Functional specificity of attentional modulation in the disparity domain

It has been suggested that absolute disparity cues, unlike relative disparity cues, are perceptually inaccessible (Chopin et al., JVis, 2016). If so, processing of absolute disparity information should be less affected by manipulation of attention. Steady-state visual evoked potentials were recorded (N=20) in response to dynamic random dot stereograms (DRDS) alternating at 2 Hz between a flat plane at zero disparity and a 0.5 cpd crossed-disparity sine-wave grating. Disparity was measured at 10 equal log steps. In one condition, participants attended to monocular nonius lines at fixation and pressed a button when a color change occurred. In a free attend condition, participants attended to the DRDS stimulus while maintaining fixation with no button presses. Evoked responses were robust at the first four harmonics of the stimulus frequency (1F, 2F, 3F, 4F). The disparity response function at 1F in the free attend condition was leftward shifted relative to the attend nonius condition, consistent with an input-gain mechanism. The 3F and 4F responses showed no effect of attention. We hypothesized that the unmodulated 3F and 4F signals came from the leading edge of the response and conducted a follow-up experiment to determine the timing of the task effect and the relative timings of absolute and relative disparity encoding (N=9). The disparity grating response (relative disparity) started at around 100 ms and attention began to modulate the response around 50 ms later. Responses to a disparity plane (absolute disparity) and grating diverged at the same ~150 ms latency. Importantly, the task had no effect on the response to the absolute disparity condition at any latency. Our results suggest that the earliest disparity responses represent the encoding of absolute disparity and that responses to absolute disparity are unaffected by attention, whilst responses to relative disparity, once they have been extracted, are modulated by attention.