Explore the intersection of cognitive neuroscience and machine learning in this 57-minute conference talk from the Models of Consciousness Conferences. Delve into the formal approaches to understanding perception and decision-making, comparing predictive coding and Bayesian brain hypothesis with reinforcement learning and the Bellman optimality principle. Examine the apparent dichotomy between the brain's optimization of beliefs about sensory causes and the governance of choices by value functions and reward. Investigate whether these formulations can be reconciled through an underlying information theoretic imperative. Consider the relationship between models of consciousness and models of decision-making. Follow the talk's structure, covering topics such as self-evidencing, statistical perspectives, free energy functionals, generative models, and engage with questions and discussions, including a senior question session.