An approach to single trial analysis of event-related potentials based on signal detection theory

The aim of the present paper was to introduce a single trial analysis to event-related potential measurement in order to illuminate the mechanisms behind an impaired P300 amplitude occurring under certain experimental conditions. For this purpose we applied tools from linear system theory and signal detection theory to single trials in an oddball paradigm in order to estimate the amplitude of the positive deflection around the P300 latency following target and nontarget stimuli. According to the density functions of these amplitude distributions we operationally defined 'P300 absent in target' (P300 amplitude smaller than an individual threshold under target conditions) as well as 'P300 present in nontarget' (amplitude greater than the threshold under nontarget conditions). Our approach appears to be appropriate to differentiate between an overall reduction of single trial event-related potentials and seldom elicited P300s.