AIBO, the robotdog

The first story is about a robot dog that was made In Sony's Research Lab in Paris in 2001. Because the robot dog does not mediate the presence of another human being, the experiments done in Sony CSL Paris highlight processes of attribution, synchronization and adaptation of a receiving person in a very clear way. The name of the dog is AIBO and he has been for sale for a few years. The dog moves his head, walks, turns, sits, gives a wink and more in response to certain triggers it perceives with its sensors. Frederic Kaplan, one of the researchers who was part of the design team of AIBO, spent a lot of time with the dog.^{note 54} Even though the dog had a fixed set of behavioural actions in response to certain predefined triggers that could be received by its sensors, and even though Frederic had been part of the dog's design team - so he knew it only consisted of hard and software and could only perform pre-programmed behaviour - he developed feelings for the dog. In a personal e-mail exchange Kaplan writes "When doing experiments with AIBO, I never had a doubt about the fact that it was a machine (I participated in the design). This does not prevent one having feelings towards it. This has been well documented with children and their imaginary friends. Many researchers argue that children know that their friend does not exist and yet they can really be sad if the friend for one reason or another is not there. But is that so surprising? When you go to see a movie, you can really laugh, or be really sad or even cry and yet you have not doubt that what you are seeing on the screen does not exist. The reality of feelings and ontogenic status are two independent things." (personal communication, October 2006). Kaplan argues that because he was with the dog in the rich social context of the lab, and not in isolation with the dog, that this process of attribution only became stronger. "My own research with AIBO does not suggest that isolation is an augmenting factor for attribution. AIBO, like natural pets, was used in a rich social context most of the time, as a new member of the family around which anyone could react and comment. (É) Research with natural pets (most dogs are owned by large families and not by single old ladies) does not suggest that. On the contrary, it could be argued that attribution and identity creation occur in a more powerful manner in rich social contexts. Robots and pets are part of a complex make-believe game shared by several persons." (personal communication, October 2006). When I translate Kaplan's findings to my research question, he argues that because people 'share the make-believe game', because people act together and perceive and witness each other's processes of attribution, the attribution becomes stronger. Witnessed presence, which I will discuss later, influences processes of attribution.

Luc Steels, the director of Sony CSL Paris and professor at the Free University of Brussels who conducted this research, explains Frederic's feelings by the principle of attribution, synchronization and adaptation. In processes of attribution an exchange of triggers can create a complete reality because the human brain fills in the rest. Frederic's brain anticipated the behaviour of the dog and made it consistent. The robot dog's behaviour is a trigger for Frederic's brain to attribute missing elements to the robot's behaviour, this then influences Frederic's perception of the dog and as Kaplan argues himself, this process only becomes stronger because he shares the make-believe game with other colleagues in his case. After a while the machine and Frederic's brain synchronize and one could conclude that they did indeed enter into a relationship. Once certain structures in an interaction are set, both Frederic and the dog adapt to these structures, which defines how new moments in their interaction are perceived and so forth.

In communication processes people 'attribute' elements to the other person or object, which were technically speaking not there, but which become real because while exchanging, communicating, triggers adapt to each other over time. Further research presented at Sony CSL in October 2004, points in the direction of the crucial role that the response to anticipated feedback plays in processes of attribution, synchronization and adaptation. Even in processes of trial and error when anticipated feedback actually occurs, the basic configuration of communication changes. It adapts. According to Steels, two systems that are in motion and that confront each other will adapt to each other and develop a mutual language. Processes of attribution are complex neurological and psychological processes in which the brain structures and psychological structures of a person add to a certain reality in such a way that the distinction between what is attributed and what is not, disappears. It all feels real; it all is perceived as real; it actually is real and contributes through processes of adaptation to the evolving reality in the moments to come.