Monday, January 16, 2023

Is ontological individualism still a viable social ontology?

image: a flat social ontology: actors and structures

Over the years I've continued to advocate for the position of ontological individualism -- the idea that social entities, powers, and conditions are all constituted by the actions, thoughts, and mental frameworks of individual human beings, and nothing else. I'm no longer entirely confident that this is an adequate view of social ontology, because I also maintain that doing good social science requires researchers to work with a rich theory of the social actors who constitute the "substrate" of the social world. In particular, I maintain that we need to view social actors as "socially constituted" and "socially situated". This means, fundamentally, that individuals develop into actors through interaction and exposure with the communities and institutions with which they interact from childhood to adulthood -- thus coming to possess various features of motivation, cognition, and reasoning on the basis of which they act in the social world. Further, individuals exist within institutional, cultural, and normative settings that establish constraints, resources, opportunities, and limitations on their actions. Actors are "socially situated" in ways that profoundly affect their actions.

The diagram above represents a flat social ontology, with individuals and social entities in a range of locally instantiated relationships. As the arrows indicate, influence flows in all directions, from actors to structures and from structures to individual actors and between both structures and actors. (Here is a post from 2015 that considers the logic of a "flat social ontology"; link.)

So if individual actors depend on the local and historically particular social environments in which they exist -- environments that are themselves embodied by other actors -- then in what sense can we legitimately say that the individual level is more fundamental than the social level?

Two ideas seem to be true, and they are in tension with each other. On the one hand, most views about "priority of individuals" over social facts are unsupportable. Individuals are not temporally prior to social relations and influences; individuals are not causally prior to social influences (since each actor is formed and constrained by social relationships and practices); individuals cannot be characterized in terms that avoid "social" characteristics (semantic priority); and individuals are not explanatorily prior to social facts (since social facts must be invoked to describe and explain the individual's mentality and action).

But likewise, social structures, practices, and institutions are not strictly prior to individuals. Social influences on individuals at a time depend upon the actions, thoughts, and relationships of an indefinitely large group of individual actors. That is to say that social arrangements work through the actions and mentalities of the individuals who make them up. Further, social structures are not strictly speaking causally prior to individuals in any absolute sense -- it is not the case that social structures determine the actors, and in fact later iterations of social structures and institutions are changed as a result of the actions and non-actions of the actors themselves.

So it seems clear that individuals are shaped by social realities (practices, institutions, normative systems) and social realities are constituted, maintained, and changed by individuals. We cannot separate them into separate and independent causal factors.

This suggests that neither "individualism" nor "holism" will do as a basis for social ontology. Neither individual mentality and action nor the dictates and constraints of social facts persist by themselves. Instead, social actors depend upon existing social relationships and arrangements, and social facts depend upon individual actors which carry and transform them. We need to conceive of both individual actors and social arrangements as part of a single, iterative and diverse process of change and continuity. And, unfortunately, the label of ontological individualism does not capture the fullness of this set of processes.

Other theorists have tried to solve this problem. Anthony Giddens addresses this complexity through his effort to undermine the strict distinction between agent and structure. He challenges the framework itself -- the idea that "agents determine structures" and the idea that "structures determine agents". He introduces a new term to capture the complexity of the relationship between actors and structures, the idea of structuration. His 1979 collection of essays, Central Problems in Social Theory: Action, Structure, and Contradiction in Social Analysis, provides a statement of some of his views. Here is how he frames his core concern in a key essay, “Agency, Structure”:

The principal issue with which I shall be concerned in this paper is that of connecting a notion of human action with structural explanation in social analysis. The making of such a connection, I shall argue, demands the following: a theory of the human agent, or of the subject; an account of the conditions and consequences of action; and an interpretation of ‘structure’ as somehow embroiled in both those conditions and consequences. (49)

Giddens faults much of sociology for having failed to conceptualize the social-structural context with sufficient nuance. He finds, for example, that Durkheim’s efforts to provide theoretical resources for describing the “external or objective” character of society were inadequate (51). The problem is that neither individualists nor structuralists have succeeded in expressing the inherent interdependence of the two poles. Give primacy to structures and the agents are “dopes” — robots controlled by structural conditions. Give primacy to individuals, and structures and institutions seem to disappear. Giddens' own view is that the two poles of structure and agency must be considered from within a common formulation:

I shall argue here that, in social theory, the notions of action and structure presuppose one another; but that recognition of this dependence, which is a dialectical relation, necessitates a reworking both of a series of concepts linked to each of these terms, and of the terms themselves. (53)

Giddens also observes that action necessarily implies a temporal framework.

‘Action’ or agency, as I use it, thus does not refer to a series of discrete acts combined together, but to a continuous flow of conduct. We may define action … as involving a ‘stream of actual or contemplated causal interventions of corporeal beings in the ongoing process of events-in-the-world’. (55)

This process view of actors-structures-actors is what Giddens refers to as "structuration".

