Reductionism is the hypothesis that science is unified by chains of intertheoretic reductions across disciplines, with theories from basic physics providing the ultimate ground. The classic work on intertheoretic reduction in the philosophy of science is chapter 11 of Ernest Nagel’s The Structure of Science. For Nagel, reduction is logical deduction (derivation) of the statements of the reduced theory TR from those of the reducing theory TB. In interesting cases from science’s history, the 7Rs contain terms that do not occur within the descriptive vocabulary of TB (e.g., equilibrium thermodynamics contains "pressure" and "temperature," which do not occur in statistical mechanics and the kinetic/corpuscular theory of matter). Such cases are especially prominent in the psychological and social sciences because their theories developed mostly independently of one another. To derive such 7Rs from Tg in something other than a trivial fashion, Nagel insisted that the premises of the derivation require bridge principles connecting terms across the theories (e.g., "temperature in a gas is mean kinetic energy of molecular constituents"). Furthermore, most historical scientific reductions are corrective—they indicate where the TR is false. Thus Nagel insisted that the premises of the deduction must contain not only the TB and the necessary bridge principles but also various limiting assumptions and boundary conditions on T^s application, some of which are contrary to fact. These elements circumscribe the falsity in the premises of the valid derivation of a false conclusion (TR) away from the (presumed true) TB. Challenges to Nagel’s account were quickly raised by philosophers of science, but virtually every alternative account of intertheoretic reduction—for example those by Kenneth F. Schaffner in 1967 and Clifford Hooker in 1981—emerged as a direct response to Nagel’s details.
With physics providing the ultimate reducing theories, reductionism is allied with physicalism about mental and social phenomena. As an explicit program, it has focused primarily on cases from biology and psychology rather than the social sciences. The rationale is straightforward: Claims about accomplished reductions in the former remain controversial, and if these controversies cannot be resolved, then the plausibility of the program for the latter seems remote. However, social phenomena do make an explicit appearance in one of the classic papers on reductionism in the philosophy of science. Paul Oppenheim and Hilary Putnam (1958, p. 7) illustrate the "working hypothesis" of the unity of science as follows:
6………. Social groups
5………. (Multicellular) living things
4………. Cells
3………. Molecules
2……….Atoms
1……….Elementary particles
Each level is related as parts (below) to wholes (immediately above), with "micro-reductions" hypothesized to obtain between theories explaining phenomena at a lower and an immediately higher level. And while Oppenheim and Putnam admit that accomplished micro-reductions from levels 6 to 5 have not advanced very far, they cite individual choice theories in economics and the "principal theoretical approaches" in sociology (Marxist, Veblenian, Weberian, Mannheimian) as examples of attempted micro-reductions to individualist psychology.
One popular general criticism of reductionism focuses on the multiple realizability of given higher-level kinds on lower-level mechanisms. (Bickle’s 2006 article "Multiple Realizability" provides a survey of these arguments and many reductionist responses, with an extensive bibliography.) An example from the social sciences often appealed to is Gresham’s law (colloquially stated, that "bad money drives out good," and more precisely stated in terms of what happens in monetary exchanges under specific exchange conditions). Monetary exchange is realized by a wide variety of physical phenomena—exchanging paper bills or minted coins or strings of wampum, signing checks, or depressing the "Enter" button on one’s desktop computer, to name just a few. It seems a safe empirical bet that no single physical kind, described and studied by some lower-level physical science, obtains in all these cases (not to mention in the myriad possible realizations of monetary exchange). One could disjoin all the lower-level realizers of such a high-level kind, but it seems an even safer bet that the resulting disjunctive kind does not occur in the explanatory laws of any physical science from which the laws of the higher-level science can be derived.
From the mid-1970s until the mid-1990s this argument held sway against reductionism. A few replies emerged during this time. Some reductionists advocated domain-specific reductions, limited to classes of realizers that shared lower-level mechanisms. Others pointed out examples of accomplished reductions from science’s history that included multiply realized kinds in the reduced theories, showing that multiple realizability is not sufficient by itself to block intertheoretic reduction. Since the late 1990s some critics have appealed to different "grains" in the characterizations of higher- and lower-level kinds. Advocates of multiple realizability tend to tolerate a wide variety of functional differences in their specification of higher-level kinds shared across distinct realizers, but they insist that a single (minute) physical difference indicates distinct lower-level kinds. When grains are matched across higher- and lower-level theories, multiple realizability vanishes. Others have noticed a dilemma for advocates of multiple realizability. The more that the physical mechanisms across distinct realizers are similar, the less likely is multiple realization of the same higher-level kind. The more that the physical mechanisms are dissimilar, the easier it is to find functional differences across realizers at the higher level and hence the less likely is multiple realization of the same higher-level kind. At the time of this writing, anti-reductionists appealing to multiple realizability owe responses to these challenges.
A more diffuse anti-reductionist argument focuses on the complexity of social phenomena. Most likely it is this intuition that lies behind critiques of reductionist approaches by critics of neoclassical economics: Reducing social phenomena to the dynamics of individual decision makers will fail to explain some crucial economic or social features. (Reductionism to individuals is often characterized as "doctrinaire" or "simplistic" in the social sciences.) But turning this intuition into rational argument is not easy. One recent attempt applies the sophisticated mathematics of complexity theory to social sciences—dynamical systems, chaotic nonlinearity, trajectories through high-dimensional state spaces, and the like. Explanations in the form of differential equations are proposed to account for the behavior of systems, but not in terms of the mechanical outputs of its components. (Timothy van Gelder’s application of complexity theory to cognitive psychology is a detailed attempt of this sort.) The principal worry is whether the "reductionism" targeted here is a straw man. If the behavior of systems requires these explanatory resources, then so will the lower-level sciences that appeal not only to the structure but also the dynamics and organization of the systems’ components. The philosopher William Bechtel (2001), for example, replies that complexity is compatible with reduction, in the form of decomposition and localization explanatory strategies. He argues not just on conceptual grounds but also by appealing to empirical work by cognitive neuroscientists studying memory.
