If you're like me and are dumb enough to take an interest in abstruse debates among social scientists, you may have heard of the "debate" between methodological individualism and holism. Here are a couple of articles describing it (also, see below). As a Marxist, I'm a holist. I don't find methodological individualism particularly interesting or substantive, though I appreciate its admonition against functionalism and that you should always look for specific causal mechanisms bringing about a state of affairs. To explain why there are always a lot of unemployed people in capitalism, it isn't enough to say, "It's in the interest of capital to have a reserve army of labor." Functionalist explanations aren't real explanations. But this seems obvious to me, and one hardly needs to be a "methodological individualist" to accept it.
Anyway, a long time ago I wrote a paper on individualism vs. holism, which I've copied below. (I've deleted some passages that I thought were confused or confusing.) I'm not sure of its value, though it certainly isn't valueless. Maybe it at least clarifies some of the stakes, or non-stakes, of the debate. More generally, its value might be that it suggests how empty, or nearly empty, a lot of philosophical and academic controversies are.
The debate between methodological individualism and holism has lasted over a century, and a consensus will probably never be reached. The questions involved have, until recent decades, tended to be conflated: ontological claims have been equated with explanatory claims, which have been equated with semantic claims. In his book Varieties of Social Explanation: An Introduction to the Philosophy of Social Science, Daniel Little distinguishes between three main questions: “What sorts of things is society composed of?”, “What do societal concepts mean?”, and “What method should be followed in explaining social phenomena?” The individualist gives the following respective answers: (1) society is simply an aggregate of individuals; (2) the meanings of such collectivist concepts as ‘state’ and ‘class’ can be reduced to concepts about (interactions between) individuals; and (3) social explanations can be—as J. W. N. Watkins puts it—“deduced from (a) principles governing the behavior of participating individuals and (b) descriptions of their situations”. The holist, on the other hand, answers, more or less, as follows: (1) society is composed not only of individuals, but also of relations between individuals (to quote Louis Althusser); (2) the meanings of societal concepts are not reducible to individual-level concepts; and (3) social relations—or, ultimately, macroscopic tendencies—have explanatory primacy. Now, the same kind of answer need not be given to all three questions. An individualist’s ontology, for example, does not entail an individualist’s theory of meaning—because the meaning of a given concept (which is determined by the ways people use it) has nothing to do with insight into the true nature of the world (because neither does conventional linguistic practice)—nor does it entail an individualistic methodology—though I shall have to argue for this claim later, due to its complexity. Daniel Little argues convincingly against the individualist’s theory of meaning; in this paper I’ll answer the ontological and explanatory questions, defending the holist’s position (while qualifying it in minor respects).
Here are two classic statements of methodological individualism, the first by Watkins, the second by Karl Popper:
....[T]he ultimate constituents of the social world are individual people who act more or less appropriately in the light of their dispositions and understanding of their situation. Every complex social situation or event is the result of a particular configuration of individuals, their dispositions, situations, beliefs, and physical resources and environment.
....[A]ll social phenomena, and especially the functioning of all social institutions, should always be understood as resulting from the decisions, actions, attitudes, etc. of human individuals, and...we should never be satisfied by an explanation in terms of so-called “collectives”...
The first point to be made is that Popper is not entitled to his conclusion. His methodological claim does not follow from his ontological one. This is because (assuming Popper’s ontology) it is possible to sum up the micro-level, atomistic facts of a given macrocosm and extrapolate tendencies that we then phrase in collectivist terms. In other words, we can legitimately abstract features of a collective without invoking the individuals who compose it. For instance, as a partial explanation of George W. Bush’s reelection in 2004 one might say, “His support among the Hispanic population was important”. “The Hispanic population” is a collective, yet the explanation offered is satisfactory. One can in this way abstract a plethora of features of any collective from the innumerable microstates that compose it, whether it be an economic class, a group of individuals with the same occupation, an ethnicity, or a social institution. “The bureaucracy of the State department is inefficient.” Even assuming an individualist ontology, this holistic claim is perfectly legitimate. Indeed, it would be absurd and otiose to say, instead, “The individuals from whom—or from whose acts—the concept of the ‘bureaucracy’ is abstracted operate in such a manner that it takes a needlessly long time for a directive by one of their superiors to be actualized.” To expunge references to collectives from explanations would be to make explanations far more verbose than they have to be.
