An implication of the lack of direct access to the past concerns the testability of historical hypotheses.
The historian cannot perform experiments on his work like the scientist in a laboratory. And if this is the case, then how can the historian test his theories? One may, however, argue that even the historian’s hypotheses are testable through examining logical consistency and their ability to explain historical evidence. We argue that the supposed lack of the historian’s direct access to the data is not a stumbling block for testing truth in history and gaining an accurate knowledge of the past.
It may be that the historian tests a hypothesis in a similar manner to a scientist. For example, there seems to be congruency between a model, the hypothetico-deductive model, that one adopts in the sciences that can also apply to the historical domain. This model formulates a hypothesis in a manner that could conceivably be falsified by a test on observable data (1). The scientist produces a hypothesis to explain available facts and then infers from the hypothesis specific conditions that would either confirm or disprove the hypothesis.
The historian can follow this same method. He can reconstruct a picture of the past constituting his hypothesis. He is then able to deduce certain conditions from his hypothesis that will either confirm or disprove his hypothesis. As the historian and archaeologist Robin Collingwood explained, “The historian’s picture of the past stands in a peculiar relation to something called evidence. The only way in which the historian can judge of its truth is by considering this relation” (2).
The historian’s hypothesis must be corroborated by the evidence, for example, archaeological and/or documentary evidence. William Debbins notes that “By treating coins, pottery, weapons, and other artifacts as evidence the historian raises his study to the level of a science. What happened in the past is what the evidence indicates as having happened” (3).
Beyond the hypothetico-deductive model, one can infer to the best explanation. Gilbert Herman explains that,
“In making this inference one infers, from the fact that a certain hypothesis would explain the evidence, to the truth of that hypothesis. In general, there will be several hypotheses which might explain the evidence, so one must be able to reject all such alternative hypotheses before one is warranted in making the inference. Thus one infers, from the premise that a given hypothesis would provide a “better” explanation for the evidence than would any other hypothesis, to the conclusion that the given hypothesis is true” (4).
Importantly, historical knowledge emphasizes probability rather than mathematical certainty. One accepts a hypothesis as historical when it corresponds to the evidence in such a way that any reasonable person ought to accept it. Essentially, one accepts something that has sufficient evidence to make it probable.
From the inference to the best explanation model, one begins with the evidence already available and infers what would provide the best explanation of that evidence. A scientist may test his hypothesis by conducting experiments while the historian can test his theory by seeing how well it explains historical evidence.
There are several elements to be considered in the historian’s reconstruction of the past through the inference to the best explanation. This was succinctly presented in C. Behan McCullagh’s book Justifying Historical Descriptions (1984) in which he outlines several points historians can use in testing historical hypotheses: (5)
1. The hypothesis, together with other true statements, must imply further statements describing present, observable data.
2. The hypothesis must have greater explanatory scope (that is, imply a greater variety of observable data) than rival hypotheses.
3. The hypothesis must have greater explanatory power (that is, make the observable data more probable) than rival hypotheses.
4. The hypothesis must be more plausible (that is, be implied by a greater variety of accepted truths, and its negation implied by fewer accepted truths) than rival hypotheses.
5. The hypothesis must be less ad hoc (that is, include fewer new suppositions about the past not already implied by existing knowledge) than rival hypotheses.
6. The hypothesis must be disconfirmed by fewer accepted beliefs (that is, when conjoined with accepted truths, imply fewer false statements) than rival hypotheses.
7. The hypothesis must so exceed its rivals in fulfilling conditions (2)–(6) that there is little chance of a rival hypothesis, after further investigation, exceeding it in meeting these conditions.
From McCullagh’s points it is clear that some historical hypotheses will be stronger than others in that they offer greater explanatory scope and power, plausibility, are less ad hoc, and so on. It could be that a hypothesis fulfills several of these conditions but is deficient in other areas, which means that applying the inference to the best explanation model can prove tricky.
Nonetheless, if one hypothesis can be shown to outmatch other explanations given McCullagh’s model, then it is likely to be true. The historian should accept the hypothesis that best explains all the evidence.
1. Thomas, B. 1993. The Philosophy Behind Physics. p. 86
2. Collingwood, R. 1956. The Idea of History. p. 246.
3. Debbins, W. “Introduction” in Essays in the Philosophy of History. p. xiv.
4. Gilbert Herman quoted by The Information Philosopher. Available.
5. McCullagh, B. 1984. Justifying Historical Descriptions. p. 19.