How not to create a classification system

Does social science require systems of classification?

From Jorge Luis Borges' description of a fictional Chinese classification of animals:

“These ambiguities, redundancies and deficiencies remind us of those which doctor Franz Kuhn attributes to a certain Chinese encyclopaedia entitled 'Celestial Empire of benevolent Knowledge'. In its remote pages it is written that the animals are divided into: (a) belonging to the emperor, (b) embalmed, (c) tame, (d) sucking pigs, (e) sirens, (f) fabulous, (g) stray dogs, (h) included in the present classification, (i) frenzied, (j) innumerable, (k) drawn with a very fine camelhair brush, (l) et cetera, (m) having just broken the water pitcher, (n) that from a long way off look like flies.” Jorge Luis Borges, Other Inquisitions: 1937-1952

The lesson? That a system of classification needs to rest upon a coherent, logical assessment of the causal and structural properties of the things being classified, so that the classification may lead us to a better understanding of the range of phenomena encompassed by the system.

The social science disciplines

The social sciences consist of a variety of disciplines, subject areas, and methods, and there is no reason to expect that these disciplines will eventually add up to a single unified theory of society. Political science, sociology, history, anthropology, economics, geography, and area studies all provide their own, largely independent, definitions of scope, research agenda, and research methods. And there is no grand plan according to which the disciplinary definitions jointly capture all that is of scientific interest about the social.

History rather than logic explains the particular configuration of social science disciplines that we now face. The major social science disciplines have grown up in the past century and a half by creating stylized answers to these topic areas: the “political” concerns institutions of coercion and governance; the “economic” has to do with production, exchange, and distribution of goods and services; the “anthropological” has to do with the cultures, values, and practices through which individuals and groups conduct their local lives. Area studies are defined according to a different axis; Asian studies or Latin American studies demand that we cut the social differently: not from the point of view of social domains, but from the point of view of geographical complexes of related social, cultural, economic, political, and normative regimes.

At the same time, we as “users” of the results of social inquiry have no inherent interest in the intra- and inter-disciplinary debates that have led to the constitution of the disciplines of the social sciences as they currently exist. The social world does not come to us labeled as “political,” “economic,” or “ethnographic.” We ordinary citizens have questions that cut across these boundaries recklessly: Why does the US state so commonly ignore the needs of poor people? Why are Indonesian rice farmers reluctant to make use of HYV rice strains? Why did the hi-tech bubble occur in the American economy in the 1990s? How do police departments succeed in recruiting good potential officers? When is the practice of charitable giving most likely to thrive or falter? Why did the Chinese Communist revolution occur? Why did it succeed? Note what a mixture of topics, human interactions, and methodologies is invoked by this collection of questions. Some of these queries raise the question of why individuals behaved as they did; some focus on group action, while others single out individual choices; some have to do with the institutions within which individuals live; some suggest turning to ethnography, comparative economics, or political science; and so forth.

The upshot is this: Users of the social sciences have a different way of parsing “the social” than is found in academic social science. We are interested in human agency and behavior — individual and collective. We are interested in the ways social relations and institutions work, and how they affect the behavior and choices of the individuals who operate within them. We are interested in how large agglomerations of human activity work — how they emerge, how they behave over time, and how they go wrong (cities, states, corporations, networks of friends, …), and we are interested in the dynamics of face-to-face social interaction.

This suggests that new thinking about the social sciences needs to start with the idea of acquiring a strong commitment to interdisciplinary study of common social topics.

(Here are several other posts on "social science disciplines" (tag).)

What kind of knowledge can philosophy offer?

Let us say that knowledge is "a set of true statements based on compelling reasons." (This is the same as the familiar definition of knowledge as "justified true belief".) Philosophers offer a variety of claims, and they offer arguments for their positions. But do they offer knowledge about anything? Is it possible to say that "Philosophical statement P is true", or only that "Philosophical statement P is supported by strong and compelling reasons"? Are philosophical topics within the domain of things concerning which we can have knowledge at all?

