Much of the research in human contingency learning has tried to dissociate associative and inferential processes. One feature that has been regarded as relevant for these dissociations is the
time available to process the cue. A brief presentation of the cue may facilitate the activation of associative processes whereas a long one may favor the activation of inferential processes. The two experiments reported here used a two-phase task. In Phase 1, four different cue-outcome relationships were programmed. In Phase 2, two of these relationships were changed. Participants knew about some of these changes through verbal instructions. The effect of these instructions was measured during Phase 2 in two groups that differed in the time available to process the cue, either 250 or 1500 ms. The results showed that the control of performance produced by verbal instructions differed depending on the time available to process the cue. Only in the 1500 ms group, the verbal instructions were able to affect what it had been learnt during Phase 1. Thus, the results are consistent with the hypothesis that a brief presentation of the cue during Phase 2 facilitates the activation of associative processes