William Grey Walter (1910-1977) was a neurophysiologist who worked at the Burden Neurological Institute in Bristol and made important contributions to the study of brain waves before turning to building machines that imitated life. He is remembered in the history of robotics for a pair of small wheeled automata he constructed in 1948 and 1949, nicknamed Elmer and Elsie, which he described to a general audience in his article “An Imitation of Life” in Scientific American in May 1950. That article, indexed here through the NASA Astrophysics Data System, is the primary source.
The machines, which Walter called Machina speculatrix because they seemed to explore, each contained only two sensory inputs, a light sensor and a touch sensor, two vacuum-tube nerve cells, two motors, and a few relays. From this tiny nervous system emerged surprisingly complex behavior. The tortoises would wander, steer toward moderate light while avoiding bright light, back away from obstacles, and return to their hutch to recharge when their batteries ran low. Walter even noted apparent self-recognition behavior when a tortoise carried a light and faced a mirror.
Walter’s tortoises matter to the prehistory of artificial intelligence because they showed that rich, adaptive, goal-seeking behavior could arise from very few components wired with feedback, without any internal model or stored program. They are widely regarded as among the first autonomous mobile robots and as an early demonstration of what later roboticists would call behavior-based control. Walter was, like Ashby and Turing, a member of the Ratio Club.
For a general reader, Elmer and Elsie are a vivid early lesson that lifelike, intelligent-looking behavior does not require a complicated brain; sometimes a couple of sensors and the right feedback are enough.