Are organizations emergent?

Do organizations have properties that are in some recognizable way independent from the behaviors and intentions of the individuals who inhabit them? In A New Social Ontology of Government I emphasized the ways in which organizations fail because of actor-level features: principal-agent problems, inconsistent priorities and goals across different working groups, strategic manipulation of information by some actors to gain advantage over other actors, and the like. With a nod to Fligstein and McAdam's theory of strategic action fields (link), I took an actor-centered approach to the workings (and dysfunctions) of organizations. I continue to believe that these are accurate observations about the workings of organizations and government agencies, but now that I've reoriented my thinking away from a strictly actor-centered approach to the social world (link), I'm interested in asking the questions about meso-level causes I did not ask in A New Social Ontology.

For example: 

(a) Are there relatively stable meso-level features of organizations that constrain and influence individual behavior in consistent ways that produce relatively stable meso-level outcomes? 

(b) Are there routine behaviors that are reproduced within the organization by training programs and performance audits that give rise to consistent patterns of organizational workings? 

(c) Are there external structural constraints (legal, environmental, locational) that work to preserve certain features of the organization's scheme of operations? 

It seems that the answer to each of these questions is "yes"; but this in turn seems to imply that organizations have properties that persist over time and through changes of personnel. They are not simply the result of the sum of the behaviors and mental states of the participants. These meso-level properties are subject to change, of course, depending on the behaviors and intentions of the individuals who inhabit the organization; but they are sometimes stable across extended periods of time and individual personnel. Or in other words, there seem to be meso-level features of organizations that are emergent in some moderate sense.

Here are possible illustrations of each kind of "emergent" property.

(a) Imagine two chemical plants Alpha and Beta making similar products with similar industrial processes and owned by different parent corporations. Alpha has a history of occasional fires, small explosions, and defective equipment, and it was also the site of a major chemical fire that harmed dozens of workers and neighbors. Beta has a much better safety record; fires and explosions are rare, equipment rarely fails in use, and no major fires have occurred for ten years. We might then say that Alpha and Beta have different meso-level safety characteristics, with Alpha lying in the moderate risk range and Beta in the low risk range. Now suppose that we ask an all-star team of industrial safety investigators to examine both plants, and their report indicates that Alpha has a long history of cost reduction plans, staff reductions, and ineffective training programs, whereas Beta (owned by a different parent company) has been well funded for staffing, training, and equipment maintenance. This is another meso-level property of the two plants -- production decisions guided by profitability and cost reduction at Alpha, and production decisions guided by both profitability and a commitment to system safety at Beta. Finally, suppose that our team of investigators conducts interviews and focus groups with staff and supervisors in the two plants, and finds that there are consistent differences between the two plants about the importance of maintaining safety as experienced by plant workers and supervisors. Supervisors at Alpha make it clear that they disagree strongly with the statement, "interrupting the production process to clarify anomalous temperature readings would be encouraged by the executives", whereas their counterparts at Beta indicate that they agree with the statement. This implies that there is a significant difference in the safety culture of the two plants -- another meso-level feature of the two organizations. All of these meso-level properties persist over decades and through major turnover of staff. Supervisors and workers come and go, but the safety culture, procedures, training, and production pressure persist, and new staff are introduced to these practices in ways that reproduce them. And -- this is the key point -- these meso-level properties lead to different rates of failure at the two plants over time, even though none of the actors at Alpha intend for accidents to occur. 

(b) This example comparing industrial plants with different safety rates also serves to answer the second question posed above about training and oversight. The directors and staff who conduct training in an industrial organization can have high commitment or low commitment to their work -- energetic and focused training programs or perfunctory and forgettable training programs -- and the difference will be notable in the performance of new staff as they take on their responsibilities. For example, training for control room directors may always emphasize the importance of careful annotation of the day's events for the incoming director on the next shift. But the training may be highly effective, resulting in informative documentation across shift changes; or it may be ineffective and largely disregarded. In most cases poor documentation does not lead to a serious accident; but sometimes it does. So organizations with effective training on procedures and operations will have a better chance of avoiding serious accidents. Alpha has weak training programs, while Beta has strong training programs (and each dedicates commensurate resources to training). Routine behaviors at Alpha lead to careless implementation of procedures, whereas routine behaviors at Beta result in attentive implementation, and as a result, Beta has a better safety performance record.

(c) What about the external influences that have an effect on the overall safety performance of an industrial plant? The corporate governance and ownership of the plant is plainly relevant to safety performance through the priorities it establishes for production, profitability, and safety. If the corporation's highest priority is profitability, then safety procedures and investments take the back seat. Local budget managers are pressed to find cost reductions, and staff positions and equipment devoted to safety are often the easiest category of budget reduction to achieve. On the other hand, if the corporation's guidance to plant executives is a nuanced set of priorities within which both production goals and safety goals are given importance, there is a better chance of preserving the investments in process inspectors, better measurement instruments, and on-site experts who can be called on to offer advice during a plant emergency. This differentiating feature of corporate priority-setting too is a meso-level property that contributes to the level of safety performance in a chemical plant, independent of the knowledge and intentions of local plant managers, directors, and workers.

These brief hypothetical examples seem to establish a fairly mundane form of "emergence" for organizational properties. They provide examples of causal independence of meso-level properties of organizations. And significantly, each of these meso-level features can be identified in case studies of important industrial failures -- the Boeing 737 Max (link), the Deepwater Horizon disaster (link), or the 2005 Texas City refinery explosion (link).

It may be noted that there are two related ideas here: the idea that a higher-level property is emergent from the properties of the constituent entities; and the idea that a higher-level feature may be causally persistent over time and over change of the particular actors who make up the social entity. The connection is this: we might argue that the causally persistent property at the meso-level is different in nature and effect from the causal properties (actions, behaviors, intentions) of the individuals who make up the organization. So causal persistence of meso-level properties demonstrates emergence of a sort.

Critical realism and ontological individualism

Most critical realists would probably think that their philosophy of social science is flatly opposed to ontological individualism. However, I think that this opposition is unwarranted.

Let's begin by formulating a clear idea of ontological individualism. This is the view that social entities, powers, and conditions are all constituted by the actions, thoughts, and mental frameworks of individual human beings, and nothing else. The social world is constituted by the socially situated individuals who make it up. This is not to question the undoubtable fact that individuals have social properties -- beliefs, values, practices, habits, and relationships -- that are integral to their consciousness and agency. But these properties themselves are the recursive effects of prior sets of socially constituted, socially situated individuals who have contributed to their formation as social actors. Fundamentally, then, social entities are constituted by individual actors; and individual actors have in turn been framed, shaped, and influenced by their immersion in prior stages of social arrangements and relationships.

