[raycast://extensions/raycast/raycast-ai/ai-chat?context=%7B%22id%22:%226C52DD57-B505-4B8B-B378-B6E2A393CA1D%22%7D] - The basics: >Hypothesis-testing mechanism: The brain is not a passive recipient of sensory information but actively formulates and tests hypotheses about the external world. It constantly generates predictions about incoming sensory data based on internal models. > >Prediction error minimization: The [[core principle]] of the theory. The brain's primary function is to reduce the difference between its predictions and the actual sensory input. This error signal drives learning and adaptation, refining the brain's models and improving future predictions. > >Perception: In [[this framework]], perception is not a direct registration of the world but an inferential process. It's the brain's best guess about the causes of sensory input, constructed by minimising prediction error. # the chapter on attention [raycast://extensions/raycast/raycast-ai/ai-chat?context=%7B%22id%22:%22F04696D8-7B48-45F2-BC0B-2BE601A9A5BC%22%7D] # Extended conversation about psychedelics, attention, [[predictive processing]] https://chatgpt.com/c/67c8159b-0df0-800d-9b47-9128ea5916e8 - if chatgpt isaesthetics and [[transparency]] - [[psychedelic aesthetics]] - idea correct then this is very cool. > Previous texts (particularly the one describing “attention as the optimisation of expected precision” in [[predictive coding]]) show that the brain selectively amplifies signals deemed relevant or reliable. In the context of **geometric hallucinations**, one might say the visual system typically **down-weights** or “de-selects” the spontaneous patterns arising from lateral excitatory/inhibitory circuits. These patterns are overshadowed by higher-precision, real-world sensory input. Under psychedelics (or sensory deprivation), the normal gating is relaxed, effectively _boosting the internal signals_ (i.e., the “precision” assigned to the cortical pattern).