Unified principle of reinforcement in a neural-network model: Reply to N. T. Calvin and J. J. McDowell