We know what kind of knowledge empirical science provides: factual knowledge about the world and inferences or theories supported by empirical observation. We also know what kind of knowledge is offered by mathematics and logic : deductive knowledge derived from a set of axioms in one or another of the fields of mathematics. And we can specify the nature of the knowledge provided by linguistics and semantics: expository knowledge of the meanings of various words and phrases in ordinary usage. Putting these areas of inquiry very crudely, we might summarize them as "inductive knowledge," "deductive knowledge," and "semantic knowledge." But what does philosophy add to our knowledge and understanding of human experience and knowledge?

Notice that knowledge and truth are interrelated. Truth, in turn, has to do with correspondence and reference. Statements refer to things and properties; statements are true or false depending on whether the things to which the statement refers in fact possesses the properties attributed to it. So a statement cannot be a piece of knowledge if it does not permit of correspondence to some independent set of facts.

Now consider the kinds of reasoning and statements that occur within philosophy.

First, philosophy offers assertions based on rigorous analysis of concepts and conceptual relationships. This exercise looks a bit like the linguistics/ semantics option above; but philosophers offer "value-added" by providing rational reconstruction of the concepts they consider. They reconstruct and improve upon the concepts of everyday language.

Second, philosophy may provide constructive analysis and further development of the methods and conceptual systems of various disciplines, including the empirical sciences. Here the philosopher is purporting to offer substantial findings that will improve upon the epistemic characteristics of existing scientific practices. Crudely, a deeper understanding of the logic of evolutionary explanation and the mathematics of natural selection is a philosophical effort, and it is a substantive contribution to the philosophy of biology and to biological theory. Here, we must ask whether there is a credible basis for the contribution. In what respect does the philosopher possess a distinctive basis for providing rational assessment and recommendations about the structure and practice of empirical science? Typically, philosophers would respond by saying that philosophical argumentation about the methods of science gains credibility through the understanding of rationality and reasoning that philosophers possess.

Third, philosophy can construct ethical theories that possess some degree of justification, beyond ordinary moral opinions. When John Rawls asks, "What is justice?", he begins with some ordinary associations with the concept, but then builds out a theory of justice that goes far beyond ordinary language. And he offers a range of philosophical justifications in support of his theory of "justice as fairness".

A major problem in answering the key question, what is the nature of philosophical knowledge, is the fact that we ordinarily divide knowledge into "empirical/contingent" and "formal/necessary" (or what Quine calls the "analytic/synthetic" distinction). But philosophers seem to advance their theories in a way that implies they are neither empirical nor purely formal. Kant puts this point in the form of a category of knowledge that is "synthetic a priori" -- that is, knowledge that is based on purely philosophical reasoning but that is nonetheless not purely formal. (Kant's claim that the statement "reality is spatially organized in Euclidean three-dimensional space" is true synthetic a priori asserts that the statement is necessarily true but is not a truth of logic.)

One possible stance that we might take is to restrict the concept of "knowledge" to statements about empirical states of affairs (where truth and falsity are possible), but to then postulate other categories of belief that have a degree of rational credibility without correspondence to the empirical world. On this approach, empirical statements, suitably supported, can constitute knowledge of the empirical world. Philosophical statements -- statements about scientific method, the nature of beauty, or ethical principles -- can be offered with rational justification but do not have the referential structure that would permit them to be "true" and do not count as knowledge. If we take this approach, then we may be justified in accepting the proposition "Justice is fairness," but we would not be justified in thinking it is "true." And therefore the statement is not an example of "knowledge".

This approach sounds a bit parallel to the position of early twentieth-century logical positivism, which maintained the verificationist theory of meaning -- the meaning of a sentence is the set of conditions that establish its truth or falsity -- with the result that theology, philosophy, or ethics are strictly speaking meaningless. However, the two views are not the same, because this view permits that philosophical discourse is meaningful and rational; it simply denies that the statements that philosophers make either correspond or fail to correspond with facts about the world.

The conclusion to this line of thought is somewhat startling: philosophy does not offer knowledge at all. However, it does provide opinions and statements that are founded on good reasons, and we have rational grounds for believing these statements.

