PAUL THAGARD AND RICHARD E. NISBETT VARIABILITY AND CONFIRMATION * (Received in revised form 21 January, 1982) Why is a single instance, in some cases, suffi- cient for a complete induction, whilein others myriads of concurring instances, without a single exception known or presumed, go such a very little way towards establishing an universal proposition? 1 Hume left us with two problems of induction. The strong problem concerns the global justification of induction: how do we know that the future will be like the past? This problem seems to be insoluble, since no a priori justifica- tions are forthcoming, and any a posteriori justification would be circular. We can do much more with the weak problem, which is to find a set of rules and principles which describe and justify, at least locally, our inductive inferential behavior. As Goodman pointed out, we can develop a set of induc- tive principles by paying attention to accepted inductive practice, while at the same time adjusting our inferential practice in the fight of new principles. 2 The procedure of mutual adjustment of principles and practice is more com- plex than Goodman indicated, 3 but Goodman's general point, that the des- cription and justification of inductive principles go hand in hand, provides a basis for pursuing Hume's weak problem. Yet even the weak problem is very difficult. There has been limited progress in specifying the principles that describe and justify inductive practice. We believe that this is partly because many philosophers concerned with induction have assumed that inductive principles would be like deductive ones in being formulatable in terms of the syntactic structure of the premises and conclusions of inductive inferences. Whereas deductive principles can be based solely on file logical form of the relevant sentences, inductive rules must, we shall argue, make essential reference to the content of the premises and conclusion of the inference. The validity of inductive inferences depends in part on the nature of the objects and events about which one is reasoning. Philosophical Studies 42 (1982) 379-394. 0031-8116/82/0423-0379501.60 Copyright 9 1982 by D. Reidel Publishing Co., Dordrecht, Holland, and Boston, U.S.A.