Error significance but not error expectancy predicts error-related negativities for different error types.

Violations of outcome expectancies have been proposed to account for error-related brain activity in the medial prefrontal cortex. The present study investigated whether early error monitoring processes are sensitive only to the expectancy of errors, or whether these processes also evaluate the significance of errors. To this end, we considered the error-related negativity (Ne/ERN), an electrophysiological marker of early error monitoring, in a modified flanker task in which errors could occur because participants responded to the flankers instead of the target (flanker error) or because a response unrelated to the stimulus was given (nonflanker error). By manipulating the onset of the flankers relative to the target, we manipulated two variables: the probability (and thus the expectancy) of flanker errors and the proportion of significant attention errors among each error type. Contrary to the predictions of outcome expectancy accounts, we found that the Ne/ERN was larger for flanker errors than for nonflanker errors only in the condition in which flanker errors were particularly frequent. Consistent with the error significance account, however, Ne/ERN amplitude mirrored the estimated proportion of significant attention errors as estimated by multinomial modeling. These results provide support for the idea that early performance monitoring as reflected by the Ne/ERN involves an evaluation of error significance.