Closely matched comparisons suggest that separable processes mediate contextual size illusions

Abstract

Previous research suggests the magnitudes of the Ebbinghaus, Delboeuf, Ponzo, and tilt illusions all depend on the cortical distance between the neural representations of target stimuli and the surrounding context. However, several psychophysical studies found no compelling association between these illusions, calling this hypothesis into question. Here we ask if these discrepant reports could arise from methodological differences between these studies. We ran a battery of visual size illusion and basic discrimination tasks with carefully matched geometric properties, using a classic forced-choice design. In our small, homogenous sample, the Ebbinghaus and Delboeuf illusion magnitudes were strongly correlated, consistent with the idea that they reflect the same underlying mechanism when other sources of individual differences are minimised. Ponzo illusion magnitude also correlated with these two illusions, although less strongly in the case of the Ebbinghaus. Interestingly, the classic arrowhead version of the Mueller-Lyer illusion did not correlate with any of the other illusions or even with the objective ability to discriminate line length. This suggests that an altogether separate process underlies this perceptual effect. We further demonstrate that presenting stimuli briefly with central fixation critically affects measurements of the Ebbinghaus illusion. Additionally, we found that measuring illusion magnitude via adjustment is less reliable compared to two-alternative forced-choice procedures. Taken together, our findings suggest that different tasks probe separable processes determining illusion measurements. They further highlight the importance of the experimental design when testing relationships between perceptual effects and their links to neural processing.