Melbourne School of Psychological Sciences - Theses
Permanent URI for this collection
Search Results
1 - 1 of 1
-
ItemPrediction and neural transmission delays in visual motion processing( 2022)Neural transmission takes time. Although this delay is seemingly insignificant (approximately 70 ms), it complicates the real-time localization of moving objects because the brain has no access to information about where the object is now. In this thesis, we study how prediction, and motion extrapolation in particular, can help overcome these neural transmission delays in visual motion processing. For motion extrapolation mechanisms to do this, they need to be able to drive sensory representations in the absence of sensory information, since sensory information arrives on a delayed time-line. In Study 1, we find that sensory representations corresponding to the anticipated next location on a motion trajectory are pre-activated before the anticipated stimulus onset and the arrival of sensory information. Correspondingly, in Study 2, we find that moving objects are subsequently represented with a shorter latency. To explain these findings, we outline a neural architecture of horizontal and feedback connections inducing pre-activations at anticipated stimulus locations that fits within the current dominant framework of cortical organization. The consequence of such an architecture however is that, when a moving object diverges from its predictable trajectory, the wrong stimulus location has been pre-activated. We show that, faced with a motion prediction violation, the brain indeed briefly represents the object in the anticipated but never presented stimulus location, resulting in a latency disadvantage for the representation of the mispredicted stimulus. In Study 3 we finally demonstrate that the overextrapolation is actively corrected for, preventing the conscious percept of the stimulus in the overextrapolated location.