Jeff Hawkins: Thousand Brains Theory of Intelligence | Lex Fridman Podcast #25

[@lexfridman]

I really enjoyed this conversation with Jeff. Here's the high-level outline: 0:00 - Introduction 1:28 - Understanding how the human brain works 5:44 - Parts of the brain 11:05 - How much do we understand? 14:20 - Nature of time in the brain 20:22 - Building a theory of intelligence 34:29 - Thousand brains theory of intelligence 40:06 - Ensembles and sensor fusion 44:00 - Concepts and language 45:38 - Memory palace and method of loci 50:20 - Reference frames 57:33 - Open problems 59:00 - Context 1:01:50 - Introspective thinking about the brain 1:04:19 - Deep learning 1:23:09 - Benchmarks 1:27:07 - Brain learning process 1:34:33 - How far are we from solving intelligence 1:38:37 - Possibility of AI winter 1:39:58 - Consciousness and intelligence 1:49:16 - Mortality 1:53:49 - Will understanding intelligence make us happy? 1:55:19 - Existential threats of AI 2:01:45 - Super-human intelligence and our future

[@5minuteteen697]

I also enjoy your podcast series, you look smarter in podcast outing

[@danielbigham]

On consciousness: @Lex, have you ever considered the notion that the modalities of consciousness (vision, hearing, etc.) correspond to "fields" (as in, field theory), just as there are fields like the Higgs field for representing mass, the electromagnetic field for EM phenomenon, etc.? Just as E = mc2 tells us that information can be exchanged between fields, perhaps the brain is able to transfer information to and from the experiential fields. If true, it's possible that color really does end up looking like a wave, just not in the EM field.

[@mrshow1011]

I still believe that cognitive psychology is the software of the brain while neuroscience is the hardware of the brain. Thus, neuroscience will lead to more efficient computing processes while cognitive psychology will lead to how the brain learns and extrapolates (most likely in a Bayesian capacity). Hence, GAI will only truly be achievable with the right cognitive models leading the charge.

[@ethiesm1]

I know you have a podcast plan but let the guy talk

[@kaleb7299]

How much did the work of Wai H. Tsang contribute to these ideas?

[@andrewhunter8325]

Lex Fridman has been my greatest find from JRE!

[@wildhunterwh]

Mine too

[@razadaza9651]

Lex was on JRE? Nice

[@drkmattrchscake4237]

Lex and ofcourse Joey Coco Diaz

[@artemissb6950]

Mine too!

[@CalvinJary]

This is because Lex does MMA. Which is the only Joe rogan stamp of approval lol. Well and doing comedy

[@Xweapon9099]

My brain still has an almost complete map of grand theft auto san andreas.... wierd how i can still just walk through a virtual city in my head almost 2 decades later but cant remember how to do some college algebra.

[@stoopidwookie1695]

Lolol. Time spent involved/depth of involvment.

[@RalphDratman]

What type of algebra?

[@Xweapon9099]

@@RalphDratman basic college. I plan on going back soon and treating it a lot more seriously though

[@akarshrastogi3682]

gta sa: i played it for 8 years straight

[@UltraK420]

I did the same with the originals on PS1 and 3 and SA. My next goal is to form a complete and accurate map in my mind of the mostly procedurally generated universe in space engine with the ability to find the milky way and even Sol and Earth from anywhere in the universe without any assistance from external indicators like labels or maps. Not really, that will probably never happen but it would be cool.

[@TheGunmanChannel]

One of the best episodes right here

[@cleon_teunissen]

40 minutes in: the view presented by Jeff Hawkins strongly reminds me of the views presented by Marvin Minsky in his 1986 book 'The society of mind'. Highly recommended, that book. Minsky offers a theory of psychology (both cognitive psychology and personality psychology). Minsky presents that the evidence suggests that we should think of the psyche as a multitude of cooperating agents. This society strives to reach decision by concensus, but that is not always possible. Minsky discusses: how is our brain able to process language so fast, given how slow neurons are? Minsky suggests massive parallellism. When words enter the auditory center they spread thoughout all of the language center, and all possible associations are generated, including the most farfetched. All those associations in the overall context are compared with each other simultaneously, one that outcompetes all others makes it to the level of _conscious_ thought. Almost always the winning assocation is in fact the correct one. All that in a fraction of a second. Minsky offers that this is why puns strike us as funny; a "wrong" meaning, that otherwise would not make it to the level of conscious thought, is suddenly justified.

