Most writers on economic methodology tell essentially the same story about "operationalism." Operationalism was the philosophy of science popularized by the Nobel Prize-winning physicist Percy Bridgman; the main text was his Logic of Modern Physics, originally published in 1927, but a number of different variants of the program appeared in the literature during the period 1930-1950. There was never a definitive rejection of the program, but because of technical difficulties and also because of its general identification with positivist philosophy of science, discussion of operationalism has effectively disappeared from the philosophical literature. Operationalism was most stridently promoted in economics by Paul Samuelson, who offered it as the methodological backdrop for many of his early theoretical contributions, particularly Foundations (1947) and revealed preference theory (1938a). Although operationalism continues to receive a certain amount of ritual endorsement from practicing economists, few, if any, actually abide by (or even attempt to abide by) its methodological maxims.
The purpose of this paper is not to replace this standard story about operationalism with an alternative, equally condensed, view. The standard story is fine as far as it goes; it just does not go very far, and there is a much more complex, and much more interesting, story to be told about operationalism in general and its relationship to economics in particular. Although the following discussion constitutes little more than a few first steps in an ongoing and much larger project concerning the reception of positivist ideas-and the corresponding demise of pragmatist ideas-within American economics during the interwar period, it does provide a very different reading of the intellectual history of operationalism and in particular how such ideas might be, well, operationalized in economics. The discussion will focus on the variation among operationalist views (hence the title), and even though the standard interpretation admits that operationalism was more of a broad general framework than a single unified position, I will argue that the variation was actually much greater than commonly recognized. In addition, when we turn beyond the philosophical literature to the question of how operationalism was interpreted within the social/human sciences, then the variation becomes even more pronounced. The bottom line is that certain supporters of Bridgman's operationalism--John Dewey in particular--considered the main operationalist message, and its methodological implications for the social sciences, to be precisely the opposite of the message promoted by Samuelson in economics and by mid-century behaviorists in psychology.
The paper is arranged as follows. The first section will review the standard interpretation of operationalism with an emphasis on its relationship to logical positivism and the so-called received view of scientific theories. The second will consider how operationalism was interpreted in economics, particularly by Samuelson. The third will examine Dewey's very different, pragmatic reading of operationalism and highlight how antithetical the pragmatic interpretation is to that of Samuelson and the behaviorists. The final section will attempt to answer the so what? question and bring the discussion home to the history and methodology of economics.
Cognitive Significance, Correspondence, and Operations
The operationalist program presented in Percy Bridgman's Logic of Modern Physics (1927) is more often viewed as a friendly amendment to logical positivism than as a new, free-standing framework for the philosophy of science. In particular, it is viewed as one specific answer to the general problem of correspondence rules that played such an important role in the Vienna Circle's characterization of the structure of scientific theories. According to the early logical positivists, scientific theories consist of three main parts--logical, theoretical, and empirical--and each of these three parts is couched in terms of its own separate vocabulary. The directly empirical part of a scientific theory is restricted to the observational vocabulary; the terms in this observational vocabulary are considered to be directly, and incorrigibly, empirically observable. On the other hand, the purely theoretical aspect of the theory involves exclusively the logical and theoretical vocabularies; it consists of a set of theoretical propositions constructed from various components of these two different vocabularies. Since these theoretical propositions are nonobservational, there must be a tight linkage, a transmission mechanism, that allows such propositions (and the terms in the theoretical vocabulary more generally) to hook up to the empirical domain: the terms and expressions within the observational vocabulary. A fourth component of the positivist view of scientific theories--correspondence rules--performs precisely this necessary linkage. The correspondence rules translate the terms in the theoretical vocabulary into the observational vocabulary and thus into the incorrigible empirical basis of science. Given the logical positivists' verifiability criterion of meaningfulness, these rules play a fundamental role in determining the cognitive significance, and thus the legitimacy, of scientific theories. Since empirical verifiability is necessary for cognitive significance, and since, sans correspondence rules, theoretical propositions are devoid of empirical content, these rules provide the essential correspondence between the theoretical propositions of science and the domain of incorrigible observations which guarantee their cognitive significance.
While later logical empiricists allowed for a "comparatively loose and imprecise" (Nagel 1961, 99) specification of the linkage between the theoretical and empirical domains, the early positivists were quite strict about the character of correspondence rules. They required every term in the theoretical vocabulary be given an explicit definition in terms of the observational vocabulary. The correspondence rules thus provided necessary and sufficient conditions for meaningful application of any theoretical predicate. As Frederick Suppe explained, "[i]nitially correspondence rules had to have the form of explicit definitions which provide necessary and sufficient observational conditions for the applicability of theoretical terms; theoretical terms were cognitively significant if and only if they were explicitly defined in terms of the observation vocabulary" (1977, 18). Within the philosophical framework of such strict correspondence rules, the question of the "cognitive significance of the theoretical terms"--an issue of much debate among later logical empiricists--never even surfaced; the correspondence rules allowed the cognitive significance of the empirical basis to be transmitted directly and unimpeded into the theoretical vocabulary. Strict correspondence guaranteed the theoretical vocabulary inherited its cognitive significance directly from the observational domain; the entire cognitive weight of science thus rested squarely on the shoulders of the correspondence rules.
While Bridgman himself was more interested in correcting what he considered to be the bad metaphysical habits of practicing physicists than in solving deep philosophical problems about the cognitive significance of formalized scientific theories, his operationalism was quickly interpreted as just one particular variant of the logical positivist interpretation of correspondence rules. Bridgman's claims that a theoretical term is meaningful if and only if it "corresponds to" a specific set of operations, and that the meaning of any theoretical term should be defined as (thus is synonymous with) its corresponding set of operations, were viewed as one particular, rather restrictive, specification of the explicit definition of theoretical terms in terms of the observational, in this case operational, vocabulary.
Of course many problems have been identified with the operationalist characterization of such correspondence rules. Perhaps the most significant was initially pointed out in a review of Bridgman 1927 by L. J. Russell appearing in Mind in 1928 (also see chapter 6 of Hempel 1965 and chapter 2 of Suppe 1977); this is the definitional problem associated with multiple operations. If we take Bridgman at his word, and every concept is "synonymous with the corresponding set of operations" (1927, 5, emphasis in original), then every operation defines a different concept. To use an economic example, if we define the macroeconomic theoretical concept of the "price level" in terms of the operations used to measure the U.S. Consumer Price Index (CPI), then we will define a different theoretical concept by the GDP deflator, the Wholesale Price Index, or for that matter the CPI of some other country. We cannot define theoretical concepts in terms of operations unless we are ready to accept multiple theoretical concepts: one for each operation. If we believe that the "price level" means something more than what is measured by the operations used in the construction of one specific price index, then we cannot accept the narrow operationalist definition of such theoretical terms. This is certainly not the only problem with the correspondence rule interpretation of operationalism, but it is a significant problem that was recognized very early and never really given an adequate response by the program's various supporters.
Like other aspects of the logical positivist program--the strict distinction between theory and observation, the foundationalist interpretation of the empirical basis, the analytic character of mathematics and logic, and so on--the original interpretation of operationalism was substantially softened by the later logical empiricists. (1) By 1945 Herbert Feigl would say that "concepts which are to be of value...