This is the synthesis and synergising of various talks about free energy principle, predictive processing and its connection with meaning. By understanding predictive processing and free-energy (as a lens) you can understand other aspects such as addiction or meaning-making (meaning in life).
- Predictive processing is an essential element of the brain to function in the world
- In order to predict something, you need to have a model of the world
- Stability is the meta principle
- We want to stay stable over long period of time (survival, homeostasis)
- Uncertainty is a state we want to avoid for stability
- Less uncertainty is better for survival
- So we need an internal model of the universe to anticipate what is going to happen based on the inputs we recieve
- Brain doesn’t have access to inputs (only infers)
- Friston’s Free Energy Principle is about this
- Free Energy, as a term, is borrowed from physics (entropy = uncertainty within a system)
- Free Energy here is used to describe the entropy in an information construct (not actual physical entropy)
- We are trying to compute how surprising a sensory input is (that’s important for survival!)
- Surprise cannot be calculated, it is intractable
- The Free energy is an ‘upper bound’ to surprise which is something that is tractable and computable and can be used as a proxy to determine if something is surprising
- It’s about the discrepancy between prediction and actual
- A model of the world is not much use if there is a huge discrepancy between actual and prediction
- We want to keep reducing this buffer (the buffer being free energy) so that the potential for uncertain situations is avoided and also work towards avoiding situations where we ‘uncertain’ (cold, wet, hungry etc)
- Analogy: we cannot know the weight of a box on the scales (we don’t have access to the number). It could be 10kg or 100kg. We’re trying to reduce potential space down to 10kg. Between 10-100 is uncertainty / ‘free energy’ If the box is 100kg, it can therefore be determined to be surprising.
- A surprise will cause action
- Two ways to react: change model or act on the world
- Changing the model: the model needs to be updated to the new reality, but you don’t want to ‘over-fit’ (not every surprise warrants an update, especially when taking into account priors)
- Act on the world: change the situation surroundings to make it less potentially surprising (walk out of a bar full of drunken angry people / work towards storing food for winter, etc)
- There are also hierarchies within our generative models. A surprise doesn’t need to work all the way up the hierarchy: an optical illusion shouldn’t cause someone to never trust their eyes ever again. (“Markov blankets”)
- Why not just sit in a dark room and avoid everything? We’re built for dynamic systems; a homogenous state is not ‘expected’ – would actually be surprising!
- Over time, a lot of mental models and lenses are crystallised, and often get stuck – unable to update their own models
- Psychadellics weaken the precision of high level priors (lower down layers don’t impose as much on the higher levels, so can have bigger belief update) and allow people to get unstuck
- Working towards lowering uncertainty feels good (surprise reduction)
- The better your prediction, the better you can accurately adapt to the situation (updating models, acting, responding)
- Our models work on relevance realisation (vervaeke) and there are infinite sensory inputs, which ones are true / relevant? There is a need for unifying theories (rules of thumbs, heuristics) to integrate across time
- Patterns are a generative model that we use towards our goals
- Happiness is doing better than expected towards our goals / in something we care about (meaning)
- Brain is not just tracking rewards, but tracking rate of reward relative to expectation (it’s a gradient) reducing uncertainty feels good
- Reverse is also true, anxiety is caused when doing worse than expected
- Our highly adaptive models can become maladaptive: bad habits, addictions, pathologies when models get stuck in destructive ways
- Coping mechanisms can trick brain into thinking that we feel that we are reducing uncertainty (a short term sensation of predictability) but over the long term increasing surprise, harm, uncertainty
- Culture across time have developed practices (ecology of practices) for intervening with foolishness, self deceptive and self-destructive behaviour – in order to enhance the connectedness.
- Result of a set of practices = wisdom (fittedness in the environment)
- So surprise minimisation and working towards what we care about is: dealing with a constantly a complex dynamical situation which requires a dynamic self-organising system of practices – an ecology of practices
- Ecology practices: mastery, meaning
Psychedelics and the Free Energy Principle: From REBUS to Indra’s Net
Perception and Order: Jordan Peterson and Karl Friston
https://podcasts.apple.com/gb/podcast/the-jordan-b-peterson-podcast/id1184022695?i=1000593321435
Meaning Crisis – and what we can do about it
https://podcasts.apple.com/gb/podcast/the-contemplative-science-podcast/id1612279474?i=1000552593031