The 1960s witnessed the development of a second generation of analytic philosophy of science inspired broadly by logical positivism and the Vienna Circle (post, post). The “received view” of the 1950s and 1960s, as expressed by philosophers such as Carl Hempel, Ernest Nagel, and Israel Scheffler (and others pictured above), presented a view of scientific knowledge as consisting of theories, general laws, observational predictions, and a logic of confirmation according to which theories are evaluated by their observational consequences (Patrick Suppes, The Structure of Scientific Theories). There was a largely unquestioned assumption that all scientific theories have the same logical structure (the unity of science doctrine; post); so physics, economics, or evolutionary biology could equally be represented as a system of theoretical terms and statements, a set of general laws, and a set of observational or experimental consequences that can be empirically evaluated. Here is an illustration of the idea in the natural sciences; the black line is the theoretical calculation of ferromagnetic properties, and the colored lines are the observed behavior of metals under difference conditions.
This general view of the logic of scientific knowledge was applied to the social sciences by some philosophers of science, and the result was a neo-positivist philosophy of social science. Richard Rudner was an important player in the formation of the received view in the 1960s, and he served as editor-in-chief of the core journal, Philosophy of Science, from 1959 to 1975. (His collected papers are held at Washington University, and the breadth of his correspondence gives some idea of his centrality in discussions of neo-positivist philosophy of science.) His 1966 textbook, Philosophy of social science, represents the high-water mark of neo-positivist philosophy of social science. So let’s review the main features of Rudner’s representation of social-scientific knowledge. (May Brodbeck’s 1968 Readings in the Philosophy of the Social Sciences is a fairly direct complement to Rudner’s book, in that it provides a fairly definitive representation of the philosophy of social science in the late 1960s.)
Several core assumptions about the social sciences are advanced in Rudner’s book.
- unity of science and naturalism; all sciences should resemble physics
- emphasis on the distinction between the context of discovery and context of justification
- insistence on the symmetry of explanation and prediction
- insistence on the essential role of lawlike generalizations in explanation
- fundamental reliance on a strict distinction between observation and theory
- advancement of “formalizability” as a desirable characteristic of a theory
Here is Rudner’s definition of a theory:
A theory is a systematically related set of statements, including some lawlike generalizations, that is empirically testable…. Accordingly, to the extent that a theory has been fully articulated in some formulation, it will achieve an explicit deductive development and interrelationship of the statements it encompasses. (10-11)
This formulation doesn’t yet make explicit the notion that “theory” is non-observational, or that theoretical concepts lack direct criteria of application to observational experience. Rather, Rudner emphasizes the characteristic of formalizability and deductive consequences. And this takes him in the direction of non-observational concepts; they are the concepts that are introduced into a scientific system as formal “place-holders” without explicit definition. Instead, their meaning is determined by the deductive consequences that the sentences in which they appear give rise to. The intended analogy here is with the axioms of geometry; the concepts of “point,” “line,” and “plane” are introduced as primitives without further definition, and their meaning is defined simply by the role they play in the full deductive system of geometry. So, notably, Rudner treats the problem of theory formation in the sciences as entirely analogous to the problem of creating a formal mathematical system — number theory or topology, for example. All concepts are either primitive — introduced without definition; or defined — introduced with strict logical definitions in terms of other concepts that already exist in the theoretical vocabulary. (It is interesting to note that Carl Hempel’s contribution to the Vienna Circle International Encyclopedia of Unified Science is on this subject; Fundamentals of Concept Formation in Empirical Science (International Encyclopedia of Unified Science, Vol 2 No 7) .)
Notice how foreign this approach is to actual theoretical work by social scientists. Weber or Durkheim do not attempt to strip away “connotation” or ordinary associations from their theoretical concepts; rather, they make full use of the complexity and open-endedness of concepts like capitalism or solidarity as a way of allowing them to describe and theorize complex social realities. Rudner’s approach seems to reflect the program of logical positivism, attempting to utilize the apparatus of formal logic to clarify scientific reasoning. But it misses the mark in application to the actual reasoning of path-breaking social thinkers such as Durkheim, Weber, Tocqueville, or Marx.
Rudner turns to the observation-theoretic distinction in short order. First, he describes observational concepts or predicates in these terms: “One set of primitives should be comprised of observational, or as we shall sometimes say, experimental terms (i.e., that such predicates should refer to observable features of the universe)” (21). And second, he introduces theoretical concepts: “The latter refer to nonobservable or nonmanifest characteristics of nonobservable entities. Thus, theoreticals include terms such as ‘electron,’ ‘superego,’ ‘institutional inertia,’ ‘cultural lag,’ which do not or are not (when the appropriate theories come to be formulated) likely to apply to observable entities at all” (23).
This is the standard view of the observation-theoretic distinction within the “received view” of neo-positivist philosophy of science. It was recognized that it is not possible to achieve the scientific goals of physics or chemistry if we are restricted to a purely observational vocabulary. We evidently need to refer to hypothetical entities and forces if we are to be able to explain observable phenomena. So hypothetical constructs must be permitted. But their “empirical content” is exhausted by their observational consequences.
