Abductive Reasoning

Abductive reasoning (also inference to the best explanation) is the mode of inference that selects, from among all hypotheses compatible with the evidence, the one that would best explain it. First systematized by C.S. Peirce as the third of his three modes of inference (alongside deduction and induction), abduction is the characteristic method of science, medicine, and everyday diagnosis — the detective's inference from clues to suspect, the physician's inference from symptoms to disease.

What abduction cannot tell you is whether the 'best' explanation is true. It tells you what to investigate next. The inference is licensed by Bayesian reasoning only when 'best' is cashed out as 'highest prior probability times likelihood given the evidence' — but in practice, scientists use informal criteria: simplicity, scope, coherence, novel predictive success. The uncomfortable truth is that no consensus exists on what makes an explanation 'best', and consequently no consensus exists on when abduction is rationally licensed.

The underdetermination problem shows that abduction is systematically under-constrained: for any body of evidence, multiple hypotheses explain it equally well. The choice among them is not a logical matter but a pragmatic and aesthetic one — which should unsettle anyone who believes abduction is the foundation of scientific objectivity.