Uncorking the bottleneck of crowding
34.15, Sunday, 17-May, 2:30 pm - 4:15 pm, Talk Room 1
Mauro Manassi1, Aaron Clarke1, Michael Herzog1; 1Brain Mind Institute, École Polytechnique Fédérale de Lausanne (EPFL), Switzerland
In crowding, target perception deteriorates when flanking elements are added. Crowding is usually thought to be an inevitable bottleneck of object recognition where (1) more flankers increase crowding strength, (2) only nearby elements interfere with the target (Bouma’s window), and (3) target-flanker interactions occur mainly within feature specific ‘channels’. Here, we show that none of these assumptions hold true. We determined offset discrimination thresholds for verniers at 9° of eccentricity. When the vernier was embedded in a square, thresholds increased compared to the unflanked threshold - a classic crowding effect. When the central square was embedded in an array of seven alternating squares and stars, crowding remained at a high level. Surprisingly, when this array was added both above and below the central array, crowding strongly decreased. When shuffling all the squares and stars in a random manner, crowding was again strong. These results show that (1) more flankers can decrease crowding strength, (2) uncrowding can be due to elements outside Bouma’s window and (3) shape rather than low-level interactions determine vernier offset discrimination. These results cannot be explained by the repetitive structure of the array. We show by Fourier analysis, that adding repeating elements does not improve the discriminability for a vernier offset classifier in the Fourier domain, but in fact, it makes the vernier offset less discriminable. We propose that visual acuity for each element in the visual scene depends on all elements in the entire visual field and, on top of that, on the overall spatial configuration. Crowding is not an inevitable bottleneck because adding elements can ‘uncork’ the bottle and recover target information.