The micro and the social

In his influential article "A definition of physicalism" (1993) Philip Pettit attempts to formulate a consistent and coherent account of physicalism as an ontology of the world.

I believe that we can define a possibly true, substantive doctrine which holds, roughly, that the empirical world 'contains just what a true complete physics would say it contains'. (213)

The resulting view is offered as an attempt to identify the "furniture of the empirical world". Can there be a corresponding theory of the "furniture of the empirical social world"?

His statement in the follow-on reply to criticisms of this essay in "Microphysicalism without contingent micro-macro laws" (1994) is helpful.

“Physicalism – better, perhaps, microphysicalism – is the doctrine that actually (but not necessarily) everything non-microphysical is composed out of microphysical entities and is governed by microphysical laws; and this, in a sense which means that the non-microphysical facts supervene contingently on the microphysical” (1994: 253).

What is the analogy for the trivial fact, “no social structures without human beings embodying them”? Let's call the corresponding view “bare ontological dependence” (BOD). Here is a formulation of BOD constructed to be exactly parallel to Pettit's definition of microphysicalism:

“[Bare ontological dependence] … is the doctrine that actually (but not necessarily) everything [at the social level] is composed out of [existent human individuals with specified features of mentality, psychology, and cognitive capacity] and is governed by [psychological and neurophysiological] laws; and this, in a sense which means that the [social-level] facts supervene contingently on the [micro-individuals]”.

Is this a credible and defensible conception of the relation between individual human actors and "social entities"?

This formulation entails, apparently, that “the features of the social institution are constituted and governed by the micro-psychological characteristics of the individual actors who constitute it” and the social “supervenes” upon those micro-individuals. This in turn implies the supervenience maxim: “no difference at the social-structure level without some difference at the micro-individual level”. It also specifies a clear sense in which “micro-particles” have primacy over ordinary physical objects and "individuals" have primacy over "social entities"; their properties “fix” the behavior of the macro-objects. The corresponding statement for "bare ontological dependence is then that "actors" have primacy over ordinary "social" objects; their properties "fix" the behavior of the macro-social-objects.

Pettit's application of these formulae to the world of physics, chemistry, and planetary motion is reasonable enough. And part of the plausibility of the view in the case of physics is that the microphysical particles can be said to have properties that are fixed and independent from the macro-level ensembles that they constitute. However, this is not the case in the situation of "socially situated individuals" who constitute "ongoing social structures and practices". Rather, there are reciprocal causal relations up and down, or back and forth, across the levels that make the ideas of "constitution" and "fixing" no longer compelling. And this implies in turn that there are no trans-historical, universal "regularities" of human behavior that might constitute the bridge between individual actions and social entities.

Further, because of the contingency of some historical sequences (for example, the invention of monotheism or the discovery of the heavy plow), and given the path-dependency of some structural or institutional outcomes, there is no reason to expect that a common beginning point of unsocialized or pre-socialized individuals (a state of nature) would gradually develop, by perhaps unknown behavioral laws, into the establishment and articulation of specific social structures. And if we accept the point that it is sometimes the case that "different structures elicit different kinds of human mentality", then we are brought face to face with contingency all the way down: contingent structures and contingent social individuals.

So Pettit's line of thought in defining physicalism is quite implausible when applied to the social world. It is true that "the Ford Motor Company" could not exist if there were not actual human beings occupying roles within and outside the company. It is not true, however, that "the 'social' world is governed by forces or regularities that [empirical psychology] is best equipped to describe" (the analog to Pettit's statement about micro-particles). The mental characteristics and processes of the actors involved in a social entity or set of social arrangements are themselves in fact shaped by past social arrangements to which the actors have been exposed. So if the foundational body of empirical knowledge is "empirical psychology" (as JS Mill indeed believed) then we must reject the view.

One of Pettit's claims about the relation between microphysical things and macrophysical things is fundamentally a reassertion of the supervenience relation between levels: "No macrophysical difference without a microphysical one" (216). How does this proposition fare when applied to "macro-social entities" and "micro-individual states"? It corresponds to this assertion: "No macro-social difference without a micro-individual one". Unfortunately, this seems to be a trivial statement when applied to the social realm. Any two social states differ at the micro-level, for the most trivial of reasons: they are different states, with different individuals, and different individuals have different action-plans and beliefs. So whether S1 and S2 are "different" or "the same" in their macro-descriptions, it is trivially true that they will differ in their micro-composition. Consider these three facts: "The price of soybeans on the Chicago Board of Trade on May 1, 2025, is 1,060.50"; "The price of soybeans on the Chicago Board of Trade on May 1, 2024, is 1,000.00"; "The price of soybeans on the Chicago Board of Trade on May 1, 2023, is 1,060.50". Each sentence describes a structural fact: the circumstances of supply and demand on the specified day led to an equilibrium price as quoted. But none of the structural facts described here corresponds to a single set of individual actors doing the same things for similar reasons. The actors have changed, their motivations have changed, their habits have changed, and their styles of dress have changed. The pathways that led to the same structural equilibrium in 2025 and 2023 were no doubt different in multiple ways; and likewise, the actors and the pathways that led to different equilibria in 2025 and 2024 were different as well. Even if we were to perform a massive experiment in "experimental economics" and assemble 1,000 traders on May 1, 2023, and then again on May 1, 2025; control the information to which they are exposed in the preceding twelve hours; and ask them to buy and sell as they normally would, there will still remain idiosyncratic differences between the series of thoughts and actions undertaken on the two days. And the same will be true of the experiment when we model 2024 and 2025: there will inevitably be vast numbers of individual-level differences. So the supervenience condition is vacuous. There are always differences across cases at the micro-individual level.