Another fruitful attempt to approach social ontology along these lines has been advanced by Margaret Archer through her concept of morphogenesis. Here are the opening paragraphs of Realist Social Theory.

Social reality is unlike any other because of its human constitution. It is different from natural reality whose defining feature is self-subsistence: for its existence does not depend upon us, a fact which is not compromised by our human ability to intervene in the world of nature and change it. Society is more different still from transcendental reality, where divinity is both self-subsistent and unalterable at our behest; qualities which are not contravened by responsiveness to human intercession. The nascent ‘social sciences’ had to confront this entity, society, and deal conceptually with its three unique characteristics.

Firstly, that it is inseparable from its human components because the very existence of society depends in some way upon our activities. Secondly, that society is characteristically transformable; it has not immutable form or even preferred state. It is like nothing but itself, and what precisely it is like at any time depends upon human doings and their consequences. Thirdly, however, neither are we immutable as social agents, for what we are and what we do as social beings are also affected by the society in which we live and by our very efforts to transform it. (1)

This view too suggests an inseparable entwining of the social and the individual. Neither is primary, and neither is causally determined by the other. Rather, we must postulate a continuing process of interaction between actors and structures, a process that she calls "morphogenesis".

Is there a good label for this more nuanced view of social ontology? "Ontological individualism", even when carefully qualified, still seems to give unwarranted priority to the individual actor. "Ontological holism" is even worse, since it implies that social facts are autonomous after all. I've used the terms "actor-centered sociology" and "methodological localism" to capture the sociality of the individuals who make up the social world, but these phrases are not entirely intuitive. "Structuration" is the label that Giddens uses for his own theory of agent-structure, and his language of "social life as process" is intended to capture the idea of a temporally extended back-and-forth involving actors, structures, norms, and practices. Margaret Archer's theorizing about "morphogenesis" captures much of the back-and-forth nature of the relationship that exists between social arrangements and individual agency. James Coleman offers the idea of institutions as a "house of cards", depending upon the coordinated expectations and actions of individual participants for their relative stability. Other thinkers have used the idea of a "dialectical" relationship between individual and social arrangements, but the idea of dialectical relationships brings a lot of baggage. Zygmunt Bauman's concept of liquid modernity seems to point towards this "process" ontology as well, without offering much theoretical articulation beyond the central metaphor. Even the idea of a flat social ontology, mentioned above, captures some aspects of the more complex view of the social world that we seem to need.

At the moment I can't think of a simple phrase that would adequately capture a view of social ontology that encapsulates the complex nature of the "individuals-within-social-world" view of the social that seems most justified. Whatever scheme we choose, it is crucial to incorporate the diachronic nature of the interdependence of actors and structures into the view. The diagram at the top captures the multiple directions of influence at work in a social environment, and we might imagine a sequence of representations iterated over time, in which both social structures and socially constituted actors have changed over time. There is more to do in the field of social ontology.


Tom Hickey said...

Agree. This is particularly evident in macroeconomics where the conventional approach is based on microfoundations, which assumes radical individualism in its theory of agents' rational choice regarding individual preferences.

Even though the models don't work, change is resisted because "the methodological debate is settled." However, so-called heterodox economists have not developed competing models capable of displacing the conventional ones that dominate academia, at least ones that measure up to conventional standards. Heterodox economists fire back that the conventional standards are the source of the problem. So progress is at an impass.

Evidence regarding theory and modeling in practice shows that the life and social sciences cannot be modeled like the natural sciences, or at least a way to do this satisfactorily has not been discovered. The reason for this appears to be based on the nature of the subject matter. Natural process are ergodic while life and social processes are not. And as Feynman reportedly remarked observed, physics would be a lot different if electrons had feelings.

It further appears that since the issue involves complex adaptive systems characterized by emergence, success in theory development necessitates mathematical modeling of complex systems. While work is progressing in this direction, we aren't there yet.

But even the edges of physics are fuzzy and many questions elude answers based on conventional modeling. While the standard model can explain the electromagnetic force, and the strong and weak forces theoretical, it cannot simultaneously explain gravity, which is required for a satisfactory unified field theory. Work continues.

So it is hardly surprising that other fields that are more subject to complexity cannot be model precisely — yet anyway. So we need to be more circumspect and indeed more humble before nature.

The dichotomy between individualism and holism likely points toward a dialectical relationship that involves a both/and relationship rather than an either/or one. But what the basis for that synthesis is, is not yet clear. One thing does seem clear though and that is pitting one against the other is not working. New insight is needed.

jed said...

I'm glad you are exploring this issue because your occasional invocation of ontological individualism has always made me uncomfortable.

On the other hand the various alternative accounts you quote seem extremely problematic because they set up structures, "social entities", institutions, or whatever as mysteriously stable and distinct from the people in any social situation.

To resolve this problem we need a way of understanding the stability and mutability of social structure as a non-mysterious phenomenon that is immanent in individuals, but transcends any given individual.

