Abstract
Previous research has demonstrated that applying onset transients to crowded targets can improve their orientation discriminability (Greenwood, Sayim & Cavanagh, 2014). We replicate this finding, but specify that the advantage conferred by onset transients occurs only 1) when there is a pre-existing feature contrast between the target and flankers 2) when attention is directed to the target near the time of transient onset. A cued RSVP procedure controlled attention--a sequence of white L’s and R’s was shown at fixation; after a variable delay, the last letter became black. Observers shifted their attention left or right, depending on the identity of the last white letter, to a flanked 3cpd target Gabor. The Gabors were shown 10 deg. left and right of fixation throughout each trial; only after the L or R cue did the observers know which would be the target. Having shifted attention, subjects reported the target’s tilt, left or right, from vertical (mean 3 deg.). The flankers were either nearly parallel or nearly orthogonal to the target. A 50ms transient ‘blink’ was applied to a subset of stimulus elements randomly: either targets, or flankers, or both, or none. Preliminary results: orientation discrimination of orthogonal targets was improved when transients were applied to the target within ±100ms of the RSVP cue. Otherwise, transients had no effect. In sum, when a target is both top-down attended and featurally salient, an additional ‘bottom-up’ attention-capturing transient can partially recover the target from crowding. Why? Transients may boost the target signal even further, or briefly reduce the zone of integration.
Meeting abstract presented at VSS 2015