Efforts to teach Alo and Kyaaro, two younger African Greys, with audio and videotapes were unsuccessful. An audiotape (the audio portion of one of Alex’s M/R sessions) did not provide reference, function or social interaction. The birds could not see any relationship between the object and its label, and because there were no people present to provide interaction, they could not view a demonstration of the label function. The videotape (which showed one of Alex’s M/R sessions) also did not provide any social interaction. There was reference because the birds could see the relationship between the verbal label and the object it represented. But the video provided only limited contextual applicability. Alo and Kyaaro could watch Alex and the people receive the object, but they could not obtain it, even if they produced the correct label.

For the experiment each bird had a series of audio and then video training sessions. Both Kyaaro and Alo failed to produce the labels presented during the sessions. But in 34 of 40 trials Alo correctly identified paper and cork, which she has learned from M/R training. Kyaaro produced his M/R trained labels, paper and cork, 34 of 40 and 35 of 40 times, respectively.

But what would happen if the birds had limited social interaction? Specifically, what would happen if Pepperberg broke down social interaction into its components and eliminated joint attention? Joint attention is shared visual attention. It is demonstrated by joint gazing and directional pointing. The importance of joint attention in learning has been recorded in studies of children. Pepperberg’s study is the first report of its effects on birds. (Effect of Avian-Human Joint Attention on Allospecific Vocal Learning by Grey Parrots [Psittacus Erithacus], 1996, Journal of Comparative Psychology, 110:286-297.)

Alo was trained on green and yellow with the M/R procedure. Kyaaro’s M/R sessions focused on wood and wool. Both birds were trained on key and block with V-1. There were fifty V-1 sessions of approximately thirty minutes in length. Pepperberg intended to use the same number of sessions as in M/R training, but the birds usually learned their labels in fewer than fifty M/R sessions. It took about 14-15 months to complete the training.

Alex often participated in Kyaaro’s M/R sessions. Kyaaro had previously learned some labels from M/R training administered by people, and Pepperberg feels that Alex’s participation did not provide Kyaaro with any advantage.

Pepperberg begins the tests when the bird’s "vocalization [is] recognized by trainers in blind trials with better than 90% inter-observer agreement." Thus, testing labels begins when the bird attains clear pronunciation of the label. It does not depend on the bird’s accuracy in associating the label and the object. That’s the purpose of the test. Since the birds didn’t seem to learn from V-1 training, Pepperberg tested those labels after 50 sessions.

Kyaaro’s M/R test results showed that in 35 of 40 trials, he correctly identified wood and wool. After V-1 training, he was unable to identify key and block in 30 trials.

Although Alo did learn green and yellow from M/R training, she was never tested. This is the period during which she began refusing to work with all but two students and she was removed from the lab. During her V-1 training she never produced the labels key and block. But again, she was never subjected to formal testing.

Pepperberg concluded that the birds do not learn without joint attention. They must have full social interaction. But this did not end her analysis of the learning process. She then asked the question, "What if there is joint attention but limited contextual applicability, or demonstration of function?"

Pepperberg thus further concluded that while joint attention and full social interaction are necessary for African Grey Parrots to learn referential labeling, they are still only part of the equation. It appears that functionality is also required. She states, "The M/R technique not only establishes joint attention among the two trainers and psittacine subject, but also explicitly provides a model for the use of the targeted vocalization."

Pepperberg is planning more studies to analyze other components of the learning process. Even though this experiment and earlier ones have demonstrated the importance of the standard M/R technique, a method that is fully referential, fully contextually applicable, and fully interactive, more research will serve to enhance our understanding of why the M/R training method is so successful.