Item

Abstract

We investigated the mechanisms of pattern masking in a scene recognition task by recording simultaneously psychophysical performance and magnetic brain activity using magnetoencephalography (MEG). Photographs of natural scenes were displayed for various durations and then immediately followed by a pattern mask. We were able to identify the transient brain activation caused by the switching from scene to mask. The latency of this mask-transient signals the earliest cortical interaction between target and mask. The scenes alone elicited a transient occipital activation starting 70ms after stimulus onset, peaking at 110ms and reaching a minimum at 160ms. By comparing psychophysical performance with the latency of the mask transient and the dynamics of the undistorted processing of the scene at various SOAs we found that the initial occipital activation peak reflects processes vulnerable to pattern masking. The additional activation from the mask falling within this target-processing activity significantly reduces recognition performance. We found no measurable impact of mask activation later than 160ms after target onset. Our results indicate that the information necessary for the reliable recognition of a scene among a set of distracters can be extracted within about 90ms of cortical processing. Our data support the view, that the effect of the pattern backward mask occurs by overwriting a visual buffer in which the scene is stored during analysis.