This paper deals with the question of whether and when it is appropriate or inappropriate to say that a social group performs an action. After some remarks on the concept of action three kinds of groups are distinguished, i.e. assemblies, institutions, and classes. It is found that in the first two of these cases predication of action is possible: an assembly can act in that all its members act, or some of them do who are interchangeable with any others; and (...) an institution can act because it has a structure and some individuals can act on its behalf. A class, however, cannot be said to act, for its concept may be freely formed, its members cannot be assembled, and it has no organs through which actions could be performed. (shrink)

Past criticisms to the contrary, methodological individualism in the social sciences is neither trivial nor obviously false. In the style of Weber's sociology, it restricts the ultimate explanatory repertoire of social science to agents' reasons for action. Although this restriction is not obviously false, it ought not to be accepted, at present, as a regulative principle. It excludes, as too far-fetched to merit investigation, certain hypotheses concerning the influence of objective interests on large-scale social phenomena. And these hypotheses, in fact, (...) merit empirical consideration. The attractiveness of methodological individualism as a regulative principle depends on two independent confusions, the conflation of an agent's reasons for action with the beliefs, needs, desires, or goals which are the reasons why he acted as he did, and the identification of explaining a phenomenon and describing its causes. (shrink)