Computer simulation of a neurological model of learning
A number of problems in psychology and neurology are discussed to orient the reader to a theory of neural integration. The importance is stressed of the comprehensive temporal and spatial integration of sensory, motor and motivational aspects of brain function. It is argued that an extended neural template theory could provide such an integration. Contemporary solutions to the problem of neural integration are discussed. The available knowledge concerning the structure of neural tissue leads to the description of a theory of neural integration which might provide such neural templates. Integrating Neurons are suggested to be organised in columns or pools. Sub-sets of Neurons are formed as a result of excitation and can preferentially exchange excitation. These sub-sets or Linked Constellations would act as spatial templates to be matched with subsequent states of excitation. Inhibition acts to restrict spike emission to the most highly activated sub-sets. An initial computer simulation represented a simple learning or classical conditioning situation. In a variety of test computer runs the performance confirmed the main predictions of the theoretical model. The model was then extended to include representation of instrumental, consummatory, motivational and other aspects of behaviour. The intention of these further simulations was not to demonstrate the predictions of prior formulations but rather to use the computer to develop simulations progressively able to represent behaviour. Difficulties were encountered which were remedied by incorporating rhythmic mechanisms. A number of different versions of the model were explored. It was shown that the models could be trained to produced a different response to discriminative cues, when those cues had previously signalled different contingencies of obtaining the opportunity to perform consummatory behaviour. A published experiment on the Spiral Illusion is reported, which confirmed predictions suggested by the model.