Abstract
Current models of attention (e.g. Reynolds & Heeger, 2009) posit that selection of task-relevant material results because the visual system is biased to better represent task-relevant than task-irrelevant inputs. Such biasing may emerge from "top-down" attentional control and/or "bottom-up" perceptual processing (McMains & Kaster, 2011). In this experiment, we investigate whether automatic perceptual organizational processes mediated by "top-down" feedback may also contribute to attentional selection. In figure-ground segregation the ground is suppressed. Ground suppression is greater when a portion of a familiar versus a novel object is suggested on the groundside of a figure; this suppression is likely mediated by feedback from high-level object representations (Salvagio et al., 2012) that lost the competition for figural status. We assessed whether this top-down suppression affects the efficiency of selective attention using a flanker task (Eriksen & Eriksen, 1974). Flanker displays were preceded by figures with either high suppression grounds (i.e., portions of familiar objects suggested but not perceived on the groundside) or low suppression grounds (SOA=110ms). Critically, flanker display elements were positioned such that the task-relevant element fell on the figure and task-irrelevant elements fell on the ground. If automatic, "top-down" perceptual processes involved in perceptual organization can influence attentional selection, flanking items presented on high suppression grounds will be less well represented and influence task performance less than those presented on low suppression grounds. We found that interference from inconsistent flanker items was indeed lower when they fell on the high suppression (9%) versus low suppression (12%) grounds (p<.05). Suppression of task-irrelevant material, then, was modulated by automatic, top-down perceptual processes, traditionally considered to be "outside" of the attentional system. Top-down automatic processes may indeed contribute to selective attention if they, like top-down attention, alter the relative strength with which task-relevant and task-irrelevant material are represented in the visual system.
Meeting abstract presented at VSS 2013