A functional model of some Parkinson's disease symptoms using a Guided Propagation Network

This paper presents a computational model of Parkinson's Disease (PD) symptoms. Based on psychophysiological data, the underlying system (Guided Propagation Network) implements coincidence detection between internal flows and stimuli, and can be dynamically controlled for representing the action of neuromodulators such as dopamine (DA). By modelling the DA deficit involved in PD through a decrease of response thresholds in the production modules of a GPN, four symptoms are observed in experiments carried out on a computer simulation, and then attributed to a lack of synchrony between 'proprioceptive stimuli' and internal flows: reduced intensity, increased rate, saccades and spontaneous repetitions.