Abstract
Recent research indicates that sustained attention could in fact process information rhythmically, as a sequence of successive cycles with its own intrinsic frequency. When two items must be attended, an intriguing corollary of this "blinking spotlight" notion could be that the successive cycles are directed alternately to each target; as a result, each item would effectively be sampled at half the intrinsic rate of attention. Here, we tested this prediction in two experiments. In an endogenous attention task, subjects (n=8) covertly monitored one or two peripheral images (one house, one face) in order to detect a brief contrast change. In the sustained occipital EEG power, attending to two vs. one item resulted in a relative increase around 4Hz and a relative decrease around 10-11Hz. In a second experiment, we tested if comparable oscillations could be observed in the stimulus-evoked EEG visual representational content. Subjects (n=9) saw a first peripheral image (house or face) displayed alone for 600ms, before a second one (face or house) also appeared for the same duration, but at a different peripheral location. In monkeys, a similar protocol was found to trigger low-frequency 5Hz oscillations in inferotemporal single-cell activity, reflecting competitive interactions between neural populations selective to the two objects (Rollenhagen & Olson, 2005). Using time-resolved MVPA on EEG evoked-responses, we were able to create item selective classifiers that constantly indicated which stimulus was on the screen (peak AUC=0.8 around 100 ms after image onset). The time-course of single-trial classifier decision values presented a relative peak around 11Hz when only one object was present, and around 4-5Hz when two objects were on the screen. Taken together, these results are compatible with a blinking spotlight of attention, sampling information periodically around 10-11Hz, and resulting in a half-frequency effective sampling (around 4-5Hz) when there are two items to attend.
Meeting abstract presented at VSS 2016