A modified model of the Hebbian synapse and its role in motor learning

Abstract The `Hebbian synapse', an old neuro-psychological concept which describes the process of associative learning at the synaptic level, is being increasingly confirmed by neuro-biological explanations. The purpose of the present paper is to show that sensorimotor learning, which is essentially non-associative, can also be explained by Hebbian learning. This requires a re-definition of the Hebbian synapse ensuring convergence of the synaptic weight. Re-defined Hebbian learning then appears as a linear adaptive filter known in the field of adaptive signal processing, which in turn is equivalent to the delta rule used to train artificial neural networks. For motor learning, the modified Hebbian synapse must be embedded into a special learning algorithm called `auto-imitation'. This is a non goal-oriented inductive learning algorithm, enabling a controller to adopt a general rule from being shown only a few examples of that rule. When applied to motor learning, the neural controller can acquire the capability to online invert the behavior of the plant to be controlled. The complete learning process then can be described as a relaxation process between the controller and the controlled system, which is governed by long-term potentiation (LTP), the neuro-physiological process underlying Hebbian synaptic shaping. PsycINFO classification: 2330