Mindscape 87 | Karl Friston on Brains, Predictions, and Free Energy

Blog post with audio player, show notes, and transcript: www.preposterousuniverse.com/...
Patreon: / seanmcarroll
If you tell me that one of the world’s leading neuroscientists has developed a theory of how the brain works that also has implications for the origin and nature of life more broadly, and uses concepts of entropy and information in a central way - well, you know I’m going to be all over that. So it’s my great pleasure to present this conversation with Karl Friston, who has done exactly that. One of the most highly-cited neuroscientists now living, Friston has proposed that we understand the brain in terms of a free energy principle, according to which our brains are attempting to model the world in such a way as to minimize the amount of surprise we experience. It’s a bit more complicate than that, but I think we made great headway in explicating some very profound ideas in a way that should be generally understandable.
Karl Friston received his medical degree from King’s College Hospital, London. He is currently Professor at the Institute of Neurology, University College London, and Wellcome Principal Research Fellow and Scientific Director of the Wellcome Trust Centre for Neuroimaging. Among his major contributions are statistical parametric mapping, voxel-based morphometry, and dynamical causal modeling. He is a Fellow of the Royal Society, of the Academy of Medical Science, and of the Royal Society of Biology. Among his awards are the Young Investigators Award in Human Brain Mapping, the Minerva Golden Brain Award, the Weldon Memorial Prize, the Charles Branch Award, and the Glass Brain Award for human brain mapping.

[@airgray.]

Sean Carroll, I've spent 2 months working through Friston's Free Energy Principle. Your conversation with him is the best by a mile. Friston didn't mention Markov Blankets until 46 mins in…amazing! The fact that you started with the problem being solved really unlocked the interview. It became clear that Friston's work in brain imaging resulted in computational models for image representations…and his leap of insight that the brain itself is conducting analogous processes. Less maths, fewer blankets but more clarity. I'm going through the rest of your podcasts as a result. Thanks!

[@FUTUREWA]

Inference engines wrapped in Markov blankets attempting to align error bars with reality in order to avoid an entropic fate.

[@erictko85]

FUTUREWA bravo

[@Czeckie]

yep, it's mindscape time

[@AG-pm3tc]

My dude, don't be afraid going deep! When we want to hear something with not to much high level science stuff, we go to joe. But here, go wild, this is your kingdom!

[@aaron2709]

Sean, love everything you do. Pound for pound, I think you're the smartest, most insightful human on Earth.

[@chemquests]

Pound of brain? Ha

[@pedrorockstar]

Sir, you did a great job translating Karl's words into something we can all understand.

[@MrPDTaylor]

I discovered Mindscape about 6 months ago and it's quickly become my favorite podcast.

[@thewiseturtle]

I will suggest that there are two kinds of brain functions (working at fractal levels) and one of them does indeed aim to narrow down information and lower entropy, which is the conservative/risk-averse/input focus type thinking. The other type of thinking is where we expand information and incease entropy, which is the creative/novelty-seeking/output focus type. The latter type design/function innovates (random mutation) new patterns (movements, emotions, ideas, philosophies) and the former selects the "best" patterns for a given goal. These variations in the fractal designs/functions of brains are what we call personality.

[@capoeirastronaut]

Seems like you are using the word fractal unnecessarily. Or unclearly. Are you thinking, Kahneman's fast/slow, or like the evidence from split-brain patients of an environment-focused hemisphere + a self-focused hemisphere?

[@thewiseturtle]

@@capoeirastronaut I wouldn't use the term fractal if that wasn't what I meant. But this isn't the place to explain complex models of consciousness and brains. So I think I'll just say that it's based on the dimensional mapping of current-to-goal states on a first, second, third, and fourth person level in the different developmental brain regions (motor cortex, limbic system, neocortex, and prefrontal cortex).

[@capoeirastronaut]

Oh really.

[@chemquests]

@@thewiseturtle then he was right to say you’re using the term unclearly

[@thewiseturtle]

@@chemquests Physics is indeed unclear to all of us, much of the time.