What is a social structure?

Are there such things as "social structures"? In what do they consist? What sorts of social powers do they exercise? This is a question I considered in greater detail in New Contributions to the Philosophy of History and Varieties of Social Explanation. But it is worth taking up here as well.

Here are a few books that have made useful contributions to the current understanding of the causal powers of social structures.



Consider a few candidates for social structures: the global trading system, the Federal government, the Chinese peasantry of the 1930s, the English class system, the Indian marriage system, race in the United States, the city of Chicago. Are these items examples of "social structures"?

What are the central assumptions we make in designating something as a social structure? (Note that the term "social structure" can be used in at least two important senses: first, as a causally operative institutional complex (the state or the market as causal social structures), and second, as a description of facets of the organization of society (demographic structure, urban-rural structure, structure of race and ethnicity, income structure). Here I will focus on the first sense of the term.)

Several ideas appear to be core features in our ordinary understanding of this concept. A social structure consists of rules, institutions, and practices. A social structure is socially embodied in the actions, thoughts, beliefs, and durable dispositions of individual human beings. A social structure is effective in organizing behavior of large numbers of actors. A structure is coercive of individual and group behavior. A social structure assigns roles and powers to individual actors. A social structure often has distributive consequences for individuals and groups. A social structure is geographically dispersed. Social structures can cause social outcomes involving both persistence and change.

We might try to reduce these intuitions to a definition: a social structure is a system of geographically dispersed rules and practices that influence the actions and outcomes of large numbers of social actors.

Now back to our original question: do such things exist? Before proceeding to a answer, a few points are evident. Any social entity must possess microfoundations in human mentalities and actions. There is no such thing as a social entity that lacks human embodiment--any more than there are works of art that lacks material embodiment. Social entities "supervene" upon human individuals.

This point also applies to any statements we might make about the putative causal powers of a social entity. So claims about the causal properties of social structures must be supplemented by a theory of the microfoundations of those powers. How does an extended social structure exert influence over the actions of located individuals?

And there is a final parallel point about claims about the geographical scope and coherence of a social entity. If we want to maintain that an entity exercises influence as a coherent and extended entity, we need to be able to specify the mechanisms through which this takes place. How does the Federal state exert its control and influence over the vast scope of the United States and its population?

So, with these qualifications about the unavoidable need for providing microfoundations--are there social structures?

Several of the instances offered above fit the terms of our provisional definition. They are large complexes of rules and practices that influence behavior and outcomes. And it is straightforward to begin to provide a description of the microfoundations upon which they exist: the social components through which these structures are embodied and through which they exercise influence on individuals and groups. The US Federal Government functions as a system of branches of government, each with its own departments governed by formal and informal rules. And the "reach" of the state down to the local and individual level is secured by the socially implemented forms of power that are locally expressed (bank inspectors, law enforcement agencies, tax auditors, ...).

This is an example of a large social structure that operates through a high degree of formal institutionalization. But some of the examples mentioned above depend primarily on informal mechanisms -- the workings of widespread beliefs and attitudes, along with a diffused willingness of individuals to "enforce" the requirements of the structure. Structures relying primarily on informal mechanisms include the Indian marriage system or the English class system.

Is "race" a structure in American society? Plainly it possesses some of the key elements identified above. The reality of race leads to an uneven distribution of opportunities and outcomes, so "race" is a social fact with distributive consequences. It has the element of coercion: racial prejudice and patterns of discrimination are imposed on individuals without an "opt-out" possibility. And we can identify many of the social mechanisms through which race and racial discrimination work; so the category possesses microfoundations. Today many of those mechanisms are "informal" rather than "formal"; but of course the legal institutionalization of racial discrimination is a recent fact in American history. So "race" is a structural feature of American society.

Several of the examples mentioned above appear to fall outside the category of social structure, however; for example, "Chinese peasantry". These examples appear to be large factors that play a role in large social structures, but are more akin to elements than systems. So the structure that defines "Chinese peasantry" is the system of property, agriculture, and kinship that defines the peasant's role and opportunities in society; the category of "peasant" identifies one node within that system or structure.

