Whewell's Theory of Hypothesis Testing and a Relational View of Evidence *

摘要: The debate between William Whewell and John Stuart Mill is not only hard in the sense that both sides are difficult to understand, but issue itself unresolved. Whewell's idea of predictive tests similar method cross validation statistics machine learning, except applies it a hierarchical way at multiple levels. Or least, how argues hypothesis testing works science. In contrast, received view theory confirmation rival hypotheses measured by their degree fit with total evidence, provided equally simple. However, there growing realization stronger many ways. What this suggests history science could be used as source examples against which theories learning may tested. purpose paper explain highlight some features relevant continuing controversies about confirmation. 1. Central Issues After read Philosophy Inductive Sciences, he decided expand his own treatise on deductive logic include book inductive logic. extracted everything made him Whewell, who was most famously historian time. end product, published Book III Mill's System Logic, logical wheat philosophy winnowed away from non-logical chaff. Or, perception those have failed understand parts left out. Eventually responded writing ('Mr. Logic', 1849, reprinted Butts, ed., 1989), documented disagreement long footnotes added subsequent editions Logic. transmutation more influential term, closer contemporary essay make ideas makes them debates scientific theories.