An Optimal Policy Model for Concurrent Uncertainty Estimation During Decision Making