Salience as a fundamental cue to musical pattern recognition: A Parallel Distributed Processing approach.

The aim of the present study is to test if a self-organized Neural network (Piat, 2000) is able to classify four classes of stimuli, previously widely tested by our team, and to replicate human results. The variables considered here are salience and tonality. We consider the first as a fundamental dimension to make the recognition of musical fragments possible, on the other hand we understand, both from our simulation and the results on human subjects, that tonal cues do not have such importance. The theorical frame of this work is Tulving’s model of memory, by the assumption that in subjects the salience carries out Remember answer and tonality Know answer respectively as an expression of the episodic and semantic store.