Consciousness is entailed by compositional learning of new causal structures in deep predictive processing systems
arXiv Preprint Archive January 17, 2023 V. A. Aksyuk
A proposed extension to predictive processing theory adds online, single-example structure learning through hierarchical binding of unpredicted inferences, enabling fast compositional learning and explaining access consciousness. The system learns new causes by connecting novel combinations of perceptions, forming working memories that become short- and long-term declarative memories retrievable by associative recall. Bound representations are unified yet differentiated, maintained by selective attention and globally available. The model explains contrast and masking manipulations, postdictive perceptual integration, and other paradigm cases. 'Phenomenal conscious experience' arises as the learning system transparently models its own functioning, producing perceptual illusions underlying the meta-problem of consciousness. The proposal unifies feature binding, recurrent processing, predictive processing, and global workspace theories.