> In brief, our proposal is that psychedelics disrupt functioning at a level of the system (sensitivity of deep-layer pyramidal neurons, power of low-frequency rhythms, and integrity of large-scale networks) that encodes the precision of priors, beliefs, or assumptions.
As SSC point out, this idea isn't really new with this paper. SSC cite "Mad Science Blog" and the Qualia Research Institute as precursors, but perhaps a better example would be renegade computational neurobiologist Andrew Gallimore.
Gallimore has a 45 part course[2] up on Youtube which, after providing an overview of the neurobiology of the predictive coding[3] model of the brain for context, delves into the detail of psychedelics' action on pyramidal neurons giving a picture so consonant with the Relaxed beliefs paper's (less biologically detailed) proposal that it's somewhat surprising his name doesn't get a mention.
The main takeaway from the predictive coding model for this purpose is that most of the information flow in the cortex is top-down, from a world-model (analogous to an ANN model) implemented in the neural structure of the cortex, downwards towards lower-level neurons in "deeper" cortical layers. The world we experience flows from the model rather than directly bottom-up from sensory input, as we might naively expect.
Data flowing downwards from the higher-level model to deeper layers appears as a set of predictions to be compared to sensory data arriving at the same layers from the bottom upwards.
Divergences between sensory data and the model's predictions then serve as a corrective signal, fed upwards back to the model, in a constantly running training cycle.
When additional divergences are introduced by the neurobiological impact of psychedelics, Gallimore theorizes, the model reorganizes in an effort to restore predictive efficacy, experienced as the characteristic perceptual and cognitive distortions of the psychedelic experience. The longer term effects of the disruption seem analogous, perhaps, to pushing the an ANN model out of its local minima.
For an in-at-the-deep-end look at Gallimore's neurobiological overview of how psychedelic action on pyramidal cells disrupts the training cycle see[4] though it relies heavily on context established by the previous 35 episodes.
1: https://pharmrev.aspetjournals.org/content/71/3/316
2: https://www.youtube.com/playlist?list=PLbqdD4EM-aEfmLvbWu8GQ...
3: https://en.wikipedia.org/wiki/Predictive_coding
4: https://www.youtube.com/watch?v=L45A6XlPRM0&list=PLbqdD4EM-a...