Unraveling the self: from bodily self-consciousness to artificial general intelligence

The subjective sense of having a self is commonly assumed to be static and anchored to the physical body. This presumption strongly preconditions how neuroscientists, philosophers, and computer scientists conceive of mind and attempt to engineer ‘intelligence’. The present thesis expounds on how the brain constructs reality through mechanisms of bodily self-consciousness (BSC) and the free-energy principle (FEP), and, notably, how this process can be manipulated and enhanced. This work shows how BSC can be manipulated endogenously-through self-regulation methods, e.g. meditation—and/or exogenously-through virtual reality and artificial intelligence technologies, with applications for healthy and clinical populations. I describe the human evolutionary emergence of selfhood and present the novel Causal Biomimesis hypothesis that tethers the formation of the sense of self with toolmaking, objective thought, technological development, and sociality. I argue that, particularly regarding artificial intelligence (AI) technologies, the externalisation of human cognition as technological artefact and its feedback effect on humans raises an ethical concern: that we are building AI, a class of artefact that is poised to strongly affect the evolution of humanity, based on a current image of ourselves and our perceived capabilities. From this ethical consideration, I make a case for the research and incorporation of non-ordinary consciousness (NOC), i.e. via self-regulation methods, into the design and engineering of AI. I present a neurophenomenological praxis (NP) based on the principles of the FEP that serves as a framework for instantiating (1st-/2nd-person) and researching (3rd- person) NOC and as a cognitive map for designing AI capable of supporting NOC. I discuss how a decentralised approach to AI could maximise the chances of a more ethical and representative AI by capturing and integrating a wider swathe of humanity’s values. Finally, I explore the farther frontiers of the implications of this thesis work: merged human-AI cognition in the form of what are herein termed “mindplexes”, where the phenomenology of NOC cultivation may elucidate the workings of mind in a way that helps humanity build a more capable and ethical human-AI future.