Looking at the Ebbinghaus illusion: differences in fixations fail to explain a classic perception-action dissociation

Visual recognition and visually-guided action present an animal with different computational requirements. Take, for example, changes in the retinal image of an object that occur with changes in the spatial relationships between the object and the observer. This retinal variation matters when reaching for the object, because estimates of the spatial relationships that drive it are useful for coding appropriate limb motor commands. At the same time, this variation must be filtered-out to extract the object’s invariant features and associate those features with stored semantic correlates. An organism with multiple visual areas for object coding can afford specialized operations to suit different cognitive and behavioural endpoints. Pictorial size-contrast illusions like the Ebbinghaus display, which influences the perceived sizes of disks embedded in it, have offered behavioural vision scientists a controversial vehicle for exploring the functional organization of human vision; in a now classic dissociation, the hand’s in-flight grasp aperture, which normally scales to target size, resists the Ebbinghaus display's influence on perceived disk size. An alternative account posits that grasp aperture resists the illusion because participants fixate on the target disk and ignore the illusion-inducing context. To test this account, we tracked both hand movements and gaze while participants made forced-choice judgments of relative disk size in the Ebbinghaus display or did so in combination with grasping or perceptually estimating the sizes of the illusory disks. We replicated the classic dissociation: grasp aperture was refractory to the measured illusory effect on perceived size, while judgments and manual estimates of disk size were not. Importantly, the number of display-wide saccades per second and the percentage of total fixations directed at the selected disk (in time and number) failed to explain the dissociation. Our findings support the contention that object perception and goal-directed action rely on distinct and parallel coding for objects.