Predicting Novel Facts

Abstract

Roughly, the idea is that other things equal, the verification of a prediction supports a theory more than the explanation of something already known, or something the theory was designed to account for. For example, Leibniz ([1969], p. i88) asserted in a letter to Conring dated 19 March I678: 'Those hypotheses deserve the highest praise (next to truth), however, by whose aid predictions can be made, even about phenomena or observations which have not been tested before. . . .' Likewise, Duhem ([1962], p. 28) wrote in 1905 that 'the highest test, therefore, of our holding a classification [theory] as a natural [reality-reflecting] one is to ask it to indicate in advance things which the future alone will reveal. And when the experiment is made and confirms the predictions obtained from our theory, we feel strengthened in our conviction ...'. Closer to our own time Popper ([I9651, pp. 241-2) required for one theory to supersede another that the new theory 'must lead to the prediction of phenomena which have not thus far been observed' and that some of these predictions should be confirmed. A contemporary school which places great emphasis on the prediction of 'novel' facts consists of the followers of Imre Lakatos. According to his 'methodology of scientific research programmes', the history of science should be discussed, not in terms of individual theories, but in terms of 'research programmes', each of which is a sequence of theories possessing a common 'hard core' of fundamental assumptions and a 'positive heuristic' guiding the construction of variant theories. Lakatos ([1970], pp. i 18, 132-4) held that one research programme will supersede another if the new programme is, and the old one is not, 'theoretically progressive' (each new theory in the series exceeds its predecessor in content and also predicts some 'novel' fact not predicted by its predecessor) and 'empirically progressive' (such predictions are from time to time confirmed). It is implicit in this claim that the only observational phenomena which have any bearing on the assessment of a research programme are those which are 'novel'. Accordingly, Lakatos ([19701], pp. 123, 137) claimed that a programme's success or failure in accounting for previously known but unexplained facts has little or no bearing in its assessment: 'the only relevant evidence is the evidence anticipated by a theory'; 'the positive heuristic forges ahead with almost