Philosophy of Science March, 1972 “WHAT'S WRONG WITH THE RECEIVED VIEW ON THE STRUCTURE OF SCIENTIFIC THEORIES ?”"* FREDERICK SUPPE University of Ilinois at Urbana-Champaign Achinstein, Putnam, and others have urged the rejection of the received view on theories (which construes theories as axiomatic calculi where theoretical terms are given partial observational interpretations by correspondence rules) because (i) the notion of partial interpretation cannot be given precise formulation, and (ii) the observational-theoretical distinction cannot be drawn satisfactorily. I try to show that these are the wrong reasons for rejecting the received view since (i) is false and it is vir- tually impossible to demonstrate the truth of (ii). Nonetheless, the received view should be rejected because it obscures a number of epistemologically important features of scientific theorizing. 1 show this by sketching an alternative analysis which reveals some of these features and gives a more faithful picture of scientific theorizing. For some time the Received View on scientific theories has been that theories are to be construed as axiomatic calculi in which theoretical terms are given a partial observational interpretation by means of correspondence rules; underlying this analysis is a strict bifurcation of the nonlogical terms of the theory into an observa- tional vocabulary and a theoretical vocabulary. Recently Putnam, Achinstein, and others have urged the rejection of the received view analysis of scientific theories because (i) the notion of partial interpretation employed cannot be given a precise formulation adequate for the purposes of the received view, and (ii) the observa~ tional-theoretical distinction cannot be drawn satisfactorily.” It is the contention of * Received February, 1971, + Cf. [19] and [1], pp. 85-91, 157-158, 197-202; [1] incorporates with minor changes Achin: stein’s earlier writings on the subject. In [19] Putnam also urges that the observational-theoreti- cal distinction is untenable because it is misleading both to label the class of nonobservational terms ‘theoretical terms’ and to characterize sentences formulated solely in terms of the observa~ tional vocabulary and sentences formulated solely in terms of the theoretical vocabulary as observational sentences and theoretical sentences respectively. While this is true, it hardly necessitates rejection of the received view. In [1], pp. 199-201, Achinstein suggests that it will be epistemologically more revealing if we avoid reliance on an observational-theoretical distinction in our analysis of theories, though he advances this as a corollary to his arguments in support of Gi In both cases, the strength of their contention that the received view should be rejected lies in the establishment of (i) and (ii), and we shall confine our attention to those arguments. would like to thank Mr. Nicholas Georgalis and my colleagues, Professors Thomas Nickles and Robert Stalnaker for helpful comments on earlier drafts of this paper. I would also like to thank this Journal’s referees for suggesting various improvements. 12 FREDERICK SUPPE this paper that the received view is unsatisfactory and ought to be rejected, but that reasons (i) and (ii) are the wrong reasons for rejecting it. Section I of this paper claims that reason (j) is false and that it is virtually impossible to establish reason (i). Part II attempts to show that the received view, nonetheless, is unsatisfactory and ought to be rejected because its reliance on the observational-theoretical dis- tinction causes it to obscure a number of epistemologically important and revealing features of the structure of scientific theories. In the process of arguing for this latter claim, a more adequate account of the epistemological structure of scientific theories is presented. I Achinstein’s and Putnam's arguments in support of reason (i) for rejecting the received view proceed by observing that the notion of partial interpretation has not been made clear by advocates of the received view, considering a number of pos- sible explications of the notion, and then showing that they are inadequate for the purposes of the received view. ({1], pp. 85-91, [19], pp. 244-248.) In [24] I have given a precise account of partial interpretation which is adequate for the purposes of the received view. Rather than duplicate that discussion here, I will merely claim on the basis of [24] that reason (j) is false, and confine my attention to reason (ii). The arguments which Achinstein and Putnam advance in support of reason (ii) ((1], chapters 5, 6; [19], pp. 240-244) attempt to show that (a) The observational-theoretical distinction cannot be drawn on the basis of the ordinary usage of scientific terms. Of course (ji) follows from (a) only if the further assumption is made that (b) To be tenable for the purposes of the received view, the observational- theoretical distinction must be drawn on the basis of the ordinary usage of scientific terms. This latter assumption neither is made explicit nor argued for in [1] or [9]. As such, Achinstein and Putnam have not made their