Save Big on Coursera Plus. 7,000+ courses at $160 off. Limited Time Only!
Explore a 25-minute conference talk from the Models of Consciousness Conferences that delves into the hypothesis that consciousness emerges from online compositional learning systems. Examine how humans excel at data-efficient compositional learning and how this process involves sparsely-structured perception and action encoding. Discover the connection between approximate Bayesian inference, predictive coding, and the learning of new sparse causal structures. Investigate the proposed functional model that combines compositional learning with predictive processing, and how it relates to conscious experiences. Learn about the formation of associatively-recallable declarative memories and their role in connecting perceptions and actions across different domains. Consider the extension of this model to include reinforcement learning of action policies for survival and reproduction, and how it accounts for the concept of valence. Analyze how this neurobiologically-plausible learning system architecture may explain major subjective and objective observations of consciousness, and understand its potential for mathematical definition and testing.