Inference to the Best Explanation

"Peter Lipton (1954–2007) was an exceptionally talented philosopher of science. His book Inference to the Best Explanation is a model of lucidity, rigorous argumentation, and philosophical depth. It came out in 1991 and had a second edition in 2004, with the second having substantially new material. Inference to the Best Explanation (IBE) is a pervasive mode of inference (or reasoning) in science. The key idea is that ‘best explanation’ is a guide to truth. It is related to what the American Pragmatist philosopher Charles Sanders Peirce called “abduction”. This is a reasoning process which proceeds as follows: “The surprising fact C is observed. But if A were true, C would be a matter of course. Hence, there is reason to suspect that A is true”. IBE is taken by scientific realists to be the way in which scientists form beliefs and accept hypotheses about unobservable entities. For instance, the best explanation of the macroscopic behaviour of gases is that they are composed of atoms. In fact, the most basic argument for realism itself is an inference to the best explanation: that scientific theories are (approximately) true is the best explanation of their predictive successes. But IBE has been notoriously hard to formalise and to defend (or justify). Many philosophers ask: what does explanation have to do with truth? “The most basic argument for realism itself is an inference to the best explanation: that scientific theories are (approximately) true is the best explanation of their predictive successes” Lipton attempted to answer this question by distinguishing between loveliness and likeliness. Loveliness is a function of the explanatory qualities of a hypothesis; that is, how simple, comprehensive, unified and natural it is. Likeliness has to do with how likely a hypothesis is. Hence, Lipton unravels two facets of IBE: inference to the Loveliest Explanation and Inference to the Likeliest Explanation , where the loveliest explanation is one which would, if true, be the most explanatory or provide the most understanding. In effect, Lipton’s strategy has been to impose two types of filter on the choice of hypotheses. One selects a relatively small number of potential explanations of the evidence as plausible, while the other selects the best among them as the actual explanation. Both filters operate on the basis of explanatory considerations. That is, both filters should act as explanatory-quality tests. Then, he argued that the loveliness of an explanation is a symptom of its likeliness. Hence, explanations that are lovely will also be likely. But what guides the inference is the loveliness (explanatory power) of an explanation. He was aware, though, that this was not the end of the story. What he called the ‘problem of matching’—the extent of the match between loveliness and likeliness—is still with us. In the second edition of the book, Lipton made an extra effort to reconcile IBE with Bayesianism—that is, the view that scientific inference is probabilistic and modelled by a famous theorem in the theory of probability, known as Bayes’s theorem. This has proven to be a very fruitful area of research that flourished after Peter’s untimely death."