Pietraszewski's proposal for the cognitive underpinnings of the “social group” concept is a remarkable attempt to place social science concepts on a lucid foundation, in this case to account for people's own concepts of their social environment in a computationally tractable manner. This good deed should not go unpunished, so I will argue for a major extension of the program.
Pietraszewski's model requires that we abandon common, entrenched intuitions about collections of agents and their interaction. These intuitions are not just the product of past social theory, they also constitute the way most lay people, in the most diverse cultures, construe the social world – what could be called a folk-sociology (Boyer, Reference Boyer2018, pp. 216–237; Hirschfeld, Reference Hirschfeld2001). For instance, one common assumption is that social norms are somehow external to the individual minds that represent them, which, for example, makes it possible to think that, for example, “marriage is…” this or that, regardless of people's thoughts about it. Another important and culturally widespread assumption is that groups are agent-like entities, which is why we talk about villages or social classes or nations as entities that want this or remember that, make decisions, and so on. Finally, folk-sociology assumes that political power is a “force” and power relations are similar to force-dynamics. Powerful exert “pressure” on others, who may “resist” or be “pushed around,” and so on (Boyer, Reference Boyer2018).
These assumptions are based on loose and misleading conventional metaphors (Lakoff & Johnson, Reference Lakoff and Johnson1980). Therefore, why are they so widespread and entrenched? A first reason is that they describe complex processes of social interaction, with emergent properties, for which we do not have adequate cognitive resources. Even in the small-scale groups typical of most human evolution, describing interaction between more than three individuals would require not just representing other agents' intentions, but their representations of each others' representations, and so forth. Therefore, folk-sociological assumptions produce a rough and ready understanding of social processes that may be sufficient in many situations. A second reason that makes folk-sociological assumptions quite natural is that they often “hijack” the computational machinery of evolved domain-specific cognitive systems. Seeing groups as agents makes it possible to activate our intuitive psychology (Leslie, Friedman, & German, Reference Leslie, Friedman and German2004), which provides a rich descriptive and explanatory arsenal for describing group interactions. As regards power, intuitive understandings of force dynamics (Baillargeon, Kotovsky, & Needham, Reference Baillargeon, Kotovsky, Needham, Sperber, Premack and James-Premack1995; Talmy, Reference Talmy1988) supply templates for describing the dynamics of influence between leaders and followers. A third factor is that other agents too share our folk-sociological assumptions, so that these partly metaphorical understandings provide coordination points, a form of mutual knowledge about our social environment.
In a very unfortunate turn of events, a great deal of social science theorizing, instead of explaining folk-sociology, endorsed it! Therefore, that social scientists tried to argue that norms really were external to people's minds (Gilbert, Reference Gilbert1989; Searle, Reference Searle1995), that power really worked like a force – see discussion in Lukes (Reference Lukes1984), and that social groups really had intentional states (Sheehy, Reference Sheehy2012; Tollefsen, Reference Tollefsen2015).
Dave Pietraszewski's model may help explain the groups-as-agents assumption of folk-sociology. Groups are seen as intentional, having both goals and memories, not just (as suggested above) because our intuitive psychology is a salient mental system with a rich inferential potential, but also because the very notion of social group requires an intentional description of the “primitive roles” that constitute non-cooperative triadic interactions. In other words, the top-down salient metaphor of groups as agents may be salient because of the actual bottom-up process of considering (within the triadic interactions) agents as constitutive of groups. It is agency all the way down, so to speak.
But one may want to know how this parsimonious model accommodates some aspects of large-scale group dynamics. Consider this. At the level of actual triadic interactions, or perhaps an extension to a few individuals in each “group-defining role,” the participants' representations of the different roles and their consequences are mostly veridical. That is, people do represent the imposition of cost as such, and the reactions of the other partners as one of the roles defined in the event grammar. By contrast, the dynamics of interaction at a larger scale may be quite different from these roles and depend on factors not described by the event-grammar. Consider for instance the dynamics of ethnic signaling, as modeled by Kuran (Reference Kuran1998). People's decision to wear ethnic garb, for instance, depends on the perceived cost of doing so, which, in turn, depends on an agent's perception of the relative size and cohesion of one's and the other groups. The decision, in turn, influences other agents, as it modifies their estimates of the relative costs of signaling or not signaling. Naturally, this cascade model is not at all the way people represent the macro-features of the situation – which they see as people beginning to signal their ethnicity because they are more convinced than before of their group's value and of its goals.
There are two possible paths to account for such situations. Pietraszewski suggests that the event-grammar scales up, preserving its constitutive roles. In this case, people's own representations of large-scale group dynamics do capture people's actual behavior. The macro-effects outlined here are adequately described as the emergent effect of numerous agents entertaining the representations described by the event-grammar.
Alternatively, we may consider that large-scale group dynamics cannot be captured by the event-grammar. In that case, the macro-dynamics are really different from what people represent. But the fact that people all see those dynamics in terms of agent-like groups provides enough coordination that the intentional description often appears to be roughly correct so that most agents (and even social scientists!) see it as adequate.
It speaks to the great value and conceptual precision of Pietraszewski's proposal, that we can formulate these questions with a level of precision that is not common in the social sciences.
