Endogenizing Cultural Preferences: A Futile Task for Social Science?
In a recent article published at PNAS—Weak Individual Preferences Stabilize Culture—Acerbi and de Courson (2025) argue that, even if we attribute cultural phenomena to “social influence,”
nonsocial preferences, differently from social forces, act consistently in the same direction, so that, even when weak, they sway social influence forces in the long run.
Put differently, exogenous preferences, whatever they may be, override social forces if they are sufficiently strong:
 |
|---|
| Figure from Acerbi and de Courson (2025) |
The authors tweaked a well-known model by Boyd and Richerson (1985) and added a dose of exogeneity to the system, and showed that this surgical addition can fundamentally change system behavior.
I find this, to put it mildly, radically interesting.
It is radically interesting, because the fact is that the article brings a fundamental theoretical tension out into the open: the problem of exogenous preferences.
Exogeneity of Preferences
One of the appealing features of the cultural evolution literature, at least for a sector of social scientists, has been its focus on population processes in determining preferences. In this approach, culture is something people “learn” from others. This literature did a remarkable job of specifying specific pathways, looking at, among other things, frequency-dependence or prestige-bias, models that specify how cultural transmission actually occurs.
To a sociologist’s eye, this is something crucial, both for theoretical and historical reasons, as it directly attacks to the problem of where people’s preferences come from.
One of the few commendable criticisms of sociology directed at micro-economic modeling is the problem of exogenous preferences, i.e., the common practice of “taking individuals as they are,” with preferences constrained to be outside the systems being modeled. This problem is at central stage in a—now neglected—classic, Talcott Parsons’s The Structure of Social Action. Parsons eloquently proposes that the “utilitarian” model of agency, with its conception of “ends” as “random,” cannot adequately explain why there is relative order rather than relative chaos and relative stability rather than relative change. A good chunk of the book is focused on what Parsons call the “utilitarian dilemma,” which describes the trade-off:
either the active agency of the actor […] is an independent factor in action, and the end element [the things we aim for] must be random; or [the actor] disappears and […] assimilated to the conditions of the situation (Parsons 1949, 64).
The Structure’s critique is simple and effective. In essence, it is a principled rejection of exogenous preferences. That said, the subsequent Parsonsian edifice fell short of being that interesting. The classic theoretical story asserted that (1) the force behind preferences was “values”—mysterious things like American individualism—and these values were shared among members of the public, and (2) people “learn” these values via socialization, a process through which culture-outside ends up located in culture-inside.
This Parsonsian edifice, in one way or another, continues in folk-theory form.
One of the more exciting debates in sociology, in fact, is an implicit revitalization of these Parsonsian ideas by people like Steve Vaisey and Omar Lizardo through what I personally call “cognitivist neo-Parsonsianism,” and a general movement in the late 2000s for a cognitively compatible conception of culture. That said, for various historical and empirical reasons, the question of why individuals choose one belief, practice, or behavior over another could not be resolved, and this cognitivist neo-Parsonsian movement could not fully move from general principles of learning to the more specific question of how one belief or behavior is preferable to another.
The sources of this problem are not clear. In any case, it is plausible to say that, decades after Parsons, sociologists are simply not good at predicting why people choose, learn, and prefer A over B. They think that it must be something social. Something arbitrary. Something constructed. Definitely learned from other people. And that’s about it.
Individuals As They Are
Acerbi and de Courson (2025) makes the issue much more explicit: influence models are not directional.
Classic socialization theories in sociology assume that the mechanism that drives people to adopt certain preferences or behaviors is conformity, and the alternative routes proposed recently are not specific and testable (see, e.g., Guhin, Calarco, and Miller-Idriss 2021). The same goes for the political socialization research: the widespread “social learning model,” in principle, boils down to the idea that what kids see is what they learn (Jennings and Niemi 1968). The issue is not exclusive to sociologists or cultural evolutionists, for that matter.1 And one payoff of the cultural evolution framework was, for exactly this particular question, its explicit focus on providing specific pathways of social learning.
1 See Bowles (1998).
For better or worse, this allowed a proliferation of transmission theories above and beyond silly conformity obsession (Wrong 1961), turning learning into a potential for social construction of preferences by postulating a population process, where successive generations learn and unlearn cultural objects through conformity and prestige, environmental ambiguity, or cultural group differentiation.
This article may be a signal, however, that our emphasis for these processes may have been too much.
If “taking individuals as they are” is still our best bet for understanding cultural convergence, then I wonder why we had so much fuss about the “economic agent.”
Somehow, that ghoul, Exogeneity, returned.