```dataview TABLE file.link AS "Note", file.cday AS "Created On" WHERE file.cday = date(this.file.name) SORT file.cday ASC ``` ## 1.1 Perception as Prediction Error Minimisation On the [[predictive processing]] framework, perception is not a passive, bottom-up process of building representations from incoming sensory data. Instead, perception involves the brain actively generating predictions about what it expects to encounter, then comparing these predictions with actual sensory input. Any mismatch between prediction and input generates "prediction error," which the brain tries to minimize. This prediction error minimization operates through a hierarchical neural architecture. At higher levels, cortical regions generate abstract, categorical predictions about environmental causes ("tree"), while lower levels produce more specific predictions about expected sensory features (shapes, textures, colors). High-level predictions cascade downward through the cortical hierarchy, while prediction errors—signals representing mismatches between predictions and actual input—propagate upward, modifying the generative model. Higher cortical areas encode invariant patterns that remain stable across contexts, while lower areas process rapidly changing sensory particulars. When [[perceiving objects]], higher cortical regions predict categorical features based on prior experience, while lower regions predict specific sensory details that should result from those categories. When incoming sensory data conflicts with predictions, error signals propagate upward, forcing model revision—identifying a different object category or updating the model of that category. Through iterative cycles of prediction and error correction, the brain constructs increasingly refined models of causal structures in the environment. The predictive system must maintain a balance between model stability and perceptual updating. In some circumstances, the system should maintain predictions despite conflicting sensory data (as with partially occluded objects), while in others, it should rapidly revise models based on unexpected input. This balance is achieved through precision-weighting—a mechanism that adjusts the relative influence of different prediction errors on model updating based on their estimated reliability. ## 1. 2 Introspection as Meta-Inferential Prediction Error Minimisation The [[predictive processing]] framework can be extended to model introspection. While perception involves inferring external hidden causes of sensory input, introspection involves inferring the hidden causes of our own experiences themselves. This reframes introspection not as a special faculty of inner perception, but as an application of the same inferential mechanisms to a different domain. For introspection to operate as prediction error minimisation, our experiences themselves must sometimes violate our expectations about them. Hohwy provides examples of such violations – when an espresso tastes unexpectedly unlike coffee, the drinker introspects the taste sensation itself, treating [[the experience]] (rather than just the coffee) as requiring explanation. These cases represent instances where normal perceptual inference about the world fails to resolve prediction error. The system then treats [[the experience]] itself as the source requiring investigation. In Hohwy's metaphor, sometimes we must examine the mirror itself rather than what it reflects. While introspection often occurs in response to unexpected experiences, we can also voluntarily direct our attention inward. This voluntary introspection uses the same precision-weighting mechanisms but is initiated differently. Rather than responding to prediction errors about our experiences, we deliberately increase the gain on prediction errors related to particular aspects of our experience. This allows us to attend to our internal states out of curiosity, enjoyment, or for purposes of meditation, [[not just]] when they violate our expectations. Importantly, we can engage in different types of introspection through these precision-weighting mechanisms: we might focus on unexpected experiential states that demand explanation, attend to our sensory experiences simply to appreciate them more fully, or deliberately observe our mental processes during meditation. This flexibility in directing introspective attention will be especially relevant for understanding the [[aesthetic dimension]] of [[psychedelic experience]]. ** Counterargument against the picture analogy:  MAYBE ADD THIS AT THE END OF THE INTRODUCTION, and restructure how the ideas are presented quite radically 1. [[aesthetic appreciation]] of perception generating artifacts requires [[the appreciation]] of DSPs 2. Psychedelics do not have [[design scene properties]] because there is no way of perceiving their design. "There are no brushstrokes"; the first place to look is in the vehicle (the tab of acid). There are perceptual properties there, but they are not the properties from which the scene emerges perceptually. 3. Therefore, there is no relation between the perceivable properties of the vehicle and the [[perceptual experience]]/scene elicited by the vehicle. 4. Therefore, [[aesthetic appreciation]] of psychedelics cannot be modelled on [[aesthetic appreciation]] of pictures.  OR stronger, [[aesthetic appreciation]] of psychedelics is not possible. ** # prescriptions [Preview attachment Promemoria 0300A4369592774_120220251113597280008205383066.pdf    Promemoria 0300A4369592774_120220251113597280008205383066.pdf 8.2 KB ](https://mail.google.com/mail/u/0?ui=2&ik=33039c19a8&attid=0.1&permmsgid=msg-f:1823846162086432284&th=194f9a6bd81f0e1c&view=att&zw&disp=inline)