Abstract
It is known that target pop-out reduces crowding (Kooi et al., 1994; Poder, 2006) and, conversely, target-flanker grouping increases it (Sayim et al., 2011). This indicates that global rather than local factors underlie crowding. Attention is one such factor. Previously, we have demonstrated a strong effect of the distribution of attention on crowding (Petrov & Meleshkevich, 2011). Intrilligator & Cavanagh (2001) explain crowding based on the resolution limit of spatial attention in the periphery, but they do not explain this limit. Here we provide evidence for a new hypothesis where the target-flanker grouping is instrumental in setting this limit. Two target Gabors were presented 8 deg. left and right of fixation, 36 flanker Gabors formed a uniform texture around each target. Steady state visually evoked potentials were used to frequency-tag target and flanker Gabors contrast reversing at two different frequencies. We measured EEG amplitude at the flankers’ second harmonic frequency in three conditions: flankers of the same orientation and contrast as the target (crowded condition), flankers of orthogonal orientation or lower contrast with respect to the target (pop-out condition), and the latter with attention directed away by a foveal task (unattended condition). In the first two conditions attention was directed to the targets by a target-related task. A separate psychophysical experiment confirmed strong crowding in the first condition and weak or no crowding in the second condition. We observed a significant decrease in flanker responses in the pop-out condition compared to the crowded condition and a similar decrease in the unattended condition. Importantly, the two effects were significantly correlated across observers (n = 22, p <0.001). We hypothesize that when target-flanker array elements group, spatial attention involuntarily spreads over the whole array (strengthening flanker responses) inducing crowding. When the target pops-out, attention narrows to the target area reducing crowding.
Meeting abstract presented at VSS 2013