Perhaps Popper would object to the latter example, since it is not an explanation per se. It is merely a description, so his stricture does not, strictly speaking, apply to it. (This would be a sophistical objection, but I’ll ignore that.) Here is a better example: “The state typically adheres to the interests of the capitalist class, because this class wields far more economic power than any other, and economic power yields political sway.” It is hard to see how Popper’s social ontology is incompatible with that statement, even though it makes no reference to individuals. Indeed, it refers only to collectives and relations between them, both of which concepts (‘collective’ and ‘relation’) are deemed not to ‘really’ exist. Its concepts could be translated into individualistic ones—thus: “Politicians generally act according to the interests of their capitalist donors...”—but there seems little point in carrying out this reduction.
However, this whole discussion is confused (mirroring, in this respect, the historical debate between holism and individualism). If individuals and their actions were indeed the only socially relevant things that existed, then obviously social explanations would necessarily invoke only individuals and their actions. If there were no such things as relations, collectives and so on, theorists could not refer to them. But then there would be no debate between individualism and holism: all theories would, by necessity, be individualistic. This is not the case, however, so relations and collectives must in some sense exist. In some ‘abstract’ way, there clearly are such things as relations between people. Not even Popper would deny this. But what does it mean? This question cannot be answered in an intuitively satisfying way, although it has preoccupied philosophers for millennia. The medieval scholastics, for example, were divided into three positions: realism, according to which such ‘universals’ as relations exist outside the mind; conceptualism, according to which they are mind-made; and nominalism, according to which they do not exist at all. In his paper “On What There Is”, W. V. Quine respectively associates each of these with the modern doctrines of logicism, intuitionism, and formalism. The point is that none of these answers (to the question of what it means for relations to exist yet not to exist in the same way that material objects do) is relevant to the philosophy of social science. That is, it does not matter which of them is true: it will not change the opinion of either an individualist or a holist. Suppose it turns out that social relations do (metaphysically) exist outside the mind; the individualist will say, justifiably, “Well, they don’t really exist! They don’t exist in the way that, e.g., people do!” On the other hand, if it turns out that relations do not exist at all (whatever that would mean), the holist will say, with justification, “Well, since we can coherently talk about them, obviously it isn’t senseless to say they exist. So I will stick to my original claim (viz., that social relations not only exist but determine the nature of social interactions).” Now, because these three ontological positions are exhaustive, there is no relevant answer to the question. Therefore, to argue for or against the doctrine that social relations exist is misguided. It is based on a misunderstanding; whatever one’s position, nothing substantive is said. Thus, the ontological question itself is misguided.
That argument may have been too quick. I have another, though. Louis Althusser goes so far as to say that society is composed only of relations: “What constitutes society is the system of its social relations in which its individuals live, work and struggle”. Similarly, Karl Marx says (in the Grundrisse) that “society is not composed of individuals”. To a dogmatic individualist these statements must seem perverse, just as, to a dogmatic holist, it must seem outrageous to say that only individuals constitute society. However, with a sufficiently open mind, both doctrines are understandable. It is possible to shift from agreeing with one to agreeing with the other simply by experiencing what Thomas Kuhn might call a “gestalt switch”. From one perspective, Marx and Althusser are right; from another, Popper and Hayek are. One might even say that society is composed of traditions, or of norms, or of institutions, or of roles, or of whatever concepts can justifiably be abstracted from people’s modes of interaction; and one would be right. A societal ontology can include virtually whatever one wants it to include (within reason).
In any case, the ontological thesis of the individualist is vacuous. We have seen that he would agree with the holist that social relations abstractly exist; by arguing, then, that only people (and their actions, thoughts, etc.) exist, he is really arguing that people are the only concrete existents. He is saying, in effect, that the only things that concretely exist are things that concretely exist, namely people. He is stating a truism. No rational person, after all, would deny that “Every complex social situation or event is the result of a particular configuration of individuals, their dispositions, situations, beliefs, and physical resources and environment”. What follows from that? Nothing.
