The body’s muscles for walking, speaking, lifting, pushing, etc., are activated by neural control signals that originate in the motor cortex of the brain. In the motor cortex, large arrays of neural signals are generated to enable the desired muscle activity. Because of their complexity, it is unlikely that these control signals are generated in real-time. More likely, these control signals are drawn from memory after training for the respective tasks. That being the case, how is the cognitive memory trained, and once trained, how are the prompt signals generated to retrieve the control signals from memory? Training is done with reinforcement learning. Training is a slow, unsupervised process. When the muscle motion is satisfactory, the memory’s neural networks are rewarded. When not satisfactory, the neural networks are punished. Prompting signals come from the physical environment. When walking, for example, the prompt signals for the next step come from sensing the current mechanical state, the angles of joints and tensions in the muscles, etc. Prompting yields the control signals that enable going from the present state to the next state, and the process continues. When listening to one’s auditory signals, the memory is prompted to provide control signals to the vocal tract to enable the next “step” of speech. Motor control signals are not generated in real-time but are drawn from memory.