Abstract
We used the motion aftereffect (MAE) to psychophysically characterize tuning of motion perception in the human visual system. Hiris and Blake (1992) measured the strength of the MAE for random dot kinematogram (RDK) adapter stimuli containing either one direction of motion or a range of directions and found that the MAE was stronger when the adapter stimulus included a moderate range of directions. Thus, the function relating MAE strength and the range of directions in the adapter stimulus provides information regarding the bandwidth of direction tuning of motion perception. We compared the directional anisotropy in MAE bandwidth to the well-known oblique effect in motion direction discrimination. In agreement with previous research, we found that subjects had lower motion direction discrimination thresholds for cardinal compared to oblique directions (Gros et al., 1998). For each subject, we then measured MAE bandwidth for a cardinal and for an oblique direction. The MAE bandwidth was consistently smaller for the cardinal direction, suggesting a fundamental similarity between motion direction discrimination and tuning of the MAE. We adapted a computational model of V1-to-MT connectivity (Rust et al., 2006), introducing anisotropies in the connections between V1 and MT that result in larger bandwidth of tuning in MT cells tuned to oblique compared to cells tuned to cardinal directions. Model simulations predict an oblique effect for both direction discrimination and MAE tuning, consistent with our experimental results. The model is also in accord with a recent report that the magnitude of the oblique effect in direction discrimination is inversely proportional to the bandwidth of the stimulus (Dakin et al. 2005). Finally, our model also predicts anisotropies in the tuning of large populations of cells in areas MT and V1, and we are currently testing this prediction using fMRI.
S.S. was supported by an NEI T35 grant, through the School of Optometry, University of California, Berkeley.