Purchase this article with an account.
Guillaume Masson, Jérome Fleuriet, Anna Montagnini, Pascal Mamassian; Predicting and computing 2D target motion for smooth-pursuit eye movements in macaque monkeys. Journal of Vision 2008;8(6):384. doi: 10.1167/8.6.384.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Smooth pursuit eye movements in primates unveil the temporal dynamics of 2D motion integration. Tracking of single, tilted bars is always initiated in a direction close to the velocity orthogonal to the bar orientation. This initial bias is gradually reduced: 300ms after visual motion onset both target and eye movement directions are perfectly aligned. This time course is believed to reflect the temporal dynamics of the neural solution for the aperture problem in macaque area MT (Pack et al., 2001). We investigated how high-level cues can influence this temporal dynamics. Eye movements were recorded in macaque monkeys using the scleral search coil technique. Long (20°) bars were drifted along the horizontal direction (speed 10°/s) with different bar orientations (−45°, 0 and +45° relative to the vertical axis). We found no effect of a static presentation (500ms) prior setting target into motion with unpredictable directions. In a second set on experiments, we blocked both bar orientation and direction conditions so that 2D motion was fully predictable. We found strong anticipatory responses along the 2D motion trajectory. However, 100ms after target visual motion onset, strong pursuit biases were again observed suggesting that predicting target motion had no influence upon solving the aperture problem. In a final set of experiment, we briefly (200ms) blanked target motion during steady-state tracking (i.e. when pursuit and target directions were aligned). Blanking the image drastically reduced eye velocity and in many cases pursuit was stopped. At target reappearance, monkeys reinitiated pursuit responses along the orthogonal direction to bar orientation, albeit with a weaker direction bias. These results suggest that high-level prediction about incoming target motion cannot be used to solve the aperture problem. This demonstrates that low-level motion computation remains impenetrable to cognitive factors to preserve the ability to quickly react to new visual motion input.
This PDF is available to Subscribers Only