Inference to the Best Explanation

Citation: Lipton, P. (2000) Inference to the Best Explanation. W.H. Newton-Smith (ed) A Companion to the Philosophy of Science (Blackwell) 184-193. (RSS)
Internet Archive Scholar (search for fulltext): Inference to the Best Explanation
Tagged:

Summary

In this article, Lipton introduces his theory of explanation, which he calls Inference to the Best Explanation (IBE). Related to the HD model, in which hypothesis are inductively drawn from existing phenomena and then tested to deduce explanations, the IBE model allows scientists to account for inductively derived inferences. Unlike previous models, such as the DN model, unificationism, causality and pragmatism, the IBE model resolves some of the problems inherent in those theories of explanation, and is aligned to the reality of scientific data - in many cases, the connections between data, hypothesis and explanation are non-demonstrative, and essentially inductive. If that is the very nature of scientific explanation, it is logical to utilize an inductive model.

The utility of this model is clearly articulated by a few of the examples that Lipton provides and other proponents of this IBE model have provided. Sherlock, well-known for his "pure deduction" method of solving mysteries, is actually the classic case of induction. Sherlock infers that Moriarty committed the crime based on the available evidence, however, it is always possible that someone else committed the crime, which cannot be determined by deduction alone. Darwin's theory of evolution is another great example. Darwin's theory was inferred, not deduced from the available evidence of evolution throughout the world.

The basic premises of IBE are that IBE provides a bank of possible explanations, from which we can induce the likeliest explanations. We can only infer the loveliest explanations from this bank. Lovely explanations are those that provide comprehension and a greater depth of understanding, while likely explanations are those that are statistically probable. Lovely explanations, by default, are usually the most likely of explanations as a matter of course. An example of a likely explanation is that "all good cooks produce good food", while a lovely explanation would indicate why it is that some cooks are able to produce even better food, and would provide you with more understanding about the nature of that explanation.