The Honeycomb illusion: Peripheral vision of contours, unlike that of objects, is refractory to predictions, extrapolation and memory effects

In the Honeycomb (HC) illusion, a uniform pattern is not perceived as such, with a visible change from the center to the periphery of the visual field (Bertamini et al., 2016). This illusion does not follow expectations and predictive perception. In this work, we specifically investigate the effects of Inattentional Blindness and Extrapolation in the HC illusion, through a series of two pairs of experiments. In the first pair, we test participants' ability to detect stimuli in two different conditions. In the illusion condition, the lines overlap with the intersections of the grid, while in the control condition, they do not. Lines could tilt left or right. In the second pair of experiments, we investigate participants' ability to discriminate these stimuli, by introducing an additional central region of varying size filled with crosses and asking them the orientation of the peripheral lines. Each experiment also had a variation with an additional foveal task to test the Inattentional Blindness phenomenon. Results of the first pair of experiments show that participants are able to detect stimuli only up to 13.44±0.24 degrees in the illusion compared to 49.23±0.21 degrees in the control condition. However, the size of the region reported as filled with stimuli decreases significantly more in the control condition when introducing the foveal task. This suggests that the HC illusion is less subject to Inattentional Blindness. In the second pair, the difference between the mean dprimes in the illusion versus the control condition remains the same whether there is the additional foveal task or not (approximately 0.6 in both cases). There is however a significant interaction effect between the addition of the foveal task, the condition and the size of the central region of degrees, further suggesting that the Honeycomb illusion responds differently to Inattentional Blindness compared to uniformly perceived patterns. Finally, a comparison between the two experiments shows that the size of the region in which participants are able to detect stimuli corresponds to the one in which they can discriminate them in the illusion condition, while it is significantly larger in the control condition. This indicates that the HC illusion is less prone to Extrapolation, thus proposing a more genuine vision. Attention modulated the responses, without changing the basic differences. A fundamental difference in the perception of contours in central and peripheral vision is refractory to expectations.