The Method and Induction: William Whewell and Current Philosophy of Science
Dissertation, The Johns Hopkins University (
1996)
Copy
BIBTEX
Abstract
In this dissertation I consider a number of central questions in philosophy of science, including: Is there a method by which theories should be invented? What is the nature and importance of inductive reasoning? When are scientists justified in concluding that their theories are highly probable? I examine the answers given to these questions by the nineteenth-century philosopher, historian of science, and scientist William Whewell as well as those given by twentieth-century philosophers of science such as Karl Popper, N. R. Hanson, Wesley Salmon, Larry Laudan, and others. ;I present a novel interpretation of Whewell's view, demonstrating that he is not, as universally charged, either a hypothetico-deductivist or a retroductivist. Whewell's methodology is an inductive one, requiring a hypothesis to have a certain inferential relation to the data. I describe how it avoids the problem faced by a narrow inductivism of the type proposed by J. S. Mill: namely, the problem of accounting for the inductive discovery of hypotheses involving unobservable properties and "theoretical" entities. I demonstrate that any inductive methodology, including Salmon's view, must be consistent with Whewell's "discoverers' induction" in order to avoid this problem. I also demonstrate, using probability theory, that the non-inductive methodologies most prominent in twentieth-century philosophy of science--Hempel's form of hypothetico-deductivism and Hanson's retroductivism--must be rejected in favor of Whewell's methodology, if we are to show hypotheses to be highly probable by empirical testing. We must also reject Popper's hypothetico-deductivism if, like Whewell, we believe it is possible to determine that a hypothesis is probably true, and not just that it is "corroborated." And I establish, contra Laudan and others, that Whewell's confirmation criteria of prediction, consilience and coherence do yield knowledge that a hypothesis is highly probable, when that hypothesis is known to have the inferential relation to the data required by discoverers' induction. Throughout the dissertation I examine important episodes in the history of science, such as Kepler's discovery of his first law of planetary motion, in order to demonstrate that Whewell's methodology is superior to other views in capturing the practice of science