Read on Twitter
Yes, good question @LitvakVladimir ! One of my motivations for writing the Markov Blanket paper was to warn students. The situation is quite simple: metaphysical conclusions require metaphysical premises. The answers to philosophical questions lie not in going deeper into 1/ https://twitter.com/LitvakVladimir/status/1334236348912427014
the math, but in figuring out how your mathematical description relates to reality. Fine if you want to have a Blanket-Oriented Ontology (BOO!), but your ontology will not follow from the math. A lot of people who should know better have uncritically accepted grotesque claims 2/
they did not fully understand, leaving the field in a very poor state. Even while getting this mapping right, I see no reason why Markov Blankets should be of any relevance to answer philosophical questions. The so-called "Markov Blanket formalism" that we try to reconstruct 3/
in our paper is full of arbitrary assumptions. It is these assumptions that shape your conclusions, not the formalism.
In a sense my frustration lies in the fact that I think FEP has so much more to offer, also philosophically. I see it as a lens to understand the selection 4/
In a sense my frustration lies in the fact that I think FEP has so much more to offer, also philosophically. I see it as a lens to understand the selection 4/
pressures at work on cognitive agents on the time-scales of perception and action. It's the most interesting exponent of a neuroscience based on selectionist principles with many connections to embodied cognition and enactivism. All of this is neither trivially nor necessarily5/
true. So, cutting a long story short: I'd strongly urge interested PhD students to stay far away from BOO. I think it is broken beyond repair. But I would recommend them to work on the softer sides of FEP, both within STEM and philosophy there is a lot of work to be done. 6/6
Like the papers, this might be complementary to @bayesianboy's thread: https://twitter.com/bayesianboy/status/1334363536559116288
