A NEURAL-NETWORK MODEL OF SENSORIMOTOR TRANSFORMATIONS IN THE LOCAL BENDING REFLEX OF THE MEDICINAL LEECH

In response to a moderate mechanical stimulus, the leech withdraws by forming a U-shaped local bend beneath the stimulus. Inter-neurons in t he local bending reflex receive multiple sensory inputs, and have outp uts to many motor neurons, suggesting that stimulus location on the bo dy is represented in a distributed fashion. However, it is possible th at as yet unidentified interneurons dedicated to the detection of spec ific spatial locations are necessary for the full range of the local b ending motor output. Using the backpropagation algorithm to optimize c onnections in a model of the reflex, we showed that the distributed pr ocessing model of sensorimotor integration was consistent with our kno wledge of the connection strengths between identified neurons in the r eflex. The model further showed that the spatial and temporal constrai nts on the linkage between sensory representations and motor output ca n be satisfied by the appropriate connectivity between layers of inter neurons and motor neurons in the leech, as well as in other systems wi th similar circuitry.