An EEG study of masking effects in RSVP [Abstract]

We present an EEG study that investigates how target items are processed when presented in Rapid Serial Visual Presentation (RSVP). The RSVP task was similar to the well-known Attentional Blink experiment [Chun and Potter, 1995], however, the stream contained only one target. To increase task difficulty, the presentation rate was doubled to 47ms SOA. Half the targets were followed by a blank to test the effects of backward masking. A skeletal task [Ward et al., 1997] allowed study of targets presented without surrounding distractors (except a following mask). Subjects' EEG waves were recorded at 20 electrode sites. The study was motivated by predictions made on the basis of the ST2 model of temporal attention and visual working memory [Wyble and Bowman, 2005] concerning the influence of masking and task difficulty on target processing. In line with our predictions, the following results are observed: We find delayed processing of targets in RSVP streams compared to targets that are presented stand-alone (skeletal). In RSVP streams we do not observe accelerated processing of masked compared to unmasked targets as suggested by Kessler et al. (2005). A categorisation of target letters based on subject performance ('hard' or 'easy') modulates target ERPs, in particular the P3 component. Finally, we investigate the 'all-or-none' issue of conscious perception [Sergent et al., 2005] by correlating subjects' task performance and P3 amplitude using EEG single-trial analysis.