case, But I wish to establish something stronger—namely that (a) is true whereas it is virtually impossible to establish (b). However, I do not want to base my claim that (a) is true on their arguments; for I do not find them wholly satisfactory: Achinstein’s arguments only show that the observational-theoretical distinction cannot be drawn on the basis of ordinary usage in the ways Carnap and others have suggested, and so establish a conclusion that is weaker than (a); and Putnam’s arguments contain numerous lacunae. Rather, I will refine the sorts of considerations they raise into a much tighter and stronger argument for (a); then I will use features of that argument to argue that (b) is virtually impossible to establish. In order to argue for my conclusions about (a) and (b), a fuller characterization of the received view is needed. Versions of the received view have been advanced by a number of authors, including Braithwaite ((2], pp. 22 ff.), Campbell ({4], Ch. 6), Carnap (e.g. [5], p. 43), Duhem ({7], p. 19), Hempel ({9], [10), Hesse [(11], (12), Kaplan ({13], pp. 298-299), Margenau ({15]), Nagel ({16], p. 90), Northrop [17],“‘WHAT’S WRONG WITH THE RECEIVED VIEW OF SCIENTIFIC THEORIES ?” 3 Ramsay ({21], pp. 212-236), and Reichenbach ((22], Ch. 8). Although there are a number of differences (some of them significant) in these various versions of the received view,? there is a substantial core of agreement among them. In particular, most proponents of the received view would agree on the following: The received view is not advanced as a descriptive account of how theories are formulated in actual scientific practice; rather it presents a canonical linguistic formulation for theories and claims that any scientific theory can be given an essentially equivalent reformulation in this canonical way. As such it is advanced as an explication of the notion of a scientific theory.* Moreover, it is claimed that this canonical linquistic formulation for theories will display the essential epistemic features of scientific theories. Such a canonical linguistic formulation will have the following episte- mologically revealing features: (1) The theory consists of theoretical laws and correspondence rules formulated in a language L'; (2) the nonlogical vocabulary of L can be exhaustively bifurcated into a observational vocabulary, Vo, consisting of terms which refer to directly observable attributes or entities, and a theoretical vocabulary, Vz, consisting of terms which refer to attributes or entities which are not directly observable; (3) the Jaws of the theory are formulated as sentences of L whose only nonlogical terms are from Vp; (4) the correspondence rules are for- mulated as sentences of L which contain terms from Vo and also from Vz, and are intended to embody various experimental procedures, etc., for applying the laws of the theory to directly observable phenomena®; (5) the meanings of the Vo terms are completely specified in terms of their corresponding directly observable attributes or entities; (6) no direct observational interpretation or meaning is given to the Vz terms, they being given an indirect partial empirical interpretation by the correspondence rules and the laws of the theory. Requirement (2) of the received view stipulates that the nonlogical terms of L be bifurcated into two disjoint classes—the observation terms and the theoretical terms. Since this distinction lies at the heart of the received view analysis, one would expect that, in advancing the received view, its proponents would have extensively discussed the nature of this bifurcation and the basis upon which it is 2 The primary disagreements are on the form of the correspondence rules. Campbell, Nagel, Hesse, and Kaplan maintain that in addition to satisfying conditions (1)-(6) below, the theory also must possess realizable or concrete models. Kaplan deviates from the other authors in that he claims that the analysis only works for one type of theory. Hempel no longer adheres to the received view, and now holds a similar position in which the observational-theoretical dis- tinction is replaced by a different bifurcation of terms not drawn on the basis of direct observa- tions; ef. [8] for details. 9 For an assessment of this claim, cf. my [23], pp. 38-46. 4 In many versions of the received view, the language L is supposed to be a symbolic language such as a first-order predicate calculus augmented by modal operators (e.g. in Carnap (5]); other authors allow L to be a natural language such as scientific English. Apparently the motiva- tion for employing symbolic languages is the increased precision and more precise deductive mechanism of symbolic logics. We will see below that the plausibility of an observational- theoretical bifurcation of terms depends to an extent on whether L is natural or artificial. ® Correspondence rules are referred to by various authors as coordinating definitions, dic- tionaries, interpretative systems, operational definitions, epistemic correlations, and rules of interpretation, Most authors impose further restrictions on the forms these correspondence rules have.