Active inference

[@stark1ll]

So the idea of active inference is that the brain is constantly trying to predict certain selected data using the models derived from previous data, then updating said models AND what information it is trying to test at the same time actively?

[@PeterIntrovert]

I recommend this video: **Maxwell Ramstead - A tutorial on active inference** kzbin.info/www/bejne/jaupgq18j86qsM0&feature=shares Brain don't process sensory input becouse it would be too many data to process. What brain process is difference between what he predicts and input signal. That information flow up the system to update higher level models. That's why for example: first you recognize growing shape/object then you index that someone is walk towards you and then you recognize that it's your friend Johny. It happen fast enough that you can don't notice this process.

[@thinginitselfclips]

I think that’s roughly right. The argent is learning this generative model because it’s trying to minimize some variational free energy objective.