Abstract
Scientific
investigation displays close similarities with a perceptual task. In both
cases, categories already stored in a semantic memory cannot be assumed as
working hypotheses for an interpretation, but they must be matched with
novel information, not deducible from the starting data set. The irruption
of non trivial novelty is the
mark of complexity as opposed to the simplicity of what can be deduced by
an algorithmic procedure.
The adaptive
strategies, whereby we cope with complexity both in scientific modeling
and in perception, represent a clear demonstration of the objective
character of reality, which is not a mental construction.
1. Introduction
A widespread attitude within the cognitive science community consists in reducing the mind-body problem to the dichotomy between software and hardware in a computer, that is, between the programs and the physical machinery which carries them on. Hence, the fact that a program can be executed by different physical machineries gives rise to the misconception of a disembodied thought, as Descartes’ “res cogitans”.
Regarding the res cogitans as an agent who categorizes the sense impressions and organizes them as thoughts, we realize that the knowledge problem reduces to the correspondence between two Cantor sets, that of the worldly events (“res extensa”) and that of our mental schemes. As a result of this attitude three consequences emerge, namely,
i) knowledge is not the creation of novelty, but just the correspondence between two sets established “ab aeterno”; God reduces to an inspector granting the good ordering between the two sets, such is indeed the “Deus sive Natura” of Spinoza;
ii) the classical truth criterion as “adaequatio intellectus et rei” is replaced by the Cartesian criterion of “clear and distinct ideas”, that is, by a requirement of self consistency of our mental procedures;
iii) the AI (Artificial Intelligence) proposal of a machine equipped with a sufficient number of “subroutines” in its archives in order to cope with whatever problem, is a presumption of solipsism; in Hofstadter’s, “Gödel, Escher and Bach: an eternal golden braid”, (Hofstadter) the golden braid is the entangled set of routines which refer to each other in a self consistent way.
In a similar way, a scientific program is currently seen as a two phase task. In the first phase, one gathers relevant features via measuring apparatuses; the features are then organized into an axiom set. The second phase consists in deducing all consequences. The deduction parallels the deterministic evolution of the world. If the first phase has captured the relevant features, then the second phase yields prediction of all future events.
Both approaches, that is, the cognitive program and the scientific modeling, fail whenever complexity occurs. We loosely denote by complexity the emergence of events not deducible from our initial data set, be it our personal set of sensations or the body of measurements on which science is built.
The aim of this paper is to show that adaptation is crucial to provide novelty in knowledge. Adaptation means considering new features whenever necessary, not just selecting a different sub-routine which is already part of the “golden braid”.
