Conference on Cognitive Computational Neuroscience

Decisions are made with different degrees of uncertainty, i.e., how confident one is that a decision is correct. Theoretical work suggests that attractor dynamics in networks can account for decision uncertainty. However, attractor dynamics and their relationship to decision uncertainty have not been examined in the biological brain. In this work, we provide evidence that the energy landscapes around attractor basins in the population neural activity space shape the uncertainty in our decisions. We trained two macaques to make accept/reject decisions based on pretrained visual cues that signaled reward offers with different magnitudes and delay-to-reward time. Monkeys showed low decision uncertainty for very good (high reward, short delay) and very bad (low reward, long delay) offers, and high decision uncertainty for intermediate offers of reward and delay. We recorded from prefrontal cortex using large scale array recordings while monkeys made decisions. Our results showed that the attractor basins have shallower energy landscapes following presentation of intermediate offers that are associated with high uncertainty in decisions. Therefore, we provide neural evidence that manifold landscapes shape decision certainty.