If this is true, why is there any question over methodological individualism in the social sciences?

Since its first formulation by a student of Weber, methodological individualism has been interpreted in a number of ways. However, they all centre on the idea that large scale (macro-level) social phenomena can be explained in terms of the actions and intentions of individuals. In effect it claims that sociology can be reduced to psychology. While there is no question that society consists of individuals - after all, if there were no individuals there would be no society - there is a question as to whether society merely consists of individuals, or whether the whole is in some way more than the sum of its parts. If so, there is a sense that in reducing the social to the individual we lose something. Thus there is a question as to whether such a reduction is possible.

In this essay I shall examine the question of whether social explanations can be reduced to individualistic explanations without loss of explanatory power. This will also entail addressing whether the social can be eliminated from individualistic explanations, and whether or not methodological individualism requires it to be.

Note that there is a difference between methodological individualism (hereafter MI) as a thesis of theory reduction, and MI as a thesis of explanation. In essence, the former requires something to reduce, the latter does not. In other words, MI as theory reduction assumes that a macro-level explanation of the relevant social phenomena is possible and claims that it can be reduced, whereas MI as explanation only assumes that a micro-level explanation is possible. However, the discussion of one can usually be applied in slightly reworded form to the other without significantly changing its substance. In the cases where it does not I shall endeavour to make it clear [Kincaid].

I shall begin by discussing supervenience and multiple realisability.

Social phenomena are multiply realisable. In other words, a given social phenomena can occur as a result of many different configurations of individuals and circumstances. The comparison is sometimes made between mental states and physical brain states. No two people's brains are identical, so an identical belief will be represented by different neural configurations for different people [Kincaid, p6; Wright et al, p62].

This presents our first problem. Reduction requires a 1-1, or many-1 relationship between the explanation to be reduced and the reduced explanation. In other words, any given phenomena must have one and only one reduced explanation. However, a given type of social phenomena might reduce to any one of a number of individualist descriptions.

A possible counter to this might be to argue that using multiple realisability to refute reduction is to compare dissimilar things. Two identical social events are nonetheless distinct. We could well be able to collect identical social events into a category and show how they all have distinct specific explanations. However, the collection is a "type" and the specific explanations are "tokens". To reduce type to token is incoherent. It is like explaining all beliefs about bananas in terms of my belief that there is a banana on the table. Instead, we should either be reducing type to type, or token to token.

Consider democracy. Democracy is certainly multiply realisable. Different countries have different forms of democracy, and each has arisen as a result of different circumstances. So it is certainly true that an explanation of what democracy is and how it arose cannot be reduced to an explanation of the actions and intentions of medieval English landowners. The argument is that it is wrong to expect such an explanation. We either explain a particular instance of democracy (e.g. British parliamentary democracy) in specific terms (e.g. of medieval English landowners), or we explain democracy generally in terms of generalised individuals.

The latter fits with Weber's original conception of methodological individualism, which was closely connected with his ideas of ideal types. In other words, idealised social phenomena were to explained by the actions of idealised individual agents [Heath].

The question now becomes whether such type-type reduction is possible, and whether it is at all plausible to believe that an explanation involving individuals - even generic, anonymous, idealised individuals - can explain "democracy" as a general concept. Unfortunately this is largely an empirical question, so I shall have to move on to token reduction.

Reducing token to token is initially plausible. After all, explaining the roots of a given historical event in terms of the people involved is what historians do every day (although see below). However, it prompts a new question. There is a sense that an explanation of a social phenomenon referencing only individuals is missing something.

Consider this example. As a result of an election a staunchly capitalist government is replaced by a socialist government. This would have a significant effect on the economy. It's possible we could explain the election result in terms of individuals' actions. It would doubtless be impractical, as it would involve explaining the motives and actions of every single voter, but it would be possible in principle. However, this explanation would not explain the subsequent economic effects. By contrast, a sociological theory could probably explain both the election results and the subsequent economic impact.

Social phenomena interact in the social sphere, while individuals interact in the individual sphere, and thus it appears that an explanation involving only individuals cannot accommodate the social interactions of social entities. Any adequate social theory will include laws that account for the interactions between social entities. If an individualistic theory cannot account for these interactions, then the social theory cannot be reduced [Kincaid, p11 & p16].

This may only be a problem if we are regarding MI as a thesis of theory reduction. If we regard MI as a thesis of explanation, we require only that a given individualistic explanation accounts for a given social phenomena. Once we've explained the election result individualistically, we may then go on to explain the subsequent economic impact in terms of the reactions of individual investors and business people.

Unfortunately, the matter is not settled there. Earlier I mentioned the activities of historians in explaining social events in individualistic terms. If historians reduce the sociological to the psychological every day, why is there any question over MI? Lets take a closer look at the kind of answers a historian - and indeed anyone attempting a token-token reduction of a social phenomenon - might provide.

Consider again an election. To explain an election result individualistically, we'd need to examine the psychology of the voters. Why would Person X vote for Party A? He might blame the incumbent party for losing his job. He might believe Party A has a better crime-reduction policy. He might be more impressed by the charisma of Party A's leader. There are any number of reasons, but one thing they will have in common is that they all reference social concepts.

An election is a social process. Political parties are social entities. The policies upon which elections are fought are social policies. Explaining the thought processes of a voter without referencing a social concept in some way will almost certainly be impossible. Society consists of individuals, but it may be impossible to eliminate the social from the individual. The question now becomes whether this is necessarily true, and how this affects MI.

