APPLICATION OF COVARIATION MODEL TO THE ANALYSIS OF PERFORMANCES OF HUMAN-SUBJECTS IN CONDITIONS OF TRAINING AND TRANSFER IN A CONDITIONAL DISCRIMINATION TASK

In a previous study, according to the requeriments of a model of covar iation, children and college students showed performance highly effici ent in the learning of tasks of conditional discrimination. These resu lts were limited since not collected data on transfer. The purpose of the present study was to explore the necessary conditions to maintain the efficiency of performance in training during different tests of tr ansfer. For this, college students received one of the two training se quence in conditional discrimination of first order, using geometric f igures of colors like stimuli presented in a computer. Each sequence c onsisted of two conditions of training (similarity/oddity or, oddity/s imilarity) and three tests of transfer. Subjects in the conditions of training were exposed to precise instruction's task and continuous fee dback on their performance. In the tests of transfer, feedback was rem oved and the instructions were vague. Each session of training and tra nsfer consisted of 36 trials of typical arrangements of matching-to-sa mple (one stimulus as sample and three as comparison). To advance of a training to another, all of the subjects had to fulfill a criterion b f performance of two consecutive Sessions with 34 correct responses or complete four training sessions as maximum, The results successfully replicated the effective performances during training. However, subjec ts did not transfer their previous level of performance in the tests. These results are interpreted in terms of the covariation model propos ed and suggest that the properties of the used stimuli, as well as the vague instructions could be relevant variables for the transfer of re lational learning.