I think a key to the needed understanding is given in your remark (paraphrasing Coleman) that "institutions ... depend... upon the coordinated expectations and actions of individual participants for their relative stability."

To move this toward an ontology, we can conceive of any given institution (in the most general sense) as a set of complementary expectations and patterns of action in the minds of its participants (whom I will call "carriers"). These sets are of course nested, overlapping, and vary along many gradients.

Also, of course, those expectations and patterns of action vary diachronically and so the institution changes -- sometimes without conscious intent by any participant, and sometimes very much due to conscious intervention.

For example any given language (a paradigmatic institution) varies across individuals geographically, by social class, and so forth. It typically contains multiple registers, is carried differently by different individuals, etc. None the less the language can be considered a single institution to the extent that all of its carriers are mutually intelligible -- which is to say, all its carriers have sufficiently complementary expectations and patterns of action.

This raises the remaining key point about the ontology of institutions: They must be self-sustaining. For example, a particular language is self-sustaining because individuals, to be understood and to understand others, must learn the complementary expectations and patterns of action that constitute the language. Enforcement of institutions for the most part isn't required, because typically individuals want to participate in institutions, so they want to maintain themselves as carriers of those institutions.

Individuals may leave an institution by adopting expectations and/or patterns of action that conflict with those that constitute the institution. Conversely, individuals may consciously or unconsciously recruit others to carry an institution (a very conscious example is evangelical faiths, a typically unconscious example is providing a word when someone else is groping for it). Individuals may create new institutions by persuading others to adopt a new, distinct, self-sustaining set of complementary expectations and patterns of action.

Stepping back from the details, I'd propose that in addition to individuals, a social ontology only needs institutions as defined above. Any given institution is carried by many individuals, and any given individual carries multiple institutions. Institutions are not mysterious intangible entities, they are immanent in their individual carriers in entirely ordinary ways.

Dan Little said...

Jed and Tom, Thanks very much for your many detailed and useful comments. Jed, the final paragraph of your comment (on institutions) is particularly apt. Tom, the connection you make to complex adaptive systems seems to me to be on the mark. I'll think carefully about both sets of comments. Dan

Anonymous said...

Organizations are instances of institutions.

Are Norm systems themselves institutions or are they components of institutions?

Do Structure nodes represent collections of institutions arranged in specific topologies?

Are individuals able to interact directly between themselves or are their interactions mediated implicitly or explicitly through institutions?

If the flat ontology image portrays a set of structural or causal linkages between elements of the social, would a stack of such images indexed by time be a plausible solution to the problem highlighted in this note?

Would a legitimate claim as social ontology consist of fleshing out the relevant transition rules from frame i to frame i+1?

Thank you Dan,


Anonymous said...

(just a few more notes)

According to this note, there are four distinct linkages between social entities that I can recognize: temporal, causal, descriptive and explanatory.

The way I see it, the later three linkages between social entities are one-to-many relations (or many-to-one depending on how one positions the relevant entities within the relation):

- a causal linkage can be drawn between those social entities that cause the respective effect, which itself is another social entity. In terms of drawing such a linkage I'd propose something like {cause,...}=>effect ie a list of entities that serve as causes to the specified effect.

- a descriptive linkage is one between an object and its properties, again a many-to-one relation. Both the "object" and its "properties" are social entities. {property,...}-|object is how I'd write it down, in shorthand.

- an explanatory linkage would be one between an explanandum and its expalanans, again a many-to-one relation: explanandum:={explanans,...}. Again, all relevant entities are social entities.

Note how, if the causal and descriptive relations are treated as many-to-one relations ({argument,...} op header), then the explanatory linkage uses a reverse notation (where the target is placed on the left and the sources are placed on the right of the operator) but that's simply notational formalism.

As far as the temporal dimension is concerned, I would think that different social entities should have different time scales that would be relevant to their kind. It would seem that a time tick should be different for different entities.

Also, if the relevant time scale is eg one second, then I'd assume that everything observed and analyzed within that second would remain constant, hence the view of the social ensemble would be one of immutability, even if that were not actually the case. It would just be an artifact of our insufficiently configured instrumentation.

On the other hand, if the relevant time scale were a geological eon, then what would probably get registered would be a punctuated equilibrium image where eons of stasis would be perturbed by a single blip of social activity but nothing more; there would be no resolution available for more details.

Entities could be endowed with a temporal tag encoding their relevant synchrony wrt to the timescale of the investigation. This means that every tick in the investigator's clock should be predetermined and associated appropriately to the clocks of available entities.

Now, would a proper social ontology be a set of rules that demonstrate how certain constellations of linkages with their respective supports, transform between successive ticks of the clock?

Wouldn't something like that entail the need to hash out the details about the interactions between the different categories of linkages? Eg, wouldn't we have to account for how causality and temporal succession mingle? Or, how the explanatory rests (or doesn't rest) on the descriptive?