[@amitkhambekar3994]

You are great Sean! From India

[@vpdabholkar]

Enjoyed this podcast! Thank Sean. I am a huge fan of Friston and really appreciate the role Sean played in translating the technical jargon and covering so many dimensions of the free energy principle framework. For me, Friston's mathematics is still quite heavy. So this podcast is a big help. One aspect that I hope gets explored with Friston in the future is the intersection of mindfulness and active inference. It got touched briefly with Sean's two cats - in-the-moment cat vs the planning cat. Perhaps Sean's summary lesson - we should aspire to bring our error bars in close alignment with reality - could be related to mindfulness? I could be wrong here.

[@danzigvssartre]

Vinay Dabholkar I would recommend the book: “Am I Dreaming” by James Kingsland. It may help elucidate your question.

[@vpdabholkar]

@@danzigvssartre Thanks for the recommendation. Will read.

[@neoepicurean3772]

I came here from another of Friston's videos which went completely over my head. At first I thought it was just me, but he really doesn't seem to explain things well. Only 20 mins in and I hope it gets easier to follow as I think there's something in here I really should be aware of as a philosopher.

[@neoepicurean3772]

30:50 THANK YOU SEAN

[@errgo2713]

Another approach to what Friston refers to as "vagueness" is the philosophical concept of IDEOLOGY. In particular the approach offered by Alain Badiou where all instances of units for us, sapient beings, are necessarily results of the ideological process, "count-as-one". What counts as a "one" is always contingent upon ideological (cultural) bias.

[@capoeirastronaut]

I was surprised Friston didn't mention that people blind from birth or shortly after seem to be immune to schizophrenia, and not just slightly - there isn't a single recorded case, although scientists have only been looking for a few decades. While formally a mystery, it does seem like a powerful clue about the role of dysfunctions in our predictive models & mental illness

[@dave4148]

minimizing error ("surprise") is probably the most obvious thing to note about any information model... defining surprise and finding out the algorithm for back-propagating surprise through the brain seems like the hard part

[@GrantBrennerMD]

Wouldn't an adaptive system optimize entropy? Rather than always minimizing? Because sometimes maintaining higher than minimum entropy is useful?

[@DdotArt]

In a similar vein. I'd be curious to hear Sean have a conversation with Anil Seth. I'd imagine given Sean's ongoing interests in consciousness it could be an interesting hour+.

[@timothykieper]

Why not just upload your questions show to KZbin ? Last time I checked it was FREE !

[@thewiseturtle]

Also, what kind of server host charges more money for having more files on the server? My server is unlimited for one flat rate (about $130 a year, I think). I could host a million podcast files for the same price. (Though it might crash their servers if everyone tried to download/stream it all at once...)

[@johnjoseph9823]

I agree.

[@capoeirastronaut]

It is done now

[@JoeyVol]

He is talking about his websites feed, btw, where you simply download the mp3 file. Hosting space is expensive for that old school model, but it's by far the best way to receive the show. You can save it forever, etc.

[@whitenoise4794]

Thank you Sean for this quality podcast. Can you please share, along with the episode's description, an outline with timestamps where we can click and jump right to that point in the episode.

[@artandculture5262]

How old are you?

[@stephanvonugron1944]

Great Podcast!!!

[@freeair9460]

Your podcast are great. Don't doubt your work

[@chrisrecord5625]

Big Picture published in 2016. This seems to be a book tour related interview. Could be mistaken,

[@mike66977]

my brain needs more energy for this, i need to sleep on it and try tomorrow again

[@palhardy4292]

Favorite episode by far

[@malcolmvargas45]

let me see if I understand this: Karl Friston's Free-energy uses the term free-energy simply because of one thing - the formula for Bayesian inference is borrowed from the free-energy concept from physics. Otherwise, his concept of Free-energy has nothing to do with ffree and nothing to do with energy... is that correct?