What about "the city of Chicago"? Is this a structure or some other category of social entity? I am inclined to say that the city of Chicago is a complex social entity, not a structure. It falls within a variety of structures in America and the world--the global trading system, the electoral process, and the politics of national funding for large cities; and it embodies within it a variety of smaller structures--the public school system, lending practices, nepotism. But the city itself does not function as a regulative system coordinating the activities of large numbers of individuals. Rather, it is a complex social entity composed of a mix of social practices, behaviors, systems, and relationships.

There are quite a few posts in the UnderstandingSociety blog on the topic of structures and agents; follow the structure label to find more.

How does rational choice theory relate to social facts and individual psychology?

Rational choice theory is a mid-level theory of human agency, intended to capture core features of human decision-making in order to provide a basis for abstract and generalized theories of the dynamics of various areas of social behavior. The disciplines of political science and economics make particularly extensive use of versions of rational choice theory (decision theory, expected utility theory, game theory, theories of parametric and strategic rationality) in designing theories and models of stylized circumstances of social action (e.g. market decision-making, production decisions, voting decisions, collective action decisions). Homo economicus -- the rationally self-interested preference- or utility-maximizing individual -- stands at the foundation of typical models in economics and political science.

The theory of economic rationality can be specified in numerous ways. In all its versions it is an abstraction from real human behavior, in two senses. The theory abstracts from idiosyncratic differences in reasoning across individuals; and it abstracts from other, perhaps systematic, factors that might influence reasoning (emotions and morals, for example). The theory allows for differences across individuals, of course, but primarily in the individual's preferences or utility function. (The nature of the decision rule is also a variable: more risk averse or less, optimizing or satisficing, maximin or maximize expected utility).

The question here is, what role does rational choice theory play in social explanations and in formal models in political science and economics, and how does it function in description of individual psychology and behavior?

Some formalistic economists and political scientists treat the assumptions of rational choice theory as purely formal axioms that can provide the basis for a mathematical treatment of a stylized problem -- solution to an n-person non-zero sum game, for example. These theorists are not concerned about the descriptive adequacy of the axioms, but rather the feasibility of employing the axioms to derive a solution to the problem.

This approach is unsatisfying, however, if we think that the results of formal economic or political analysis are supposed to be explanatory in some sense of actual social phenomena. Suppose we explain the low level of public contribution to public radio, by saying that the "contribution/non-contribution" game is an n-person prisoners' dilemma, and the mathematical solution to such a game demonstrates an equilibrium of low contribution. This mathematical finding is only potentially explanatory if we have some reason to think that the model bears a relevant relationship to the system of behavior it is modeling; the model assumes rationally self-interested maximizers; so the model is potentially explanatory of the public radio result only if the listeners are to some degree approximately well-described as "rationally self-interested maximizers." In other words, the rational-choice explanation of this real social situation appears to require that there be some degree of realism in the assumption of individual rationality. The purely formalistic interpretation of the axioms would render the account devoid of explanatory value.

This suggests a different approach to the question. It suggests that we should regard the assumption of rationality as an approximately true description of most human beings. Rational choice theory is an empirical theory of human behavior, at a high level of abstraction. "Abstract" in this context means "disregarding of interfering or contrary factors" -- as the theory of ideal gases is abstract in its disregard of intermolecular forces. In other words, the theory of rationality might be construed as a particularly abstract part of an empirical theory of human psychology.

If we take this approach, then we are naturally invited to test and improve upon this theory of human decision-making. Are there other factors of deliberation and action that need to be introduced? Are there differences across human groups and cultures with respect to the system of reasoning that individuals employ in practical decisions? Is the theory of maximizing rationality true of at least a certain range of individuals and decision-making problems? (Perhaps, for example, we are maximizing decision-makers when it comes to choosing a loaf of bread, but are influenced by emotions and commitments when it comes to choosing a political party to support.)

