We largely support the main argument of Smaldino's article and think his ideas about GLTs will be useful for understanding the evolution and origin of complex organizations and societies. We especially welcome the insight that many important GLTs are emergent properties that may be difficult, or even impossible, to partition into units of individual selection. However, we also think Smaldino is too quick to dismiss existing institutional frameworks as useful ways to think about GLTs.
In multilevel selection (MLS) models of cultural evolution, the fitness consequences of traits are partitioned between at least two components, typically group-level and individual-level components (Henrich Reference Henrich2004a). For example, if groups of agents play public goods games, the group benefits of the public good may be accounted for in between-group fitness, and the individual costs may be accounted for in within-group fitness. The typical argument against MLS accounting is that partitioning at the group level can be unnecessary because selection can often be more parsimoniously accounted for at the individual level. However, Smaldino takes the opposite track, arguing that group traits are emergent and, by definition, not easily reduced to aggregates of individual-level traits.
Although we agree with the broad strokes of Smaldino's thesis, we think he errs in denying that many of these GLTs can be usefully classified as “institutions” and studied using existing tools of institutional analysis (e.g., North Reference North1990). Institutions are often defined as the “rules” that are the properties of groups called “organizations.” Institutions range from informal norm-based rules to formally codified laws of complex societies, and a key feature of institutions is that they are often maintained by systems of rewards and punishments that suppress individual-level deviation and stabilize behavior at a game theoretic equilibrium. Because the space of stable institutions is vast in both theory and in practice, institutional change can often be appropriately modeled as between-group “equilibrium selection” (Boyd & Richerson Reference Boyd and Richerson1990). In fact, it is relatively common for institutional economists and political scientists to treat formal institutions, such as the rules of political and economic organizations, as group-level traits in direct evolutionary competition. This is the sort of simplification to group-level traits for which Smaldino seems to advocate.
“Constitutional democracy” is a good example of an institutional GLT that can spread between countries as a result of selection-like processes (Cederman & Gleditsch Reference Cederman and Gleditsch2004). However, breaking something as complicated as the rules governing a constitutional democracy down to units of individual-level selection seems prohibitively difficult – a constitutional democracy is more than just the aggregate sum of individual-level costs and benefits. In short, it is emergent.
Smaldino gives three examples of where he does not consider institutions to qualify as group level traits:
Group-level traits are related, but not equivalent to institutions… a group-level trait is the phenotypic effect of social organization. Thus, examples of group-level traits are the music rather than the rock band, the election of a leader who reflects the public interest rather than the democratic voting system, the sailing ship's voyage rather than the crew positions, the economic surplus rather than the market economy. (sect. 2, para. 2)
We disagree with Smaldino and think that an institutional framework is more useful for understanding the spread of GLTs than a focus on phenotypic effect.
Taking Smaldino's suggestion that a GLT should be considered the “economic surplus rather than the market economy” (sect. 2, para. 2), one might ask how does economic surplus spread? Does a poorer country look to a richer country and say “we should copy their economic surplus?” That probably would not get them very far. Instead, a country might more successfully emulate the economic or political institutions that may underlie another country's economic surplus. We might therefore think of “economic surplus” as an indication of the success of the GLT, not the trait itself.
Taking Smaldino's other examples; a country is more likely to emulate a successful country's political institutions than it is to elect their political leaders. Similarly, during the Meiji restoration, the Japanese slavishly emulated British naval institutions, including standard crew positions (Goldman Reference Goldman, Farrell and Terriff2002). However, they did not attempt to recreate particularly successful British voyages.
Although the institutional framework is more useful than Smaldino supposes, not all GLTs need be institutions. A lot of bands emulated the music of the Beatles, for example. However, although the sound of a Beatles' song can be emulated directly, it is most efficient to do it with the same numbers and types of musicians. Some styles of music have an institution-like structure enforced by fans. Fans can usually tell you their favorite style of music – classical, country and western, folk, rhythm and blues, rock, etc. Musicians who change styles may lose all or part of their audience when they do so.
In summary, we think Smaldino's article highlights a future direction for modeling the evolution of social complexity and institutional selection. He convincing argues that GLTs are useful simplifying concepts. We hope his insights will be integrated into the study of both simple and complex emergent institutions.
