As the naturalization of capitalism places selfishness as the default, and cooperation as the anomaly to be explained, so Stanford places subjectivized phenomenology as the default and seeks to explain what evolutionary pressures lead to moral judgments feeling external and objective. We agree with Stanford that the difference between externalized and subjectivized phenomenology of motivations needs an explanation and that current evolutionary theories have not resolved this question. However, we disagree with him on etiology–that is, which side of this distinction is to be taken as default and which emerges later and needs further explanation. We side with Joyce's alternate proposal (mentioned by Stanford in sect. 3, para. 8; see sect. 4.4 of Joyce Reference Joyce2006) that externalization is the norm. Thus, subjectivization (e.g., our feelings towards ice cream) is the exceptional ability in need of an adaptationist account. For us, the general question becomes: How did humans evolve a phenomenology of the subjective world as separate from the default phenomenology of an external world?
In the second half of section 3, Stanford dismisses the idea that an externalized phenomenology “might well be an ancestral condition from which no shift to an externalized or objectified moral psychology was ever required” (sect. 3, para. 8) because he believes that cooperation requires correlated interactions and correlated interactions require active maintenance. We think that his dismissal is unjustified on at least four points.
First, if we are looking for ancestral states from which our current moral psychology developed, then, even if moral capacities are required for cooperation, that doesn't necessarily mean that cooperation is required for their ancestral states. For example, we don't appeal to considerations of cooperation if we want to understand why the phenomenology of colors is external.
Second, although correlated interactions are needed for the maintenance (or emergence) of cooperation in games like Prisoner's Dilemma and Stag Hunt, those are not the only games relevant to our distant ancestors. In games like Harmony, for example, it is rational to cooperate, but active mechanisms like ethnocentrism can actually undermine this cooperation and replace it with defection (Kaznatcheev Reference Kaznatcheev2010). Note that the example of Harmony and ethnocentrism also casts doubt on Stanford's claim that active mechanisms and morality primarily promote hypercooperation. This makes empirically determining the games played by our distant ancestors an important unresolved question. Many researchers mistakenly conclude that, because games like Prisoner's Dilemma and Stag Hunt are the most studied games (due to their undoubted theoretical interest), this implies that these are the games that were important in early human evolution. There is no solid justification to believe this, as it is a case of the (interestingness) tail wagging the (pervasiveness) dog. Instead, game types should be diagnosed from more direct evidence of a game's ability to explain important phenomena, regardless of whether the game seems theoretically interesting.
Third, for evolutionary games, we believe that activity is required to break correlation rather than to create or maintain it. The classic example would be spatial structure, which is known to facilitate cooperation. Although uncorrelated inviscid models are easier to analyze (and were thus analyzed first), they are not more natural because humans (like all other organisms) are embedded in space. To undo this inherent correlation, we need the quintessential active process of locomotion for de-correlation. The other big example requiring active mechanisms to de-correlate would be the genetic proximity in inclusive fitness, and in mammals such as ourselves, parental care.
Finally, in his argument for active mechanisms for building correlations, Stanford speaks of payoffs of evolutionary games as if they are known by the agents making decisions to cooperate or not. But there is no reason to believe that a person's perception of evolutionary payoffs should align with those payoffs’ actual effect on individual fitness. In particular, Kaznatcheev et al. (Reference Kaznatcheev, Montrey and Shultz2014) show examples in which individuals act rationally on their perceptions of payoffs, but due to evolution those perceptions track inclusive fitness instead of the game's effect on individual fitness. In such cases, agents would not know that defection increases their individual payoff. These useful delusions create a social interface, making it easier to cooperate. And just like the individual interfaces (Hoffman Reference Hoffman, Dickinson, Tarr, Leonardis and Schiele2009) (e.g., color making it easier to choose ripe fruit) which they extend, such social interfaces can feel external and objective.
Having made a negative case against Stanford's dismissal of Joyce, we also propose a positive argument for externalized phenomenology being the default. Just as biologists turn to embryology for inspiration on the etiology of evolution, we turn to developmental psychology for an initial sketch of the order of evolution of morality. Given the limits on paleontological data for morality, we believe that this is the best available option. And current understanding of child development largely supports objectivity as the default. Emotional arousal and emotional contagion are observed even in newborns (Dondi et al. Reference Dondi, Simion and Caltran1999), but toddlers don't develop a sense of self and the ability to differentiate between their own and others’ internal states until around 2 years of age (Decety Reference Decety2010; Roth-Hanania et al. Reference Roth-Hanania, Davidov and Zahn-Waxler2011). Children acquire basic theory of mind capacities for inferring subjective states like desires, intentions, and beliefs between 3 and 5 years (Wellman et al. Reference Wellman, Cross and Watson2001). This suggests that an understanding of experience as subjective both in oneself and others develops from an objectivized phenomenological precursor.
This pattern also occurs with more explicit tests of moral objectivism. Preschool children exhibit higher levels of moral objectivism than 9-year-olds and adults (Schmidt et al. Reference Schmidt, Gonzalez-Cabrera and Tomasello2017). It has also been observed that the subjectivization of experience occurs at different times for different classes of experience, with 4- to 6-year-olds treating properties like fun and icky as more response-dependent than moral properties like good and bad (Nichols & Folds-Bennett Reference Nichols and Folds-Bennett2003). Overall, this suggests that it is reasonable to view externalized phenomenology as an ancestral condition. Consequently, we should also take this direction as the default for evolutionary etiology unless future empirical evidence on the evolution of morality is found to suggest otherwise.
