Error-Related Negativity on a Reinforcement Learning Task
Specific ERPs such as error- related negativity (ERN) and feedback- related negativity (FRN) reflect error processing/performance monitoring. The reinforcement learning theory of error processing suggests that ERN reflects the first detection of error commission. In order to elicit ERN, participants must perceive a response as erroneous. This implies that as learning occurs, ERN amplitude increases. A previous study (Horst, Johnson, & Donchin, 1980) examining the P300 component on a learning task demonstrated increased amplitude in response to violations of participants’ expectations about task events. The current study sought to replicate this finding while extending the analysis to include ERN and FRN amplitudes on incorrect trials. Participants completed a complex learning task in which they were required to learn 3 lists of paired nonsense syllables. Participants viewed a stimulus syllable, typed what they believed to be the correct response, then indicated their confidence level on a scale from 0- 100. Feedback was then presented regarding response accuracy. Each list was considered learned after 10 consecutively correct trials. Data from electrodes Fz, Cz, Pz, and Oz showed an increased ERN amplitude 50ms after error commission on incorrect trials. The ERN and FRN amplitudes were greater for incorrect trials than for correct trials. Larger P300 amplitudes were observed for trials in which participants’ outcome expectation was violated. This extension of previous findings provides further insight into the role of error processing in the context of reinforcement learning.
Ridley, E. A.; Jones, M. R.; Ashworth, E. C.; and Sellers, Eric W., "Error-Related Negativity on a Reinforcement Learning Task" (2019). ETSU Faculty Works. 892.