Non-linear dynamics and Operant discrimination.
Yannick Miossec & Jean-Claude Darcheville
Université de Lille, Ch. de Gaulle, France

Tuller et al (1995) showed that speech categorization deals with perceptual non linear dynamics, which is characterized among other, by multistability, hysteresis, enhanced contrast. In this view, perceptual forms are the attractors of acoustic variations in time. Dynamical approach is interesting because it can help us to understand the great variability observed in speech perception, in particular the great sensibility to context of phonemes perception. However, this approach does not say anything about the emergence of the attractors, in particular how acoustical events acquired their discriminative value as phonemes in young children. Speech categorization can be considered as an operant process. Our hypothesis is that the discriminated operant is the attractor, and that dynamics patterns cannot be revealed if the discrimination isn't established. Artificial stimuli, analogous to speech but not perceived as such, were used. Subjects were learned to discriminate two different sounds and then were tested on generalization and dynamical procedure. Results revealed that dynamical effects observed by Kelso (1995) cannot be clearly established if stimulus have not acquired their discriminative value. In such a case, attractors are not stable enough to capture the percept. These results showed that a dynamic approach associated with a selectionnist one is able to explain speech perception as a context dependant behavior and as an adaptative time dependant behavior.

Keywords:


 Back to program

 Retour au programme

 Back to contributors

 Retour aux contributeurs

 Back to summary

 Retour au sommaire