Rudner attempts to provide a positivistic interpretation of Weber’s concept of ideal types. He writes, “Our concern is with the logical character of idealizations and their methodological uses. In this latter respect, concepts that happen to be idealizations have, as we shall see, no special methodological role–regardless of the heuristic or suggestive value they may have in leading theorists to form hypotheses or frame theories about related phenomena” (56). This is a surprising exercise, since it involves the requirement that the social scientist provide a precise specification of the scope and implications of the term. I’ve always understood Weber’s emphasis to be on the open-endedness of ideal-type concepts; and certainly Weber’s use of ideal-type concepts did not remotely resemble the axiomatic, formalistic use that Rudner attributes to scientific concepts generally.
So what is the logic of scientific explanation in the social sciences, according to Rudner? His answer is straight from Hempel’s covering law theory:
The formal structure of a scientific explanation of some particular event has three parts: first, a statement E describing the specific event to be explained; second, a set of statements C1 to Cn describing specific relevant circumstances that are antecedent to, or otherwise causally correlated with, the event described by E; third, a set of lawlike statements L1 to Ln, universal generalizations whose import is roughly ‘Whenever events of the kind described by C1 through Cn take place, then an event of the kind described by E takes place.’ (60)
Perhaps the key philosophical assumption that Rudner defends is the idea that the logic of science is everywhere the same. Concept formation, deduction, confirmation, and explanation all involve the same formalizable operations. A theory should be formalized as a first-order deductive system with primitive terms and defined terms; some of its terms should have criteria of application to observable outcomes; explanation proceeds by deriving a description of the explanandum from the theory; and the theory should be tested by evaluating the truth or falsity of its observational consequences.
In addition to these extended discussions of the putative logic of social-science reasoning, Rudner also addresses the topics of “objectivity” in social-science research and the concepts of functionalism and teleology in the social sciences.
What are the strengths and weaknesses of this approach to the philosophy of the social sciences? The strength of the book is also its debilitating weakness: it is an exceptionally clear exposition and development of the central ideas of logical positivism in application to social-science theorizing. Unfortunately, this approach proves to be singularly unhelpful when it comes to actually understanding and criticizing the social sciences as they actually exist. It has the effect of attempting to force a style and form of reasoning onto social researchers that deforms the ways in which they attempt to theorize and explain the social world.
Key weaknesses include these points.
First, this formulation pays virtually no attention to the actual content and methods of existing social science disciplines. The foundational premise is that all sciences have the same fundamental logic. So it isn’t necessary to examine sociology, political science, or economics in detail in order to trace out the particular characteristics of inquiry, explanation, and theory in these disciplines. Either the social-science disciplines conform to the received view, or they do not and for that reason show themselves to be defective as science. So the philosophy of a special science is simply the special case of the more general theory of scientific knowledge represented by the received view. This is a bad way of beginning the philosophical study of any science, whether the social sciences or the biological sciences.
Second, by highlighting the issues that are central to the exposition of the received view — for example, the observation-theoretic distinction — the neo-positivist philosopher of social science is drawn away from consideration of other, more substantive problems to which philosophers could make a useful contribution. For example, the assumptions of purposive rationality underlie many explanations in areas of political science and economics; and it turns out to be very productive to examine the intricacies that arise when we try to give a careful explanation of the concept of rationality and to link this assumption to particular areas of social explanation. The intellectual disposition created by the neo-positivist view is to reduce this question to a simple matter of “concept formation.” “Rationality” becomes simply another theoretical construct to be introduced into scientific theories. But as Sen, Harsanyi, Margolis, and dozens of other philosophers and social scientists have demonstrated, we need to spend quite a bit of intellectual energy to the task of unpacking the theory of rationality if it is to be of use in the social sciences.
Third, this approach imposes an inappropriate simplification on the social sciences when it comes to empirical evaluation of social-science hypotheses. It presupposes the comprehensive generality of the hypothetico-deductive model of confirmation theory. Once again, the intellectual error derives from the assumption that the logic of scientific reasoning must be the same in every area of science. The social sciences sometimes involve the sorts of theoretical systems that are found in physics; but more commonly a social science analysis is comprised of a number of relatively independent models and mechanisms — theories of the middle range — that are amenable to piecemeal evaluation. Social science analyses are not generally not unified theoretical systems along the lines of the theory of thermodynamics or genetics. And there are other ways of providing empirical evaluation and support for these sorts of analyses — for example, process-tracing, piecemeal empirical evaluation, and applying the logic of comparative causal analysis.
In hindsight, the neo-positivist paradigm for analyzing the social sciences just didn’t go very well. It pushed an esoteric theology of logical analysis onto the enterprise of understanding society that really didn’t conform very well to the reasoning and explaining that talented sociologists, anthropologists, and political scientists were creating for themselves. And it did not succeed in focusing productive attention by philosophers on the difficult problems of theory construction and explanation that social scientists really confronted.