Consider a trivial illustration of the kind of recursive individual-social-individual process that I have in mind here. Consider the habit and norm of queuing in waiting for a bus, boarding a plane, or buying a ticket to a popular music concert. Queuing is not a unique solution to the problem of waiting for something. It is also possible for individuals within a group of people to use their elbows and voices to crowd to the front in order to be served earlier. But in some societies or cultural settings children have been given the example of "waiting your turn", lining up patiently, and conforming to the norms of polite fairness. These norms are internally realized through a process of socialization and maturation, with the result that the adult in the queuing society has both the habit and the norm of waiting for his or her turn. Further, non-conformists who break into the queue are discouraged by comments, jokes, and perhaps a quick jab with a folded umbrella. In this case adults were formed in their social norms by the previous generation of teachers and parents, and they in turn behave according to these norms and transmit them to the next generation. (Notice that this norm and behavior differs from the apparently similar situation of bidding on a work of art at an auction; in the auction case, the individuals do not wait for their turn, but rather attempt to prevail over the others through the level, speed, and aggressiveness of their bids.) Here we might say that the prevailing social norms of queueing-courtesy are a social factor that influences the behavior of individuals; but it is also evident that these norms themselves were reproduced by the prior behaviors and trainings offered by elders to the young. Further, the norm itself is malleable over time. If the younger generation develops a lower level of patience through incessant use of Twitter and cell phones, rule breakers may become more common until the norm of queueing has broken down altogether.

This example illustrates the premises of ontological individualism. The queueing norm is promulgated, sustained, and undermined by the various activities of the individuals who do various things throughout its life cycle: accept instruction, act compliantly, instruct the young, deviate from the norm. And the source of the causal power of the norm at a given time is straightforward as well: parents and teachers have influence over the behavior of the young, observant participants in the norm have some degree of motivation towards punishing noncompliant individuals, and ultimately other sources of motivation may lead to levels of noncompliance that bring about the collapse of the norm altogether.

Several points are worth underlining. First, ontological individualism is fully able to attribute causal powers to social assemblages, without being forced to provide reductionist accounts of how those powers derive ultimately from the actions and thoughts of individuals. OI is not a reductionist doctrine. Second, OI is not "atomistic", in the sense of assuming that individuals can be described as purely self-contained psychological systems. Rather, individuals are socially constituted through a process of social formation and maturation. This person is "polite", that person is "iconoclastic", and the third person is deferential to social "superiors". Each of these traits of psychology and motivation is a social product, reflecting the practices and norms that influenced the individual's formation. (This isn't to say that there is nothing "biological" underlying personality and social behavior.)

Several points can be drawn from this account. First, OI is not a reductionist doctrine -- any more than is "physicalism" when it comes to having a scientific theory of materials. We do not need to derive the properties of the metal alloy from a fundamental description of the atoms that constitute it. Second, OI is not an atomistic doctrine; it does not postulate that the constituents of social things are themselves pre-social and defined wholly in terms of individual characteristics. In a perfectly understandable sense the socially constituted individual is the product of the anterior social arrangements within which he or she developed from childhood to adulthood. And third, OI does not compel us to take an "as-if" stance on the question of the causal properties of social assemblages. The causal powers that we discover in certain kinds of bureaucratic organization are real and present in the world -- even though they are constituted and embodied by the actions, thoughts, and mental frameworks of the social actors who constitute them.

Now let's turn to critical realism and the position its practitioners take towards "individualism" and the relationship between actors and structures. Roy Bhaskar addresses these issues in The Possibility of Naturalism.

First, the ontological question about the relationship between "actors" and "society":

The model of the society/person connection I am proposing could be summarized as follows: people do not create society. For it always pre-exists them and is a necessary condition for their activity. Rather, society must be regarded as an ensemble of structures, practices and conventions which individuals reproduce or transform, but which would not exist unless they did so. Society does not exist independently of human activity (the error of reification). But it is not the product of it (the error of voluntarism). Now the processes whereby the stocks of skills, competences and habits appropriate to given social contexts, and necessary for the reproduction and/or transformation of society, are acquired and maintained could be generically referred to as socialization. It is important to stress that the reproduction and/ or transformation of society, though for the most part unconsciously achieved, is nevertheless still an achievement, a skilled accomplishment of active subjects, not a mechanical consequent of antecedent conditions. This model of the society/ person connection can be represented as below. (PON, 39)

This is a complicated statement. It affirms society exists as an ensemble of structures that individuals "reproduce or transform" and that "would not exist unless they did so". This is the key ontological statement: society depends upon the myriad individuals who inhabit it. The statement further claims that "society pre-exists the individual" -- that is, individuals are always born into some set of social arrangements, practices, norms, and structures, and these social facts help to form the individual's agency.

Here is the diagram to which Bhaskar refers (PON 40):

The cyclical relationship between social arrangements and individual "socially constituted action" is represented by the rising and falling dashed lines.

It seems, then, that Bhaskar's view is fundamentally similar to the view of methodological localism developed in earlier posts (link, link, link). Methodological localism affirms that there are large social structures and facts that influence social outcomes. But it insists that these structures are only possible insofar as they are embodied in the actions and states of socially constructed individuals. The “molecule” of all social life is the socially constructed and socially situated individual, who lives, acts, and develops within a set of local social relationships, institutions, norms, and rules. And these supra-individual structures and norms are, in turn, maintained by the actors who inhabit them.

Francesco Di Iorio addresses many of these points in his contribution to Research Handbook of Analytical Sociology through his analysis of the relationship between critical realism and methodological individualism. Much of Di Iorio's analysis seems entirely correct. But, following Bhaskar, Di Iorio seems to postulate the absolute (temporal) priority of the social over the individual; and this seems to be incorrect.

According to critical realists, MI cannot account for the fact that the social world and its bounds exist independently of the individual interpretation of this world, that is, independently of the individual’s opinion about what she is free or not free to do. (Di Iorio 141)

It seems apparent, rather, that the relationship between the social world and the particular constitution of human actors at a given time is wholly recursive: social arrangements at ti causally influence individuals at ti; actions and transformations of individuals at ti+1 lead to change in social arrangements in ti+1; and so on. So neither social arrangements nor individual constitution are temporally prior; each is causally dependent upon the other at an earlier time period.

So it seems to me that there is nothing in the core doctrines of critical realism that precludes a social ontology along the lines of ontological individualism. OI is not reductionist; rather, it invites detailed investigation into the ways in which social arrangements both shape and are shaped by individual actors. And these relationships are sufficiently complex and iterative that it may be impossible to fully trace out the connections between surprising features of social institutions and the underlying states of the actors who constitute them. As a practical matter we may have confidence about beliefs about the properties of a social structure or institution, without having a clear idea of how these properties are created and reproduced by the individuals who constitute them. In this sense the social properties are weakly emergent from the individual-level processes -- a conclusion that is entirely compatible with a commitment to ontological individualism (link).

One of the most prominent critics of ontological individualism is Brian Epstein. His arguments are considered in earlier posts (link, link, link). Here is the conclusion I draw from his negative arguments about OI in the supervenience post:

Epstein's analysis is careful and convincing in its own terms. Given the modal specification of the meaning of supervenience (as offered by Jaegwon Kim and successors), Epstein makes a powerful case for believing that the social does not supervene upon the individual in a technical and specifiable sense. However, I'm not sure that very much follows from this finding. For researchers within the general school of thought of "actor-centered sociology", their research strategy is likely to remain one that seeks to sort out the mechanisms through which social outcomes of interest are created as a result of the actions and interactions of individuals. If Epstein's arguments are accepted, that implies that we should not couch that research strategy in terms of the idea of supervenience. But this does not invalidate the strategy, or the broad intuition about the relation between the social and the actions of locally situated actors upon which it rests. These are the intuitions that I try to express through the idea of "methodological localism"; link, link. And since I also want to argue for the possibility of "relative explanatory autonomy" for facts at the level of the social (for example, features of an organization; link), I am not too troubled by the failure of a view of the social and individual that denies strict determination of the former by the latter. (link)

It is evident that the concept of microfoundations has a close relationship to ontological individualism. Here are several efforts at reformulating the idea of microfoundations in a more flexible way (link, link). And here are several effort to provide an account of "microfoundations" for practices, norms, and social identities (link, link). This line of thought is intended to provide greater specificity of the recursive nature of "structure-actor-structure" that is expressed in the idea of methodological localism.

