L’interazione tra umani e robot come gioco di finzione

An issue that is usually neglected in the literature on the attribution of mental states in human-robot interaction is the ontological commitment of users to these states. In this context, we argue the need to decouple the concept of attributing mental states from the act of uttering sentences with mentalistic content. We will perform an analysis of the different types of ontological commitment underlying the user’s utterance of a sentence with mentalistic content, in order to highlight the importance of estimating the user’s belief set for understanding the psychological dynamics of human-robot interaction. The different attitudes range from realism to eliminativism or the agnostic view, within which two positions can be distinguished: the reductionist position, which traces mentalistic statements back to beliefs about the physical state of the system, and the fictional position, according to which the user’s beliefs are true in the context of a make-believe play. In relation to the latter position, we consider a recent proposal by Clark and Fischer (2022), the so-called ‘depiction theory’, which was created to solve the social artefact puzzle, whereby, during human-robot interaction, users interact with the robot as if it were a real social agent, even though they are aware of its artefact nature. By reading the robot as a superimposition of several scenes, including the raw artefact and the depicted character, the theory not only explains the aforementioned paradox, but also presents itself as a fictional interpretation of human-robot interaction, since in this framework beliefs about the robot’s mind can be interpreted as beliefs about the depicted character’s mind. Nevertheless, other ontological attitudes can explain the social artefact puzzle, and future work is needed to determine user’s beliefs about robots’ beliefs from experimental data.