Text this: Delayed reward information is underweighted in reinforcement learning with dispersed feedback.