EEG activity reveals a trial-by-trial representation of the correctness of decisions

• Main Authors: Jose L Pardo-Vazquez, Isabel López Padrón, Jose eFernandez-Rey, Carlos eAcuña
• Format: Article
• Language: English
• Published: Frontiers Media S.A. 2014-03-01
• Series: Frontiers in Behavioral Neuroscience
• Subjects: Feedback; perceptual decision making; Performance monitoring; event-related potential (ERP); trial-by-trial
• Online Access: http://journal.frontiersin.org/Journal/10.3389/fnbeh.2014.00105/full

Performance monitoring is an executive function, which we depend on for detecting and evaluating the consequences of our behavior. Although event related potentials (ERPs) have revealed the existence of differences after correct and incorrect decisions, it is not known whether there is a trial-by-trial representation of the accuracy of the decision. We recorded the electroencephalographic activity (EEG) while participants performed a perceptual discrimination task, with two levels of difficulty, in which they received immediate feedback. Receiver Operating Characteristic (ROC) analyses were used to reveal two components that convey trial-by-trial representations of the correctness of the decisions. Firstly, the performance monitoring-related negativity (PM-N), a negative deflection whose amplitude is higher (more negative) after incorrect trials. Secondly, the performance monitoring-related positivity (PM-P), a positive deflection whose amplitude is higher after incorrect trials. During the time periods corresponding to these components, trials can be accurately categorized as correct or incorrect by looking at the EEG activity; this categorization is more accurate when based on the PM-P. We further show that the difficulty of the discrimination task has a different effect on each component: after easy trials the latency of the PM-N is shorter and the amplitude of the PM-P is higher than after difficult trials. Consistent with previous interpretations of performance-related ERPs, these results suggest a functional differentiation between these components. The PM-N could be related to an automatic error detection system, responsible for fast behavioral corrections of ongoing actions, while the PM-P could reflect the difference between expected and actual outcomes and be related to long-term changes in the decision process.