Scientific realism, Ramsey sentences and the reference of theoretical terms

Abstract

It is often thought that questions of reference are crucial in assessing scientific realism, construed as the view that successful theories are at least approximately true descriptions of the unobservable; realism is justified only if terms in empirically successful theories generally refer to genuinely existing entities or properties. In this paper this view is questioned. First, it is argued that there are good reasons to think that questions of realism are largely decided by convention and carry no epistemic significance. An alternative conception of realism is then proposed, which focuses on the Ramsey sentences of scientific theories, constructed in the manner of David Lewis's ‘How to define theoretical terms’. It is argued that because the Ramsey sentence of a theory preserves the epistemically significant part of the theory's content without generating commitments to any particular conclusions about reference, the realism issue is better addressed by asking whether Ramsey sentences of theories, rather than the theories themselves, are approximately true.

Notes

Correspondence: Department of Philosophy, King's College London, The Strand, London WC2R 2LS, UK. E‐mail: [email protected]

A related claim is also defended in Cruse and Papineau (2000). In the present paper I present a slightly different argument for identifying the potentially knowledge‐expressing content of a theory with the content of its Ramsey sentence.

Another reason is that it is thought that realism requires terms in successive theoretical paradigms to be capable of co‐reference, to avoid incommensurability worries. In fact, I think that this is false, for much the same reasons that I will discuss in this paper, but there will not be room to make this point in any detail.

The theory was put forward initially by Saul Kripke (1980) as a theory of proper names and then elaborated by Putnam (1973, 1975).

One might think that ‘principles of charity’ would help here, steering us towards generous conclusions when there is a choice (cf. Putnam 1978, p. 23). However, this would make successful reference a matter of default, undermining the claim that reference plays any substantial role in explaining theoretical success (cf. the points made against the referential generosity suggested by Hardin & Rosenberg, cited in Laudan, 1984).

According to traditional usage, the Ramsey sentence of a theory does not include a uniqueness claim; Lewis (1972) calls the Ramsey sentence with uniqueness the ‘modified Ramsey sentence’ (p. 254). However, I will use the term ‘Ramsey sentence’ to refer to Ramsey sentences with uniqueness.