The advantage of a simple, abstract theory of deliberation (the theory of economic rationality) is that it provides a mathematically tractable way of modeling certain common situations of social action. If we advance a theory of deliberation and agency that postulates more nuance at the individual level and more variation across individuals and cultures, we may lose the ability to put forward models that result in equilibrium. More empirical adequacy may reduce the theoretical or deductive adequacy of the disciplines. However, it seems unavoidable that social science is better served by a theory of human agency and deliberation that is somewhat more faithful to actual human practical cognition. And it would remain possible that the narrow assumptions of economic rationality emerge as being adequately descriptive in certain well defined problem circumstances (the circumstances of anonymous markets, for example). Finally, it may be that other methods of aggregating from individual-level assumptions to models of collective behavior are feasible -- for example, the agent-based models that are advocated by the Santa Fe Institute.

There may be one additional reason why rational choice theory asserts a centrality that separates it from the rest of social psychology -- that is the role that rationality plays in a normative sense. We might think that reason and rationality are especially important human creations, and that it is valuable to have a theory of pure rationality (and its implications) even if the typical human reasoner falls far short of its assumptions.

Why are tastes and aversions beyond rational control?

Rational choice theory is a powerful foundation for thinking about social behavior (at least in its moderate versions). People have beliefs and goals; they survey the environment of choice and they arrive at actions that are intelligently related to the achievement of their goals.

However, as an exclusive, abstract description of human behavior, rational choice theory confronts a few major problems. A minor problem is the fact that people do not always act deliberatively. They may act out of impulse, habit, or example -- without devoting the thought processes to their current choices that would be necessary in order to formulate a rational plan in the circumstances. This is a minor problem, because this behavior itself may be rational in the "satisficing" sense: the stakes may be so low that it doesn't make sense to commit time and energy to the task of calculating what the best overall solution to the problem is. (Shall I take the triple cappuccino or the iced mocha today?)

More challenging is a much more substantial fact about human action and choice: there is a powerful set of psychological imperatives and constraints that make certain choices psychologically impossible even when they are "goal-maximizing". Take food and clothing choices as examples: the fact that fish heads are perfectly nutritious and cheap will never induce most Americans to order them on the menu. The fact that a pair of Bermuda shorts is a perfectly suitable attire in some settings will not induce someone to wear them to a black tie gala in place of the tuxedo. It is as if the decision-making system is loaded with constraints of taste, preference, aversion, and norm that make some decisions simply impossible.

This fact might be described as the workings of "culture"; it might be categorized under the heading of "norms and values"; or it might be assigned to the "workings of personality and affect." But whatever description we offer, it is plainly a factor that needs to be incorporated into a full theory of human agency and choice.

It is not particularly difficult to model a "decision-making system" that incorporates the workings of tastes, aversions, and norms as well as goal-directed rationality. The more challenging task lies at the level of psychological theories of individual development (how does the individual acquire and incorporate these cultural and normative constraints?) and sociological theories of transmission of these sorts of constraints. What are the institutional and cultural mechanisms through which a system of tastes, aversions, and norms are transmitted throughout a group or population?

And, finally, there is the task for behavioral science of incorporating the empirical findings that this set of observations permits concerning the ways in which real agents process experience and arrive at decisions, into aggregate models of collective behavior. How should the abstract models of rational choice theory be extended to include these factors of affect, taste, and aversion? Are these non-rational factors so idiosyncratic that they cannot be incorporated into behavioral science? Or are there regularities at the level of individuals that permit aggregate consequences at the level of group behavior? (And is it possible that we have arrived here at the theoretical nexus of "marketing science" -- the discipline aimed most directly at understanding and controlling the behavior of groups of consumers?)

(An interesting edited volume by Lupia, McCubbins and Popkin explores some of these issues: Elements of Reason: Cognition, Choice, and the Bounds of Rationality .)

Basis for the social science disciplines

Is there a reason or rationale underlying the scope and methods of the social science disciplines? Or is the current division of the subject matter arbitrary and contingent on accidents in the history of thought (as perhaps Michel Foucault believes in The Archeology of Knowledge and The Discourse on Language)?