In our example of the voter, it seems clear that any explanation will at least initially reference social concepts, but perhaps those concepts could in turn be reduced to individualistic terms, or reduced to concepts that could be. If we'd explained the voter's actions in terms of his fear of crime, we'd then need to explain crime as a concept, which would probably in turn entail explaining the concepts of property and social status.

This leads us to the idea of an explanatory chain, which Ruben makes much use of [Ruben 1985]. MI needs only claim that this explanatory chain ends with all social terms being eliminated. It says nothing about how long the chain must be. It could be that any explanatory chain would have to trace back through time to a state of nature when there were no social concepts beyond basic interactions. This may seem implausible, but it is worth reiterating that MI is an in principle thesis.

Society did not spontaneously arise unbidden. There must have been, in the distant past, a point when there was no society. Then, beginning with the most basic levels of co-operation between individuals, it would have built up gradually over countless generations. These basic social concepts would be the result of individual innovation, but might in turn give rise to more complex social concepts. For example, a justice system might arise as a result of the emergence of theft, which would be a response to the emergence of the concept of property, which would be down to individuals coming to an understanding of who possessed what.

I shall reformulate this into a proof by iteration. Assume there was a time when there were no social concepts. Assume also that any social concept arises either as a result of individual action, or as a result of previously existing social concepts. Therefore the first social concept was a result of individual action, and thus every subsequent social concept can trace its ancestry back to an individual action.

This is not the place for a detailed anthropological study of the emergence of society, but hopefully my very cursory illustration will demonstrate how a social explanation about voting could - in principle - ultimately be reduced to an explanation referencing only individual action.

However, one could question whether such a course is really necessary. MI is not necessarily an eliminativist thesis - that's the domain of atomism [Wright et al, p56]. The content of a belief does not stop it being the belief of an individual, so it could be argued that an explanation in terms of an individual's beliefs is fully individualist regardless of whether or not those beliefs reference social concepts [Tuomela, p3].

An opponent could point out that these beliefs have explanations, that these explanations will reference social concepts, and that MI demands that any explanatory chain ends with all social concepts being eliminated [Ruben 1985, pp9-10]. This just returns us to Ruben's explanatory chains and the previous discussion. We are now claiming that the reduction to individual beliefs is enough. In effect, we are denying Ruben's formulation of MI, and claiming we merely need to get to a point on the explanatory chain where there are no social concepts, but not necessarily where there are no further social concepts further along the chain.

Ruben's initial formulation of MI is this weaker version. His argument for strengthening it rests on the interpretation that MI is about the explanatory primacy of the non-social over the social [Ruben 1985, pp4-5]. MI has been interpreted in many ways though, and whereas Ruben sees "the social must be explained in terms of the individual", others may read "the social can be explained in terms of the individual", which makes no assertion of explanatory primacy.

Indeed it is this ambiguity of interpretation that is the source of many of the questions surrounding MI. I have focussed on one of the more central questions of whether macro-level social explanations can be reduced to micro-level individualist explanations, but there are others. MI has been interpreted as a thesis of ontology that raises questions about what social concepts actually are. It has been interpreted as an epistemological thesis, which raises questions about what we can and cannot truly know about social concepts [Lukes, pp14-15]. Even the interpretation of explanation that I have focussed on can be formulated in strong terms (e.g. all social explanations must be solely in terms of individuals) or weak terms (e.g. all adequate social explanations must allow for an account in terms of individuals) [Kincaid, p20]. There is a sense that modern writers have lost sight of the fact that MI was originally intended as nothing more than a guard against the perceived dangers of social holism and subjective teleology [Heath, sect 1; Elster, p24].

In this essay I hope I have shown that, while society may unquestionably consist of individuals, methodological individualism's claim that explanations of the social can be reduced to explanations about individuals is not unquestionable.

Bibliography

• ELSTER, JON (2003), "Marxism, Functionalism and Game Theory: a case for methodological individualism". In Matravers & Pike (Ed.) Debates in Contemporary Political Philosophy. London: Routledge
• HEATH, JOSEPH (2005). "Methodological individualism". Stanford Encyclopaedia of Philosophy. Retrieved March 16, 2007 from http://plato.stanford.edu/entries/methodological-individualism/
• KINCAID HAROLD (1986), "Reduction, Explanation, and Individualism", Philosophy of Science, Vol. 53, No. 4. (Dec., 1986), pp. 492-513.
• LUKES, STEVEN (2003). "Methodological individualism". In Matravers & Pike (Ed.) Debates in Contemporary Political Philosophy. London: Routledge
• RUBEN, DAVID-HILLEL (1985). "Methodological individualism". The Metaphysics of the Social World. London: Routledge & Kegan Paul, 150-72
• RUBEN, DAVID-HILLEL (1998). "Explanation in history and social science". Routledge Encyclopedia of Philosophy. London: Routledge. Retrieved March 16, 2007, from http://www.rep.routledge.com/article/R007
• TUOMELA, RAIMO (1990). "Methodological Individualism and Explanation", Philosophy of Science,Vol.57,No.1.(Mar.,1990),pp.133-140.
• WRIGHT et al (2003), Wright, Levine and Sober. "Marxism and Methodological individualism". In Matravers & Pike (Ed.) Debates in Contemporary Political Philosophy. London: Routledge