The individualist’s methodology is more substantive than his ontology. Admittedly, at first glance it may seem equally truistic: to say, for example, that a theorist must look for the explanation of a bureaucracy’s inefficiency in the behavior of the individuals who compose it is, in a sense, to say not much at all. No holist would disagree with that statement. He would insist, however, that the bureaucrat’s behavior is itself explainable through more fundamental facts—facts rooted not in the bureaucrat’s psychology but in his location within a nexus of relations that extends through the whole of society. The individualist apparently denies this—or, at most, qualifies his agreement by saying, perhaps, that these relations are explained by human psychology and people’s reactions “toward others....guided by their expected behavior”. The holist retorts, e.g., that psychology cannot account for the differences between societies and that the nature of people’s “reactions” depends on where they are located in the system of social relations. Then the individualist responds by invoking environmental conditions, etc., as an explanation of the differences between societies, and says that a person’s location in the system of social relations depends on his psychological characteristics and past actions, etc. And in this way the debate goes around in circles, beginning to seem uncomfortably similar to the “chicken or the egg” problem.
To find a way out, we must clarify wherein lies the confusion. This entails clarifying the disagreement. What does it really mean to say that the individualist’s perspective is atomistic? What does “atomism” mean? Hayek might say it means that “individual actions” are the fundamental level of analysis. But what does that mean? In itself, the principle that “there is no other way toward an understanding of social phenomena but through our understanding of individual actions directed toward other people and guided by their expected behavior” says nothing controversial, surely. One can be a holist—a structuralist, a functionalist, a Marxist—yet subscribe to it, as far as it goes. It does not contradict the methodological injunction that relations, and the macroscopic structures they constitute, should be the focus of analysis, because the holist can consistently maintain both that a person’s behavior is causally influenced by the character of the social relations in which he is embedded and that the properties of a given social phenomenon supervene on the actions of individuals. Indeed, to deny that they do so “supervene” would be ridiculous: it is obvious that if the behavior of a group of people changes, the nature of the social phenomena that are related to it will change (and the latter will change only if the former does). This claim is essentially equivalent to Hayek’s idea that in order to understand a social phenomenon one must understand the actions in which it is instantiated (—or, at any rate, each idea implies the other); and neither idea implies that actions cannot be understood from a holistic perspective.
So we still do not know exactly what the individualist is saying. Let’s look at Watkins’ account of the theory: social explanations can be “deduced from (a) principles governing the behavior of participating individuals and (b) descriptions of their situations”. This statement is stronger than Hayek’s. A holist would not subscribe to it. It states that there is a set of principles (by which Watkins means psychological principles) that governs the behavior of all individuals, and that in order to understand a given social phenomenon the theorist must apply this set of principles to each relevant individual’s behavior, on the basis of a description of his situation. Now we finally know wherein consists the atomism of methodological individualism: social phenomena are explained by ‘summing up’ the psychological explanations, and the histories, of each relevant individual’s behavior. This interpretation is supported by Watkins’ frequent use in his aforementioned paper of the concept of a person’s “dispositions”. Dispositions (to act a certain way), he seems to say, are what matter most. Thus, methodological individualism, if it is to have any substance at all, is essentially a familiar form of reductionism: the reduction of sociology to psychology.
Reductionism in most forms is not currently very fashionable among scientists or philosophers. It isn’t hard to see why: its program is unrealizable, due to practical and theoretic difficulties. Practically, the task of collecting enough data to explain, for example, the beating of a heart in terms of the movements of its molecules—or, more relevantly, a social phenomenon in terms of the psychology and the actions of all its participants—is impossible. But theoretically, most forms of reductionism are indefensible anyway, simply because when lower-level constituents combine to create a new entity, the latter has “emergent” properties that do not fit into the framework of the micro-level theory that describes the entity’s constituents.
[Here there was a muddled discussion of reductionism in the physical sciences, which I've deleted. Good riddance to bad rubbish.]
The situation is different on the level of social science. While the states of a given macrocosm still supervene on the behavior of individuals, the latter do not exist in the same kind of physical, deterministic network in which lower-level constituents exist in biological, chemical and physical entities. This is why methodological individualism is espoused only in relation to social systems: it is palpably false vis-à-vis deterministic systems, in which the interconnections between things are obvious. The existence of a psyche—that is, consciousness and will (free or otherwise)—in humans and animals somehow ‘softens’ environmental causality, so that it becomes mere influence. Thinkers are thus able to refer to the actions of “atomistic individuals”, which would be an absurd turn of phrase in the natural (i.e., deterministic) sciences.
