Perception de la difficulté, entropie et performance

Abstract

De nombreux modèles psychologiques du comportement utilisent le concept de difficulté perçue, sans pour autant en étudier la nature, ni les modalités d'élaboration. L'expérience présentée vise à éclairer, dans une perspective psychophysique, les processus à l'œuvre dans la perception de la difficulté. Certains travaux réalisés dans les années 70 sur ce sujet n'ont pu donner de résultats consistants (Borg et al, 1971). Cela peut être expliqué par un manque de rigueur dans l'utilisation des méthodes psychophysiques et de leurs propriétés mathématiques. L'expérience présentée utilise les tâches de pointage de Fitts (1954), avec cinq niveaux de difficulté. Les sujets (n = 27), après réalisation de chaque tâche, en évaluent la difficulté selon la méthode d'estimation des grandeurs. Une relation exponentielle est mise en évidence entre difficulté objective et difficulté perçue. La relation difficulté objective-performance est également de type exponentiel. En revanche, la relation performance-difficulté perçue est de type puissance, mais ce type de fonction n'est mis en évidence qu'en ce qui concerne les données moyennées. Ces résultats suggèrent que la perception de la difficulté ne se base ni sur les caractéristiques objectives de la tâche, ni sur une évaluation de la performance, mais sur une variable intermédiaire dont la nature demeure à déterminer. Many psychological theories use the concept of perceived difficulty as a mediator of emotion, motivation and behavior. Nevertheless, these models do not give any explanation about its nature and its elaboration. The aim of this experiment is to investigate, according to a psychophysical point of view, how difficulty is perceived in perceptivo-motor tasks. Earlier research, conducted in the seventies by Borg et al on the problems of perceived difficulty, did not give any consistent results 1971. This could be explained in part by a lack of rigour in the use of the psychophysical methods and of their mathematical properties. The present experiment uses the Fitts (1954) tapping tasks with five levels of difficulty. The entropy of each task is calculated according to the formula proposed by Fitts. Subjects (n = 27) perform each task and rate its difficulty according to the magnitude estimation method. The easiest tasks serve as a modulus. Performance is recorded as movement time. An exponential function is evidenced between objective and perceived difficulty, for averaged data as well as for individual data. It is concluded that objective difficulty cannot represent the stimulus taken in to account, as the magnitude estimation method generally yields to power functions between stimuli and responses. However, it could be predicted that the true stimulus is an exponential transformation of objective difficulty. The relationship between objective difficulty and performance appears to be exponential, for averaged data as well as for individual data. Many authors have previously emphasized the linearity of the relation entropy-performance (Hick, 1952; Fitts (1954)). However, a re-examination of the original data of Fitts (1954) confirms the exponentiality of the relation. This issue is briefly discussed. On the other hand, the relationship performance-perceived difficulty can be fitted by a power function. Nevertheless, the last result is obtained for averaged data only. The relationship between performance and perceived difficulty, for individuals, is largely inconsistent. This indicates that performance equally cannot represente the stimulus taken into account. It is concluded that this stimulus should be an intermediate variable, of which performance could represent a probabilistic indicator.