Text this: An MRP formulation for supervised learning: generalized temporal difference learning models