Resolution of outcome-induced response conflict by humans after extended training

Studies of incongruent discrimination learning, where the outcome event of one response acts as the discriminative stimulus for the opposite response, suggest that humans rely on habitual stimulus–response (S–R) associations when outcome–response (O–R) associations would cause response conflict. Here, two experiments were conducted to investigate the robustness of this habitual strategy. In Experiment 1, we found that extensive instrumental discrimination training supported learning about the incongruent R → O contingencies, as assessed by an outcome devaluation test. Differential representations of the stimulus and the (associatively retrieved) outcome may have allowed for goal-directed incongruent performance. Experiment 2 failed to provide evidence for this possibility; direct presentation as well as associative retrieval of the incongruent events (by Pavlovian stimuli) activated the response that was associated with each event in its role of stimulus as opposed to outcome. We did find that participants successfully acquired explicit knowledge of the incongruent contingencies, which raises the possibility that propositional encoding allowed them to overcome the response conflict caused by O–R associations. Alternative associative and propositional accounts of successful goal-directed incongruent performance with extensive training will be discussed.