Philosophy of History Part XXII: Carl Gustav Hempel and the Return of Positivist History

If you cannot predict, you have not explained. –Carl Gustav Hempel

Carl Gustav Hempel (1905–1997) was a German-American philosopher of science and a prominent member of the Vienna Circle. These philosophers were called logical positivists—“logical” because they argued that scientific laws could and should be reformulated as logically necessary deductions from a given set of observations; and “positivists” because they believed that only science produced genuine (i.e., “positive”) knowledge. In their view, most philosophy was metaphysical or speculative, and not worth pursuing. Instead, philosophers should assist scientists by clarifying the logical structures of explanations, so that invalid inferences could be spotted quickly (thus saving scientists time) and valid inferences recast as logically necessary deductions (thus bolstering the work of scientists).

Although this project has since become defunct, largely due to the work of Willard van Orman Quine, it was extremely influential in the first half of the twentieth century. In 1942, C.G. Hempel published a paper titled “The Function of General Laws in History,” in which he extended the logical positivist program to history. Debates over its merits absorbed considerable attention in the historical profession over the next three decades, and quite a few historians, persuaded by Hempel’s arguments, began to think of their work as a branch of the sciences rather than the humanities. His major points can be summarized as follows.

All valid explanations take the form of the Deductive-Nomothetic (D-N) model, according to which

1. we establish empirically that an event occurred, or in other words, the initial conditions
2. describe the event in terms of general laws, or in other words, make a prediction
3. demonstrate that, in light of the foregoing, the event was logically necessary, and then
4. describe those circumstances in which the explanation does not hold, and why, or in other words, the boundary conditions.

For instance, an explanation of the trajectory of an artillery shell would require that we

1. demonstrate that the artillery shell was fired and that it charted a certain trajectory
2. describe the operations of gravity, combustion, aerodynamics, and so on, which govern the trajectory of the shell, and then give an equation specifying their effects
3. demonstrate the logically necessary connection between (1) and (2), and then
4. specify which events might invalidate the predictive apparatus (the shell was a dud, the barrel melted from heat, etc.).

According to Hempel, this is the only valid form of an explanation possible—it must demonstrate the logically necessary connection between general laws and particular instances, and it must do it in a form that enables prediction. An explanation that does not fulfill these criteria is either a “pseudo-explanation” or an “explanation sketch.” A pseudo-explanation is simply no explanation at all, it merely appears to be one. An explanation sketch, on the other hand, fulfills some of the requirements of the D-N model, but is incomplete. So, for instance, in history a genuine explanation of the American Revolution would demonstrate by way of documentary evidence that it occurred, by way of general laws that revolutions always occur in certain circumstances, that the American Revolution bears a logically necessary connection to a general law of revolutions, and that the boundary conditions that might have prevented a revolution from breaking out were absent. A pseudo-explanation, on the other hand, would explain it in terms of “the spirit of the times,” “the will of the American people,” “the necessary consequence of the dialectic process,” or some other unempirical and unverifiable abstraction. Because these explanations are not about things that can be observed, and because they make no substantive or testable predictions, they are not explanations at all. An explanation sketch, which, according to Hempel, is what we usually encounter in history, would take the form:

The Revolution broke out because the colonists believed they were being taxed unfairly, and had come to see the British as foreign occupiers.

The explanation-sketch implicitly asserts that unfair taxation and the perception of foreign occupation act as causal mechanisms in producing, not just the American Revolution, but revolutions in general—but it is not specific enough, and it does not have the necessary logical form. It’s close to an explanation, but it hasn’t arrived yet.

Hempel was not the first person to take historians to task for producing vague and untestable explanations. At least as far back as Henri St. Simon (the originator of the term “positivist”), it has been claimed that all phenomena of any description whatsoever must be explicable in terms of scientific and predictive laws, and ultimately reducible to the laws of physics. Henry Thomas Buckle and Karl Marx continued in this vein, and couched their explanations in just the sort of predictive terms that Hempel advised. However, the more common reaction among historians has been to reject positivism as naively reductionist and irrelevant to their practice. If there are laws of history, it is far from clear how their existence could be demonstrated, since history offers no laboratory in which to conduct experiments, and is not concerned with the future, but with the past, and therefore does not admit of the possibility of prediction (only retrodiction); or to what they would apply, since historical events are so complex and contingent that each one constitutes a unique, non-repeatable phenomena, which does not admit of any general, but only particular, explanations; or that they exist at all, since the determinism that positivists take for granted is itself a piece of metaphysical explanation that can never be demonstrated from the evidence, and that, indeed, seems wholly inapplicable to human beings, whose choices are the subject matter of history.

Hempel, aware of these criticisms, answered that successful explanation is a matter of logical form. A good explanation is not necessarily one that makes a successful prediction possible in the present, but one which would have made a successful prediction possible at some point prior to the event. In physics, for instance, an explanation of the origins of the sun and the planets is not a prediction, but it is nonetheless given in a form that would make prediction possible beforehand, whereas a historian who explains the American Revolution in terms of grievances over taxation has hardly demonstrated that the Revolution was a necessary consequence of those grievances, since he has not specified wherein the difference lies between tax grievances that do, and do not, make revolution inevitable. Similarly, the physical sciences too deal with very complex phenomena (the human body, for instance), but they do so by dividing up the labor in such a way that only one particular aspect of that phenomena has to be explained at any given time. A physiologist who explains spinal degeneration has not, and need not claim to, explain the color of the patient’s hair or their height. He has nonetheless successfully explained a part of the human body, to the extent that he has followed the D-N model. In history, also, one cannot appeal to the unique character of the American, as distinct from all other, revolutions, or to the complexity of the topic—the task is not to explain the whole of American society at that time, but only a particular aspect of it. The contention that human beings make choices, and are thus qualitatively different from protons or tectonic plates, and so require different forms of explanation, Hempel dismisses as mystical obfuscation. Human beings are physical entities, and all physical entities are governed by the laws of physics—ergo, their decisions are the consequence of laws, not choices. Determinism is the only valid explanation of the world because, while it may not itself be demonstrable, it is demonstrable that all other forms of explanation are false. It must therefore be our default position.

Few historians today would describe themselves as positivists or as scientists (at least not without qualification), but Hempel’s arguments were highly influential for a time, and they continue to provide material for philosophers of history to consider. Hempel probably set the bar higher than we can realistically expect to reach, at least at this moment, and positivism in history has all (perhaps more) of the same problems that it does in science, philosophy, or anywhere else. Nevertheless, it is difficult to deny that history is far less rigorous, from a methodological or explanatory point of view, than we would like, and it seems reasonable to ask what we can do to improve it. By forcing historians and philosophers to consider these questions, he almost certainly improved the quality of discussion in that field. As we will see in a later article, one of the most powerful responses to Hempel is not that history has no logical structure (though this too has been argued), but that its structure is that of a narrative sentence.

This post is the twenty-second is a series on the philosophy of history; the previous article in the series is here, the next is here.

Daniel Halverson is a graduate student studying the History of Science and Technology of nineteenth-century Germany. He is also a regular contributor to the PEL Facebook page.