Nevertheless, emergent phenomena abound in human communities (though less so in animal ones, since social relations there are extremely rudimentary). Wherever there is a social structure—a hierarchy, a network of relations, a system of roles—there is emergence. This fact in itself implies the falsity of the individualist approach. One wonders why so many thinkers have overlooked it. Have they actually failed to see, for example, that the market economy has structural tendencies? Or that the behavior of social classes follows patterns? Or that a bureaucracy is, in a sense, more than the sum of the bureaucrats’ activities? Obviously not. They have been misled, I think, by the irresistible, albeit shallow, logic of their methodology. For instance, upon reading the following statement from the economist Kenneth Arrow, one’s first inclination is to think its ideas are self-evident:
A full characterization of each individual’s behavior logically implies a knowledge of group behavior; there is nothing left out. The rejection of the organism approach to social problems has been a fairly complete, and to my mind salutary, rejection of mysticism.
“Of course!" one thinks. “What could be more obvious?! Groups are nothing but people; once you understand the people, you understand the group!” It is strange, though, that none of these thinkers had the suspicion that this reasoning was too pat, too easy. It does suggest a useful methodological principle (on which, see below), but, as I have already noted, it does not rule out the existence of structural factors which, while supervening on people’s behavior, constrain and guide their behavior as well. I think it is the invisibility of this dialectic between relations and agents that mystifies individualists and causes them to ridicule holism. My task, then, must be to elucidate it.
The methodological principle I referred to just now is something we have already encountered: namely, the injunction to find concrete mechanisms whenever there is causation due to the emergent properties of a social group. The reason for this principle is that the emergent properties do not exist in some mysterious Platonic realm, but rather in the concrete mechanisms themselves. (Or, more accurately, they are extrapolations from the latter.) For example:
Consider the proto-capitalist market, the ur-market of the early classical economists like Adam Smith. Its essential features are that it is a collection of small entrepreneurs in a competitive market, a homogeneous collection of traders engaged in competition with one another for resources or markets. The crucial fact about this situation is that it is unstable. In fact, as the degree of competition sharpens, the homogeneous situation becomes less and less stable. Many factors contribute to this situation.
Garfinkel proceeds to outline four such factors, one of which is the existence of economies of scale: “large-scale production is more efficient than small-scale; more automation can be used.... Hence a small advantage tends to become bigger.” The phenomenon being explained, viz., the instability of an economy, is emergent, in that (1) it is a macro-state that is unintentionally produced as people pursue their own private ends, and (2) it cannot be derived from the sum of atomistic descriptions of each individual’s behavior. It is exclusively a relational fact: it exists ‘in’ the relational interstices between entrepreneurs. What matters, therefore, is not how this or that entrepreneur behaves, but how his behavior affects others. We want to know his behavior not for its own sake (as the individualist assumes), but for the sake of how it makes someone else behave; and we want to know this other person’s behavior only so that we can see how it bears on another’s behavior. And so on. The holist’s methodology, then, is a sort of negative methodology: he wants descriptions of the ‘positive’ or the ‘present’ (i.e., observable behavior) purely so he can understand the ‘negative’ or ‘absent’ (i.e., unobservable relations)—because, although it is the positive/present that concretely exists, it is the negative/absent that determines the dynamics of society. And this is because the negative determines the positive—or, less obscurely: relations ‘determine’ behavior. Thus, the individualist has it backwards: he adds up activities, when what must be ‘added up’ is how the activities bear on one another. He forgets that if social phenomena were mere sums of parts, mere aggregates—in the way that, for example, “taxonomic collectives” are, or a herd of buffalo is—there would be no substantive relations between people, and so there would be no society.
