3.1 General laws in history?
The philosopher of science Carl Hempel stimulated analytic philosophers’ interest in historical knowledge in his essay, “The Function of General Laws in History” (1942). Hempel’s general theory of scientific explanation held that all scientific explanations require subsumption under general laws. Hempel considered historical explanation as an apparent exception to the covering-law model and attempted to show the suitability of the covering-law model even to this special case. He argued that valid historical explanations too must invoke general laws. The covering-law approach to historical explanation was supported by other analytical philosophers of science, including Ernest Nagel (1961). Hempel’s essay provoked a prolonged controversy between supporters who cited generalizations about human behavior as the relevant general laws, and critics who argued that historical explanations are more akin to explanations of individual behavior, based on interpretation that makes the outcome comprehensible. Especially important discussions were offered by William Dray (1957), Michael Scriven (1962), and Alan Donagan (1966). Donagan and others pointed out the difficulty that many social explanations depend on probabilistic regularities rather than universal laws. Others, including Scriven, pointed out the pragmatic features of explanation, suggesting that arguments that fall far short of deductive validity are nonetheless sufficient to “explain” a given historical event in a given context of belief. The most fundamental objections, however, are these: first, that there are virtually no good examples of universal laws in history, whether of human behavior or of historical event succession (Donagan 1966: 143–45); and second, that there are other compelling schemata through which we can understand historical actions and outcomes that do not involve subsumption under general laws (Elster 1989). These include the processes of reasoning through which we understand individual actions—analogous to the methods of verstehen and the interpretation of rational behavior mentioned above (Dray 1966: 131–37); and the processes through which we can trace out chains of causation and specific causal mechanisms without invoking universal laws.
A careful re-reading of these debates over the covering-law model in history suggests that the debate took place largely because of the erroneous assumption of the unity of science and the postulation of the regulative logical similarity of all areas of scientific reasoning to a few clear examples of explanation in a few natural sciences. This approach was a deeply impoverished one, and handicapped from the start in its ability to pose genuinely important questions about the nature of history and historical knowledge. Explanation of human actions and outcomes should not be understood along the lines of an explanation of why radiators burst when the temperature falls below zero degrees centigrade. As Donagan concludes, “It is harmful to overlook the fundamental identity of the social sciences with history, and to mutilate research into human affairs by remodeling the social sciences into deformed likenesses of physics” (1966: 157). The insistence on naturalistic models for social and historical research leads easily to a presumption in favor of the covering-law model of explanation, but this presumption is misleading.