INFERENCE TO THE BEST EXPLANATION

In an inductive inference, we acquire a belief on the basis of evidence that is less than conclusive. The new belief is compatible with the evidence, but so are (possibly many) competing hypotheses that we are unwilling to infer. Such is the situation for a great number of the inferences we make, and this raises a question of description and a question of justification. What principles lead us to infer one hypothesis rather than another? And do we have any reason to believe that these principles are good ones, leading us to accept hypotheses that are true and to reject those that are false? Inference to the Best Explanation offers partial answers to both questions.

According to this model, explanatory considerations are a guide to inductive inference. We decide which of the competing hypotheses the evidence best supports by determining how well each of the competitors would explain that evidence. Many inferences are naturally described in this way. Seeing the ball next to the broken vase, I infer that my children have been playing catch in the house because this is the best explanation of what I see. Charles Darwin inferred the hypothesis of natural selection because, although it was not entailed by his diverse biological evidence, natural selection would provide the best explanation of it. When astronomers infer that a galaxy is receding from the Earth with a specified velocity, they do this because the supposition of such a recession would provide the best explanation of the observed red-shift of the galaxy's characteristic spectrum. When the detectives infer that it was Moriarty who committed the crime, they does so because this hypothesis would best explain the fingerprints, blood stains, and other forensic evidence. Sherlock Holmes to the contrary, this is not a matter of de duction. The evidence will not entail that Moriarty is to blame, since it always remains possible that someone else was the perpetrator. Nevertheless, Holmes is right to make his inference, since the supposition of Moriarty's guilt provides a better explanation of the evidence than does the supposition of anyone else's guilt.

Inference to the Best Explanation can be seen as an extension of the idea of "self-evidencing" explanations, where the phenomenon that is explained in turn provides an essential part of the reason for believing the explanation is correct. In the example above, the speed of recession explains the red-shift, but the observed red-shift may at the same time be an essential part of the reason astronomers have for believing that the galaxy is receding at that speed. Self-evidencing explanations exhibit a curious circularity, but this circularity is apparently benign. The recession is used to explain the red-shift and the red-shift is used to determine the recession, yet the recession hypothesis may be both explanatory and well supported. According to Inference to the Best Explanation, this is a common situation: Hypotheses are supported by the very observations they are supposed to explain. Moreover, on this model, the observations support the hypothesis precisely because it would explain them.

Inference to the Best Explanation thus partially inverts an otherwise natural view of the relationship between inference and explanation. According to that natural view, inference is prior to explanation. First we must decide which hypotheses to accept; then, when called upon to explain some observation, we will draw from our pool of accepted hypotheses. According to Inference to the Best Explanation, by contrast, it is only by asking how well various hypotheses would explain the available evidence that we can determine which hypotheses merit acceptance. In this sense, Inference to the Best Explanation has it that explanation is prior to inference.

Although it gives a natural account of many inferences in both science and ordinary life, the model needs further development. What, for example, do we mean by "best?" It is sometimes taken to mean "likeliest" or "most plausible," but Inference to the Likeliest Explanation would be a disappointingly uninformative model, since the main point of an account of inference is to say what leads one hypothesis to be judged likelier than another, that is, to give the symptoms of likeliness. A more promising approach construes best as "loveliest." In this view, we infer the hypothesis that would, if correct, provide the greatest understanding.

The model should thus be construed as "Inference to the Loveliest Explanation." Its central claim is that loveliness is a guide to likeliness, that the explanation that would, if correct, provide the most understanding, is the explanation that is judged likeliest to be correct. This at least is not a trivial claim, but it faces at least three challenges. The first is to identify the explanatory virtues, the features of explanations that contribute to the degree of understanding they provide. There are a number of plausible candidates for these virtues, including scope, precision, mechanism, unification, and simplicity. Better explanations explain more types of phenomena, explain them with greater precision, provide more information about underlying mechanisms, unify apparently disparate phenomena, or simplify our overall picture of the world. But analyzing these and other explanatory virtues is not easy, and it also leaves the other two challenges. One of these is to show that these aspects of loveliness do indeed match judgments of likeliness, that the loveliest explanations tend also to be those that are judged likeliest to be correct. The remaining challenge is to show that, granting the match between loveliness and judgments of likeliness, the former is in fact our guide to the latter.

In addition to offering a description of our inductive practices, Inference to the Best Explanation has been used to justify them, to show that those hypotheses we judge likely to be correct really are so. For example, it has been argued that we have good reason to believe that our best scientific theories are true, since the truth of those theories is the best explanation of their wide-ranging predictive success. Indeed, it has been claimed that the successes of a theory would be inexplicable unless it were at least approximately true. This argument has considerable plausibility, but it faces serious objections. If scientific theories are themselves accepted on the basis of Inferences to the Best Explanation, then to appeal to an argument of the same form to show that those inferences lead to the truth seems to beg the question. Moreover, it is not clear that the truth of a theory really is the best explanation of its predictive success. For one thing, it seems no better an explanation than would be the truth of any other competing theory that happens to share those particular predictions. For another, to explain why our current theories have so far been successful may not require an appeal to truth if scientists have a policy of weeding out unsuccessful theories.

See also Epistemology; Naturalized Epistemology; Realism.

Bibliography

Ben-Menahem, Yemima. "The Inference to the Best Explanation." Erkenntnis 33 (1990): 319–344.

Day, Timothy, and Harold Kincaid. "Putting Inference to the Best Explanation in Its Place." Synthese 98 (1994): 271–295.

Harman, Gilbert. "The Inference to the Best Explanation." The Philosophical Review 74 (1965): 88–95.

Lipton, Peter. Inference to the Best Explanation, 2nd ed. London: Routledge, 2004.

Okasha, Samir. "Van Fraassen's Critique of Inference to the Best Explanation." Studies in the History and Philosophy of Science 31 (2000): 691–710.

Psillos, Stathis. Scientific Realism: How Science Tracks Truth. London: Routledge, 1999.

Salmon, Wesley. "Explanation and Confirmation: A Bayesian Critique of Inference to the Best Explanation." In Explanation: Theoretical Approaches and Applications, edited by Giora Hon and Sam Rakover. Dordrecht: Kluwer, 2001.

Thagard, Paul. "The Best Explanation: Criteria for Theory Choice." The Journal of Philosophy 75 (1978): 76–92.

Peter Lipton (1996, 2005)