My proposal was simple: I said I wanted to replicate the linguistic and cognitive skills that had been previously achieved with chimps in a Grey parrot, an animal with a brain the size of a shelled walnut, but one that could talk. My confidence that I could do it was based on two things. The first was my experience growing up with talking birds, and the sense that they are indeed smart. Second were the facts that Greys, like apes, live a long time, and that their social groups are large and complex. Both these factors were thought to account for at least some of the brainpower that apes so obviously possess. Why not a similar kind of brainpower for Greys?
My plans for training Alex differed from the accepted standards of the time. Under the prevailing psychological dogma known as behaviorism, animals were seen as automatons, with little or no capacity for cognition, or thought. Biology was little better, dominated by theories claiming that much of animal behavior was innately programmed. Experimental conditions for working with animals were very tightly prescribed. Animal subjects were actually supposed to be starved to 80 percent of their body weight so they would be eager for the food given for a "correct" response. They were also to be placed in a box so that the appropriate "stimuli" could be very tightly controlled and their responses precisely monitored. The technique was known as "operant conditioning." This was, to me, completely crazy, not to put too fine a point on it. It was contrary to all my gut instincts and commonsense understanding of nature.
For a start, isn't it blindingly obvious that communication is a social process, and that learning to communicate is a social process, too? It seemed clear to me that putting an animal in a box and expecting it to learn to communicate could not succeed. Several researchers had attempted to do this with mimetic birds and had failed spectacularly. They blamed the failure on a supposed deficiency in the brains of the birds, whereas I felt strongly that it was due to a deficiency in the researchers' assumptions and approach.
In fact, the first people working on human-animal communication in chimps in the late sixties and early seventies had not followed the behaviorism model. For the most part they had adopted much more naturalistic training techniques. Nevertheless, I still felt something was missing. And I couldn't quite treat a parrot the way the researchers were able to treat a baby chimp like a baby human, living with it 24/7 and still maintaining some objectivity. While mulling over this conundrum in 1975, I came across a paper by the German ethologist Dietmar Todt, published in what to me was an obscure German journal. In it he described his so-called model/rival program of training, which I adapted for working with Alex.
As I said earlier, under this system, instead of having one trainer, an animal subject had two. The principal trainer, A, would ask the secondary trainer, B, to name an object A showed to her. If B answered correctly, A would reward her; an incorrect answer would result in scolding. Trainer B is the "model" for the animal subject and its "rival" for the attention of trainer A. From time to time, trainer A would ask the animal subject to name the object, and it would be rewarded or scolded accordingly. Todt reported that Greys had learned speech very rapidly under this approach.