Item

Abstract

It is still not very clear how the visual system compensates for self-induced visual motion. This mechanism is crucial to convey visual stability, and also to recognize motion in the external world. There are two possibilities how an object can change its position in your visual field. Either it moves in the outside world or we move our eyes. In both cases its image will move across the retina. The mechanisms enabling us to discriminate between these two options are still not well understood. It is thought that efference copies of eye movement commands are integrated with visual input, allowing to separate self-induced retinal motion from external objective motion. A recent fMRI study showed that area V3A encodes visual motion almost exclusively in world-centered (objective) coordinates, while being almost unresponsive to retinal motion per se. Conversely, the human motion complex (V5/MT and MST) encoded both, objective and retinal motion with equal strength (Fischer, Bülthoff, Logothetis and Bartels, 2012). In the present study we asked two related questions. First, we asked whether human motion regions differentiate between outside objective motion being faster or slower than eye movements with different speeds (i.e. resulting in either positively or negatively signed retinal motion). Second, do these regions encode retinal and objective motion in absolute units or in units relative to the velocity of eye movements? To answer these questions, we created 2D random dot stimuli that moved either slower or faster than a fixation dot. All velocities were chosen so that we could examine neural responses to slower, matched and faster background motion relative to 0°/s, 2°/s and 3°/s eye movement speed. Moreover, we ran a functional localizer scan for each subject, allowing us to identify areas V5/MT, MST, V3A, V6, and CSv for region of interest (ROI) analyses. In our analysis, we tested each ROI’s response using a separate set of general linear models (GLM). The GLMs incorporated each of the above hypothesized response properties, and F-tests were used to identify which of the competing models accounted for significantly more variance. We found that all regions encoded both, retinal and objective motion in absolute, not in relative units, and that V3A, but not CSv or V5/MT does differentiate between negatively and positively signed retinal motion. These results suggest that motion is encoded in absolute units throughout the visual motion system, and that V3A has indeed a special role among human motion processing regions in that it represents motion signed with respect to eye movement direction.