Visual and non-visual contributions to perception of object movement during observer movement

Abstract

Accurate estimation of object movement is difficult for moving observers, since it is unclear whether to attribute retinal motion to movement of the object or observer. Recently, we have proposed a purely visual neural mechanism (?flow parsing?) which aids in this process by attempting to globally filter retinal motion due to observer movement. Here we investigate the contributions of visual and non-visual information to object movement perception during pursuit eye movements. Participants tracked a pursuit target amongst a background of stereoscopically presented wireframe objects and made judgements about the trajectory of a probe dot moving vertically upwards on the retina (assuming perfect tracking of the target). Eye movements were recorded and trials on which tracking was poor were excluded. Extra-retinal (E) movement information was manipulated by changing the target speed. Retinal (R) movement information was manipulated by changing the relative speed between the target and background. Using a 2IFC procedure, on each trial participants saw the reference stimulus, for which R and E were in conflict (E = 3 deg/s; R = 5 deg/s ? achieved by moving the background in the opposite direction to the target at 2 deg/s) and a comparison stimulus, for which R and E were equal (E = R = 1, 2,..., 7 deg/s). Participants judged the interval in which probe movement was more vertical. If probe trajectory perception was based on extra-retinal information alone then the PSE of the recovered psychometric function would be 3 deg/s. PSEs above 3 deg/s indicate some contribution of the retinal information. For all (N=3) observers, PSEs were considerably greater than 3 deg/s and, assuming a linear combination rule, the associated estimated weights for R were between 25% and 49%.We conclude that visual self movement information is involved in the estimation of object movement, even when extra-retinal information is present.