Elkan’s ‘Paradox’ and the Correctness of Fuzzy Logic
Phil Serchuk
In [19] Charles Elkan claimed that despite its popularity, fuzzy logic could only be successfully used in controllers; expert systems, he claimed, required the kind of reasoning that fuzzy logic could not provide. I reject this claim, along with Elkan’s ‘paradox’, on the grounds that it is rooted in an unfounded distinction between controllers and expert systems. But a rejection of Elkan’s ‘paradox’ does not automatically license the use of fuzzy logic, and the burden is on the fuzzy community to give a positive account of why fuzzy logic should be used to model vagueness and uncertainty. I begin this task in the second half of the paper, where I evaluate the correctness of using fuzzy logic to model vagueness and two kinds of uncertainty.