Perceptual decisions rely on accumulating sensory evidence. This computation has been studied using phenomenological models, e.g. the drift diffusion model, or neurobiological network models exhibiting attractor dynamics. It remains unclear whether the dynamics of both models are qualitatively equivalent and whether attractor models can integrate evidence optimally. Here, I will present distinctive features of attractor models that allow them to perform flexible temporal weighting of stimulus evidence. In the discrete attractor model, this is due to transitions between decision states that can reverse initially-incorrect categorizations. Moving from categorical choices to continuous perceptual judgments, I will show that a continuous bump attractor network can integrate a circular feature, such as stimulus direction, nearly optimally. As required by optimal integration, the population activity of the network unfolds on a two-dimensional manifold, in which the position of the network’s activity bump tracks the stimulus average, and, simultaneously, the bump amplitude tracks stimulus uncertainty. Moreover, the model can flexibly switch between different temporal weighting profiles by changing a single control parameter, the global excitatory drive. Predictions of the models are validated with psychophysical data. Finally, I will outline how these attractor models can provide a comprehensive and experimentally testable computational framework to study the neural mechanisms underlying stimulus integration and bias effects in combined discrimination-estimation tasks.

If you would like to attend the seminar, please register here.