What is the nature of self? Apparently, it’s just some sensors and actuators, with a dash of artificial intelligence.
A new paper published in the scientific journal Science Robotics argues that robots can be used to understand the development of the human sense of self, as well as related disorders.
The review, signed by researchers from the University of Sheffield in the UK, the University Hospital Cologne in Germany, and the Italian Institute of Technology, says that robots can stand in as models of self.
“An emerging view is that key phenomena of self can be generated in robots with suitably configured sensor and actuator systems and a layered cognitive architecture involving networks of predictive models,” the paper reads.
While thinkers like René Descartes considered self as simple, undivided, and distinct from the body, and David Hume understood it as a bundle of experiences, contemporary views see organisms, including humans, as complex dynamical systems.
This means recognizing that self is dependent on the body, the wider environmental context, and on interactions with others.
“In this review, we understand self as a complex system that binds across sub-systems to form an emergent and unified whole constituted by (but not limited to) core phenomena of ownership, agency, and transtemporal unity,” the authors say.
All can be replicated in a robot by assembling appropriate modules within its cognitive architecture, they suggest, leading to a sense of self as “having a unique perspective through its sensors, and as having agency through its actuators.”
“Moreover, such self, may also infer its own persistence from the predictability and consistency of that embodiment over time,” the paper argues.
Why does it matter?
As a result, robotics and AI methods such as generative modeling can provide a “useful testbed” to further explore the concept of the self and related behavioral phenomena, including disorders related to the aspects of self.
These may include disorders of body representation such as an “alien hand” syndrome, where patients might experience a limb as belonging to someone else, or a “phantom limb” symptom, where they still feel a missing limb.
Robotics could also be used in researching self-related aspects of dementia and depression, in addition to conditions such as depersonalization disorder and schizophrenia.
“Various schizophrenia subsyndromes, and their relation to self, could be explored through robotics, by constructing a cognitive architecture that matches aspects of the healthy case then altering the contributions of different sub-systems,” the researchers say.
The paper proposes three different ways in which robotics could be made useful for the scientific study of self. First, robotics allows both the systemic construction of self by integrating different components in computational models, and its deconstruction.
Second, robots can be used as experimental apparatus that study properties of self, and, finally, they could be used to understand the diversity of human selves.
“The human experience of self is broad and diverse. Although existing work has largely considered the self under conditions of mental health, robotic modeling offers the potential to explore the diversity of selves and so could contribute to a better understanding of interindividual differences and, potentially, to the diagnosis and/or treatment of disorders of the self,” the paper says.
Your email address will not be published. Required fields are markedmarked