Purchase this article with an account.
Hörmet Yiltiz, Denis Pelli; Noise masking and crowding reveal two very different kinds of spatial integration.. Journal of Vision 2017;17(10):802. doi: https://doi.org/10.1167/17.10.802.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Recognition of a simple target, e.g. a letter, can be prevented by adding visual noise or nearby clutter. Both effects involve integration of visual information over space. The literature says that crowding integrates "features," and noise-masking integrates contrast energy. Here we compare the spatial extent of the two kinds of integration. For crowding, this spatial extent is called the "crowding distance" (or "critical spacing") and is well-known. For noise masking, we made new measurements. We added white noise (independent pixels) with a Gaussian spatial envelope centered on the letter target. This is like critical-band masking, but varying the extent ("band") in space. Suppose a detector that integrates over a fixed area. Its threshold would increase with noise radius until the radius goes beyond the fixed area. Indeed, that is what we found for letters varying in size from 0.5 to 16 deg, at eccentricities of 0 to 32 deg. We expected this noise-integration radius to match the crowding distance, but they are very different. The crowding distance is one third of the eccentricity, independent of letter size. The noise-integration radius is roughly the letter radius (half the size), independent of eccentricity. For a small letter at large eccentricity, the noise-integration radius is less than 1/10 the crowding distance. For a large letter at small eccentricity, the noise-integration radius is more than than 100 times the crowding distance. This reveals two fundamentally different kinds of spatial integration.
Meeting abstract presented at VSS 2017
This PDF is available to Subscribers Only