We largely support the main argument of Smaldino's article and think his ideas about GLTs will be useful for understanding the evolution and origin of complex organizations and societies. We especially welcome the insight that many important GLTs are emergent properties that may be difficult, or even impossible, to partition into units of individual selection. However, we also think Smaldino is too quick to dismiss existing institutional frameworks as useful ways to think about GLTs.
In multilevel selection (MLS) models of cultural evolution, the fitness consequences of traits are partitioned between at least two components, typically group-level and individual-level components (Henrich Reference Henrich2004a). For example, if groups of agents play public goods games, the group benefits of the public good may be accounted for in between-group fitness, and the individual costs may be accounted for in within-group fitness. The typical argument against MLS accounting is that partitioning at the group level can be unnecessary because selection can often be more parsimoniously accounted for at the individual level. However, Smaldino takes the opposite track, arguing that group traits are emergent and, by definition, not easily reduced to aggregates of individual-level traits.
Although we agree with the broad strokes of Smaldino's thesis, we think he errs in denying that many of these GLTs can be usefully classified as “institutions” and studied using existing tools of institutional analysis (e.g., North Reference North1990). Institutions are often defined as the “rules” that are the properties of groups called “organizations.” Institutions range from informal norm-based rules to formally codified laws of complex societies, and a key feature of institutions is that they are often maintained by systems of rewards and punishments that suppress individual-level deviation and stabilize behavior at a game theoretic equilibrium. Because the space of stable institutions is vast in both theory and in practice, institutional change can often be appropriately modeled as between-group “equilibrium selection” (Boyd & Richerson Reference Boyd and Richerson1990). In fact, it is relatively common for institutional economists and political scientists to treat formal institutions, such as the rules of political and economic organizations, as group-level traits in direct evolutionary competition. This is the sort of simplification to group-level traits for which Smaldino seems to advocate.
“Constitutional democracy” is a good example of an institutional GLT that can spread between countries as a result of selection-like processes (Cederman & Gleditsch Reference Cederman and Gleditsch2004). However, breaking something as complicated as the rules governing a constitutional democracy down to units of individual-level selection seems prohibitively difficult – a constitutional democracy is more than just the aggregate sum of individual-level costs and benefits. In short, it is emergent.
Smaldino gives three examples of where he does not consider institutions to qualify as group level traits:
Group-level traits are related, but not equivalent to institutions… a group-level trait is the phenotypic effect of social organization. Thus, examples of group-level traits are the music rather than the rock band, the election of a leader who reflects the public interest rather than the democratic voting system, the sailing ship's voyage rather than the crew positions, the economic surplus rather than the market economy. (sect. 2, para. 2)
We disagree with Smaldino and think that an institutional framework is more useful for understanding the spread of GLTs than a focus on phenotypic effect.
Taking Smaldino's suggestion that a GLT should be considered the “economic surplus rather than the market economy” (sect. 2, para. 2), one might ask how does economic surplus spread? Does a poorer country look to a richer country and say “we should copy their economic surplus?” That probably would not get them very far. Instead, a country might more successfully emulate the economic or political institutions that may underlie another country's economic surplus. We might therefore think of “economic surplus” as an indication of the success of the GLT, not the trait itself.
Taking Smaldino's other examples; a country is more likely to emulate a successful country's political institutions than it is to elect their political leaders. Similarly, during the Meiji restoration, the Japanese slavishly emulated British naval institutions, including standard crew positions (Goldman Reference Goldman, Farrell and Terriff2002). However, they did not attempt to recreate particularly successful British voyages.
Although the institutional framework is more useful than Smaldino supposes, not all GLTs need be institutions. A lot of bands emulated the music of the Beatles, for example. However, although the sound of a Beatles' song can be emulated directly, it is most efficient to do it with the same numbers and types of musicians. Some styles of music have an institution-like structure enforced by fans. Fans can usually tell you their favorite style of music – classical, country and western, folk, rhythm and blues, rock, etc. Musicians who change styles may lose all or part of their audience when they do so.
In summary, we think Smaldino's article highlights a future direction for modeling the evolution of social complexity and institutional selection. He convincing argues that GLTs are useful simplifying concepts. We hope his insights will be integrated into the study of both simple and complex emergent institutions.