The ART1_Stable update function updates the neurons activation
and output values until a stable state is reached. In one propagation
step the activation of all non-input units is calculated and then the
calculation of the output of all neuron follows. The state is
considered stable if the 'classifiable' or the 'not classifiable'
neuron is selected. 'Classifiable' means that the input vector
(pattern) is recognized by the net. 'Not classifiable' means that that
there is no neuron in the recognition layer which would fit the input
pattern. The required parameter is in field1.