Visual Motion Processing Deficits in Alzheimers Disease Patients can be Modeled by Delayed Feedback
16.425, Friday, May 10, 5:30 - 8:00 pm, Orchid Ballroom
N. Andrew Browning1; 1Center for Computational Neuroscience and Neural Technology, Boston University, MA
Alzheimers disease patients (AD) have reduced ability to perceive behaviorally-relevant visual information. Verlarde et al. (Journal Alzheimers Disease, 2012) found that AD patients required a higher signal to noise ratio for perception of heading compared to controls. It is likely that this contributes to some of the navigational deficits observed in AD patients. If you cannot tell where you are going, it is hard to get to your destination. AD patients suffer from late-stage myelin breakdown as a precursor to plaque build-up, resulting in slower conduction of signals within the brain. This work explains heading perception deficits in terms of the conduction of signals within and around heading selective brain area MST. We model MST with a variant of the ViSTARS model (Browning et al. Cognitive Psychology, 2009) of primate heading perception: motion vectors in model area V1 feed into model area MT which pools over space and time to resolve the aperture problem, MT then feeds in to model MST where heading is perceived. Lateral feedback within MST forms part of the neural circuit for discriminating heading, and feedback from MST to MT helps MT resolve noise in the optic flow estimates. Perceived heading direction is represented as a peak in the population activity in MST, the sharper the peak the more confidence the circuit has in the direction of heading. Our experiments show that feed-forward behavior of the network is slowed by reduced conduction speeds but does not result in differences in noise tolerance. In the absence of noise, feedback is largely irrelevant. However, as noise in the global motion estimate increases, feedback becomes increasingly important for high confidence discrimination of heading. In high noise environments with slow feedback, the neural circuit becomes confused, indicating that delayed feedback signals may be a significant contributor to AD visual-motion impairments.