Source separation of ERP components in brain responses to reward evaluation

An event related potential (ERP) called feedback related negativity (FRN) is believed to encode reward prediction error (RPE) in the human brain. In this paper we conduct an experiment to study the influence of effort on the size of RPE through an investigation on the magnitude of the corresponding FRN. However, FRN is inevitably distorted by the neighboring P2 and P300 ERP components, and this makes the evaluation of FRN difficult. To resolve this issue, we apply blind source separation to decompose the ERP signal into isolated ERP components, adopting two-step temporal and spatial decompositions, using the Promax and information-maximization algorithms, respectively. Experimental results show that FRN can be reliably isolated and the influence of effort on RPE can then be evaluated.