Right but Still Lousy: Correct Responses With an Unfavorable Outcome Elicit an Error Positivity

The error positivity (Pe) is a neural correlate of performance monitoring that is observed after errors in choice tasks. The results of previous studies suggest that the Pe reflects a monitoring process that goes beyond the mere distinction between correct and incorrect responses. Here, we investigated the idea that the Pe represents a higher-order error signal reflecting an inference-based outcome evaluation. To this end, we created a multistage task whose overall outcome depended on the correctness of each individual stage and was revealed not until the last stage. This implied that the final response could lead to an unfavorable outcome even if it was objectively correct. Our results replicated the general finding that a Pe occurs immediately after errors within each stage. Crucially, we also obtained a Pe after correct responses associated with an unfavorable outcome at the final stage. Moreover, a pattern classifier trained to decode this higher-order Pe successfully decoded the Pe for incorrect responses. These results suggest that the Pe represents an evaluative process that infers the outcome by integrating multiple error signals and taking context into account.