Abstract

The new riddle of induction proposed by Goodman can be seen as related to the old problem of induction as proposed by Hume. It is shown that this new riddle infects not only those theories that endorse induction, but is also common to any theory that takes into consideration the relation of observed instances, be they positive or negative, with the projected hypothesis or hypotheses. ;Various proposed solutions based on the semantic or syntactic differences between "green" and "grue" are presented and shown to have weaknesses. The use of the distinction of lawlike and unlawlike statements is also shown to involve unsurmountable difficulties. The analysis then turn to the attempt to place the paradox within a certain formalized system of inductive logic. It is shown that no matter how detailed the information put into the confirmation functions of the competing hypotheses, both would acquire the same probability. Nor could the subjectivist's notion of exchangeability give better reasons in this case. ;The nature of the paradox becomes clear in the consideration of imaginary cases in which "grue", but not "green", is reasonably believed to be projectible under certain situations. These show the importance of the function of background knoweldge in our practice of projecting hypotheses. A consideration of the concept of randomness is related to this concept of projectibility. ;Goodman's theory of entrenchment is shown to be in the right direction for the solution of this paradox. His employment of the notion of entrenchment is one way to utilize the credibility of our background knowledge. The difficulties of his concepts of entrenchment and projectibility are explored and analyzed; the practical difficulties which would be encountered in applying his concept of entrenchment are also discussed. ;In the final chapter, I propose a distinction with respect of the scope of gruification, which takes place only in the empircal portion of our background knowledge. I call a gruification local if it is limited in its scope, and a gruification global if it is unlimited in its scope. In the case of local gruification, we have at least some empirical items in our background knowledge which will be allowed by both the green hypothesis upholder and his grue opponent to be immune, at least temporarily, from gruification. Consequently we would utilize this portion of empirical knowledge to differentiate the two competing hypotheses, and this would usually bring forth strong preference of the normal hypothesis to its grue competitor. Thus, Goodman's paradox can be said to be solved in the case of local gruification. ;If the process of gruification continues, it would finally land in global gruification in which no piece of empirical knowledge could be used. I also contend that neither would any formal or theoretical measures, such as simplicity, be able to help us out of this impasse. It becomes a kind of paradigm competition. We have to wait for the outcome of further confirmation information to decide the case. Hence, Goodman's paradox is insoluble in its global species