Conclusions
So, the main principles, which create premises for the
consecutive self-organization of the neuronal automaton, are as
follows:
1. The availability of object function (energy
consumption), which is evaluated subjectively through the self-feeling,
and basic opportunity of minimization of this function (optimization of
self-feeling). The following points meet this principle in the concept:
|
ability of neurons to
renew functionally; |
|
existence of
energetically optimal frequency of neuronal spikes. |
2. Constructive diversity at the uniformity of elements,
which permits in either case to bypass restrictions and to reduce
energy consumption.
In the concept described all elements are uniform – only
neurons and plastic interneuron links. The constructive diversity is
provided due to the following:
|
the availability of
interneuron links, which enable neurons to organize mutual aid; |
|
the plasticity of
interneuron links, which is controlled by neurons themselves with the
purpose of achievement of optimal spike frequency; |
|
the availability of
inhibitory links, which creates the effect of ensembles' alternativity; |
|
the existence of
activity modulator – the emotional center. |
3. The impossibility to achieve the minimum of object
function by simple means because of numerous restrictions.
The choice of restrictions can be arbitrary enough, as
far as it is determined by desired direction of self-organization and
by desired secondary effects of self-organization, which we plan to
receive. In given concept, aimed on the growing of intelligence, these
restrictions are:
|
insufficiency of
external feed; |
|
unattainability of
energetically optimal frequency of generation for isolated neuron; |
|
impossibility to fix
the beneficial operating modes (that is instability of decisions
because of growing old effects). |
The neuroenergetic concept of intelligence may be of
special interest in two senses: applied, as one of directions in AI
problem, and conceptual, i.e., in the sense of theoretical biology.
In AI the basic direction of investigations is the
info-logical one, the main lack of which is the absence of active
memory and inductive procedures. The above-considered direction of AI
investigations one may define as biological one (or, more exactly,
bio-logical or "the biologic"), and its main feature is the internal
activity of the system and the presence of induction. In the outlook
both directions may complete each other.
The intelligent automaton in the conceptual sense supposes
the relation to itself as to the model of biological system. This
means, that one should evaluate the potential application and
sufficiency of model hypotheses by comparison of complete properties of
biological system (in particular, of man) and of the automaton.
Some hypotheses in the concept may seem controversial
enough. Nevertheless, such a union of hypotheses provides
self-organization and interesting integral properties of the automaton,
and should be evaluated from these positions. Also one should take into
account, that the self-organization has two components – relatively
fast ontogenetic, and slow philogenetic. The latter can consist either
in natural selection of the organisms, most adapted to the environment,
or in the external teleological interference of experimenter. Some
aspects of the automaton's functioning were simulated on the computer,
and received results confirm the productivity of proposed ideas.