Stevan Harnad was invited to present a talk at the recent Online Consciousness Conference. His talk is titled Minds, Brains, and Turning (PDF format). In it, Stevan explains how and why he thinks Turning got it wrong when he set the modern agenda for cognitive science. He starts with a description of the famous Turing Test (which he calls T2). Along the way he offers comments on Searle's Chinese Room thought experiment and brings in the idea of "what it feels like" - an idea on which he places a great deal of importance:
"Yet, although it may be an illusion that some of the things I do, I do because I feel like it, it is certainly not an illusion that it feels like some of the things I do, I do because I feel like it. And that feeling is as real as the feeling that I have a toothache even when I don’t have a tooth."
Later he proposes the "Robotic Turing Test" (T3) as an improvement on Turing's original. Stevan believes an entity cannot be intelligent if it merely communicates intelligently; it must also have human-like "sensorimotor capacities". Finally he proposes the neurobehavioral Turing Test (T4) which tests the for human-likeness in communication, sensorimotor capacities, and "neurobehavioral performance capacity". The apparent downside to this is that T4 ends up as more of a test for identifying human intelligence than determining whether non-human entities are intelligent. It's easy to imagine a silicon-brained robot with alternate sensorimotor capabilities that's as intelligent as a human, yet would have no chance of passing T4. In fact, most well-known fictional robots and non-human life forms (e.g. HAL, Mr. Spock, etc) would fail T4 despite general agreement they seem intelligent. A variety of interesting responses and comments to the talk have been posted.