A THEORY OF THEORIES
Abstract
The aim of the document is to set out a bookkeeping procedure by which a scientist (using the term flexibly) may compare the conclusions of a theory with facts obtained by reduction of observational data with the aim of assessing the hypothesis on which the theory is based. It is argued that the appropriate formalism is probability theory, and that the key process is the inductive process as represented by Bayes' theorem which indicates how the degree of belief in a hypothesis should be adjusted in response to new information. The following model of the inductive process in science is adopted. Between observation and theory there is an 'interface' which comprises a set of independent items: each item comprises a complete set of mutually exclusive statements. It must be possible to assign two probabilities to each of these statements, one by 'reduction' of observational data, and the other by theoretical analysis of the considered hypothesis. Formulas are derived which show (a) how the assumed probability of the hypothesis should be adjusted in response to information concerning one item, and (b) how such estimates concerning more than one item may be combined. The method is illustrated by a 'worksheet' indicating the way in which a few observational facts about pulsars may be used to appraise both the neutron-star hypothesis and the white-dwarf hypothesis.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 01, 1969
- Accession Number
- AD0697780
Entities
People
- Peter A. Sturrock
Organizations
- Stanford University