Abstract:
Psychological theories and empirical evidence suggest that the human performance-monitoring system processes feedback or detects errors in order to learn new tasks and improve behavior. However, much of the research identifying a link between the performance monitoring system and behavioral improvement have been limited to situations in which the inter-trial-interval (ITI) is relatively long (>1000 ms). This dissertation presents two studies that utilized electroencephalography (EEG) and the event-related potential (ERP) technique in order to investigate the consequences of performance-monitoring when ITIs were relatively short (<533 ms) or long (>866 ms). Both studies employed a difficult perceptual decision-making task in order to facilitate investigation of performance-monitoring in relation to task-related sensory attention. In study one, it was found that task-related attention was impaired following errors at short ITIs, as indexed by a reduction in the P1 ERP component. Further, depth of error processing, as indexed by the error-evoked Pe component, negatively correlated with task-related attention (indexed by modulation of the P1 component) on the trial after an error was made; critically, this negative relationship was only present when ITIs were short. In study two, the magnitude of unexpected feedback processing, as measured by the reward positivity (RewP) ERP component, was negatively correlated with task-related attention and accuracy following unexpected feedback; this negative correlation was also present only when ITIs were short. Collectively, studies one and two suggest that two aspects of the human performance-monitoring system, error-detection and feedback processing, cause an attentional bottleneck, which can impair task performance when the next trial occurs in short succession. That is, the very system dedicated to monitoring and improving behavior, the performance-monitoring system, can cause performance decrements when decisions must be made rapidly. Given that the real world often requires decisions to be made in a rapid, sequential fashion, the current results suggest an important limitation of the human performance-monitoring system within ecologically valid contexts.
