Information relevant for making a decision may arise from multiple sources and it may be extended across time. In this thesis I investigate the mechanisms by which Rhesus macaques combine information to make decisions about the direction of visual motion in a random-dot kinetogram. First, I analyze existing neural data from the middle temporal area (area MT) and the lateral intraparietal area (area LIP) to show that temporal integration of motion-selective neural signals from MT accounts for many of the observations obtained from LIP neurons during this task. These findings suggest that LIP represents a decision variable, the accumulated evidence for one direction, using mechanisms that resemble diffusion of a particle toward a barrier. Next, I describe experiments in which I manipulate the prior probability of motion direction in the discrimination task. I show that increasing the prior probability of a given direction results in more choices toward the more likely direction and faster choices in that direction; the opposite effects are seen when the prior probability is reduced. Quantitative analysis of these behavioral data using a diffusion-based model shows that prior probability affects the decision variable in the same manner as a change in stimulus intensity. Finally, I record the activity of neurons in area LIP while monkeys discriminate the direction of motion under different prior probabilities. I show that changing the prior probability of which direction is correct alters the responses of LIP neurons such that their activity reflects a combination of both motion information and prior probability. Specifically, changing the prior probability alters the LIP response in the same way as changing the intensity of stimulus motion, consistent with the findings from the behavioral data. These experiments show that various sources of decision-related information are combined using a common mechanism and that neurons in area LIP appear to reflect this decision mechanism.
展开▼