You have
Access
As the naturalization of capitalism places selfishness as the default, and cooperation as the anomaly to be explained, so Stanford places subjectivized phenomenology as the default and seeks to explain what evolutionary pressures lead to moral judgments feeling external and objective. We agree with Stanford that the difference between externalized and subjectivized phenomenology of motivations needs an explanation and that current evolutionary theories have not resolved this question. However, we disagree with him on etiology–that is, which side of this distinction is to be taken as default and which emerges later and needs further explanation. We side with Joyce's alternate proposal (mentioned by Stanford in sect. 3, para. 8; see sect. 4.4 of Joyce Reference Joyce2006) that externalization is the norm. Thus, subjectivization (e.g., our feelings towards ice cream) is the exceptional ability in need of an adaptationist account. For us, the general question becomes: How did humans evolve a phenomenology of the subjective world as separate from the default phenomenology of an external world?
In the second half of section 3, Stanford dismisses the idea that an externalized phenomenology “might well be an ancestral condition from which no shift to an externalized or objectified moral psychology was ever required” (sect. 3, para. 8) because he believes that cooperation requires correlated interactions and correlated interactions require active maintenance. We think that his dismissal is unjustified on at least four points.
First, if we are looking for ancestral states from which our current moral psychology developed, then, even if moral capacities are required for cooperation, that doesn't necessarily mean that cooperation is required for their ancestral states. For example, we don't appeal to considerations of cooperation if we want to understand why the phenomenology of colors is external.
Second, although correlated interactions are needed for the maintenance (or emergence) of cooperation in games like Prisoner's Dilemma and Stag Hunt, those are not the only games relevant to our distant ancestors. In games like Harmony, for example, it is rational to cooperate, but active mechanisms like ethnocentrism can actually undermine this cooperation and replace it with defection (Kaznatcheev Reference Kaznatcheev2010). Note that the example of Harmony and ethnocentrism also casts doubt on Stanford's claim that active mechanisms and morality primarily promote hypercooperation. This makes empirically determining the games played by our distant ancestors an important unresolved question. Many researchers mistakenly conclude that, because games like Prisoner's Dilemma and Stag Hunt are the most studied games (due to their undoubted theoretical interest), this implies that these are the games that were important in early human evolution. There is no solid justification to believe this, as it is a case of the (interestingness) tail wagging the (pervasiveness) dog. Instead, game types should be diagnosed from more direct evidence of a game's ability to explain important phenomena, regardless of whether the game seems theoretically interesting.
Third, for evolutionary games, we believe that activity is required to break correlation rather than to create or maintain it. The classic example would be spatial structure, which is known to facilitate cooperation. Although uncorrelated inviscid models are easier to analyze (and were thus analyzed first), they are not more natural because humans (like all other organisms) are embedded in space. To undo this inherent correlation, we need the quintessential active process of locomotion for de-correlation. The other big example requiring active mechanisms to de-correlate would be the genetic proximity in inclusive fitness, and in mammals such as ourselves, parental care.
Finally, in his argument for active mechanisms for building correlations, Stanford speaks of payoffs of evolutionary games as if they are known by the agents making decisions to cooperate or not. But there is no reason to believe that a person's perception of evolutionary payoffs should align with those payoffs’ actual effect on individual fitness. In particular, Kaznatcheev et al. (Reference Kaznatcheev, Montrey and Shultz2014) show examples in which individuals act rationally on their perceptions of payoffs, but due to evolution those perceptions track inclusive fitness instead of the game's effect on individual fitness. In such cases, agents would not know that defection increases their individual payoff. These useful delusions create a social interface, making it easier to cooperate. And just like the individual interfaces (Hoffman Reference Hoffman, Dickinson, Tarr, Leonardis and Schiele2009) (e.g., color making it easier to choose ripe fruit) which they extend, such social interfaces can feel external and objective.
Having made a negative case against Stanford's dismissal of Joyce, we also propose a positive argument for externalized phenomenology being the default. Just as biologists turn to embryology for inspiration on the etiology of evolution, we turn to developmental psychology for an initial sketch of the order of evolution of morality. Given the limits on paleontological data for morality, we believe that this is the best available option. And current understanding of child development largely supports objectivity as the default. Emotional arousal and emotional contagion are observed even in newborns (Dondi et al. Reference Dondi, Simion and Caltran1999), but toddlers don't develop a sense of self and the ability to differentiate between their own and others’ internal states until around 2 years of age (Decety Reference Decety2010; Roth-Hanania et al. Reference Roth-Hanania, Davidov and Zahn-Waxler2011). Children acquire basic theory of mind capacities for inferring subjective states like desires, intentions, and beliefs between 3 and 5 years (Wellman et al. Reference Wellman, Cross and Watson2001). This suggests that an understanding of experience as subjective both in oneself and others develops from an objectivized phenomenological precursor.
This pattern also occurs with more explicit tests of moral objectivism. Preschool children exhibit higher levels of moral objectivism than 9-year-olds and adults (Schmidt et al. Reference Schmidt, Gonzalez-Cabrera and Tomasello2017). It has also been observed that the subjectivization of experience occurs at different times for different classes of experience, with 4- to 6-year-olds treating properties like fun and icky as more response-dependent than moral properties like good and bad (Nichols & Folds-Bennett Reference Nichols and Folds-Bennett2003). Overall, this suggests that it is reasonable to view externalized phenomenology as an ancestral condition. Consequently, we should also take this direction as the default for evolutionary etiology unless future empirical evidence on the evolution of morality is found to suggest otherwise.