tbray@mprvaxa (11/13/83)
x Laura has just raised one of my favourite philosophical questions with her recent discussion of reality, and must now bear the windy consequences. The question is: on what basis, if any, can we be confident of our knowledge of reality? The relevance of this question to many other questions, including those currently under debate in this group, are obvious. (Even "objectivism" - yecch). There is general agreement that there are two general flavours of reasoning and of knowledge - deductive and inductive. Examples: DEDUCTION: Axiomatic definition of syllogistic "There are infinitely number, primeness, and ==> reasoning ==> many prime numbers" the relation ">" INDUCTION: Putting unprotected flesh inductive "Hot places are painful in a region of high ==> reasoning ==> and damaging" temperature has always caused me pain and damage. The distinguishing characteristic of deductive reasoning is that no NEW knowledge is produced. The infinitude of primes is intrinsic in the specification of the axioms and rules of deduction. The distinguishing characteristic of inductive reasoning is that new "knowledge" is produced, but that there is apparently no purely rational basis for confidence that that knowledge will remain reliable. The scientific method is based on inductive reasoning. Conclusion: those who put forward weighty aguments based, they say, on nothing but "rationality" and "reality" should be aware that there is at least one potential gaping hole in their defenses. If this topic is of interest, I will present some of the famous historical arguments on it. The first point for debate, of course, must be: Is this a real problem or a red herring? Tim Bray ...decvax!microsoft!ubc-vision!mprvaxa!tbray
tbray@mprvaxa (11/21/83)
x <-- netnews insecticide Mr. Fritz proposes the following (Epminides' paradox) as an example of deduction leading to new knowledge: -------- All statements are either true or false ==> "This statement is false" is either true or false. One may reasonably conclude that some statements are not true and not false under the definitions of true and false used. -------- The first assertion is an example of deductive reasoning, the process of "reasonable conclusion" in the second is the worst kind of level- mixing and has nothing to do with deduction. Hofstadter has beaten this issue to death in GEB. Mr. Fritz also wonders in what fashion humans really do reason. This of course is an issue at the very crux of AI, and he can snap up an easy Ph. D. if he can settle it convincingly. My personal conviction is that human reasoning is overwhelmingly inductive and that the human mind is little more than a superb pattern-matching apparatus with some deductive ability on the side. Oops. Shouldnta said that. Better duck. ...decvax!microsoft!ubc-vision!mprvaxa!tbray