Scientific Discovery on Positive Data via Belief Revision

Contemporary approaches to inductive logic divide into two schools distinguished by how each perceives the investment of a rational agent in a given hypothesis. For the probabilist school, investments are degrees-of-belief, scattered over all the hypotheses in play. For advocates of Formal Learning Theory, an investment has a more absolute character, and is designated as \acceptance." The di erence between belief and acceptance is thoroughly treated in [Cohen, 1992]. SuÆce it here to say that acceptance can be just as provisional as belief, but is more qualitative. It represents deliberate selection of a particular hypothesis as the one to elaborate and test, and to represent as the most interesting and viable at a particular moment of investigation. Belief is an involuntary feeling of conviction and carries no intellectual commitments. The distinction between belief and acceptance is parallel to the moral distinction between desire and consent. In both cases, the former is often but not always a precursor to the latter, and only the latter involves some kind of personal accountability.