Generativity and emergence

Social entities and structures have properties that exercise causal influence over all of us, and over the continuing development of the society in which we live. Schools, corporations, armies, terror networks, transport networks, markets, churches, and cities all fall in this range -- they are social compounds or entities that shape the behavior of the individuals who live and work within them, and they have substantial effects on the broader society as well.

So it is unsurprising that sociologists and ordinary observers alike refer to social structures, organizations, and practices as real components of the social world. Social entities have properties that make a difference, at the individual level and at the social and historical level. Individuals are influenced by the rules and practices of the organizations that employ them; and political movements are influenced by the competition that exists among various religious organizations. Putting the point simply, social entities have real causal properties that influence daily life and the course of history.

What is less clear in the social sciences, and in the areas of philosophy that take an interest in such things, is where those causal properties come from. We know from physics that the causal properties of metallic silver derive from the quantum-level properties of the atoms that make it up. Is something parallel to this true in the social realm as well? Do the causal properties of a corporation derive from the properties of the individual human beings who make it up? Are social properties reducible to individual-level facts?

John Stuart Mill was an early advocate for methodological individualism. In 1843 he wrote his System of Logic: Ratiocinative and Inductive, which contained his view of the relationships that exist between the social world and the world of individual thought and action:

All phenomena of society are phenomena of human nature, generated by the action of outward circumstances upon masses of human beings; and if, therefore, the phenomena of human thought, feeling, and action are subject to fixed laws, the phenomena of society can not but conform to fixed laws. (Book VI, chap. VI, sect. 2)

With this position he set the stage for much of the thinking in social science disciplines like economics and political science, with the philosophical theory of methodological individualism.

About sixty years later Emile Durkheim took the opposite view. He believed that social properties were autonomous with respect to the individuals that underlie them. In 1901 he wrote in the preface to the second edition of Rules of Sociological Method:

Whenever certain elements combine and thereby produce, by the fact of their combination, new phenomena, it is plain that these new phenomena reside not in the original elements but in the totality formed by their union. The living cell contains nothing but mineral particles, as society contains nothing but individuals. Yet it is patently impossible for the phenomena characteristic of life to reside in the atoms of hydrogen, oxygen, carbon, and nitrogen.... Let us apply this principle to sociology. If, as we may say, this synthesis constituting every society yields new phenomena, differing from those which take place in individual consciousness, we must, indeed, admit that these facts reside exclusively in the very society itself which produces them, and not in its parts, i.e., its members.... These new phenomena cannot be reduced to their elements. (preface to the 2nd edition)

These ideas provided the basis for what we can call "methodological holism".

So the issue between Mill and Durkheim is the question of whether the properties of the higher-level social entity can be derived from the properties of the individuals who make up that entity. Mill believed yes, and Durkheim believed no.

This debate persists to the current day, and the positions are both more developed, more nuanced, and more directly relevant to social-science research. Consider first what we might call "generativist social-science modeling". This approach holds that methodological individualism is obviously true, and the central task for the social sciences is to actually perform the reduction of social properties to the actions of individuals by providing computational models that reproduce the social property based on a model of the interacting individuals. These models are called "agent-based models" (ABM). Computational social scientist Joshua Epstein is a recognized leader in this field, and his book Growing Artificial Societies: Social Science From the Bottom Up provides developed examples of ABMs designed to explain well-known social phenomena from the disappearance of the Anasazi in the American Southwest to the occurrence of social unrest. Here is his summary statement of the approach:

To the generativist, explaining macroscopic social regularities, such as norms, spatial patterns, contagion dynamics, or institutions requires that one answer the following question: How could the autonomous local interactions of heterogeneous boundedly rational agents generate the given regularity?Accordingly, to explain macroscopic social patterns, we generate—or “grow”—them in agent models. 

Epstein's memorable aphorism summarizes the field -- "If you didn't grow it, you didn't explain its emergence." A very clear early example of this approach is an agent-based simulation of residential segregation provided by Thomas Schelling in "Dynamic Models of Segregation" (Journal of Mathematical Sociology, 1971; link). The model shows that simple assumptions about the neighborhood-composition preferences of individuals of two groups, combined with the fact that individuals can freely move to locations that satisfy their preferences, leads almost invariably to strongly segregated urban areas.

There is a surface plausibility to the generativist approach, but close inspection of many of these simulations lays bare some important deficiencies. In particular, a social simulation necessarily abstracts mercilessly from the complexities of both the social environment and the dynamics of individual action. It is difficult to represent the workings of higher-level social entities within an agent-based model -- for example, organizations and social practices. And ABMs are not well designed for the task of representing dynamic social features that other researchers on social action take to be fundamental -- for example, the quality of leadership, the content of political messages, or the high degree of path dependence that most real instances of political mobilization reflect.

So if methodological individualism is a poor guide to social research, what is the alternative? The strongest opposition to generativism and reductionism is the view that social properties are "emergent". This means that social ensembles sometimes possess properties that cannot be explained by or reduced to the properties and actions of the participants. For example, it is sometimes thought that a political movement (e.g. Egyptian activism in Tahrir Square in 2011) possessed characteristics that were different in kind from the properties of the individuals and activists who made it up.

There are a few research communities currently advocating for a strong concept of emergence. One is the field of critical realism, a philosophy of science developed by Roy Bhaskar in A Realist Theory of Science (1975) and The Possibility of Naturalism (1979). According to Bhaskar, we need to investigate the social world by looking for the real (though usually unobservable) mechanisms that give rise to social stability and change. Bhaskar is anti-reductionist, and he maintains that social entities have properties that are different in kind from the properties of individuals. In particular, he believes that the social mechanisms that generate the social world are themselves created by the autonomous causal powers of social entities and structures. So attempting to reduce a process of social change to the actions of the individuals who make it up is a useless exercise; these individuals are themselves influenced by the autonomous causal powers of larger social forces.

Another important current line of thought that defends the idea of emergence is the theory of assemblage, drawn from Gilles Deleuze but substantially developed by Manuel DeLanda in A New Philosophy of Society: Assemblage Theory and Social Complexity (2006) and Assemblage Theory (2016). This theory argues for a very different way of conceptualizing the social world. This approach proposes that we should understand complex social entities as a compound of heterogeneous and independent lesser entities, structures, and practices. Social entities do not have "essences". Instead, they are continent and heterogenous ensembles of parts that have been brought together in contingent ways. But crucially, DeLanda maintains that assemblages too have emergent properties that do not derive directly from the properties of the parts. A city has properties that cannot be explained in terms of the properties of its parts. So assemblage theory too is anti-reductionist. 

