The rewards pass through the hierarchy from manager to employee and create an effect of backward propagation similar to the learning methods of a neural network. But unlike the simple switches and numeric functions of a neural network, the MDL agents can be arbitrarily complex programs or ...