Abstract
In everyday life, we feel that our perception of the world is detailed, while our reports of it are poor (Simons & Chabris, 1999, Simons & Levin, 1997). Cohen, Dennett, and Kanwisher (2016) proposed that our rich perceptual experience can be explained by the ability of our visual system to extract ensemble statistics from groups of objects. Ensemble statistics computation overcomes many limits of perception - for instance, it is robust to inattention (Alvarez & Oliva, 2008). However, it is unknown whether ensemble statistics can be computed without awareness. In the present study, we tested the hypothesis that masked ensemble statistics can prime individual stimuli. The participants (n = 45) were presented with an ensemble of oriented gratings and asked to report whether a subsequently presented test grating was tilted to the right or to the left. The test grating had either the mean ensemble orientation, one of the members’ orientations, or an explicit non-member orientation. The ensemble was backward and forward masked and we manipulated the stimulus-onset asynchrony between the masks and the ensemble. In the aware condition, there was a 160 ms blank between the masks and the ensemble. In the unaware condition, there was no blank. Additionally, the participants completed two control tasks to assess whether the awareness manipulation was successful. The results show a main effect of the test orientation: the reaction times were lowest for the mean orientation, higher for member orientation, and highest for non-member orientation. The control tasks revealed that the ensemble was harder to detect in the 0 ms SOA condition, but it was not completely invisible. These results show that priming by masked ensembles is possible, yet it is unclear whether the priming effects stem from unconscious processing.