Aqsa Hassan¹, Heiko H. Schütt; ¹University of Luxembourg

Masking experiments are an important basis of our understanding of early visual processing. Masking data allows us to constrain the non-linear interactions between spatial frequency and orientation-tuned channels like divisive normalization. It is well established how these interactions work across spatial frequencies and orientations. However, it is not well understood how these interactions work across space, and we lack data for different mask shapes and sizes. Therefore, we set up a contrast masking experiment to test how strongly different sizes of masks increase the contrast threshold. We use small Gabor targets (SD=0.5 cycles) centrally superimposed over a 15% contrast grating mask of the same spatial frequency and orientation. The circular aperture of the mask is composed of a flat top and a gradual border with the same fall off as the Gabor target. We vary the size of the flat top of the mask for changing its size. The spatial frequencies of both stimuli are 1, 2, or 4 cycles per degree, which are displayed at 5° and 10° eccentricity. The observer’s task is to judge on which side the test stimulus is superimposed on the mask. In our preliminary experiments, contrast thresholds increase with increasing mask size and saturate at some point as expected. However, the shape of the masking curve is surprisingly independent of eccentricity and spatial frequency, saturating always at a radius of about 1° visual angle. We plan to extend our results with a foveal condition and masks elongated along or orthogonal to the target grating by the time of the conference. Our results have important consequences for our understanding of early visual processing as the prevalent expectation based on neural recordings is that normalization pools are proportional to receptive field sizes and should thus be strongly eccentricity and spatial frequency dependent.