Attentional control of spatial scale: effects on self-organized motion patterns.

Prior to the presentation of a test stimulus, subjects' attentional state was either narrowly focused on a particular location or broadly spread over a large spatial region. In previous studies, it was found that broadly spread attention enhances the sensitivity of relatively large spatial filters (increasing the perceiver's spatial scale), thereby diminishing spatial resolution and enhancing sensitivity to global stimulus structure. In this study it is shown that attentional spread also affects the self-organization of unidirectional versus oscillatory motion patterns for the directionally ambiguous, counterphase presentation of rows of evenly-spaced visual elements (lines segments; dots); i.e. qualitatively different motion patterns can be formed for the same stimulus at different spatial scales. Although the degree to which attention is spread along a spatial axis can be controlled by the perceiver, the effects of spread attention are not limited to a single axis. These results, as well as previously observed effects of attentional spread on spatial resolution, are accounted for by a neural model involving large, foveally-centered receptive fields with co-operatively interacting subunits (probably at the level of MST or higher).

The effect of attentional spread on spatial resolution.

The effects of attentional spread were studied by having subjects detect a luminance increment along a row of evenly spaced dots. The increment could occur for the central, fixated dot (Narrow Attention) or for either the fixation dot or one of the four dots to its left or right (Broad Attention). Narrow Attention enhanced the detection of luminance increments for the fixated dot, and also enhanced spatial resolution near the fixation dot for judgments of vernier alignment and separation. This indicated that the sensitivity of small spatial filters in the fovea was increased more by narrowly focused than broadly spread attention. Effects of attentional spread on spatial resolution were not obtained for judgments of the separation between two peripherally located targets, perhaps because of their dependence on eccentricity (position) rather than separation.

Cooperative interactions and the perception of motion and stationarity for directionally ambiguous apparent-motion stimuli.

Evidence is reported that stationarity rather than motion can be perceived for displaced stimuli, not because of insufficient motion energy for the stimulus to activate individual motion detectors, but because of cooperative interactions that actively suppress the perception of motion. A long row of evenly spaced dots was presented in counterphase; the dots presented during each 180 ms frame were located midway between the dots presented during the previous frame. When either a blank interval as brief as 15 ms was inserted between successive frames or the luminance polarity of the dots was reversed on successive frames, the unidirectional motion pattern perceived for small interdot distances (small displacements) was replaced by the perception of stationarity. However, when under the same conditions a single dot was displaced over the same small distances, motion rather than stationarity was perceived. The contrasting results for the long row of displaced dots and the single displaced dot indicated that when the activation of motion detectors is weakened (by nonzero interframe intervals and/or the reversal of luminance polarity), the perception of motion can be actively suppressed by the collective effects of inhibitory interactions among the large ensemble of detectors that is activated by the long row of dots.

Spatial scale dependent in-phase and anti-phase directional biases in the perception of self-organized motion patterns.

A long row of evenly spaced dots is displaced on successive frames by half the distance between the dots. Although these stimuli are directionally ambiguous, spatially and temporally coherent unidirectional and oscillatory motion patterns are perceived as a result of the temporal persistence of competing in-phase and anti-phase directional biases, respectively. The perceiver's spatial scale is critical is determining whether dots are near enough to favor an in-phase bias or far enough apart to favor an anti-phase bias. The results are explained by a differential-gradient model of cooperative interaction, which specifies that the strength of facilitating (excitatory) interactions among motion detectors with similar directional selectivity falls off with distance at a greater rate than the strength of competing inhibiting interactions.