To return to Garfinkel’s example: the question we are trying to answer is how emergent properties manifest themselves. It must be in concrete mechanisms, such as those that account for the frequent superiority of economies of scale to small-scale production. It is easy to imagine the specifics of how an economy organized around the latter could evolve to one organized around the former: an entrepreneur who had the benefit of economies of scale could lower his prices, thereby forcing his competitors either to follow suit or to go out of business. This process would gradually spread through the entire economy. The individualist might invoke psychology to explain why certain entrepreneurs were successful (as in a Social Darwinistic way), or he might note the importance of luck, or he might invoke rational choice theory or game theory; and from each person’s perspective, his account might (possibly) be correct. The problem is that it ignores the essential question, namely, why is there a pattern? If he answers, “The pattern is simply an aggregate of all these microstates” he is begging the question. The point is, what makes it possible that all these microstates can occur (in such a way that a pattern results)? Granted that, in each case, the more efficient entrepreneur will ‘drive out’ the less efficient; what is important is that whatever factors explain the possibility of this relation between entrepreneurs are the factors that explain the virtual necessity of the pattern. The individualist, if he is to be consistent with his methodology, cannot even ask the relevant question (due to his hyperspecific focus, which prevents his seeing the phenomenon), much less answer it; the holist answers it by appeal to emergent properties, which manifest themselves in—one might almost say, as—the very microstates we are trying to explain. There is no other answer to the question that began this paragraph. Thus, an irreducible circularity characterizes society, between micro-level relations and emergent properties. This circularity may be a philosophically obscure notion, but as a methodological postulate, it is necessary.
For further illustrations of the inadequacy of methodological individualism, I refer the reader to Garfinkel’s book Forms of Explanation, which includes scores of arguments from a holistic perspective. The ideas set forth in this paper have been, admittedly, far from complete, consisting mostly of suggestions and hints, but I hope they have at least cast doubt on individualism—the only valid doctrines of which, I have argued, are the idea of supervenience, the injunction to look for concrete mechanisms, and the emphasis on “human nature” as setting the range of possible social organizations. (This emphasis, however, should not, as it were, be over-emphasized.)
 (Boulder, CO: Westview Press, 1991), p. 183, http://www.questia.com.  J. W. N. Watkins, “Ideal Types and Historical Explanation,” in The Philosophy of Social Explanation, ed. Alan Ryan (New York: Oxford University Press, 1973), p. 88. Friedrich Hayek expresses the principle differently: “there is no other way toward an understanding of social phenomena but through our understanding of individual actions directed toward other people and guided by their expected behavior”. (Ibid., p. 121.) Both Watkins’ and Hayek’s statements amount to the idea that explanations must begin with the acts of individuals considered atomistically. They are then ‘added up’.  Quoted in Steven Lukes, “Methodological Individualism,” ibid., p. 121.  Incidentally, that translation presupposes social relations. It might be possible to express these relations in terms only of individuals’ behavior towards one another, but, again, no explanatory purpose would be served in doing so. Invoking relations is a legitimate and useful way of expressing—and explaining—human interactions. –But I’m getting ahead of myself.  For universals either do not exist, exist only in the mind, or exist outside the mind. There are no other options.  To repeat: both sides accept that social relations exist but do not exist (in the ‘concrete’ way of material objects). The individualist merely emphasizes their non-existence, while the holist emphasizes their existence.  Louis Althusser, Essays on Ideology (London: Verso, 1984).  Ibid., p. 85.  I’ll explain below what that means.  Or, to satisfy Popper et al: the behavior of the individuals from whom the notion of a ‘group’ is abstracted.  Recall too that Adam Smith, another methodological individualist [well, that’s arguable], grounded his system in the psychological postulate of a “propensity to truck and barter”.  It is ironic that Popper, in one of his rational moments, praised Marx for helping to establish the autonomy of sociology from psychology. He seemed not to realize that, in his advocacy of methodological individualism, he was effectively fighting against this autonomy.  Alan Garfinkel, Forms of Explanation (New Haven: Yale University Press, 1981), p. 53.  Ibid., p. 120.  Andrew Sayer, Method in Social Science: A Realist Approach (London: Hutchinson & Co., 1984), p. 109.  Garfinkel refers to a study by C. Jencks, the conclusion of which is that luck is more important than intelligence and education in explaining the economic success of particular individuals. Chance acquaintanceships, for example, or the range of jobs that happen to be available at a particular time, or various other accidents. Garfinkel’s point is that, from the individualistic perspective, luck is essential. The individualist’s “hyperspecific” focus precludes his explaining patterns, which are necessarily holistic.  Namely, “What is it that makes all these microstates possible?”—or, more clearly, “What is the nature of this economic structure? (And why does it exist?)”  As a Marxist, I would say there is a third circularity: between the economic base and the cultural superstructure. (It is an ‘asymmetric’ circle, however—if that term makes any sense.)