Abstract
Purpose. A recent study (Meng & Tong, 2004) indicated attention can selectively bias the intended percept for ambiguous figures (i.e. Necker cube) but not binocular rivalry displays. For Necker cubes, attentional control (or “top-down” influence) may operate by enhancing the desired representation, or suppressing the alternative representation. In addition, low-level cues (or “bottom-up” influence) such as eye fixation can also boost the desired percept by increasing the stimulus strength. However, the magnitudes of top-down vs. bottom-up influences on Necker cube reversals have rarely been compared in the past. Our present goal is to directly compare the magnitude of top-down vs. bottom-up influence with a random noise paradigm.
Methods. The Necker cube (width= 7°), with a red fixation crosshair (= 0.3°) in the center, was presented on a black background of Acer 17″ LCD monitor. Experiment 1 measured bottom-up influences in 28 naïve observers with passive viewing. The strength of bottom-up cue was manipulated by adding 0%–99% random noises exclusively in the lower square (biasing the “top-view” percept) or the upper square (biasing the “bottom-view” percept). Observers received 27 trials in a mixed order (9 conditions × 3 repeats). Experiment 2 measured the extent of selective attentional control over the 0 % noise Necker cube in the same observers. Three instructions were given: (1) passive viewing; (2) perceive the cube from top; and (3) perceive the cube from bottom. In each 20-s trial, observers were instructed to maintain fixation while monitoring their perceptual state and reported perceptual switches by pressing one of two keys. Results. First, the percentage of dominance duration of a percept increased reliably as a function of the noise density. Second, the attentional modulation was significant. Third, the magnitude of bottom-up cue was comparable to that of attentional modulation and was affected by the individual's initial bias.
This project is supported partly by NSC 95-2413-H-002-001 to C.C. Chen and by CMU96-108 to S. H-L. Chien.