Prerequisites for an Artificial Self

Abstract

Traditionally investigated in philosophy, body ownership and agency-two main components of the minimal self-have recently gained attention from other disciplines, such as brain, cognitive and behavioral sciences, and even robotics and artificial intelligence. In robotics, intuitive human interaction in natural and dynamic environments becomes more and more important, and requires skills such as self-other distinction and an understanding of agency effects. In a previous review article, we investigated studies on mechanisms for the development of motor and cognitive skills in robots (Schillaci et al., 2016). In this review article, we argue that these mechanisms also build the foundation for an understanding of an artificial self. In particular, we look at developmental processes of the minimal self in biological systems, transfer principles of those to the development of an artificial self, and suggest metrics for agency and body ownership in an artificial self.

The Embodied Crossmodal Self Forms Language and Interaction: A Computational Cognitive Review

“…This includes physical interaction and multiple sensory modalities to develop systems that understand language more like humans (Anderson, 1972;Wermter et al, 2009;McClelland et al, 2020). Additional prerequisites for modeling a communicative self requires curiosity, body representations, and predictive processes (Hafner et al, 2020;Eppe and Oudeyer, 2021). In reinforcement learning, there is a body of research (Pathak et al, 2017;Dean et al, 2020;Nguyen et al, 2020;Röder et al, 2020), containing these components.…”

“…This demonstrates that language is an important cognitive tool to convey meaning (Mirolli and Parisi, 2011;Colas et al, 2020;Eppe and Oudeyer, 2021). However, the self described in recent literature (Schillaci et al, 2016;Hafner et al, 2020;Nguyen et al, 2021) is also important for embodied dialog. The self builds upon the actor's capabilities to sense its own body and the environment.…”

“…To address these questions, and as our main contribution, we look into recent articles defining the prerequisites of an artificial self (Schillaci et al, 2016;Georgie et al, 2019;Hafner et al, 2020;Nguyen et al, 2021) and relate these prerequisites with verbal and non-verbal dialog methods for computational agents and reinforcement learning. In section 2, we survey the developmental processes of humans to ground language in embodied sensorimotor representations of the self and its surrounding world.…”

“…The self builds upon the actor's capabilities to sense its own body and the environment. It is, therefore, characterized by the response to actions and predictions of the internal model (Schillaci et al, 2016;Hafner et al, 2020). Grounded language in the context of the self refers to the context of these senses.…”

“…Therefore, they fail to learn robust and general behaviors. We suppose that this gap is due to a lack of understanding of "the self " (Hafner et al, 2020), and how it is defined in the context of "the other." Specifically, we suggest that a self-other projection model is critical for empathy and a theory of mind to map an observed other agent, along with its semantic properties and relations, to the self and its semantic properties and relations.…”