Single-trial decomposition: a new approach to the analysis of neuronal reactions during psychological tasks

Neurophysiological and statistical assumptions underlying conventional procedure of averaging the neuronal responses to sequentially presented stimuli or psychological tasks are critically discussed. A new approach to the analysis of the neuronal evoked reactions is developed based on the decomposition of the single-trial discharge rate patterns into the separate peak-shaped components characterised by the latent periods, amplitudes and durations. It is demonstrated that using only latent period and signs of each component's amplitude is sufficient to detect reliably all the latency intervals where the conventional peristimulus-time histograms reveal significant deviations of the neuronal discharge rate from the mean background level. The approach suggested enables the analysis of the real physiological events within the single trials, obscured by the conventional averaging into the peristimulus-time histogram curve.