Status Updates From Surfing Uncertainty: Predic...

And not only that, if you take active inference seriously and still wanna deploy a Cartesian skepticism argument, you'll have to grant that the very environment which the active PP-agent exploits is something besides the brain doing its guessing, something Matrix-style.

... all that matters to the fundamental role embodied engagement with the world is to show that this is more empirically adequate than the "reconstructive inner model". if one wants one can still entertain Cartesian doubts, but embodied-PP lends no extra support here.

Great treat: he's dispelled a philosophical spin sometimes given to PP, namely to take it as showing that we can't "really know" the world because all the brain does is inferential work about hidden causes. He argues this is an illicit move from the scientific context to the philosophical one: ...

LONG overdue, but finally got myself around re-starting it. As is known, it's much more technical than The Experience Machine, but all the more fascinating because of that. The nitty-gritty is indispensible for a real taste of just how powerful and promising PP is. Clark's constant emphasis on action-prediction loops and on integration with enactive/embodied cognition are major pluses here too. Beautiful stuff.

Normally, we attenuate the proprioceptive impact of our self-generated movements, but this goes wrong in schizophrenia, which leads schizophrenic patients to hypothesize about alien agency ("when I press my finger against my hand, *someone* is also pushing my hand against my finger!")

About mirror neurons: it’s interesting to say that the perception of an action has to be influenced by context to be understood, and that this translates into adjustments of the prediction error weightings in the PP model, but that doesn’t tell us how the adjustments themselves materialize in neural terms