A novel conception of consciousness as a complex-adaptive-system of Markov blankets

The study of consciousness has faced two major problems: lack of consensus regarding what consciousness is, and incomplete understanding of conscious and unconscious mechanisms responsible for the emergence of mental states. This thesis is an effort to circumvent these problems by proposing a novel schema of consciousness based on principles from complex-adaptive-systems, predictive processing and emergentism. The central claim is that consciousness can be defined by the interactions of conditionally independent component sub-systems, referred to as agents, which organise themselves hierarchically. The proposed schema of consciousness comprises three primary sub-systems: 1) a reservoir of states that manifest in unconscious form, including potential knowledge, referred to as the unconscious Markov blanket; 2) a conscious Markov blanket that functions to convert some unconscious information into conscious form, generating a conscious lower-base (the state of the world), and; 3) a (first person) mental component that functions to capture the conscious lower-base, convert it into phenomenological experiences and implement mental action to add novelty to the state of the world.

The schema shows that consciousness sub-systems are conditionally interdependent, separated by their statistical membranes. They communicate through a two-phased bi-directional cycle. During the top-down causality phase, the consciousness system’s higher-level information influences the sensory-state of the membrane of lower-level component parts, enslaving them. In the bottom-up emergent phase sub-systems process this information, adding novelty to it and, through the active-state of their membranes generate individuals’ conscious mental states. I conclude that individuals do not ‘have’ a consciousness but rather occupy a mental position within the consciousness system. Finally, I argue that the strength of the model lies in its well-researched principles; but it is weak in that is based, by and large, on ‘theory-to-fact’ abductive inferences that require validation.