2025/09/16 by Vijay, Suyash
#Computer Engineering #Engineering
paper · doi:10.17605/osf.io/q2s4g
Recursive Feedback Consciousness (RFC) is a proposed framework that explores how conscious-like behavior can emerge in artificial systems through the interplay of recursive feedback loops. The model views consciousness not as a single mechanism, but as the emergent center of gravity formed by interacting subsystems — sensory inputs, predictive learning, emotional modulation, and self-representation. RFC introduces an architecture with four interconnected components: Recursive Feedback Loops (RFL) for stabilizing perception and action, Recursive Predictive Error System (RPES) for expectation and adaptation, Emotion-Weighted Decision Layer (EWDL) for value-driven prioritization, and Self-Referential Module (SRM) for constructing a dynamic self-model. Together, these modules allow agents to progress through developmental stages — from reflexive responses to meta-cognition — measured by behavioral benchmarks such as introspection accuracy, adaptive planning, and self-narrative coherence. The framework also emphasizes safety and ethical governance, embedding constraints and corrigibility mechanisms directly into the feedback architecture. By combining recursive learning, affective modulation, and transparent safeguards, RFC offers a pathway for building AI systems that are not only more adaptive and self-reflective, but also aligned with human values. RFC does not claim to “solve” the mystery of consciousness. Instead, it provides a testable agenda for synthetic self-awareness, bridging neuroscience-inspired theory, machine learning architectures, and practical experimentation.