|
Evolution of stability of equilibrium
state in function of learning. Viviane Kostrubiec Université de Lille, Ch. de Gaulle, France According to nonlinear dynamics, to say that an equilibrium state is stable means that a response persists even if stochastic fluctuations are encountered : if that state looses its stability, then there is an increase in the probability of response switching (Hock, Schöner and Hochstein, 1996). The current study explores the evolution of stability. The experimental material consists of 19 boards continuously evolving from dark to light. During the learning, subjects learn to associate responses on the right button to the three lightest boards and responses on the left button to the three darkest boards. In the condition "long learning", seven subjects were exposed to a FR3, while in the condition "short learning", seven other subjects were only exposed to a FR1. During the test phase, subjects were exposed twenty times to boards 1 to 19 in that order and twenty times to boards 19 to 1 in that order. For each board, they had to decide whether they press the left button or the right button. Results shows that the subjects switch from one response to the other for the boards in the middle of the dark-ligh continuum. For "long learning", the number of the switchings increases in function of repetition of the task, suggesting that the system has lost its stability. Inverse results were observed for the "short learning" conditions. Keywords: stability, stochastic fluctuation, perceptual categorisation, non linear dynamics, hysteresis |
|
|
|
|
|
|
|
|
|