Abstract

For the last three decades, the discussion on Hilary Putnam’s provocative suggestions around the issue of realism has raged widely. Putnam’s various formulations of, and arguments for, what he called internal realism in contrast to what he called metaphysical realism have been scrutinised from a variety of perspectives. One angle of attack has been missing, though: the view from the social sciences and the ontology of society. This perspective, I believe, will provide further confirmation to the observation that Putnam’s two concepts of realism are all too aggregative in that they conflate elements that had better be kept distinct, at least for many important purposes (e.g. Niiniluoto 1996). The present essay can be read as an argument for a topic-specific examination of realism. This means that it challenges the overall validity of Richard Boyd’s arguments against what he calls “realism about”, that is, against “a certain fragmented conception of scientific realism, one according to which realism is deeply topic specific” (Boyd 1990, p. 175). Boyd takes the ‘x’ in realism about x” to designate two kinds of things: entities postulated in scientific theories (“realism about the ether” and “realism about higher taxa”) and scientific disciplines (“realism about physics” and “realism about biology”) (ibid., pp. 175, 190). He also suggests that “it seems possible cogently to accept realism about the natural sciences while denying it about at least some of the social sciences” (ibid., p. 191). In Boyd’s view, the possibility of realism about social sciences is undermined by their weak instrumental success; it is the instrumental reliability of method that he takes as the basis for a realist account of science. I do not underwrite this connection between realism and instrumental success, thus I am not compelled to reject realism about social sciences on these grounds; and, as we shall see, social sciences have other peculiarities as well (see Mäki 1996). I think we should, at least to some degree, disaggregate and relativize the issue of realism. I do have sympathy with Boyd’s strategic arguments against “realisms about”, but yet I believe realists will be better off by adopting a more concrete and localized approach to the issue (Mäki 2005). Social objects and social sciences deserve a separate treatment. There are two strategies that can be followed in examining a realism about x (type of entity, theory, discipline). One is to first fix the meaning of realism’ and then specify the extension of ‘x’. This is Boyd’s strategy. The other is to fix some part of the extension of ‘x’ and then adjust the meaning of ‘realism’ so as to accommodate whatever peculiarities the x in question may have (and finally to check whether the adjusted meaning of ‘realism’ meets the criteria of a minimal notion of realism; it would be here that we encounter a relatively fixed idea of some minimal realism, or a set of realist intuitions). The latter is my approach here. I begin with the interconnected ideas of “realism about social objects” andrealism about social sciences”. I am not only interested in what comes after ‘about’; I am also interested in what goes before it, namely ‘realism’ itself. Here is one premise of the discussion: the issue of proper forms of realism for dealing with specific problems is an issue to be settled at least partly a posteriori. Here is the problem for which a solution will be sought: what constraints, if any, does what we know about the social sciences and society (as depicted by the social sciences and our commonsense views) impose upon the forms of realism that can justifiably be adopted about them? This paper suggests to offer only some partial insight into this issue. It seeks to do so indirectly by discussing Putnam’s characterizations of internal and metaphysical realism. The focus will be on two aspects of Putnam’s realisms: [1] the role and kinds of independence and dependence (in relation to the human mind and related things); and [2] the possibility of error. It is shown that the nature of social objects has implications concerning the appropriate constraints on [1] and [2], and thereby on the kinds of realism that are available to social scientists.