SHOULD A RELIABLE INFORMATION PROCESSOR BE CHAOTIC?

Brain‐like structures have evolved by performing signal processing initially by minimizing “tracking errors” on a competitive basis. Such systems are highly complex and at the same time notoriously “disordered”. The functional trace of the cerebral cortex of the (human) brain is a good example. The Electroencephalogram (E.E.G) appears particularly fragmented during the execution of mental tasks, as well as during the recurrent episodes of R.E.M. sleep. A stochastically regular or a highly synchronized E.E.G on the other hand, characterises a drowsy (relaxing) or epileptic subject respectively and indicates—in both cases—a very incompetent information processor. We suggest that such behavioral changeovers are produced via bifurcations which trigger the thalamocortical non‐linear pacemaking oscillator to switch from an unstable limit cycle to a strange attractor regime (i.e. to chaos), or vice versa. Our analysis aims to show that the E.E.G's characteristics are not accidental but inevitable and even necessary and, therefore, functionally significant.