Cybernetic Analysis of Natural Intelligence

Abstract

Over thirty years ago, we conjectured that the bilateral brain capable of introreflection is a multi-level disystem (a sufficient condition) for generating natural intelligence and consciousness. Cascading introreflections can generate a hierarchy of semantic systems wherein each level contains emergent structures (e.g. new cognitive categories and new higher level semantic systems) capable of controlling lower level systems. During the thirty-third annual meeting of this Society we presented experimental evidence that such cascading introreflections may be generated by neural network metabolism.A cybernetic approach to human development in which heredity and environment are joined by a self-determined (teleogenic) component provides insight into the so-called “hard problems” associated with studying natural intelligence and consciousness. Standard local analysis discloses that a general continuous control system with m inputs, n outputs, and p state variables is (1) controllable iff the matrix governing the relation between the inputs and the canonical state variables has no rows which are zero and (2) completely observable if the matrix governing the relation between the canonical variables and the outputs has no columns which are zero. These two theorems may be applied to disystems with cascading introreflections to (1) identify state variables that are self-determined (a necessary condition for teleogenesis) or (2) become unobservable to an external observer. From the point of view of the external observer, systems that become unobservable may experience a period similar to the Verschrankung coined by Erwin Schrodinger in the context of quantum entanglement but measured in much longer “intelligence bits” (“ibits”). Unlike many current conjectures about consciousness, this hierarchy is neither reductionist nor mystical.