A neuronal network model for the detection of single bands and cruciform figures

The model of artificial neuronal network has been proposed, which detects not only single bar orientation but also has selective sensitivity to cross-like figures. Properties of such "detector of cross" have been studied at the model constructed on the basis of convergence of excitatory connections with different weights from neurons with different orientation tuning. It has been shown that the amplification mechanism (reverberator) is necessary for the detector of cross with monomodal orientation tuning.