Genetic Algorithms in Scientific Discovery: A New Epistemology?
Ioan Lucian Muntean · PhilSci-Archive (University of Pittsburgh) · 2014
Philosophers argue that scientific discovery is far from being a rule-following procedure with a general logic: More likely it incorporates creativity and autonomy of the scientist, and probably luck.Others think that discovery can be automatized by some computational process.Based on a concrete example of Schmidt and Lipson Schmidt and Lipson (2009), I argue that the bottom-up discovery is computable and that both aspects of creativity and autonomy can be incorporated.The bio-inspired evolutionary computation (genetic algorithms) are the most promising tool in this respect.The paper tackles the epistemology of applying a evolutionary computational and genetic algorithms, to the process of discovering laws of nature, invariants or symmetries from collections of data.Here i focus on more general aspects of the epistemology of evolutionary computation when applied to knowledge discovery.These two topics: computational techniques applied in science and scientific discovery taken separately are both controversial enough to raise suspicions in philosophy of science.The majority of philosophers of science would look with a jaundiced eye to both and ask whether there is anything new to say about discovery and computers in science.This paper is a first stab to the philosophical richness of computational techniques applied to the context of discovery.I discuss the prospect of using this type of computation to discover laws of nature, invariants or symmetries and appraise their role in future scientific discoveries.