1. Introduction

The number of neurons in the mature brain is enormous, as is the problem of understanding their functioning and development. To arrive at the very organized and complicated structure of the full-grown brain, mechanisms other than random formation of connections are needed. Many connections seem to be predetermined in the way that they always turn out to connect specific parts of the brain. This predetermination points in the direction of a genetic mechanism controlling the formation of connections. However, in what way this genetic control leads to the formation of specific connections is still a matter of experimental exploration. Changeux and Danchin (1976) pointed out that the information content of the DNA is far too limited to account for all the connections to be specified. They state:

"The DNA(...) may, at most code for a few million proteins. (...), moreover a significant proportion of it (more than 60 %)(a) does not correspond to genuine genes".

The number of synapses in the central nervous system of a high vertebrate is estimated as 1014 (Barnes,1986).  In addition, the relative strengths of these connections (the synaptic efficacies) must be specified. When comparing the number of genes with the number of connections to be made during development, the influence of genetic mechanisms on the fate of an individual connection seems negligible. One could argue that also combinations of genes can be responsible for specific connections. In this way, more connections can be specified with the same limited number of genes. Such a mechanism has already been demonstrated for the genetic control of the immune system. Here, a few hundred genes are shuffled and recombined to make billions of different antibodies (Leder, 1982). Furthermore, evidence is accumulating that 'gene shuffling' mechanisms are operating outside the immune system, while cell recognition during embryonic development has been put forward as a possible candidate. Therefore, the argument that the information content of the DNA is too limited to account for all the connections to be specified is not as self-evident as it seems.

        Whether or not it is possible in principle that every individual connection is genetically specified, such a strong predetermination is very unlikely in view of the adaptive and learning capabilities of the brain. Genetic control probably plays a major role in connecting the nervous system with the periphery (the muscles and sensory organs). The connections in the spinal cord and brainstem regions probably also are predetermined to a great extent, because the functions of these brain parts are not very susceptible to adaptive changes. In contrast, the higher brain regions, in particular the cortical areas, have been demonstrated to undergo plastic changes during life and their development has been shown to be sensitive to environmental conditions (Wiesel, 1982). This neuronal plasticity is not confined to the biochemical level (changes in synaptic efficacies), but also has anatomical consequences. Therefore, a stringent genetic specification of the connections is very unlikely in these regions. Other, less rigid, mechanisms can come into play.

         Development of the brain never really comes to a standstill, but smoothly changes into the phenomenon of 'plasticity', which is strongly coupled to the functioning of the brain. General mechanisms relating development, plasticity and function may therefore deepen our insight into the functioning of the brain. Understanding higher brain functions is one of the major challenges of current research. The function of sleep patterns, the mechanism of memory and learning, the associative capabilities and sensory processing are phenomena which we are only beginning to understand. In spite of tremendous experimental effort to find the cause(s) of epilepsy, there is still no satisfying explanation for the sudden and unpredictable interruptions of normal brain function which characterize this disease.

          A class of non-genetic mechanisms which could play a role in the development, functioning, plasticity and pathology of the brain is comprised by the term 'selforganization'(b). By selforganization is meant a spontaneous (physico-chemical) process by which an organized spatio-temporal structure develops in a system out of an initially non-organized state. The order which is arrived at is intrinsic: it is not forced upon the system by the environment. Moreover, it is stable: the effects of small perturbations (either spontaneous internal fluctuations or external disruptions) eventually die out. The phenomenon of selforganization was first described for physical and chemical systems and a thermodynamic theory explaining the onset of selforganization was developed. At the same time, exciting developments took place in branches of mathematics and physics dealing with nonlinear dynamical systems: 'strange attractors', 'deterministic chaos', 'catastrophes' are a few terms used to describe the bewildering variety of complex behaviours displayed by a majority of these systems.

          In Part 2 some examples of selforganizing physical and chemical systems will be given together with the general mechanism. Part 3 sketches the developments in mathematics relating to selforganization. In Part 4 the use of this term in the context of brain function and development will be investigated and a possible role of selforganization in the brain is discussed.

(a) The amount of non-coding DNA in humans is currently estimated at 98%. How much of this non-coding DNA has a biological function is under intense debate.

(b) The term 'selforganization' has aquired different meanings since 1988, many of them outside the realms of physics, chemistry and mathematics.