Zagadnienia metodologiczne w bayesowskiej teorii konfirmacji

Bayesian confirmation theory is conceived here as a model of inductive inference. A methodological facet of Goodman's grue paradox (MWP) constitutes a major difficulty for this theory. This could only be evaded on the grounds of philosophical assumptions such as the existence of laws of nature or natural kinds. These would unwittingly bias the theory towards some scientific hypotheses which would contradict the character of the theory. The analysis of J. Earman's and R. Chuaqui's formal frameworks demonstrates, however, that weaker assumptions, e.g. seemingly anti-realist admittance of the existence of symmetries alone, will not suffice to cope with (MWP). Finally then Bayesian confirmation theory faces a dilemma: either to allow for ad-hocness in narrowing down the class of its models or to defend a version of scientific realism, e.g. structural realism.