Abstract
Background: Two mechanisms have been proposed to explain how attention affects signal processing: gain (signal enhancement) and tuning (external noise exclusion). Gain and tuning can be distinguished by using the equivalent noise paradigm, which measures threshold at different levels of external noise. Whereas the gain model only predicts attentional benefits with low external noise, the tuning model predicts attentional benefits with high external noise. Using this paradigm, we have shown that spatial attention only increases the gain of motion selective channels (VSS 2006). In the present study we assessed the influence of feature-based attention on the gain and tuning of motion selective channels by cueing observers to a global motion direction under different external noise levels.
Methods: In each trial, observers were shown a moving dot cinematogram at fixation (100 ms). The local motion directions of individual dots were drawn from circular gaussian distributions centered on a direction, either clockwise or counterclockwise from the cue direction. In the Attended condition a small cue at fixation pointed to one of four directions 600 ms prior to the stimuli presentation, instructing observers to attend to that upcoming direction. In the Neutral condition, the same cue appeared simultaneously with the stimulus. Observers performed a 2AFC direction discrimination task, reporting whether the global motion was clockwise or counterclockwise relative to the cue direction. External noise was manipulated by varying the coherence of the dot fields, which was defined as the variance in dot trajectories. We obtained direction thresholds for different levels of external noise.
Results: For all observers, attention reduced thresholds across all motion coherence levels, i.e., for both low and high external noise levels. These findings are consistent with a model where attention both increases the gain (benefit at low noise) and narrows the tuning (benefit at high noise) of direction selective channels.
NRSA to S.L. (F31 NS051111-03) & NIH to M.C. (R01 EY016200-01A2)