential signaling costs, or who stand to make differential gains from being believed. Some types can afford to send signals that are uneconomic for other types, and, hence, any sufficiently costly signal is credible. Costly signaling mechanisms have been proposed to explain phenomena like unconditional sharing, costly punishment, apology, and guilt (Gintis et al. Reference Gintis, Smith and Bowles2001; Jordan et al. Reference Jordan, Hoffman, Bloom and Rand2016; Ohtsubo & Watanabe Reference Ohtsubo and Watanabe2009). Could externalizing moral psychology be a form of costly signal also?
It is conceivable. By rigid commitment to moral attitudes, an agent might incur a cost that a defector would not be willing to pay. But for this hypothesis to be plausible, we need an explanation for why externalized moral attitudes are especially burdensome or costly. Prima facie, it is hard to see that they are. Compared to better established examples of costly signals, including rituals such as fasting, bodily mutilation, and animal sacrifice, it does not seem as if having the belief that moral requirements are objective features of the world is especially costly at all.
The green beard hypothesis is that altruistic traits are genetically linked to a distinctive phenotypic trait. If only altruists can develop green beards, then a strategy of cooperating with fellow green beards is a plausible evolutionary outcome. A genetic barrier to green-bearded defectors keeps the world safe for cooperation. Recent work (cf. Gardner & West Reference Gardner and West2010) has shown that green beard mechanisms can sustain cooperation even if the link between beard and altruism is not strict but permits some plasticity. The result is an unstable dynamic, in which the population cycles through different beard colors, with bursts of cooperation in the beginning of each beard cycle, followed by invasion by defectors, followed by development of a new beard color (Jansen & van Baalen Reference Jansen and van Baalen2006; Traulsen & Nowak Reference Traulsen and Nowak2007). Perhaps the best elaborated account of how these sorts of dynamics might explain actual human cooperation is the case of accents, which are certainly hard to fake and are also not tightly linked to any particular genes (Cohen Reference Cohen2012).
Both costly signaling and green beard accounts undermine the indispensability thesis, however, because both imply that the cooperative correlating mechanism is arbitrary. Stanford emphasizes that the moral psychology, which he seeks to explain, appears to be cross-culturally robust. This makes it quite unlike rituals or accents, which vary dramatically across time and space.
Finally, the most promising mechanism to underlie Stanford's account is social selection. Suppose that a competitive mating environment exists in which fitness is enhanced by finding reliable long-term cooperative partners. In such a market, it is adaptive to have a reputation for being reliable. So, we predict adaptations that make one sensitive to reputation (cf. Haley & Fessler Reference Haley and Fessler2005). Cooperative, altruistic behaviors may then be adaptive because they enhance one's reputation – even if those behaviors are done cynically, for reputation-enhancing reasons. The competitiveness of the mating market may then drive this process so that ever better demonstrations of reliability are required in order to obtain a mate. In this setting, it may be more cost-effective to be reliable than to merely appear reliable (Sperber & Baumard Reference Sperber and Baumard2012). One particular way to be reliable is to take moral facts as external, objective demands. Notably, this process explains the development of “high-quality” types as emerging from a competition among lower-quality types who are more self-serving in their pursuit of reputation.
This better explains the adaptive function of externalizing psychology in particular, but without more explicit modeling, it remains open that a population in equilibrium may have only a small minority who display this phenotype. The main mechanism – reputation-sensitive coordination with the prevailing norms – may explain most observed cooperation. Indeed, this is plausible. Consider the evidence of social norms prevailing over moral commitments, revealed both in history and in experimental settings (e.g., Haney et al. Reference Haney, Banks and Zimbardo1972). Externalizing psychology is apt to fascinate philosophers, who are in the grip of related meta-ethical puzzles dating back to Plato's Euthyphro, but this is not yet evidence that it plays a significant role in achieving actual cooperation.
Debates about the genesis of human cooperation are unlikely to make significant advances without comparison of models “in the field” – using disciplines such as archaeology, ethnography, genetics, and experimental economics. This work, however, requires models that can deliver testable predictions. We wait with interest to see whether Stanford's proposal constitutes a novel mechanism, with novel predictions, or if it can be assimilated to existing mechanisms. We have suggested here that if it is assimilated to existing mechanisms, there is little hope for the indispensability thesis. The “categorical imperative” nature of moral commitments may be a contingent artefact of our idiosyncratic evolutionary history.