Networks modeling the involvement of the frontal lobes in learning and performance of flexible movement sequences

Network architectures for classifying spatiotemporal patterns are developed. These architectures combine the adaptive resonance architecture, which classifies spatial patterns, with the avalanche, which generates sequences. The primary spatiotemporal processing area is identified with all area of the corpus striatum. The prefrontal cortex is identified with higher-order controls of functions of this sequence-classifying area. Some of these controls make possible the formation of complex rules for sequence classification. Others cause competitive biases among sequence representations, favoring longer over shorter sequences to facilitate attention to a motor plan. The network models experimental results showing that frontal damage impairs the performance of flexible movement sequences but not of invariant movement sequences