There are abbreviated definitions of the subject matters of sociology, political science, of economics -- "Sociology is the study of groups and norms," "political science is the study of the institutions and behavior through which political decisions are made," "economics is the study of rational behavior in the context of scarcity". And likewise for the core methods of inquiry and explanation. But the questions remain: Is this a logical way of covering the domain of possible inquiry?Are there areas of phenomena that are overlooked in this taxonomy? And pragmatically--is this combination of topics, scopes, and methods likely to solve the problems of explanation and policy to which we would like the social sciences to contribute?

It is worth noticing in this discussion that each of the core disciplines of the social sciences is actually an umbrella encompassing a wide range of sub-disciplines, and these sub-disciplines themselves show significant variation in assumptions about subject matter, ontology, and methodology. Within political science, for example, there are wide rifts between new institutionalists, rational choice theorists, comparativists, area specialists, and American politics specialists, and there is no common perspective that gives substantial coherence across these divides. Similar observations are true of economics, anthropology, and sociology as well.

So how should we begin to analyze the status and adequacy of the current definitions of the social science disciplines?

What would we want from a division of the disciplines in a large field of study (biology, physics, social science, psychology)? This is an epistemic question: we want to know how best to go about acquiring knowledge about a broad domain of phenomena. We would want to be sure that the most fundamental processes and the most practically important phenomena are given close scrutiny. We would want to be sure that there are a variety of methods of inquiry that are well designed and effective in probing the nature of the phenomena under study. And, of course, we would want the disciplines to be organized in such a way that practitioners are motivated and evaluated towards valid discovery. Finally, it would appear logical that we would want there to be mechanisms of scientific communication through which discoveries and insights in one discipline may inform or deepen the discoveries of other neighboring disciplines.

This last point deserves emphasis. Disciplines provide focus; they encourage researchers to dig deeply into the specialized problems and methods that have been defined by the discipline. But it is also true that specialization creates tunnel vision -- with the result that scientific understanding of a complex phenomenon may miss essential parts of the process because they overlap disciplinary definitions of subject matter. So a logical expectation of the organization of scientific knowledge would presumably include a process through which interdisciplinary sharing is encouraged and facilitated.

It needs also to be noted that it is entirely possible for a discipline to be mis-conceived. The founders may have taken a wrong turn in highlighting what appeared to be the most important phenomena and mechanisms; the discipline may have adopted methodological commitments that were fashionable at a time but poorly fitted to the research problems of the subject matter (extreme behaviorism in psychology, for example); the disciplinary institutions (journals, universities) may have designed procedures of peer review that were discouraging to genuine progress in the discipline's ability to gain knowledge about the subject matter.

There are many threads that this subject opens. But a few preliminary observations are justified. First, it is worth observing that the social world does not come to us with a prior division of phenomena into "economic," "ethnographic," or "sociological." So the social world itself does not dictate the way in which we parcel out research problems across specialists. Almost all social phenomena have dimensions of each of these characteristics, and the methods of each of these disciplines are relevant to inquiry and explanation of a single phenomenon. So the insistence on using only the methods of one's discipline in analyzing a complex social phenomenon is a problem--along the lines of the saying that "to a person with a hammer, everything looks like a nail". Second, it is noteworthy that each of the major social science disciplines has undergone harsh "methodology wars" in the past several decades--between rational choice theory and more empirical research in political science, between quantitative and qualitative researchers in sociology, or between interpretive and materialist ethnography methods. Are these disputes productive of further discovery into the workings of the social world? It is not evident that they are. And third, the points made above about the general utility of interdisciplinary communication and collaboration seem especially applicable to the social sciences. Given the inherently complex and mixed nature of social phenomena, surely the social sciences would benefit from a greater degree of interdisciplinarity and mutual respect. Pluralism in theory and method would seem to best serve the goals of the social sciences.

(See Andrew Abbott, Chaos of Disciplines, for a lively and illuminating examination of the social science disciplines.)

Are social facts reducible to something?

What is the relationship between facts about society and facts about individuals?

