Group-level traits cannot be expressed by individuals (by definition groups are more than one individual); however, whether group-level traits can be encoded by individuals relates more directly to the main claim of the article: that we need a conceptual extension to the theory of cultural evolution to include emergent group-level traits as a unit of selection. The illustrative example of division of labor helps to examine this claim.
When discussing the division of labor in the context of group-level traits, Smaldino raises two questions that are supposed to be difficult to reconcile within standard cMLS framework: (1) “How are individuals selected for specific roles?” [within the group] and (2) “Where do the available roles come from in the first place?” (sect. 5.2). It is possible that both of these questions can be answered, in principle, at the level of the individual, for example (1) by showing that individuals can be selected for their role through a combination of a good-stay bad-shift strategy in combination with inherited variation in task-aptitude, and (2) by showing that naturally occurring problems of coordination provide affordances for different roles and need not be socially constructed as Smaldino appears to suggest. To expand on these points, say individuals behave in a certain way, which we can call a good-stay bad-shift phenotype. If individuals are good at a particular task (or at least better than their neighbor), they stay put. If they are bad (or worse than their neighbor), they shift what they are doing and try something else. Given some variation in the aptitudes for different tasks in a population, this strategy, defined at the level of individual, ensures the best person for the job by self-sorting.
One can see how this tendency could form the basis of a structured group. To use the example of hunting, one member of the group might be better at running, one member better at wrestling prey, one with better eyesight, one better at identifying different prey, and so on. The efficiency saving bought about by the division of labor (law of comparative advantage; Ricardo Reference Ricardo1817) means that those groups with self-sort division of labor will tend to do better than groups that do not behave in this way. The social structure is expressed as a group (one cannot benefit by trading skills with oneself) but encoded at the level of the individual (the good-stay bad-shift strategy). Once this structure is in place, one can imagine second-order and third-order hierarchal roles becoming available – for example, coordinating leadership roles – but still emergent from lower-level properties of the group. To answer Smaldino's second question, where available roles come from in the first place, again consider the hunt example above. Importantly, the roles themselves need not necessarily come from existing human societal structure, as is implied with the examples used by Smaldino (although they might do). Organization emerges in the above example from the affordances of a hunt event and the self-sort strategy.
I agree that it is important to recognize that an organized and structured group is qualitatively different from an aggregation of individuals and an important unit in the evolution of human culture. (However, the distinction between aggregate and emergent systems might not be as straightforward as this suggests, captured by the complexity science aphorism: more is different). All of this is not to claim the reducibility of all group-level traits to the individual. More that, because the standard model is so powerful one needs to be sure to rule out individual-level explanations first (Sumpter Reference Sumpter2006). In the division of labor examples above, the answers to both of Smaldino's questions can be sensibly grounded at the level of the individual. Moreover, because of Ricardo's law of comparative advantage, what is good for the group is also good for the individual. In this context, it might make sense to talk of the social structure that emerges from a good-stay bad-shift mutation as an extended phenotype (Dawkins Reference Dawkins1989). If this is right, then it is not so clear that we need a conceptual extension to cMLS, or even that a between-levels perspective dissolves the equivalency between MLS and inclusive fitness theory, as Smaldino proposes.
You have
Access
Group-level traits cannot be expressed by individuals (by definition groups are more than one individual); however, whether group-level traits can be encoded by individuals relates more directly to the main claim of the article: that we need a conceptual extension to the theory of cultural evolution to include emergent group-level traits as a unit of selection. The illustrative example of division of labor helps to examine this claim.
When discussing the division of labor in the context of group-level traits, Smaldino raises two questions that are supposed to be difficult to reconcile within standard cMLS framework: (1) “How are individuals selected for specific roles?” [within the group] and (2) “Where do the available roles come from in the first place?” (sect. 5.2). It is possible that both of these questions can be answered, in principle, at the level of the individual, for example (1) by showing that individuals can be selected for their role through a combination of a good-stay bad-shift strategy in combination with inherited variation in task-aptitude, and (2) by showing that naturally occurring problems of coordination provide affordances for different roles and need not be socially constructed as Smaldino appears to suggest. To expand on these points, say individuals behave in a certain way, which we can call a good-stay bad-shift phenotype. If individuals are good at a particular task (or at least better than their neighbor), they stay put. If they are bad (or worse than their neighbor), they shift what they are doing and try something else. Given some variation in the aptitudes for different tasks in a population, this strategy, defined at the level of individual, ensures the best person for the job by self-sorting.
One can see how this tendency could form the basis of a structured group. To use the example of hunting, one member of the group might be better at running, one member better at wrestling prey, one with better eyesight, one better at identifying different prey, and so on. The efficiency saving bought about by the division of labor (law of comparative advantage; Ricardo Reference Ricardo1817) means that those groups with self-sort division of labor will tend to do better than groups that do not behave in this way. The social structure is expressed as a group (one cannot benefit by trading skills with oneself) but encoded at the level of the individual (the good-stay bad-shift strategy). Once this structure is in place, one can imagine second-order and third-order hierarchal roles becoming available – for example, coordinating leadership roles – but still emergent from lower-level properties of the group. To answer Smaldino's second question, where available roles come from in the first place, again consider the hunt example above. Importantly, the roles themselves need not necessarily come from existing human societal structure, as is implied with the examples used by Smaldino (although they might do). Organization emerges in the above example from the affordances of a hunt event and the self-sort strategy.
I agree that it is important to recognize that an organized and structured group is qualitatively different from an aggregation of individuals and an important unit in the evolution of human culture. (However, the distinction between aggregate and emergent systems might not be as straightforward as this suggests, captured by the complexity science aphorism: more is different). All of this is not to claim the reducibility of all group-level traits to the individual. More that, because the standard model is so powerful one needs to be sure to rule out individual-level explanations first (Sumpter Reference Sumpter2006). In the division of labor examples above, the answers to both of Smaldino's questions can be sensibly grounded at the level of the individual. Moreover, because of Ricardo's law of comparative advantage, what is good for the group is also good for the individual. In this context, it might make sense to talk of the social structure that emerges from a good-stay bad-shift mutation as an extended phenotype (Dawkins Reference Dawkins1989). If this is right, then it is not so clear that we need a conceptual extension to cMLS, or even that a between-levels perspective dissolves the equivalency between MLS and inclusive fitness theory, as Smaldino proposes.