[@NumentaTheory]

Here is Jeff's response to a similar comment: www.reddit.com/r/askscience/comments/bowie2/askscience_ama_series_were_jeff_hawkins_and/enn73xf/

[@nickking6371]

They brought that up on reddit ama also

[@UtraVioletDreams]

Thanks for the recommendation!

[@lnc-to4ku]

Very interesting! Thanks for the referral of the book you mentioned!

[@user-zu6kj8fv2w]

I was just thinking today about how thinking feels like holding a council. I'm always so excited when your videos hit my suggested feed!

[@marlinwilkes]

How about you subscribe and turn on notifications

[@karimjedda]

So far my favorite. Excellent knowledge in this one.

[@alicethornburgh7552]

I like what he's saying, but his voice and cadence is so irritating.

[@Hexanitrobenzene]

@@alicethornburgh7552 Yeah, at first I was sceptical about this man. He sounded like he is trying "to sell what he doesn't have". For example, all those "I feel confident about this...". Luckilly, I suspended my disbelief and he redeemed himself shortly after :)

[@vkataevify]

Did you play different 3D games as a child many years ago? Today I still remember all the levels, different locations, hidden items, etc. I also remember many geographical places I visited many years ago. I think there's something really interesting in how brain remembers so much so efficiently related to that kind of what he calls a frame of reference. Really interesting discussion.

[@MorrisonEnterprise]

Makes me wonder how living between VR and the real world is going to mess with our heads.

[@skyacaniadev2229]

That is probably more related to you being a child at that time (and still have lots of "room" to establish new synapses). Probably it will become harder and harder over time as people grow.

[@venture3800]

It sounds weird for the first half a second when he explains that language and concepts are built on the same circuits you use to map all the input from your senses but then once you think about it for a few seconds it's like "oh, duh" 😂

[@thesomeonetwo]

Nobody in the comments mentioned that this is the man who invented Palm devices. Only true 90s/2000s folks know what I'm talking about!!

[@KravMagoo]

@Leonardo Laprickpossibl I reported YOU because I am going to be born in 2020...which is why I have such great hindsight. Too bad for you.

[@UtraVioletDreams]

Really? That's the guy? Once upon a time, in a land far far away.... I used to support those devices for business and consumer marked... Seems so long ago.

[@robfranzo2238]

And before starting Palm and creating Graffiti, Jeff wrote the best Penpoint note pad that used all of the hyperlink/text features of the platform. The rest is history.

[@plantraco]

Rob Franzo remember the Apple Newton?

[@SergioPelin]

And he also invented the first smart phone: Treo

[@pragnakar1]

15 min into the video and already finding it so enlightening.

[@Strelnikov10]

Jeff Hawkins' TED Talk (back in 2006 I believe) was the singular moment that led me into a career in theoretical neuroscience. So cool to see him on here.

[@boxedowl]

Allow me to collect my jaw from the floor...

[@Gi-Home]

I have always enjoyed Jeff Hawkins work at Numenta and I pretty much catch all of your interviews, they are excellent. You did an outstanding job with this interview and I hope you bring Jeff back in the future.

[@underpowerjet]

My mind is literally blown right now! This is how my brain, how all of our brain works! Amazing! Thank you Lex for this awesome interview!!!

[@TheMrXDion]

I love these podcasts so much!

[@koningsbruggen]

Same

[@spasibushki]

потрясающий разговор, у гостя интересные и свежие взгляды на очень много вещей, спасибо, Лекс!

[@weedeeohguy]

Same to you buddy!

[@rks7828]

Probably the best interview in this series. Thanks Lex for this one!

