Abstract
Observers viewing a periodic motion stimulus frequently experience illusory reversals of the direction of motion despite continuous lighting conditions (the continuous wagon-wheel illusion or c-WWI). This illusion is maximal at temporal frequencies around 10 Hz, and is limited to the object of attention. It has been proposed, therefore, that this phenomenon reflects periodic sampling of motion information by attentional processes. What is not known, however, is how such periodic sampling behaves when multiple objects are attended simultaneously. In order to address this question, we presented participants with 40 s sequences of rotating gratings during which several actual motion direction reversals occurred, and asked them to report the number of reversals in each sequence. Both the number of gratings (set sizes 1 to 4), and the temporal frequency at which they rotated, were varied from trial to trial. We considered overestimations of the number of reversals as manifestations of the c-WWI, and fitted the temporal frequency tuning of the illusion for each set size with a Gaussian function. Preferred frequency decreased consistently with increasing set size, from approximately 10 Hz for set size one to about half that for set size four. The fact that preferred frequency (which we take to reflect the rate at which attention samples each individual stimulus) decreases when attention is divided constrains the underlying mechanisms of attentional deployment. Neither a simple parallel model, whereby sampling rate would be independent of set size, nor a strictly serial model, whereby sampling rate would be inversely proportional to set size, can account for this finding. An imperfect serial strategy, in which a variable number of samples are taken from each successively attended stimulus, could be one of several alternative models.
This research was supported by grants from the ANR (project ANR 06JCJC-0154), the Fyssen Foundation and the EURYI to R.V. The authors wish to thank P. Cavanagh and F. Gosselin for useful discussions.