[@boden8138]

Go deep. That said, a huge amount of information is lost when you are not able to see the speaker. Nonverbal information is necessary for deep learning.

[@perjespersen4746]

Wow. This guy is a badass.

[@hmdshokri]

i'm waiting for David Deutsch ,he's absent for sometime i guess

[@wduncani]

Intro ends at 6:20

[@HarryNicNicholas]

odd you mention compartments, i was thinking just last night, does everything have a skin? does this evolve from single cells? it's pretty obvious why we have skin, but everything from fruit and veg to rhino's have skin. my girlfriend is schizophrenic and is quite hard to understand pretty much all the time, would've liked to heard more about schizophrenia, i have "explained" to her that she and i live in different domensions which is why she sees and hears things i don't, it a non judgemental way of asking about hallucinations (she is very, very defensive). still, interesting talk nonetheless.

[@pleiadesglow]

Hmmm...I never knew "Energy" had so many meanings.

[@danzigvssartre]

pleiadesglow Not energy. Entropy has many different meanings. Plenty of these physical scientists will tell you that entropy means energy and that entropy also means “information.” Unfortunately, they won’t tell you that their notion of “information” has nothing to with meaningful information as we understand it in our commonsensical way.

[@LeeLightfoot]

1 hour 14 sounds more like a conversation to have with Nick Lane

[@Czeckie]

this was extremely interesting, I need to learn some non-equilibrium thermodynamics. Btw, Karl often uses the word 'deflationary' and I have no idea what it means.

[@capoeirastronaut]

This is funny, because even the Wikipedia page on says "Non-equilibrium thermodynamics is a work in progress, not an established edifice." When Shaun says it is the coming paradigm, he is not being explicit about how allergic physicists are to this field - it rapidly gets like 'messy' chemistry & biology, rather than the beautiful physics of old. Hella more people need to get interested in this field, but you won't find much to chew on

[@emilylowrance7930]

outside your comfort zone is where magic happens!

[@TheReferrer72]

I must have a lot of magic in my life!

[@emilylowrance7930]

@@TheReferrer72 well i think there's more to magic than just being uncomfortable, heh.

[@thewiseturtle]

Yeah, this guy is a very conservative thinker on many levels, and so he doesn't see the other brain types, which are high entropy functions, innovating and loving every moment of it, finding delight in surprise.

[@gr500music6]

It is podcasts like this that cause me to remind myself" "Mind SC, Ape!"

[@plaguedoct0r]

I am captain minds cape, I wear my cape around my head

[@gr500music6]

Wishing you the Cape of Good Hope !

[@israelallen6430]

Lmao y’all funny

[@UtraVioletDreams]

Thank you professor Caroll, wish I have the means to support you. But please keep going I love these podcast that are, well quit new and getting towards a year by now. Maybe you can make the you tube vid less boring. Why no life cam? I know you're not camera shy :D

[@hmdshokri]

there is no free lunch,except the whole universe

[@andrewandrus3296]

yes!

[@user-zy8lm4mu5v]

兀的内容丰富多彩

[@DivinitySaid]

i wonder if deja vu is us inferring with 100% accuracy.

[@danzigvssartre]

div inity I’m not if sure “inferring” with 100% accuracy is still an inference? This is one of the (many) philosophical problems with Bayesian Inference Models of Perception. Does one “infer” that one is eating breakfast in the morning?

[@DivinitySaid]

danzigvssartre - honestly, i'm not smart enough to even be on this channel. so at the risk of ridicule, i'll reply anyway :) it sounded to me like our brains use inference to effectively guess what should happen next, and after things happen as subconsciously assumed (there is no false inference) then the brain continues on inferring the future to mitigate surprise/uncertainty. that's how i heard it, and that's why i veered out of my lane and made this comment. :) to me, it seemed like he was saying that our brains sort of auto-pilot the inference process. you've* had deja vu, right? suddenly you're doing something you've 'seen before' - did our minds infer a scenario with such accuracy that we are able to recall it? i don't think we are inferring eating our breakfast? seems like inference happens before we ever move a muscle. but then again, i've never heard of a Bayesian* anything. thanks for the reply. 🤍 *edit: grammar

