Functional data analysis of cognitive events in EEG

This paper investigates a data-driven approach to the detection of statistically reliable fluctuations in electroencephalographic (EEG) brain waves. The main advantage of this approach is the ability to automatically identify meaningful sources of variation in signals dominated by large spatiotemporal fluctuations due to noise. The approach is tested on artificial data as well as on EEG signals obtained in a cognitive experiment. Results not only show that the approach can robustly identify meaningful variations in brain activity, but also open possible applications to brain-computer interfaces based on single-trial analyses.