The danger of the past was that men became slaves. The danger of the future is that men may become robots.
Erich Fromm
The first robots that were programmed to “think” the way biological brains do and meant to have free will were built in the late 1940s by William Grey Walter (1910–1977)—an American-born British neurophysiologist, and cybernetician, who wanted to prove that rich connections between a small number of brain cells could give rise to very complex behaviors, essentially that the secret of how the brain worked lay in how it was wired up. We don’t know if Walter knew the pioneering work of Thomas Ross and Stevenson Smith, but his first robots looked quite similar to the Robot Rat of Smith.
The first two robots of Walter, so-called Elmer and Elsie (short for ELectroMEchanical Robot, Light-Sensitive) were often labeled as tortoises because of how they were shaped and the slow rate at which they moved. They were capable of phototaxis which is the movement that occurs in response to light stimulus. Constructed between 1948 and 1949 using war surplus materials and old alarm clocks, they had a single light or touch sensor hooked up to two different paths that ran two different motors acting as two separate neuron brains. The robots had a plastic shell that was phototropic in that it could follow light and act as a bumper sensor. Interestingly, Walter stressed the importance of using purely analog electronics to simulate brain processes at a time when his famous contemporaries Alan Turing, John von Neumann, and Norbert Wiener were all turning towards a view of mental processes in terms of digital computation.
The tortoises were three-wheel machines (which Walter called Machina Speculatrix, because “it explores its environment actively, persistently, systematically as most animals do”), designed to show the interaction between both light-sensitive and touch-sensitive control mechanisms which were basically two nerve cells with visual and tactile inputs. These systems interacted with the motor drive in such a way that the tortoises were actually finding their way around obstacles. They were allowed to randomly wander around the floor in no specific pattern and when they were presented with two light sources equally distanced from their sensor, they’d head towards whichever light they saw as a consistent part of the scanning process.
In one experiment a light was placed on the nose of one of the tortoises. It appeared that the robot was looking at itself in a mirror. Its light began flickering and the robot started shaking as if excited to see itself in the mirror. Walter argued that if this behavior were seen in an animal it “might be accepted as evidence of some degree of self-awareness.”
When presented with certain stimuli, even outside of their programmed range of experience, the robots responded consistently, as if they had a personality. Naturally, they had their quirks and odd behaviors the way living things do. Using only two neurons they exhibited much of the same behaviors and oddities that any biological beings have.
After building the tortoises, Walter added additional capabilities naming his new robot, Machina Docilis. Based on the same physical design, this robot included a Conditioned Reflex Analogue (CORA) which demonstrated simple Pavlovian learning. The new robot had three sensors: a sound detector, a light detector, and a bump switch. The CORA circuitry could be trained to establish learned connections between the three sensors and the motor drive oscillators. This resulted in an ability to learn different behaviors which were initiated by sounds, light, or the bumpers. This kind of circuitry is structurally very similar to neural circuitry most recently identified in marine snails.