[@danzigvssartre]

@@DivinitySaid That's a good summary of Friston's theory. This principle of unconscious inference is modelled statistically by using Bayesian statistics. In the 1800s, Bayes' Theorem was used by Helmholtz to suggest that sensory perception was a process of 'unconscious inference.' I am sympathetic to an alternate theory of perception called 'ecological perception.' EP contends that when we perceive events, we directly experience them, unmediated by some kind inference mechanism in the brain. In this theory, perception by definition, is 100 percent accurate (as opposed to illusion or hallucination). Incidentally, if deja vu were "100 percent accurate" would it not be reincarnation, rather than a faulty inference?🤔 Just a speculative thought? Incidentally, the best essay written on Deja Vu (in my opinion) was written by Henri Bergson: "Memory of the present and false recognition."

[@DivinitySaid]

danzigvssartre you are the absolute best!!! I look forward to reading that essay. I cant say enough how much I appreciate your time and thoughtful reply. As to your speculative thought, again I find myself in unfamiliar waters. I've been rolling the word 'reincarnate' around in my brain and I suppose you're onto something here... mostly because i cant differentiate between reincarnation and deja vu anymore haha. Our thought potential is blowing the lid off my mind right now. thank you thank you thank you

[@danzigvssartre]

div inity 😊 Thank you too.

[@gamblinguru33]

This podcast is like lifting weights for your brain

[@nINJAP3RUNA]

The podcast is the weights and the lifting happens inside your head :O

[@jynxkizs]

You should invite Lee Smolin.

[@chiphill4856]

I love Mindscape! I was excited about this episode but simply could not stay with Dr. Friston because of his accent and tendency to mumble the end of his sentences. No mal intent here, just relating my experience.

[@ericjohansson721]

へくててねほねつほもねねももすねもつねてもねもつねてもほもつねもねつももねねつほすほねももねつもねもつねねつもねつももねもつねもつねもつねもつねもねすほすねつもねもつねすすほねすほねほすねほもねつもねつもねつほねもねつももねほすももねつもねつももねねつもねつもねもねつももねつほもねもつももねつもももつねつほもねつもねつもねすほつもねつもねももねもつもねもてねももねもねつもねすほねすねもすほねもねもねほすねつほすねすほねつもねつもねもつねつもねほねつももねすほねももずもほねつもねつもねすほほねもつねもつもつつもねもねすほもつねほねももつもねつもつもねもつねつもねつすねほつもねももねもすほねももつねつもねすねもねつほつもねもつねつほねつもねもねもねつほもねもつねすほねもつねもねもねもねもねほつもねもねつももつつもねつもももつねつももねもねもつもねもつねもねつもほすねもつもねもつつもねもねもつねももねつもねつもねもつねもつもつもねもつもねもねもつねもてねもねもつねもねつもすねもつねもつねつもねもつねつもつもねもすねほつねつもねもつもつもつねもつねほすねもねほすねほすつほねほすねつもねもつねもつねもつねつねつもねほつねすもねほつねつもねもてねすほねもつねつほほねすねすほねすほねもつねつもねすつももねもねもつねつもねすねもつねももねもつすねほすもねほすほもつねもつねすほもねつほすすほねもつねほすもねつもつもねもねつもねもねもねもねもつねももすねもつつもすねもつねほすねほねつもねつもねもつねつもねつもねもねもつねもつねつもねつもつもねすほねもすねすほねほすもねもねつもつもねもつねつもねすねもつねすねすねもつねつもねつもねもつねつもねもつねすももつもつねもねもねももつねもねつもすねつもねすねもつねもつねほすねすほねもつねすほねもつほすねほすもねもつねすねほすもねもつつほねももつねつもねすねほすねもつつもねもつねももねほつねもつもねつもすほすねもねほすすねもつほすねもつすねもつねもつねほねもつもねつもねつもねもねもつもねほすねほねほすほすねももねもつもつもねもつつもねほつもほねもねももてねもつねもつねつもねつもねもつもねもねもつねほすほつもももねつほねもつねもねもつもつねほすずぽもつもねもねすほもほすねもねもつねもねもつすほねすほねすねもつねつもねもつねもつつももつねももねもねもつほすもつねすねもねほつももつももてねもつねもねもねもつねもつねもつねつもつもねつもねほねつももつもねつももねほすねつもねつすほねすほねももねもねつもすねすほつもねすねほねほねつもねつもねもつねつもももねつももつもねもねもつもつねねもねつほねつももねもつもつつもももねほもつもつつねすねももつねもつももねすほつもねもねすねつもねほもねもつほつももねつもねつももねほつねつもねもねつもねつもねもつもねつもつもねもつもねつもねつもねつももつねつもつつもねつもねもねもつもねつもつもねつもねもねもねもつねつもねつもつももねつほねつもねほすつもねもねつほもねほもねつほねもつねもつもねもねつももねつもねもつもねつもねつもねもねもねもねもねつももねもつねもつもねすほねつもすねもねもつねつももねつほつもねもねもつねもねもねね

