II. THE PARADOXES OF CONFIRMATION ? A SURVEY

Abstract

rT,HE purpose of this article is to survey and set * in perspective the various solutions which have been proposed for Hempd's paradoxes of confirmation, and, among them, to defend one particular solution. The bibliography is not ex? haustive, but it does include, I would claim, the major contributions to the discussion. The para? doxes were first brought to our notice in [10] and then, more fully, in [n]. The paradoxes are generated by the fact that three highly plausible principles of confirmation prove incompatible. The first principle is known as Nicod's criterion. It states that evidence of the existence of a which is \js confirms all which is iff or a ~ which is r^t/j is irrelevant to, that is neither confirms nor disconfirms, all 's are 0. Put symbolically, ?z & 0?z confirms, a & ^?/??z disconfirms while was not a ifs?could also confirm or disconfirm the hypothesis. Neither Nicod nor Hempel explicitly discusses under what conditions Nicod's criterion is supposed to hold, but, since their discussion does not suggest any limitation to its operation, we shall suppose that it is meant to apply, whatever the background knowledge (unless that evidence entails all and to everything which is or ^ . Different paradoxes are generated by con? sidering different claims of the scientific laws condition. Thus given the equivalence condition and the first claim of the scientific laws condition, whatever confirms all ravens are black also confirms all non-black objects are non-ravens. But this contradicts Nicod's criterion, which claims that (if we use CR' for raven and '?' for black) ifa & Ba confirms the former, but is irrelevant to the latter, while ~Ra & ~Ba confirms the latter but is irrelevant to the former.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 26 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.