The claim of emergence too has a superficial appeal. It is clear, for one thing, that social entities have effects that are autonomous with respect to the particular individuals who compose them. And it is clear as well that there are social properties that have no counterpart at the individual level (for example, social cohesion). So there is a weak sense in which it is possible to accept a concept of emergence. However, that weak sense does not rule out either generativity or reduction in principle. It is possible to hold both generativity and weak emergence consistently. And the stronger sense -- that emergent properties are unrelated to and underivable from lower level properties -- seems flatly irrational. What could strongly emergent properties depend on, if not the individuals and social relations that make up these higher-level social entities?

For this reason it is reasonable for social scientists to question both generativity and strong emergence. We are better off avoiding the strong claims of both generativity and emergence, in favor of a more modest social theory. Instead, it is reasonable to advocate for the idea of the relative explanatory autonomy of social properties. This position comes down to a number of related ideas. Social properties are ultimately fixed by the actions and thoughts of socially constituted individuals. Social properties are stable enough to admit of direct investigation. Social properties are relatively autonomous with respect to the specific individuals who occupy positions within these structures. And there is no compulsion to perform reductions of social properties through ABMs or any other kind of derivation. (These are ideas that were first advocated in 1974 by Jerry Fodor in "Special sciences: Or: The disunity of science as a working hypothesis" (link).)

It is interesting to note that a new field of social science, complexity studies, has relevance to both ends of this dichotomy. Joshua Epstein himself is a complexity theorist, dedicated to discovering mathematical methods for understanding complex systems. Other complexity scientists like John Miller and Scott Page are open to the idea of weak emergence in Complex Adaptive Systems: An Introduction to Computational Models of Social Life. Here is how Miller and Page address the idea of emergence in CAS:

The usual notion put forth underlying emergence is that individual, localized behavior aggregates into global behavior that is, in some sense, disconnected from its origins. Such a disconnection implies that, within limits, the details of the local behavior do not matter to the aggregate outcome. (CAS, p. 44)

Herbert Simon is another key contributor to modern complexity studies. Simon believed that complex systems have properties that are irreducible to the properties of their components for pragmatic reasons, including especially computational intractability. It is therefore reasonable, in his estimation, to look at higher-level social properties as being emergent -- even though we believe in principle that these properties are ultimately determined by the properties of the components. Here is his treatment in the third edition of The Sciences of the Artificial - 3rd Edition (1996):

[This amounts to] reductionism in principle even though it is not easy (often not even computationally feasible) to infer rigorously the properties of the whole from knowledge of the properties of the parts. In this pragmatic way, we can build nearly independent theories for each successive level of complexity, but at the same time, build bridging theories that show how each higher level can be accounted for in terms of the elements and relations of the next level down. (172)

The debate over generativity and emergence may seem like an arcane issue that is of interest only to philosophers and the most theoretical of social scientists. But in fact, disputes like this one have real consequences for the conduct of an area of scientific research. Suppose we are interested in the sociology of hate-based social movements. If we begin with the framework of reductionism and generativism, we may be led to focus on the social psychology of adherents and the aggregative processes through which potential followers are recruited into a hate-based movement. If, on the other hand, we believe that social structures and practices have relatively autonomous causal properties, then we will be led to consider the empirical specifics of the workings of organizations like White Citizens Councils, legal structures like the laws that govern hate-based political expressions in Germany and France, and the ways that the Internet may influence the spread of hate-based values and activism. In each of these cases the empirical research is directed in important measure to the concrete workings of the higher-level social institutions that are hypothesized to influence the emergence and shape of hate-based movements. In other words, the sociological research that we conduct is guided in part by the assumptions we make about social ontology and the composition of the social world.

Generativism

There is a seductive appeal to the idea of a "generative social science". Joshua Epstein is one of the main proponents of the idea, most especially in his book, Generative Social Science: Studies in Agent-Based Computational Modeling. The central tool of generative social science is the construction of an agent-based model (link). The ABM is said to demonstrate the way in which an observable social outcome of pattern is generated by the properties and activities of the component parts that make it up -- the actors. The appeal comes from the notion that it is possible to show how complicated or complex outcomes are generated by the properties of the components that make them up. Fix the properties of the components, and you can derive the properties of the composites. Here is Epstein's capsule summary of the approach:

The agent-based computational model -- or artificial society -- is a new scientific instrument. It can powerfully advance a distinctive approach to social science, one for which the term "generative" seems appropriate. I will discuss this term more fully below, but in a strong form, the central idea is this: To the generativist, explaining the emergence of macroscopic societal regularities, such as norms or price equilibria, requires that one answer the following question: 

The Generativist's Question 

*How could the decentralized local interactions of heterogeneous autonomous agents generate the given regularity?  

The agent-based computational model is well-suited to the study of this question, since the following features are characteristic: [heterogeneity, autonomy, explicit space, local interactions, bounded rationality] (5-6)

And a few pages later:

Agent-based models provide computational demonstrations that a given microspecification is in fact sufficient to generate a macrostructure of interest. . . . To the generativist -- concerned with formation dynamics -- it does not suffice to establish that, if deposited in some macroconfiguration, the system will stay there. Rather, the generativist wants an account of the configuration's attainment by a decentralized system of heterogeneous autonomous agents. Thus, the motto of generative social science, if you will, is: If you didn't grow it, you didn't explain its emergence. (8)

Here is how Epstein describes the logic of one of the most extensive examples of generative social science, the attempt to understand the disappearance of Anasazi population in the American Southwest nearly 800 years ago.

The logic of the exercise has been, first, to digitize the true history -- we can now watch it unfold on a digitized map of Longhouse Valley. This data set (what really happened) is the target -- the explanandum. The aim is to develop, in collaboration with anthropologists, microspecifications -- ethnographically plausible rules of agent behavior -- that will generate the true history. The computational challenge, in other words, is to place artificial Anasazi where the true ones were in 80-0 AD and see if -- under the postulated rules -- the simulated evolution matches the true one. Is the microspecification empirically adequate, to use van Fraassen's phrase? (13)

Here is a short video summarizing the ABM developed under these assumptions:

The artificial Anasazi experiment is an interesting one, and one to which the constraints of an agent-based model are particularly well suited. The model follows residence location decision-making based on ground-map environmental information.

But this does not imply that the generativist interpretation is equally applicable as a general approach to explaining important social phenomena.

Note first how restrictive the assumption is of "decentralized local interactions" as a foundation to the model. A large proportion of social activity is neither decentralized nor purely local: the search for muons in an accelerator lab, the advance of an armored division into contested territory, the audit of a large corporation, preparations for a strike by the UAW, the coordination of voices in a large choir, and so on, indefinitely. In all these examples and many more, a crucial part of the collective behavior of the actors is the coordination that occurs through some centralized process -- a command structure, a division of labor, a supervisory system. And by its design, ABMs appear to be incapable of representing these kinds of non-local coordination.

Second, all these simulation models proceed from highly stylized and abstract modeling assumptions. And the outcomes they describe capture at best some suggestive patterns that might be said to be partially descriptive of the outcomes we are interested in. Abstraction is inevitable in any scientific work, of course; but once recognizing that fact, we must abandon the idea that the model demonstrates the "generation" of the empirical phenomenon. Neither premises nor conclusions are fully descriptive of concrete reality; both are approximations and abstractions. And it would be fundamentally implausible to maintain that the modeling assumptions capture all the factors that are causally relevant to the situation. Instead, they represent a particular stylized hypothesis about a few of the causes of the situation in question.  Further, we have good reason to believe that introducing more details at the ground level will sometimes lead to significant alteration of the system-level properties that are generated.