[@islandbuoy4]

hello Sean Carroll ... let us apply one Karl insight to yet another Karl insight? ... i.e. *Karl Popper's falsifiability as applied to Karl Friston's theory?* what we love about Karl Friston's insights is how it defaults in a sense to the law of thermodynamics as applied to entropy [a pet peeve of mine] ... now it was not until 2012 that the following was stumbled upon by the physics community i.e. *entropy + particle shape = chirality* [suggesting right hand dominance in humans does not need to be found in our genes necessarily] ... thus yours truly being the son of yet another Karl asks the obvious ... is the following algorithm hinting at what set the world into 'motion' falsifiable? the algorithm is thus expressed as *ACHIRAL >>> CHIRAL >>> ACHIRAL >>> CHIRAL >>> forever and ever* and ever till death do us part and the chiral anomaly [part of a feedback loop] gives birth to yet another ever emergent manifestation of what 'just is' ... *BTW Karl Popper's insights to Plato's elementary square and elementary equilateral as being approximations of sqrt2 + sqrt3 = pi merged with the squaring of the circle archetype clearly demonstrate the ACHIRAL >>> CHIRAL algorithm...herr Carl Jung yet another Karl called squaring of the circle the 'archetype of wholeness'*

Did anybody else not understand a word of this? Lots of jargon and unnecessary complexity in expressing what seems a very basic theory. When Sean Caroll has to continuously explain what you are saying it’s a worry.

[@diggitus]

Well, if you didn't understand a word of it than how can you call it "very basic"? I didn't understand it either but I didn't get the sense Friston is being obfuscatory, I think this field is just super technical. Yes, Carroll has rare gifts in explicating things, he's incredible.

[@iruleandyoudont9]

a lot of people like being surprised

[@chromeome5245]

I guess I am not smart enough to understand any of this

[@benjaminandersson2572]

He comes off as a little dogmatic. I don´t think the Free-Energy principle explains EVERYTHING about sentient behavior. Does free-energy explains mathematical acitivity? Clearly it can´t, since the paradigm USES mathematics.

[@danzigvssartre]

Benjamin Andersson I would contend that the Free-Energy principle says NOTHING about “sentient” behaviour. Friston has thrown the baby out with the bath water. Good luck to him in applying any of this to his patients with schizophrenia.

[@Gerardemful]

I used to admire Mr. Friston until he outed himself as a cheap nationalist in his Guardian interview.

[@benwhite8145]

Link to the interview?

[@Gerardemful]

@@benwhite8145 www.theguardian.com/world/2020/may/31/covid-19-expert-karl-friston-germany-may-have-more-immunological-dark-matter

[@benwhite8145]

@@Gerardemful Help me out here, I don't see anything here that points toward nationalism. Maybe I'm missing something?