It is pertinent to observe that some physical processes are path dependent as well, which means that the initial states of the micro-particles by themselves are not sufficient to "determine" the outcome. Rather, the outcome depends in part on the process of transition from one state to another. Suppose a physicist observes two vessels of pure water over a bunsen burner. One container is boiling vigorously, while the other, at the same observed temperature, is not boiling. The difference is that the first vessel was heated quickly while the second was heated slowly. The process made a difference in the outcome, even though the micro-constituents were indistinguishable. But this is the relevant point: there is a difference in the two states of containers of water, even though there is currently no difference in the states of the micro-particles.

Pettit considers a possibility that he considers to be inconsistent with his understanding of "physicalism":

Another [opponent of physicalism] will be the person, perhaps difficult to imagine accepts microphysical composition but thinks that the composition involved is not necessarily conservative: it allows, without further need of explanation, that two entities that are composed in the same way, and of the same materials, may yet differ intrinsically from one another. (217)

Pettit appears to think this is an absurd contention; how could these two macro-entities differ, without any difference in the composing microphysical parts?

Let's ask first what Pettit means by the phrase "composed in the same way". One natural reading is asynchronous and structural. A bar of iron may consist of precisely the same number of iron atoms, but the arrangement of the atoms is coherent in the first case and incoherent in the second arrangement. In that case the first iron bar has a property that the second bar lacks; it is "magnetic". And yet the two bars consist of exactly the same kind of microphysical parts. The solution in this case is that the arrangement of the parts makes a difference; when the iron atoms are coherently aligned, their magnetic fields aggregate to a macro-scale magnetic field.

The second possible meaning of "composed in the same way" is diachronic and historical. To be "composed in the same way" is to have undergone precisely the same set of processes of material transformation, heat transfer, application of pressure, etc. From the short explanation of path-dependent processes above, we know that differences in physical processes of material transformation can indeed lead to differences in macro-physical outcome for ensembles of precisely similar microphysical constituents; in the current case, one process leads to a magnetic bar of iron, while the second process leads to a non-magnetic bar.

Now return to the relation between individuals and social structures. Is it possible for two ensembles of individuals to be exactly similar in the current psychological characteristics of the individuals involved in the two cases but to nonetheless differ in some important way at the macro-social level? Much turns on how finely we expect to interpret "exactly similar" here. But suppose we assume an abstract conception of the individual's psychology along these lines: each individual wants a situation where he or she can satisfy as many preferences as possible, while avoiding catastrophic failures. And each individual has a set of "social emotions" that permit the emergence of social relationships based on trust and mutual solidarity. Now consider two thought experiments involving the emergence of a "wannabe" strong man dictator in the political system. In the first instance the potential dictator has the good fortune that his first efforts at taking power are generally unopposed because individuals recognize his intentions but find resistance to be too risky; as the dictator gains followers and successes this population becomes more passive; and after ten years the dictator is applauded and supported throughout much of the population. In the second instance the dictator has bad fortune. A few of his supporters are overly willing to use violence against dissidents and resisters, which stimulates a higher degree of alarm in ordinary citizens; a few of the ordinary citizens recognize the discontent present in others and form relations of trust. These "trust" circles expand over time and incidents of resistance become more frequent; the dictator and his supporters become more willing to turn to violent suppression; the dictator's violence tips more citizens into trust circles of their own; and after ten years the dictator's hold on power is precarious. His rallies are attended by his own militias and inner circle; but acts of resistance continue and proliferate.

The two scenarios begin and end with the same kinds of actors -- persons with their own interests and a capacity for forming social relationships. And yet the properties of the two regimes at the end of the experiment are quite different. This seems to correspond to the terms of Pettit's original reductio ad absurdum: different macro-arrangements constituted by the same kinds and arrangements of microparticles. What differentiates the two cases is the contingency, path dependence, and "triggering" of individual capacities that occurred during the processes involved in the two scenarios. In the first case, no events arose to trigger and encourage the emergence of trust networks; whereas in the second case, there were such events. The historical processes in the two cases were different, and the properties of the ensemble continued to evolve in different directions. The social capacities of the individuals were present in both scenarios, but they were only triggered in the second scenario.

What all of this suggests is fairly simple: the idea of "physicalism" as a fundamental model of ontology is not a suitable framework for thinking about social ontology. The individuals who "constitute" social arrangements are not analogous to the microparticles that Pettit considers; rather, they are actors whose states of agency are altered dynamically by events, processes, and structures that emerge historically, and the inherent contingency and path dependence of the social world guarantee that no version of foundationalism or individualism will suffice for social ontology.