The following is a guest post by Natalie Aviles.
The language of social “mechanisms” has become so widespread in recent decades that it seems sometimes to operate as a stand-in for any mid-range theoretical proposition about a given social phenomenon. Despite the ubiquity of “mechanism-talk”, there is little agreement over what social mechanisms are and how they should be used in sociological explanation.
In “Ratio via Machina: Three standards of mechanistic explanation”, I (with Isaac Reed) argue that debates over how we might explain social phenomena mechanistically must contend with the existence of (at least) three separate practices of mechanistic explanation that have emerged over the years, each of which assumes different standards of what a mechanism is and what it can achieve.
- The substantial standard argues that mechanisms manifest relations between ontologically real social entities according to their inherent causal capacities. In plain English, this means that mechanisms are conjunctures of real social things that behave in predictable (though not strictly deterministic) ways by virtue of being those types of things and not others. To take a contemporary example inspired by Phil Gorski, we might imagine “democracy” to be a real social thing that tends to manifest in particular ways, but can be distorted when it interacts with other real social things, such as “discipline” or “tyranny”. Today’s sociologists are most familiar with the substantialist mechanisms advocated by critical realists like Gorski, but in our paper we draw upon a more lucid account of realism by philosopher of science Nancy Cartwright to clarify what these debates entail for the relationship between ontology and explanation more generally.
- The formal standard is a pragmatic (as in, non-metaphysical) approach to explaining how one variable might affect another in terms of causal pathways that run through intervening variables. In terms of formal propositions of mechanisms, this is the most common form sociologists see. For example, we might wish to explain the connection between parent’s socioeconomic status and their children’s occupational outcomes by looking at the intervening variable of education. Where a clear relationship between these variables exists and can be measured or modeled, we say we have found a social mechanism and can argue among ourselves the relative weight of this or that mechanism in explaining the outcome of interest. What is pragmatist about formal mechanisms is that they work from disciplinary conventions of where explanations for social phenomena might “bottom out”, or stop being relevant to our explanations. So, while adherents to formal mechanisms may be concerned about national education reform or individual student outcomes as they relate to this mechanism, they are unconcerned as to what makes a person or a policy a social thing in the world as distinct from other kinds of things.
- The metaphorical standard assumes that mechanisms are those social processes that maintain durability or regularity in social life when measured interpretively against features of the social world that are not regular or durable. Here the concept of a mechanism is deployed heuristically and relationally to aid in interpretation of complex social conjunctures by highlighting how certain features of the social world remain constant in ways that allow us to understand which variable features of causal significance might help us explain a phenomenon of interest. We draw upon Ricoeur’s rhetorical theory to defend this interpretation, which provides more systematic support for the standard as a distinct form of explanation.
We end our paper with a plea for nuance: “each of the three standards presupposes a different approach to the construction and evaluation of explanation, and conflating these standards in an effort to enforce a single universal definition of mechanism will lead to misunderstandings of explanatory practice in sociology.”
Beyond advertising the paper, this post provides as a means for me to discuss how the things I had hoped to accomplish with our tripartite distinction have grown in scope since the manuscript was accepted for publication. I think the following addenda would be productive in advancing the conversation to whatever extent it remains necessary to discuss mechanisms as a vehicle for sociological explanation.
First, I want to address an understated point we make about the relationship between method and mechanistic standard—that last bit about the importance of understanding explanatory practice in sociology. There is more to be said to this, and more work to be done in fleshing out the affinities between, for instance, the metaphorical standard and interpretive historical and ethnographic research. Equally, the dominance of quantitative analysis, network analysis, and computational modeling in formal mechanistic explanation is not accidental; it aligns with many of the epistemic virtues of these models for adding complexity to correlational studies. Finally, what it is exactly that draws sociologists to the ontological focus of critical realism might have something to do with the existential dread that accompanies self-conscious reflection upon our current position in the long arc of late capitalism, where serious alternatives for organizing society are seriously lacking.
My second addendum follows directly from this last thought. In response to informal critique, I think it is worth my defending Christian Smith as an “exemplar” of critical realism despite the increasing space taken up by more moderate proponents of the approach. For example, Smith consistently defends a normative approach to social justice by arguing that things like “human dignity” are objectively real, and stem from capacities inherent to human beings as particular natural kinds. Smith’s critical realism is both internally fairly consistent and faithful to the substantialist ontology to which it is necessarily bound qua critical realism. This is something, as many have pointed out (Healy 1998, Sawyer 2005) that most other critical realist accounts fail to accomplish, and this makes Smith’s program productively challenging to think through.
The more it seems advocates of CR attempt to mainstream the philosophy, the more they lose the ambition of the enterprise as first proposed. This ambition is a moral one—and, I think, ultimately a theological one, of a pair I believe with Whitehead’s attempts to locate God in a world upturned by Einstein’s general theory of relativity. (By this I mean that an appreciation of the ecological complexity of social phenomena in light of increasingly emergentist and processual theories of social organization that have reversed the order of our thinking so that we presume change and ask after how stability is even possible is similar to Whitehead’s spiraling meditations on the relationship between eternal objects and events-in-the-becoming.) This is utterly fascinating and worth thinking through for anyone whose sociology guides their reflections on moral philosophy, metaphysics, and cosmology.
The thing of it is, metaphysics has no place in the explanatory enterprise of empirical social science. Claims about what is or is not a manifestation of the intransigent forms of the universe are wholly unanswerable by the methods we have available to us. I stand firmly by the argument that critical realism’s attempt to anchor empirical explanation in metaphysics is, epistemologically speaking, wrong-thinking at best and counter-productive at worst.
Finally, some of the inconsistencies we discussed as plaguing cases that propose mechanisms that straddle the substantial and formal standards of mechanistic explanation in Proposition 1, which states “Debates about ‘realism’ are really debates between the substantial and the formal standards” (pp. 727-729), have been addressed in far greater detail in a recent article in Sociological Theory by Francisco León-Medina. It is not coincidental that most of the problematic cases we highlight come from proponents of Analytical Sociology (AS). León-Medina considers the tensions between AS’s stated theoretical goals of describing mechanisms as generative processes on the one hand and “nuts and bolts” version of mechanisms often deployed in agent-based modeling practices. This article is a much-needed deep dive into an area of practice Isaac and I barely touch upon in our article, and I suggest it as a companion to “Ratio via Machina.”
In the years since this article was accepted for publication, these standards remain durable. What the pre-print helped accomplish, partly by way of its debut as a response to Healy’s “Fuck Nuance” on that infamous ASA panel, had been to stoke conversation around the role of causal pluralism in sociological explanation. This was, if I can speak for Isaac at least to the extent memory serves, the initial motive of the manuscript, far and above demonstrating the disunity of practice that prevails among different proponents of mechanisms. I hope that the paper continues to generate interest in causal pluralism and philosophy of science proper, particularly the work of Nancy Cartwright, from whom sociologists have a lot to learn.
Natalie Aviles is a Postdoctoral Fellow in Sociology at Colby College.
 The opinions reflected in this post are my own. Isaac bears no responsibility for any errors of reasoning or diplomacy I might make herein.
 I thank Andreas Glaeser for this insight.