Reducibility means that the statements of one scientific discipline should be logically deducible from the truths of some other "more fundamental" discipline. It is sometimes maintained that the truths of chemistry ought in principle be derivable from those of quantum mechanics. A field of knowledge that is not reducible to another field R is said to be "autonomous with respect to R". Philosophers sometimes further distinguish "law-to-law" reduction, "type-to-type" reduction, "law-to-singular-fact" reduction, and "type-to-token" reduction.

Are social sciences such as economics, sociology, or political science reducible in principle to some other more fundamental field--perhaps psychology, neurophysiology, or the theory of rationality?

To begin to answer this question we must first decide what items might be reduced: statements, truths, laws, facts, categories, or generalizations. Second, we need to distinguish several reasons for failure of reduction: failure in principle, because events, types, and laws at the social level are simply not fixed by states of affairs at "lower" levels, and failure for reasons of limits on computation. (The motions of a five-body system of stars might be determined by the laws of gravitation even though it is practically possible to perform the calculations necessary to determine future states of the system.)

So now we can consider the question of social reduction in a reasonably clear form. Consider first the "facts" that pertain to a domain of phenomena--whether these facts are known or not. (I choose not to concentrate on laws or generalizations, because I am doubtful about the availability of strong laws of social phenomena.) Do the facts of a hypothetically complete theory of human psychology "fix in principle" the facts of economics or sociology, given appropriate information about boundary conditions?

One important approach to this problem is the theory of supervenience (Jaegwon Kim, Physicalism, or Something Near Enough). A level of description is said to supervene upon another level just in case there can be no differences of state at the first level without there being a difference of state in the second level. The theory is first applied to mental states and states of neurophysiology: "no differences in mental states without some difference in neurophysiology states." Supervenience theory implies an answer to the question of whether one set of facts "fixes in principle" the second set of facts. (It has been taken as obviously true that social facts supervene upon facts about individuals; how could it be otherwise? What other constitutive or causal factors might influence social facts, beyond the actions and ideas of individuals?) If the facts about social life supervene upon facts about the psychological states of individuals, then it follows that the totality of facts about individual psychology fixes in principle the totality of facts about social life. (Otherwise there would be the situation that there are two total social worlds corresponding to one total "individual psychology" world; so there would be a difference at the social level without a difference at the level of individual psychology.)

So this provides the beginnings of an answer to our question: if we believe that social facts supervene upon facts about individuals, then we are forced to accept that the totality of facts about individuals "fix" the facts about society.

However, supervenience does not imply "reducibility in principle", let alone "reducibility in practice" between levels. In order to have reducibility, it is necessary to have a system of statements describing features of the lower level which are sufficient to permit deductive derivation (or perhaps probabilistic inference) of all of the true statements contained in the higher-level domain. If it is a social fact that "collective action tends to fail when groups are large", then there would need to be set of statements at the level of individual psychology that logically entail this statement. Two additional logical features would appear to be required for reduction: a satisfactory set of bridge statements (linking the social term to some construction of individual-level terms; "collective action" to some set of features of individual agents, so there is a mapping of concepts and ontologies between the two domains), and at least some statements at the lower level that have the form of general laws or law-like probabilistic statements. (If there are no general statements at the lower level, then deductive inference will be limited to truth-functional deduction.)

Now it is time for a speculative leap: a judgment call on the question of whether we ought to look for reductive links between social facts and individual-level facts. My intuition is that it is not scientifically useful to do so, for several reasons. First is the point about computational limits: even if the outcome of a riot is "fixed" by the full psychological states of participants ex ante and their strategic interactions during the event--it is obviously impossible to gather that knowledge and aggregate it into a full and detailed model of the event. So deriving a description of the outcome from a huge set of facts about the participants is unpromising. Second, it is telling that we need to refer to the strategic interactions of participants in order to model the social event; this means that the social event has a dynamic internal structure that is sensitive to sub-events that occur along the way. (Jones negotiates with Smith more effectively than Brown negotiates with Black. The successful and failed negotiations make a difference in the outcome but are unpredictable and contingent.) Third, the facts at the social level rarely aggregate to simple laws or regularities that might have been derived from lower-level laws and regularities; instead, social outcomes are contingent and varied.

