Rethinking Consciousness: The Self-Organising Information-Processing Systems Hypothesis

Abstract

Across biology, artificial intelligence and social systems, interacting components develop specialised functions and collective capabilities that were not individually specified in advance. Examples range from reorganised living cells to functionally differentiated neural networks and collective decisions in animal groups. What organisational principles might account for these outcomes—and for consciousness?We propose the Self-Organising Information-Processing Systems (SOIPS) Hypothesis. Within physical and environmental constraints, causally accessible, information-mediated interactions among changeable components can generate organisational structures that alter subsequent interactions and enable new functions. SOIPS distinguishes a generative engine from Dimension A, describing informational organisation from functional signalling to context-sensitive coordination, and Dimension B, comprising the elementary functional families of retention, inference and regulation. Particular configurations of these elements support composite capabilities, including perception, memory, learning, agency and cognition. Developed capabilities can reshape their own organisational conditions, combine with other functions and contribute to systems at new organisational scales.Within this framework, consciousness is proposed as a configuration in which awareness-related information participates in flexible cognition. Awareness refers to the organised availability of information about relevant internal or external conditions. Consciousness is thus neither an inevitable endpoint of increasing complexity nor a capability restricted by definition to human biology. If experience is constituted by particular organised neural activity, SOIPS may also help investigate the origins of phenomenal experience.SOIPS offers a shared language for interpreting biological development, learned organisation in AI and collective behaviour. It also motivates testable predictions: changing information access, constraints or elementary functional contributions should alter the capabilities they support. Cross-domain interventions could assess whether analogous organisational relationships produce comparable outcomes, informing both scientific explanation and the design of artificial agents and institutions.