El efecto del reforzamiento parcial en la extinción del aprendizaje predictivo: una evaluación del modo de respuesta

Abstract

The partial reinforcement effect (PRE) is observed when a partially reinforced cue shows more extinction resistance than a continuous reinforced cue. In two experiments using a predictive learning task the effect of response mode in PRE was studied. Human participants were trained in a trial-by-trial, phase or global response mode in a predictive task in which they had to predict the probability of an outcome (appearance of a rabbit) following different cues (magic words). During an acquisition phase a cue EC25 was paired with the outcome on 25 % of the trials (partial reinforcement), while cues EC100 and ECcont were paired with the outcome on all the trials (continuous reinforcement) During an extinction phase, cues EC25 and EC100 were extinguished but cue ECcont not. Judgments were required for each cue in the Global group at the end of training, in the Phase group at the end of each phase and in the Trial group, each 3 trials. These results showed PRE for cue EC25 in all groups and can not be explained by associative or probabilistic models.

Keywords

• Partial reinforcement effect
• humans
• predictive learning
• judgment frequency effect
• associative theory