Abstract
The visual system can represent the information about multiple objects in the form of ensemble statistics, such as mean feature per group. A question as to whether visual statistical summaries require attention to be calculated is under debate. We investigated this question using an ERP marker of automatic sensory discrimination under diverted attention, visual mismatch negativity (vMMN). In Experiment 1, observers attended to a central cross for detecting oddball events. The cross changed it orientation from vertical to horizontal in 10% of trials. Participants should respond to this change. The cross was surrounded by background sets of 64 lines of different orientations (variation range was 50 degrees) that were unattended. The oddball events in background stimuli included changes in lines mean orientation by 20 degrees. In Experiment 2, observers had to attend the change of mean orientation in the background stimuli. Central cross orientation was constant across all trails. No evidence for the vMMN to unattended ensemble changes in the selected temporal and spatial ROI was found. However, when ensembles were attended we found the P300 component correlated with conscious change detection. The latter was supported by good behavioral rate of oddball detection when the ensembles were attended. These results let us suggest that attention is critical for ensemble discrimination, at least within the range of stimulus differences that we tested.