9 Emotional tint of
memory
The variation of links inside and between the ensembles
plays an important part at the selection of the wish in the paragraphs.
Conductivities of excitatory links vary due to their dynamic
components, which increase in the active state and gradually decrease
in the rest state. It was mentioned above, that conductivity variation
algorithm submits to the problem of spike frequency optimization,
however the rate of conductivity variation depends also on the
emotional state: the better is emotional state, the greater is rate. As
it was shown, the emotional state is defined by the age distribution of
EC neurons and depends on the qc movements and on the accords. The automaton has one
more possibility to influence strongly the emotional state: the direct
influence on the quantity of external EC feed. This possibility is
realized by means of an EC feed valve, which can regulate the feed
throughput depending on the emotional tint of working ensembles in
memory. The feed valve is controlled also by two channels of external
education: the channel of penalties, closing the valve, and the channel
of encouragement, opening the valve. Emotional tints of memory
ensembles are formed in the process of accumulation of experience, when
the automaton interacts with environment and teachers.
If, for example, while learning chess, the loss of a
figure is accompanied by a penalty, and this penalty is variable for
the various figures, then ensembles, corresponding to the notions of
"lose my knight", "lose my queen" and so on, acquire a negative tint.
This means that they get the inhibitory links to the feed valve, and
these links have the conductivity values, proportionate to the values
of penalties. Thus, at the choosing of the best behavior variant by
dQ(t)/dt value, the selection toward variants with positively tinted
EW's will take place.