Performance monitoring: Error detection and the error-related negativity in choice-related time tasks.

摘要

Supervisory control is an integral part of the human information processing system. A component of the event-related brain potential, the error-related negativity (ERN), appears to be a manifestation of a process that monitors the accuracy of performance. Current evidence suggests that the component is generated in the anterior cingulate cortex. In choice reaction time tasks, the monitoring process compares a representation of the appropriate (correct) response with a representation of the response actually made. The ERN amplitude reflects the degree of mismatch, i.e. the error signal. In the experiments reported in this thesis, subjects performed a visual search, a memory search and an Eriksen flankers task. The error-related negativity (ERN) was smaller with high search loads, with non-automatized stimulus-response mappings, with extended wakefulness, and with reduced stimulus contrast. These data are consistent with the view that difficulties with target identification compromise the computation of the representation of the appropriate response and the error detection process. Uncertainty about the response that should have been made was also reflected in ratings of performance accuracy. The ERN amplitude was large when subjects believed they were in error. The dependence of performance monitoring on the outcome of stimulus-related processing had implications for the effectiveness of error detection. The data suggest that performance accuracy in conditions of data-limited processing was undetermined, because the representation of the appropriate response was ambiguous. However, errors due to impulsive responding were detected, because continued processing of the stimulus after the response was selected prematurely permitted the representation of the appropriate response to be completely specified. The idea that the ERN is associated with performance monitoring was contrasted with the crosstalk monitoring hypothesis, developed to explain functional blood flow changes in the anterior cingulate cortex. The insensitivity of the ERN to the presence of response competition in the Eriksen flankers task provided little support for the model of crosstalk monitoring. Investigations of supervisory control using the methods of cognitive neuroscience promise to further our understanding of the cognitive and neural mechanisms that underlie this critical aspect of human information processing.