On the Reward-Loss Duality in Time Consistent Decision Making