Abstract Details

Romancing the Oxymoron: The 'Hardware Problem' of Machine Consciousness  Stephen Deiss (Institute for Neural Computation, appliedneuro.com, La Jolla, CA )   C22

THE fundamental question is how consciousness relates to mechanistic processes studied in the sciences. Many advocate one of two extremes: either consciousness is not mechanistic at all or it is an emergent property from mechanistic processes of a certain type or order of complexity (e.g., spiritual traditions vs IIT, ORCH-OR, GW). Intuitive philosophers often emphasize the 'aboutness' of consciousness as it refers to something beyond itself while others emphasize the self which is doing the referring, or a high order thought about the two together. A very basic intuition is that conscious experiences involve sensory and cognitive qualia, and without either only a zombie remains to sing the blues. The presumption that engineers cannot give such qualitative sensations to machines is a reason many refuse to entertain the possibility of machine consciousness. They argue that such systems simply execute rules or algorithms following natural laws devoid of sensation, and they are automata devoid of reflexive self experience in doing so. In contrast neuroscientists use the word 'mechanism' probably more than any other to describe the complex but approachable biochemical, genetic and cellular interactions that make a conscious mind possible, often avoiding the need to explain why a blob of fat should hear itself think or feel anything at all. Theories of consciousness grow ever more sophisticated and quantitative. But this fundamental gap in the explanations remains. I have maintained for over a decade that the crux of the problem is the assumption that there are laws operating on nature from a higher mathematical or divine realm. If one is able to abandon this view for a radically secular view of nature, the question becomes how natural systems do what they do from intrinsic principles and constraints rather than as externally directed. I will present this viewpoint using counterintuitive arguments precisely because what is giving us the hard problem is our intuitions about what a mechanism is accompanied by a lack of clarity about what consciousness is. Neuroscience is unraveling the so-called mechanisms that accompany living systems experience. Methods such as neuromorphic engineering and deep learning, and models such as free energy theory and Bayesian inference are inevitably leading to the ability to engineer machines that can do more than we can. I will argue that all manner of systems, from atoms to brains and beyond, are conscious with highly variable perceptual skills and a spectrum of self-reference.