Apprentissage en situation aléatoire dans le cas d'un ensemble fini de réponses multiples
P. Crombé · L’Année psychologique · 1982
Summary : Probability learning experiment in a discrete case with multiple responses. When the learning of probabilities was extended to include the continuum case by rneans of a continuum of responses, the results stated prouved to be somewhat different from these obtained in the discrete case. However, it should be noted that the comparisons were based on results obtained in situations with very limited sets of responses and that the number of events was likely to aller the adjustment observed in the binary case. It was therefore necessary to propose data representative of the probability learning situation in the discrete case while not limiting the experiment to the particular case of a set reduce of to 2 or 3 elements, but extending the classical situation to a finite set of multiple responses. The experiment proposed here includes 24 elements arrangea in circle. 20 adults subjects were asked to predict the disappearance of one of the elements. The experiment included 360 trials and was run with a unimodal noncontingent reinforce-ment. The probability distribution chosen here is the triangular one, in order to allow us to check the stability of the results previously obtained in the continuum case. Some tests specifically related to Markov chains were used to analyse the data. The results obtained do not show any definite difference in behavior as a function of the continuum vs discrete character of the respome sets. Key-words : learning, probability learning.