You have
Access
Pietraszewski's proposal for the cognitive underpinnings of the “social group” concept is a remarkable attempt to place social science concepts on a lucid foundation, in this case to account for people's own concepts of their social environment in a computationally tractable manner. This good deed should not go unpunished, so I will argue for a major extension of the program.
Pietraszewski's model requires that we abandon common, entrenched intuitions about collections of agents and their interaction. These intuitions are not just the product of past social theory, they also constitute the way most lay people, in the most diverse cultures, construe the social world – what could be called a folk-sociology (Boyer, Reference Boyer2018, pp. 216–237; Hirschfeld, Reference Hirschfeld2001). For instance, one common assumption is that social norms are somehow external to the individual minds that represent them, which, for example, makes it possible to think that, for example, “marriage is…” this or that, regardless of people's thoughts about it. Another important and culturally widespread assumption is that groups are agent-like entities, which is why we talk about villages or social classes or nations as entities that want this or remember that, make decisions, and so on. Finally, folk-sociology assumes that political power is a “force” and power relations are similar to force-dynamics. Powerful exert “pressure” on others, who may “resist” or be “pushed around,” and so on (Boyer, Reference Boyer2018).
These assumptions are based on loose and misleading conventional metaphors (Lakoff & Johnson, Reference Lakoff and Johnson1980). Therefore, why are they so widespread and entrenched? A first reason is that they describe complex processes of social interaction, with emergent properties, for which we do not have adequate cognitive resources. Even in the small-scale groups typical of most human evolution, describing interaction between more than three individuals would require not just representing other agents' intentions, but their representations of each others' representations, and so forth. Therefore, folk-sociological assumptions produce a rough and ready understanding of social processes that may be sufficient in many situations. A second reason that makes folk-sociological assumptions quite natural is that they often “hijack” the computational machinery of evolved domain-specific cognitive systems. Seeing groups as agents makes it possible to activate our intuitive psychology (Leslie, Friedman, & German, Reference Leslie, Friedman and German2004), which provides a rich descriptive and explanatory arsenal for describing group interactions. As regards power, intuitive understandings of force dynamics (Baillargeon, Kotovsky, & Needham, Reference Baillargeon, Kotovsky, Needham, Sperber, Premack and James-Premack1995; Talmy, Reference Talmy1988) supply templates for describing the dynamics of influence between leaders and followers. A third factor is that other agents too share our folk-sociological assumptions, so that these partly metaphorical understandings provide coordination points, a form of mutual knowledge about our social environment.
In a very unfortunate turn of events, a great deal of social science theorizing, instead of explaining folk-sociology, endorsed it! Therefore, that social scientists tried to argue that norms really were external to people's minds (Gilbert, Reference Gilbert1989; Searle, Reference Searle1995), that power really worked like a force – see discussion in Lukes (Reference Lukes1984), and that social groups really had intentional states (Sheehy, Reference Sheehy2012; Tollefsen, Reference Tollefsen2015).
Dave Pietraszewski's model may help explain the groups-as-agents assumption of folk-sociology. Groups are seen as intentional, having both goals and memories, not just (as suggested above) because our intuitive psychology is a salient mental system with a rich inferential potential, but also because the very notion of social group requires an intentional description of the “primitive roles” that constitute non-cooperative triadic interactions. In other words, the top-down salient metaphor of groups as agents may be salient because of the actual bottom-up process of considering (within the triadic interactions) agents as constitutive of groups. It is agency all the way down, so to speak.
But one may want to know how this parsimonious model accommodates some aspects of large-scale group dynamics. Consider this. At the level of actual triadic interactions, or perhaps an extension to a few individuals in each “group-defining role,” the participants' representations of the different roles and their consequences are mostly veridical. That is, people do represent the imposition of cost as such, and the reactions of the other partners as one of the roles defined in the event grammar. By contrast, the dynamics of interaction at a larger scale may be quite different from these roles and depend on factors not described by the event-grammar. Consider for instance the dynamics of ethnic signaling, as modeled by Kuran (Reference Kuran1998). People's decision to wear ethnic garb, for instance, depends on the perceived cost of doing so, which, in turn, depends on an agent's perception of the relative size and cohesion of one's and the other groups. The decision, in turn, influences other agents, as it modifies their estimates of the relative costs of signaling or not signaling. Naturally, this cascade model is not at all the way people represent the macro-features of the situation – which they see as people beginning to signal their ethnicity because they are more convinced than before of their group's value and of its goals.
There are two possible paths to account for such situations. Pietraszewski suggests that the event-grammar scales up, preserving its constitutive roles. In this case, people's own representations of large-scale group dynamics do capture people's actual behavior. The macro-effects outlined here are adequately described as the emergent effect of numerous agents entertaining the representations described by the event-grammar.
Alternatively, we may consider that large-scale group dynamics cannot be captured by the event-grammar. In that case, the macro-dynamics are really different from what people represent. But the fact that people all see those dynamics in terms of agent-like groups provides enough coordination that the intentional description often appears to be roughly correct so that most agents (and even social scientists!) see it as adequate.
It speaks to the great value and conceptual precision of Pietraszewski's proposal, that we can formulate these questions with a level of precision that is not common in the social sciences.
Conflict of interest
None.