So the idea that an agent-based model of civil unrest could demonstrate that (or how) civil unrest is generated by the states of discontent and fear experienced by various actors is fundamentally ill-conceived. If the unrest is generated by anything, it is generated by the full set of causal and dynamic properties of the set of actors -- not the abstract stylized list of properties. And other posts have made the point that civil unrest or rebellion is rarely purely local in its origin; rather, there are important coordinating non-local structures (organizations) that influence mobilization and spread of rebellious collective action. Further, the fact that the ABM "generates" some macro characteristics that may seem empirically similar to the observed phenomenon is suggestive, but far from a demonstration that the model characteristics suffice to determine some aspect of the macro phenomenon. Finally, the assumption of decentralized and local decision-making is unfounded for civil unrest, given the important role that collective actors and organizations play in the success or failure of social mobilizations around grievances (link).

The point here is not that the generativist approach is invalid as a way of exploring one particular set of social dynamics (the logic of decentralized local decision-makers with assigned behavioral rules). On the contrary, this approach does indeed provide valuable insights into some social processes. The error is one of over-generalization -- imagining that this approach will suffice to serve as a basis for analysis of all social phenomena. In a way the critique here is exactly parallel to that which I posed to analytical sociology in an earlier post. In both cases the problem is one of asserting priority for one specific approach to social explanation over a number of other equally important but non-equivalent approaches.

Patrick Grim et al provide an interesting approach to the epistemics of models and simulations in "How simulations fail" (link). Grim and his colleagues emphasize the heuristic and exploratory role that simulations generally play in probing the dynamics of various kinds of social phenomena.

Menon and Callender on the physics of phase transitions

In an earlier post I considered the topic of phase transitions as a possible source of emergent phenomena (link). I argued there that phase transitions are indeed interesting, but don't raise a serious problem of strong emergence. Tarun Menon considers this issue in substantial detail in the chapter he co-authored with Craig Callender in The Oxford Handbook of Philosophy of Physics, "Turn and face the strange ... ch-ch-changes: Philosophical questions raised by phase transitions" (link). Menon and Callender provide a very careful and logical account of three ways of approaching the physics of phase transitions within physics and three versions of emergence (conceptual, explanatory, ontological). The piece is technical but very interesting, with a somewhat deflating conclusion (if you are a fan of emergence):

We have found that when one clarifies concepts and digs into the details, with respect to standard textbook statistical mechanics, phase transitions are best thought of as conceptually novel, but not ontologically or explanatorily irreducible. 

Menon and Callendar review three approaches to the phenomenon of phase transition offered by physics: classical thermodynamics, statistical mechanics, and renormalization group theory. Thermodynamics describes the behavior of materials (gases, liquids, and solids) at the macro level; and statistical mechanics and renormalization group theory are theories of the micro states of materials intended to allow derivation of the macro behavior of the materials from statistical properties of the micro states. They describe this relationship in these terms:

Statistical mechanics is the theory that applies probability theory to the microscopic degrees of freedom of a system in order to explain its macroscopic behavior. The tools of statistical mechanics have been extremely successful in explaining a number of thermodynamic phenomena, but it turned out to be particularly difficult to apply the theory to the study of phase transitions. (193)

Here is the mathematical definition of phase transition that they provide:

Mathematically, phase transitions are represented by nonanalyticities or singularities in a thermodynamic potential. A singularity is a point at which the potential is not infinitely differentiable, so at a phase transition some derivative of the thermo­dynamic potential changes discontinuously. (191)

And they offer this definition:

(Def 1) An equilibrium phase transition is a nonanalyticity in the free energy. (194)

Here is their description of how the renormalization group theory works:

To explain the method, we return to our stalwart Ising model. Suppose we coarse­grain a 2­D Ising model by replacing 3 × 3 blocks of spins with a single spin pointing in the same direction as the majority in the original block. This gives us a new Ising system with a longer distance between lattice sites, and possibly a different coupling strength. You could look at this coarse­graining procedure as a transformation in the Hamiltonian describing the system. Since the Hamiltonian is characterized by the coupling strength, we can also describe the coarse­graining as a transformation in the coupling parameter. Let K be the coupling strength of the original system and R be the relevant transformation. The new coupling strength is K′ = RK. This coarse­graining procedure could be iterated, producing a sequence of coupling parameters, each related to the previous one by the transformation R. The transformation defines a flow on parameter space. (195)

Renormalization group theory, then, is essentially the mathematical basis of coarse-graining analysis (link).

The key difficulty that has been used to ground arguments about strong emergence of phase transitions is now apparent: there seems to be a logical disjunction between the resources of statistical mechanics and the findings of thermodynamics. In theory physicists would like to hold that statistical mechanics provides the micro-level representation of the phenomena described by thermodynamics; or in other words, that thermodynamic facts can be reduced to derivations from statistical mechanics. However, the definition of a phase transition above specifies that the phenomena display "nonanalyticities" -- instantaneous and discontinuous changes of state. It is easily demonstrated that the equations used in statistical mechanics do not display nonanalyticities; change may be abrupt, but it is not discontinuous, and the equations are infinitely differentiable. So if phase transitions are points of nonanalyticity, and statistical mechanics does not admit of nonanalytic equations, then it would appear that thermodynamics is not derivable from statistical mechanics. Similar reasoning applies to renormalization group theory.

This problem was solved within statistical mechanics by admitting of infinitely many bodies within the system that is represented (or alternatively, admitting of infinitely compressed volumes of bodies); but neither of these assumptions of infinity is realistic of the material world.

So are phase transitions "emergent" phenomena in either a weak sense or a strong sense, relative to the micro-states of the material in question? The strongest sense of emergence is what Menon and Callender call ontological irreducibility.

Ontological irreducibility involves a very strong failure of reduction, and if any phenomenon deserves to be called emergent, it is one whose description is ontologically irreducible to any theory of its parts. Batterman argues that phase transitions are emergent in this sense (Batterman 2005). It is not just that we do not know of an adequate statistical mechanical account of them, we cannot construct such an account. Phase transitions, according to this view, are cases of genuine physical discontinuities. (215)

The possibility that phase transitions are ontologically emergent at the level of thermodynamics is raised by the point about the mathematical characteristics of the equations that constitute the statistical mechanics description of the micro level -- the infinite differentiability of those equations. But Menon and Callender give a compelling reason for thinking this is misleading. They believe that phase transitions constitute a conceptual novelty with respect to the resources of statistical mechanics -- phase transitions do not correspond to natural kinds at the level of the micro-constitution of the material. But they argue that this does not establish that the phenomena cannot be explained or derived from a micro-level description. So phase transitions are not emergent according to the explanatory or ontological understandings of that idea.

The nub of the issue comes down to how we construe the idealization of statistical mechanics that assumes that a material consists of an infinite number of elements. This is plainly untrue of any real system (gas, liquid, or solid). The fact that there are boundaries implies that important thermodynamic properties are not "extensive" with volume: twice the volume leads to twice the entropy. But the way in which the finitude of a volume of material affects its behavior is through the effects of novel behaviors at the edges of the volume. And in many instances these effects are small relative to the behavior of the whole, if the volume is large enough.