So for a variety of reasons, it is reasonable to take the view that social facts supervene upon facts about individuals, but that social explanations are autonomous from laws of psychology. (This final point might be paraphrased as "the laws of psychology underdetermine social outcomes.")

How does transportation function as a mid-range social cause?

Transportation systems function to move people, goods, and ideas. Rail systems, road networks, airline systems, and water transport provide links between places that permit more reliable and low-cost movement of people and goods from point to point than previously available. The history of transportation is simultaneously a history of technology change, population movement, colonialism, economic growth, business development, and the spread of disease.

Transportation systems are particularly interesting when we consider their capacity for conveying social causation. Consider these examples of causal relations mediated by transportation systems:

• Extension of a rail network stimulates the growth of new towns, villages and cities in North America in the 1880s.
• Establishment of a direct air travel link between A and B causes the more rapid spread of disease between these locations.
• Breakdown of the administration of the rail system leads to logistics bottlenecks and military defeat of the French army in the Franco-Prussian War.
• Regular river travel throughout the Canton Delta in China leads to the rapid spread of revolutionary ideas during the Republican Revolution, as travelers and merchants move easily from place to place.
• Commodity price correlation increases between Chicago and New York as a result of regular and cheap rail transport between these two cities.
• New business institutions (grain futures markets and grain elevators) are created to take advantage of cheap regular rail transport (Nature's Metropolis: Chicago and the Great West).
  

Examples can be multiplied. But the central point is that transportation is a robust causal mechanism that mediates many important social processes and outcomes. And its causal effectiveness is fairly transparent: new transportation opportunities create new options for social actors, who take advantage of these opportunities in choosing a place to live and work, in pursuing political goals, in moving armies, and in generating income. So transportation is a causal mechanism whose microfoundations are especially visible, and whose causal consequences are often very large.

(See "Transportation as a Large Causal Factor" for more on this subject.)

Varieties of social causation

What are some chief mechanisms through which social behavior is shaped and social outcomes are caused? Ideally the best answer to this question would result from a survey or inventory of many explanations. But consider some high level possibilities that could serve as a causal mechanism bringing about a social phenomenon of interest.

Selection mechanisms. Why are passengers on commercial aircraft better educated than the general population? Because most airline passengers are business travelers, and high-level and mid-level business employees tend to have a higher level of education than the general population.

Evolutionary explanations. Why does the level of firm efficiency tend to rise over time? Because the net efficiency of a firm is the product of many small factors. These small factors sometimes change, with an effect on the efficiency of the firm. Low efficiency firms tend ultimately to lose market share and decline into bankruptcy. Surviving firms will have features that produce higher efficiency.

Imitation mechanisms. Why did the no-huddle offense become so common in the NFL in the 1980s? Because it was successful for a few teams, and others copied the offense in the hope that they too would win more games.

Rational-intentional explanations. Why do rebellions often fizzle out? Because participants are rational agents with private goals, and they make calculating decisions about participation.

Aggregative explanations (aggregate consequences of individual-level strategies). Why does technological innovation occur continuously within a market-based society? Because manufacturers are constantly looking for lower-cost and higher-value-added methods of manufacturing, and this search leads them to innovations in products and technologies.

Conspiracy explanations (covert strategems of the powerful). Why did the United States move away from passenger railroads as the primary form of intercity transportation? Because powerful interests took political actions to assure that private automobiles would be encouraged as the primary form of transport.

Mentality explanations (behavior is changed by changing beliefs and attitudes). Why were so many Quaker men conscientious objectors at great personal cost during World War II? Because their religious beliefs categorically rejected the violence in war and they refused to participate in this immoral activity.

Historical / path-dependency explanations. Why do we still use the very inefficient QWERTY keyboard arrangement that was devised in 1874? Because this arrangement, designed to keep typists from typing faster than the mechanical keyboard would permit, was so deeply embodied in the typing skills of a large population and the existing typewriter inventory by 1940 that no other keyboard arrangement could be introduced without incurring massive marketing and training costs.