Stochastic Biasing Theory (SBT): A Six-Layer Architecture of Conscious Agency

This paper introduces Stochastic Biasing Theory (SBT): A Six-Layer Architecture of Conscious Agency, formalizing consciousness as the real-time, intentional biasing of stochastic neural processes. The theory begins from the premise that physical dynamics are inherently stochastic, but constrained and biased by the laws and structures of the universe, producing non-uniform variability in which many macroscopic outcomes remain highly predictable.

Biological systems exploit this structured stochasticity through evolved mechanisms that regulate which properties are preserved and which are allowed to vary. Replication, mutation, and selection operate by controlling degrees of stochastic freedom, providing the fundamental engine of biological evolution. Over evolutionary time, the capacity to regulate stochastic processes becomes increasingly sophisticated, culminating in nervous systems in which intrinsically stochastic neural events, such as vesicle release, are biased in real time by internal and external constraints.

SBT proposes that this real-time control over stochastic neural dynamics constitutes the core mechanism of consciousness. Consciousness is not identified with behavior, representation, or subjective report, but with the emergence of active control over probabilistic state transitions within a system. On this basis, the theory traces an evolutionary pathway from basic physical constraint, through biological regulation and neural control, to higher-order forms of agency.

The paper further introduces a six-layer architectural framework that classifies forms of agency according to how stochastic processes are constrained, biased, and hierarchically regulated. This framework provides a unified account of conscious agency across biological systems and offers principled criteria for evaluating artificial systems, independent of task performance or intelligence benchmarks.