Journal of Experimental Psychology. Animal Behavior Processes issue:submitted
Using a computer task measuring online conditioned suppression, we investigated the effect of US-unpredictability on subsequent learning of new CS-US associations. Three between-subjects conditions were included: One predictable (Paired) and two unpredictable conditions (US-only and Unpaired). These conditions were followed by exposure to a partially reinforced CS (50%). For half of the participants this CS was the same as that used previously, for the other half a new CS was used. In addition, a context switch was introduced in half of the participants. The results showed that subsequent learning of new CS-US associations was generally retarded in the unpredictable conditions compared with the predictable condition. The retardation effect was not mediated by context conditioning as expected from an associative learning account. In addition, the retardation effect did not seem to be modulated by cue-specificity. The present data pattern rather suggests that unpredictability has a general debilitating effect on subsequent learning (cognitive account).