4. Selforganization in the Brain
4.1 Introduction
Can the brain be considered as a dynamic system?
The requirements mentioned in part 2 are met: the brain does consist of a
large number of more or less similar units, the neurons, interacting with
each other. Neurons interact by 'exciting' and 'inhibiting' one another,
thereby changing each other’s functional state, as reflected for instance by
the membrane potential or the mean firing rate. Neurons therefore exchange
'levels of excitation'. Unlike energy and momentum, this property is not
strictly conserved. But the
maintenance of the excited states and changes therein caused by synaptic
interaction lean heavily on
metabolic energy. This strong coupling of function and energy
consumption
is clearly demonstrated by the experiments with
14C-deoxyglucose uptake (Sokoloff, 1977). On a larger timescale,
neurons are capable of changing each other’s biochemistry by the process of
'long-term potentiation' (Bliss and Lomo, 1973). This interaction between
units gives rise to the emergence of
macroscopic properties which do not depend on the state of individual
neurons, but only on the average state of a large set of neurons. One such a
macroscopic parameter could be the 'neuronal activity' defined as:
the number of neurons
firing per unit time in a small volume. The rates of change in these
macroscopic parameters are equivalent to the 'flows' mentioned in
Part 2. A constant supply of energy
(glucose) keeps the system in a dynamic state, sufficiently far from
thermodynamic equilibrium.
Are processes of selforganization to be expected in the brain?
Positive answers to this question were given by Katchalski
et al. (1974) and Freeman (1975).
According to Katchalski, dynamic patterns can be expected in free energy
dissipating systems containing a large number of nonlinear interacting
elements. The dissipation (=
conversion to heat) of free energy in the brain is
high and special mechanisms exist to guarantee a sufficient supply of oxygen
and glucose, in order to maintain the high level of dissipation. The
interaction between the neurons is strongly nonlinear, due to threshold and
saturation effects. Furthermore, the different inputs to a neuron are not
simply summated or
integrated: coincident inputs are much more effective than asynchronous
inputs (Abeles, 1982), an effect which adds to the non-linearity. Moreover,
the neurons are diffusely coupled, as a consequence of the large degree of
divergence and convergence, and therefore interactive in a continuum.
According to Freeman (1975), macroscopic entities can be expected when a
large number of highly nonlinear elements such as neurons mutually influence
each other diffusely over an extended range in time and space.
The macroscopic form of neuronal activity is
reflected by the electroencephalogram (EEG) and Freeman stresses the
importance of EEG measurements for the understanding of higher brain
functions. Experimental evidence for the existence of macroscopic forms of
neural activity is presented (Freeman and Skarda, 1985; Freeman, 1986).