In Defense of Unrealistic Assumptions*

s the common and crucial elements from the mass of complex and detailed circumstances surrounding the phenomena to be explained and permits valid prediction on the basis of them alone.” 8 This is why assumptions of a theory must be unrealistic, and further, the more unrealistic the assumption set the better, ceteris paribus. In an attempt to increase the fruitfulness of the theory, one must reduce the number of assumptions and, as a result, make the assumption set unrealistic and incomplete as a description of complex empirical reality. One does not purport to make assumptions unrealistic for its own sake; it is just a necessary and unavoidable consequence of making the theory more fruitful. From this perspective, it is important to recognize that assumptions of a theory are not its scope conditions (Cohen 1989:183–89; Markovsky 1994:16–20). Scope conditions are universal statements that define the class of circumstances to which a theory applies (Cohen 1989:83). Assumptions do not define the theory’s empirical range of applicability; scope conditions do. When certain empirical phenomena fall outside a theory’s scope conditions, one cannot use the theory to explain them, and one cannot use them to confirm or refute the theory (Walker and Cohen 1985; Cohen 1989:82–84; Markovsky 1994:19–20). However, when a theory’s assumptions are not realistic and accurate descriptions of a certain part of the empirical world, it does not mean that one cannot use the theory to explain phenomena in that part of the empirical world. For assumptions are always empirically inaccurate as descriptive statements because they are necessarily simplifications. Many of the criticisms of rational choice theory and game theory quoted above reflect this confusion of assumptions and scope conditions (see especially the quote from Kollock). When theorists talk about “relaxing the assumptions” to make the theory more applicable, what they really mean is “relaxing (or broadening) the scope conditions.” Assumptions have nothing to do with the theory’s applicability. RECONCILING FRIEDMAN’S ARGUMENT WITH THE REALIST VIEW OF SCIENCE My argument that a theory’s assumption set ought to be as unrealistic as possible closely resembles Milton Friedman’s position in his 1953 classic article “The Methodology of Positive Economics.” Friedman’s view stems from an instrumentalist philosophy of science (Boland 1979). Instrumentalists believe that theories are neither true nor false in any ontological sense. They aver instead that theories are useful computational devices that produce empirical predictions. If its predictions are supported by empirical data, the theory is a good instrument; if the predictions are empirically false, then the theory has no merit (Keat and Urry 1975:63–65; Boland 1979:210–11). 8In his critique of rational choice theory, Smelser (1992:392–93) discusses Friedman’s (1953) argument as a possible defense for its “unrealistic assumptions.” It is instructive to note that Smelser fails to discount this argument by Friedman when he discounts all other possible defenses for rational choice theory. 9Another potential reason for descriptive incompleteness of a theory’s assumptions is the theorist’s cognitive limitations. As human beings, theorists have limited capacity to comprehend and describe the infinite complexity of the empirical world and must therefore limit themselves (and their theories) to a manageable range of the empirical complexity. I thank Trond Petersen for pointing out this possibility to me. 10“What Smelser means by realism [in his critique of rational choice theory] amounts to descriptive accuracy” (Farmer 1992:413). 11Musgrave (1981:237–39) calls scope conditions “domain assumptions” and attempts thereby to subsume them under assumptions. However, assumptions and scope conditions have different functions in a scientific theory and thus one needs to separate them completely (Cohen 1989; Markovsky 1994). 12Even Jasso (1988:5) exhibits some evidence of this confusion in her otherwise brilliant article when she states: “The more general the postulate set, the greater will be the scope of the theory.” 13Friedman himself did not characterize his argument as instrumentalist, but did in private communication acknowledge that Boland’s characterization of it as such is “entirely correct” (Caldwell 1980:226). 198 SOCIOLOGICAL THEORY