Does this fact imply that there is a great mystery about extensivity, that extensivity is truly emergent, that thermodynamics does not reduce to finite N statistical mechanics? We suggest that on any reasonably uncontentious way of defining these terms, the answer is no. We know exactly what is happening here. Just as the second law of thermodynamics is no longer strict when we go to the microlevel, neither is the concept of extensivity. (201-202)

There is an important idealization on the thermodynamic description as well -- the notion that several specific kinds of changes are instantaneous or discontinuous. But this assumption can also be seen as an idealization, corresponding to a physical system that is undergoing changes at different rates under different environmental conditions. What thermodynamics describes as an instantaneous change from liquid to gas may be better understood as a rapid process of change at the molar level which can be traced through in a continuous way.

(The fact that some systems are coarse-grained has an interesting implication for this set of issues (link). The interesting implication is that while it is generally true that the micro states in such a system entail the macro states, the reverse is not true: we cannot infer from a given macro state to the exact underlying micro state. Rather, many possible micro states correspond to a given macro state.)

The conclusion they reach is worth quoting:

Phase transitions are an important instance of putatively emergent behavior. Unlike many things claimed emergent by philosophers (e.g., tables and chairs), the alleged emergence of phase transitions stems from both philosophical and scientific arguments. Here we have focused on the case for emergence built from physics. We have found that when one clarifies concepts and digs into the details, with respect to standard textbook statistical mechanics, phase transitions are best thought of as conceptually novel, but not ontologically or explanatorily irreducible. And if one goes past textbook statistical mechanics, then an argument can be made that they are not even conceptually novel. In the case of renormalization group theory, consideration of infinite systems and their singular behavior provides a central theoretical tool, but this is compatible with an explanatory reduction. Phase transitions may be “emergent” in some sense of this protean term, but not in a sense that is incompatible with the reductionist project broadly construed. (222)

Or in other words, Menon and Callender refute one of the most technically compelling interpretations of ontological emergence in physical systems. They show that the phenomena of phase transitions as described by classical thermodynamics are compatible with being reduced to the dynamics of individual elements at the micro-level, so phase transitions are not ontologically emergent.

Are these arguments relevant in any way to debates about emergence in social system dynamics? The direct relevance is limited, since these arguments depend entirely on the mathematical properties of the ways in which the micro-level of physical systems are characterized (statistical mechanics). But the more general lesson does in fact seem relevant: rather than simply postulating that certain social characteristics are ontologically emergent relative to the actors that make them up, we would be better advised to look for the local-level processes that act to bring about surprising transitions at critical points (for example, the shift in a flock of birds from random flight to a swarm in a few seconds).

Coarse-graining of complex systems

The question of the relationship between micro-level and macro-level is just as important in physics as it is in sociology. Is it possible to derive the macro-states of a system from information about the micro-states of the system? It turns out that there are some surprising aspects of the relationship between micro and macro that physical systems display. The mathematical technique of "coarse-graining" represents an interesting wrinkle on this question. So what is coarse-graining? Fundamentally it is the idea that we can replace micro-level specifics with local-level averages, without reducing our ability to calculate macro-level dynamics of behavior of a system.

A 2004 article by Israeli and Goldenfeld, "Coarse-graining of cellular automata, emergence, and the predictability of complex systems" (link) provides a brief description of the method of coarse-graining. (Here is a Wolfram demonstration of the way that coarse graining works in the field of cellular automata; link.) Israeli and Goldenfeld also provide physical examples of phenomena with what they refer to as emergent characteristics. Let's see what this approach adds to the topic of emergence and reduction. Here is the abstract of their paper:

We study the predictability of emergent phenomena in complex systems. Using nearest neighbor, one-dimensional Cellular Automata (CA) as an example, we show how to construct local coarse-grained descriptions of CA in all classes of Wolfram's classification. The resulting coarse-grained CA that we construct are capable of emulating the large-scale behavior of the original systems without accounting for small-scale details. Several CA that can be coarse-grained by this construction are known to be universal Turing machines; they can emulate any CA or other computing devices and are therefore undecidable. We thus show that because in practice one only seeks coarse-grained information, complex physical systems can be predictable and even decidable at some level of description. The renormalization group flows that we construct induce a hierarchy of CA rules. This hierarchy agrees well apparent rule complexity and is therefore a good candidate for a complexity measure and a classification method. Finally we argue that the large scale dynamics of CA can be very simple, at least when measured by the Kolmogorov complexity of the large scale update rule, and moreover exhibits a novel scaling law. We show that because of this large-scale simplicity, the probability of finding a coarse-grained description of CA approaches unity as one goes to increasingly coarser scales. We interpret this large scale simplicity as a pattern formation mechanism in which large scale patterns are forced upon the system by the simplicity of the rules that govern the large scale dynamics.

This paragraph involves several interesting ideas. One is that the micro-level details do not matter to the macro outcome (italics above). Another related idea is that macro-level patterns are (sometimes) forced by the "rules that govern the large scale dynamics" -- rather than by the micro-level states.

Coarse-graining methodology is a family of computational techniques that permits "averaging" of values (intensities) from the micro-level to a higher level of organization. The computational models developed here were primarily applied to the properties of heterogeneous materials, large molecules, and other physical systems. For example, consider a two-dimensional array of iron atoms as a grid with randomly distributed magnetic orientations (up, down). A coarse-grained description of this system would be constructed by taking each 3x3 square of the grid and assigning it the up-down value corresponding to the majority of atoms in the grid. Now the information about nine atoms has been reduced to a single piece of information for the 3x3 grid. Analogously, we might consider a city of Democrats and Republicans. Suppose we know the affiliation of each household on every street. We might "coarse-grain" this information by replacing the household-level data with the majority representation of 3x3 grids of households. We might take another step of aggregation by considering 3x3 grids of grids, and representing the larger composite by the majority value of the component grids.

How does the methodology of coarse-graining interact with other inter-level questions we have considered elsewhere in Understanding Society (emergence, generativity, supervenience)? Israeli and Goldenfeld connect their work to the idea of emergence in complex systems. Here is how they describe emergence:

Emergent properties are those which arise spontaneously from the collective dynamics of a large assemblage of interacting parts. A basic question one asks in this context is how to derive and predict the emergent properties from the behavior of the individual parts. In other words, the central issue is how to extract large-scale, global properties from the underlying or microscopic degrees of freedom. (1)

Note that this is the weak form of emergence (link); Israeli and Goldenfeld explicitly postulate that the higher-level properties can be derived ("extracted") from the micro level properties of the system. So the calculations associated with coarse-graining do not imply that there are system-level properties that are non-derivable from the micro-level of the system; or in other words, the success of coarse-graining methods does not support the idea that physical systems possess strongly emergent properties.

Does the success of coarse-graining for some systems have implications for supervenience? If the states of S can be derived from a coarse-grained description C of M (the underlying micro-level), does this imply that S does not supervene upon M? It does not. A coarse-grained description corresponds to multiple distinct micro-states, so there is a many-one relationship between M and C. But this is consistent with the fundamental requirement of supervenience: no difference at the higher level without some difference at the micro level. So supervenience is consistent with the facts of successful coarse-graining of complex systems.

