Complexity versus complex systems: a new approach to scientific discovery
Authors: Arecchi F.T.
Autors Affiliation: Istituto Nazionale di Ottica Applicata, Largo E. Fermi 6, 50125 Firenze, Italy;
Department of Physics, University of Firenze, 50125 Firenze, Italy
Abstract: Extraction of quantitative features from observations via suitable measuring devices M means that the words of science are coded as numbers, and the syntaxis is a set of mathematical rules. Once general premises are available all consequences can be worked in a purely deductive way. This characteristic of science displays two orders of drawbacks, namely, undecidability of deductive procedures, and intractability of computer modelings of complex situations. The way out of such a crisis consists in an adaptive strategy, that is, in a frequent readjustment of M suggested by the observed events. As a consequence, M provides different data streams (words) for the same observed events, as it is tuned to different resolutions. The adaptive strategy here introduced should by no means be confused with the adaptivity of a learning machine, which—inputted by a data stream—readjusts itself over a class of theoretical explanations in order to select the optimal one, thus providing knowledge conditional on the assigned input. On the contrary, physics aims at extracting regular patterns out of things, by a trial and error procedure which includes not only modifications of the explanations for fixed data sets, but also exploring different data sets via modified M”s. This M-adjustment is a pre-linguistic endeavour, not expressible by a formal language. Such an essential characteristic of the physical program means that physics can not be performed by a machine.
Volume: 5 (1) Pages from: 21 to: 35
KeyWords: nonlinear dynamics; complexity; epistemology; cognitive sciences;