Weighted bandits or: How bandits learn distorted values that are not expected