EEG-based neural decoding of Intelligence Questions

Decoding mental processes in single trials is one of the prerequisites for tailoring learning paradigms, which aim at improving performance in cognitive tasks. In this study user choices are predicted in a matrix reasoning task. By employing multivariate analysis techniques we are able to show that it is possible to decode the subjects' answers prior to their response by means of ERP-based EEG data. Our results indicate that parieto-occipital ERP components are most discriminative and lead to an average single-trial accuracy of 67.7 ± 6.3% across 5 subjects.