What coarse-graining is inconsistent with is the idea that we need exact information about M in order to explain or predict S. Instead, we can eliminate a lot of information about M by replacing M with C, and still do a perfectly satisfactory job of explaining and predicting S.

There is an intellectual wrinkle in the Israeli and Goldenfeld article that I haven't yet addressed here. This is their connection between complex physical systems and cellular automata. A cellular automaton is a simulation governed by simple algorithms governing the behavior of each cell within the simulation. The game of Life is an example of a cellular automaton (link). Here is what they say about the connection between physical systems and their simulations as a system of algorithms:

The problem of predicting emergent properties is most severe in systems which are modelled or described by undecidable mathematical algorithms[1, 2]. For such systems there exists no computationally efficient way of predicting their long time evolution. In order to know the system’s state after (e.g.) one million time steps one must evolve the system a million time steps or perform a computation of equivalent complexity. Wolfram has termed such systems computationally irreducible and suggested that their existence in nature is at the root of our apparent inability to model and understand complex systems [1, 3, 4, 5]. (1)

Suppose we are interested in simulating the physical process through which a pot of boiling water undergoes sudden turbulence shortly before 100 degrees C (the transition point between water and steam). There seem to be two large alternatives raised by Israeli and Goldenfeld: there may be a set of thermodynamic processes that permit derivation of the turbulence directly from the physical parameters present during the short interval of time; or it may be that the only way of deriving the turbulence phenomenon is to provide a molecule-level simulation based on the fundamental laws (algorithms) that govern the molecules. If the latter is the case, then simulating the process will prove computationally impossible.

Here is an extension of this approach in an article by Krzysztof Magiera and Witold Dzwinel, "Novel Algorithm for Coarse-Graining of Cellular Automata" (link). They describe "coarse-graining" in their abstract in these terms:

The coarse-graining is an approximation procedure widely used for simplification of mathematical and numerical models of multiscale systems. It reduces superfluous – microscopic – degrees of freedom. Israeli and Goldenfeld demonstrated in [1,2] that the coarse-graining can be employed for elementary cellular automata (CA), producing interesting interdependences between them. However, extending their investigation on more complex CA rules appeared to be impossible due to the high computational complexity of the coarse-graining algorithm. We demonstrate here that this complexity can be substantially decreased. It allows for scrutinizing much broader class of cellular automata in terms of their coarse graining. By using our algorithm we found out that the ratio of the numbers of elementary CAs having coarse grained representation to “degenerate” – irreducible – cellular automata, strongly increases with increasing the “grain” size of the approximation procedure. This rises principal questions about the formal limits in modeling of realistic multiscale systems.

Here K&D seem to be expressing the view that the approach to coarse-graining as a technique for simplifying the expected behavior of a complex system offered by Israeli and Goldenfeld will fail in the case of more extensive and complex systems (perhaps including the pre-boil turbulence example mentioned above).

I am not sure whether these debates have relevance for the modeling of social phenomena. Recall my earlier discussion of the modeling of rebellion using agent-based modeling simulations (link, link, link). These models work from the unit level -- the level of the individuals who interact with each other. A coarse-graining approach would perhaps replace the individual-level description with a set of groups with homogeneous properties, and then attempt to model the likelihood of an outbreak of rebellion based on the coarse-grained level of description. Would this be feasible?

Phase transitions and emergence

Image: Phase diagram of water, Solé. Phase Transitions, 4

I've proposed to understand the concepts of emergence and generativeness as being symmetrical (link). Generative higher-level properties are those that those that can be calculated or inferred based on information about the properties and states of the micro-components. Emergent properties are properties of an ensemble that have substantially different dynamics and characteristics from those of the components. So emergent properties may seem to be non-generative properties. Further, I understand the idea of emergence in a weak and a strong sense: weakly emergent properties of an ensemble are properties that cannot be derived from the characteristics of the components given the limits of observation or computation; and strongly emergent properties are ones that cannot be derived in principle from full knowledge of the properties and states of the components. They must be understood in their own terms.

Conversations with Tarun Menon at the Tata Institute for Social Sciences in Mumbai were very helpful in allowing me to broaden somewhat the way I understand emergence in physical systems. So here I'd like to consider some additional complications for the theory of emergence coming from one specific physical finding, the mathematics of phase transitions. 

Complexity scientists have spent a lot of effort on understanding the properties of complex systems using a different concept, the idea of a phase transition. The transition from liquid water to steam as temperature increases is an example; the transition happens abruptly as the system approaches the critical value of the phase parameter -- 100 degrees centigrade at constant pressure of one atmosphere, in the case of liquid-gas transition. 

Richard Solé presents the current state of complexity theory with respect to the phenomenon of phase transition in Phase Transitions. Here is how he characterizes the core idea:

In the previous sections we used the term critical point to describe the presence of a very narrow transition domain separating two well-defined phases, which are characterized by distinct macroscopic properties that are ultimately linked to changes in the nature of microscopic interactions among the basic units. A critical phase transition is characterized by some order parameter φ( μ) that depends on some external control parameter μ (such as temperature) that can be continuously varied. In critical transitions, φ varies continuously at μc (where it takes a zero value) but the derivatives of φ are discontinuous at criticality. For the so-called first-order transitions (such as the water-ice phase change) there is a discontinuous jump in φ at the critical point. (10)

So what is the connection between "emergent phenomena" and systems that undergo phase transitions? One possible connection is this: when a system undergoes a phase transition, its micro-components get rapidly reconfigured into a qualitatively different macro-structure. And yet the components themselves are unchanged.  So one might be impressed with the fact that the pre- and post-macro states correspond to very close to the same configurations of micro-states. The steaminess of the water molecules is triggered by an external parameter -- change in temperature (or possibly pressure), and their characteristics around the critical point are very similar (their mean kinetic energy is approximately equal before and after transition). The diagram above represents the physical realities of water molecules in the three phase states. 

Solé and other complexity theorists see this "phase-transition" phenomenon in a wide range of systems, including simple physical systems but also biological and social systems as well. Solé offers the phenomenon of flocking as an example. We might consider whether the phenomenon of ethnic violence is a phase transition from a mixed but non-aggressive population of individuals to occasional abrupt outbursts of widespread conflict (link).

The disanalogy here is the fact that "unrest" is not a new equilibrium phase of the substrate of dispersed individuals; rather, it is an occasional abnormal state of brief duration. It is as if water sometimes spontaneously transitioned to steam and then returned to the liquid phase. Solé treats "percolation" phenomena later in the book, and rebellion seems more plausibly treated as a percolation process. Solé treats forest fire this way. But the representation works equally for any process based on contiguous contagion.

What seems to be involved here is a conclusion that is a little bit different from standard ideas about emergent phenomena. The point seems to be that for a certain class of systems, these systems have dynamic characteristics that are formal and abstract and do not require that we understand the micro mechanisms upon which they rest at all. It is enough to know that system S is formally similar to a two-dimensional array of magnetized atoms (the "Ising model"); then we can infer that phase-transition behavior of the system will have specific mathematical properties. This might be summarized with the slogan, "system properties do not require derivation from micro dynamics." Or in other words: systems have properties that don't depend upon the specifics of the individual components -- a statement that is strongly parallel to but distinct from the definition of emergence mentioned above. It is distinct, because the approach leaves it entirely open that the system properties are generated by the dynamics of the components.

