how do you teach them what the name of an object is?
We use a modeling technique. It includes three-way interactions between two human trainers and the parrot subject, such that neither human is a principal trainer, and the roles of the human questioner and the human respondant can be (and often are) reversed: this avoids training parrots to respond only to one trainer, and allows them to witness that all involved can both ask and answer questions, and no participant is limited to one role. So, one humans asks the other “What toy?”, the second responds “Key” and gets the key and can play with it; the bird watches the interaction and realizes that the novel sound causes transfer of the object it wants. The technique makes use of the objects that are being used for training as the reward itself, rather than an external, irrelevant food reward: Again, after labeling an object or its attribute, the subject (initially, the human model) receives the object itself to play with, chew, or examine, in practice assigning a referential and contextual use to the label being trained. It also allows for the parrot to witness “corrective feedback”, as trainers occasionally respond incorrectly or unintelligibly on purpose and so demonstrate that not any novel noise is functional….the bird sees the consequences of incorrect answers (mild scolding and no reward), and sees how the model receives subsequent encouragement to try again (for more details see The Alex Studies).
Dr. Irene Pepperberg