Abstract
Several studies suggest that attentional modulation can act very early on the visual system by changing the properties of units tuned to targets and/or distractors. If attention acts as early as in V1, either by enhancing the gain of the units seeing the target or by reducing the spread of the tuning function of such units, or both, then the signature of the responsible mechanism(s) can be revealed by using a masking paradigm. We devised a task that mixes the classical Posner and masking paradigms. The test was a vertical Gabor patch varying in contrast displayed in one of two intervals (2IFC) and in one of two locations on either side of fixation. The mask, a ‘barcode’ patch varying in contrast and orientation, was presented in both locations and overlapped the test. Attention was summoned through peripheral cues signaling the location of the test patch in the signal interval, and a random location in the noise interval. Comparing thresholds for parallel, orthogonal and no mask in the two attentional conditions showed a significant difference only with a parallel mask, which disrupted performance strongly. When the mask is orthogonal or absent, there was no difference between cued and uncued conditions. We then measured a tuning function by systematically varying the orientation of the mask. Interestingly, the width of the tuning function is unaffected by the attentional condition, whereas sensitivity increased by 50% when the cue signaled the test location. The results indicate that uncertainty plays little or no role in our task, as the cue is effective only under particular angle relationships between test and mask. Comparing the tuning functions under the two attentional conditions suggests that attention produces a pure signal enhancement. In fact, the width of the tuning function does not differ in the two conditions, showing a spread of about 40°, similar to previous physiological estimates.
This work was made possible by a NEI grant R01EY12038 to PV.