This idea is fundamental to Solé's analysis, when he argues that it is possible to understand phase transitions without regard to the particular micro-level mechanisms:

Although it might seem very difficult to design a microscopic model able to provide insight into how phase transitions occur, it turns out that great insight has been achieved by using extremely simplified models of reality. (10)

Here is how Solé treats swarm behavior as a possible instance of phase transition.

In social insects, while colonies behave in complex ways, the capacities of individuals are relatively limited. But then, how do social insects reach such remarkable ends? The answer comes to a large extent from self-organization: insect societies share basic dynamic properties with other complex systems. (157)

Intuitively the idea is that a collection of birds, ants, or bees may be in a state of random movement with respect to each other; and then as some variable changes the ensemble snaps into a coordinated "swarm" of flight or movement. Unfortunately he does not provide a mathematical example illustrating swarm behavior; the closest example he provides has to do with patterns of intense activity and slack activity over time in small to medium colonies of ants. This periodicity is related to density. Mark Millonas attempted such an account of swarming in a Santa Fe Institute paper in 1993, "Swarms, Phase Transitions, and Collective Intelligence; and a Nonequilibrium Statistical Field Theory of Swarms and Other Spatially Extended Complex Systems " (link).

This work is interesting, but I am not sure that it sheds new light on the topic of emergence per se. Fundamentally it demonstrates that the aggregation dynamics of complex systems are often non-linear and amenable to formal mathematical modeling. As a critical variable changes a qualitatively new macro-property "emerges" from the ensemble of micro-components from which it is composed. This approach is consistent with the generativity view -- the new property is generated by the interactions of the micro-components during an interval of change in critical variables. But it also maintains that systems undergoing phase transitions can be studied using a mathematical framework that abstracts from the physical properties of those micro-components. This is the point of the series of differential equation models that Solé provides. Once we have determined that a particular system has formal properties satisfying the assumptions of the DE model, we can then attempt to measure the critical parameters and derive the evolution of the system without further information about particular mechanisms at the micro-level.

Are emergence and microfoundations contraries?

image: micro-structure of a nanomaterial (link)

Are there strong logical relationships among the ideas of emergence, microfoundations, generative dependency, and supervenience? It appears that there are.

The diagram represents the social world as a laminated set of layers of entities, processes, powers, and laws. Entities at L2 are composed of or caused by some set of entities and forces at L1. Likewise L3 and L4. Arrows indicate microfoundations for L2 facts based on L1 facts. Diamond-tipped arrows indicate the relation of generative dependence from one level to another. Square-tipped lines indicate the presence of strongly emergent facts at the higher level relative to the lower level. The solid line (L4) represents the possibility of a level of social fact that is not generatively dependent upon lower levels. The vertical ellipse at the right indicates the possibility of microfoundations narratives involving elements at different levels of the social world (individual and organizational, for example).

We might think of these levels as "individuals," "organization, value communities, social networks," "large aggregate institutions like states," etc.

This is only one way of trying to represent the structure of the social world. The notion of a "flat" ontology was considered in an earlier post (link). Another structure that is excluded by this diagram is one in which there is multi-directional causation across levels, both upwards and downwards. For example, the diagram excludes the possibility that L3 entities have causal powers that are original and independent from the powers of L2 or L1 entities. The laminated view described here is the assumption built into debates about microfoundations, supervenience, and emergence. It reflects the language of micro, meso, and macro levels of social action and organization.

Here are definitions for several of the primary concepts.

• Microfoundations of facts in L2 based on facts in L1 : accounts of the causal pathways through which entities, processes, powers, and laws of L1 bring about specific outcomes in L2. Microfoundations are small causal theories linking lower-level entities to higher-level outcomes.
• Generative dependence of L2 upon L1: the entities, processes, powers, and laws of L2 are generated by the properties of level L1 and nothing else. Alternatively, the entities, processes, powers, and laws of A suffice to generate all the properties of L2. A full theory of L1 suffices to derive the entities, processes, powers, and laws of L2.
• Reducibility of y to x : it is possible to provide a theoretical or formal derivation of the properties of y based solely on facts about x.
• Strong emergence of properties in L2 with respect to the properties of L1: L2 possesses some properties that do not depend wholly upon the properties of L1.
• Weak emergence of properties in L2 with respect to the properties of L1: L2 possesses some properties for which we cannot (now or in the future) provide derivations based wholly upon the properties of L1.
• Supervenience of L2 with respect to properties of L1: all the properties of L2 depend strictly upon the properties of L1 and nothing else.

We also can make an effort to define some of these concepts more formally in terms of the diagram.

Consider these statements about facts at levels L1 and L2:

1. UM: all facts at L2 possess microfoundations at L1. 
2. XM: some facts at L2 possess inferred but unknown microfoundations at L1. 
3. SM: some facts at L2 do not possess any microfoundations at L1. 
4. SE: L2 is strongly emergent from L1. 
5. WE: L2 is weakly emergent from L1. 
6. GD: L2 is generatively dependent upon L1. 
7. R: L2 is reducible to L1. 
8. D: L2 is determined by L1. 
9. SS: L2 supervenes upon L1. 

Here are some of the logical relations that appear to exist among these statements.

1. UM => GD 
2. UM => ~SE 
3. XM => WE 
4. SE => ~UM 
5. SE => ~GD 
6. GD => R 
7. GD => D 
8. SM => SE 
9. UM => SS 
10. GD => SS 

On this analysis, the question of the availability of microfoundations for social facts can be understood to be central to all the other issues: reducibility, emergence, generativity, and supervenience. There are several positions that we can take with respect to the availability of microfoundations for higher-level social facts.

1. If we have convincing reason to believe that all social facts possess microfoundations at a lower level (known or unknown) then we know that the social world supervenes upon the micro-level; strong emergence is ruled out; weak emergence is true only so long as some microfoundations remain unknown; and higher-level social facts are generatively dependent upon the micro-level.   
2. If we take a pragmatic view of the social sciences and conclude that any given stage of knowledge provides information about only a subset of possible microfoundations for higher-level facts, then we are at liberty to take the view that each level of social ontology is at least weakly emergent from lower levels -- basically, the point of view advocated under the banner of "relative explanatory autonomy" (link). This also appears to be roughly the position taken by Herbert Simon (link). 
3. If we believe that it is impossible in principle to fully specify the microfoundations of all social facts, then weak emergence is true; supervenience is false; and generativity is false. (For example, we might believe this to be true because of the difficulty of modeling and calculating a sufficiently large and complex domain of units.) This is the situation that Fodor believes to be the case for many of the special sciences. 
4. If we have reason to believe that some higher-level facts simply do not possess microfoundations at a lower level, then strong emergence is true; the social world is not generatively dependent upon the micro-world; and the social world does not supervene upon the micro-world. 

In other words, it appears that each of the concepts of supervenience, reduction, emergence, and generative dependence can be defined in terms of the availability or inavailability of microfoundations for some or all of the facts at a higher level based on facts at the lower level. Strong emergence and generative dependence turn out to be logical contraries (witness the final two definitions above).