[@Fanofjambi]

Fantastic, I love the human side of these new ideas, i.e hearing the concepts explained by the originator of the work and defending them from pushback in real time, just excellent content

[@delatroy]

I love listening to Jeff Hawkins. I remember watching him on MIT over 10 years ago about Numenta. 🙏

[@felixfunk6816]

There were a lot of great ideas in this podcast! Will definitely check out his papers. I personally find the idea of modelling the temporal properties of the brain into neural networks remarkable and positively challenging.

[@highburyful]

Brilliant podcast as usual, listening to it without visuals sounds like a lecture from Kevin Costner. Great content Lex

[@Rybek]

Thank you, I really enjoyed this one. Because of Jeff book I started be interested in all topics around neural networks, AI, brain structure.

[@VladislavKurmaz]

Thanks a lot Lex, absolutely amazing video, 2 hours of pure science and great ideas.

[@samuelec]

Wow!! This talk was so enlightening! I got captured almost immediately about his ideas and the insights. As a programmer and a curious being, some parts of the video, I will have to watch again to better understand the implications, the model and process he describes. Lex you literally were asking same questions I would ask him. I have also appreciated how you were driving the conversation. I hope to find more material on this subject. Thank you both

[@Retrograde6]

What a great interview. Thoroughly enjoyed that.

[@gedankenthesis]

Jeff is a really special human and his ideas should be studied, cherished and celebrated. Applaud your efforts in making this interview happen, Lex. Thank you man.

[@sra-cu6fz]

This is, far and away, the best one. Brain learning process @1:27:07 sounded a lot like a spec to me. Next stop: Jeff's papers.

[@jonatan01i]

The thought of neurons fireing in advance of the input arriving could be onto something. And after the fireing if the input does not arrive, that's a feedback that the neuron doesn't do as well.

[@user-ol5bj4dm2v]

@@jonatan01i this has been experimentally proven to occur in the grid and place cells.

[@jonatan01i]

@@user-ol5bj4dm2v can you show me a study on this one, please? Or at least point me to the right direction?

[@nanotech_republika]

Great interview about mesoscale brain computing - I love it! 1. Each hypercolumn creating models of the world (the thousand brains theory). 2. Short term memory works by converting silent synapses to an active one.

[@danielisaacson264]

best interview i've seen on your channel!

[@ArthaOne]

Truly fascinating! I will definitely try to follow his work in the future.

[@jonacacarr3839]

I will definitely watch this a second time, so helpful. Great questions Lex. Thank you

[@Stwinky]

Awesome you had Jeff on! Him and his group definitely deserve more attention!

[@sidaliu8989]

It feels so great to listen to Jeff Hawkins after reading his book! This conversation makes the book come to life to me. Thanks Lex!

[@fedsummer90]

His book "on intelligence" is so amazing.

[@carlostorres656]

Jeff Hawkins is a brilliant person, it is amazing the way he express so many points connected in a rhythm that makes you feel you are smart, I believe that a talk about the brain and the capabilities it has makes you in some conscious of the powerful and incredible part of your body that makes so many outstanding things.

[@Ideophagous]

One of the greatest interviews on this channel!

[@wadewilson7320]

Thank you for these wonderfully informative videos, Lex!!

[@johangodfroid5285]

thank you for this post it is so interesting

[@fornasm]

Many thanks Jeff and Lex! This is one of the most interesting/stimulating podcasts I have heard in last years. Regards, Marco

[@TheThunderwesel]

The way he talks about your brain working by recalling reference frames really helps explain what Leonard suskind was talking about when he was saying that he felt like it was so hard to envision any dimensions other than 3. It's most likely because our entire collection of reference frames is in 3d space.

[@kentvandervelden]

Fantastic introduction. Loved the interview

[@fotoyartefotoyarte1044]

ive been following jeff for the last 5 yrs or so, maybe more, very interested in his theory since i heard of it, this is awesome to me, thanks !!

[@szghasem]

I really enjoyed Jeff's point of view on life and intelligence. It was a profound aha moment for me. Thanks for sharing this interview!

[@juniorrealtordavesoffice2435]

Great Work Lex!

[@JohnPretto]

Lex, I love the opening of your show with pitch back background in your mobster black suite. Love the show....

[@BodoHoenen]

@lex, I really appreciate all of these conversations you are sharing!

[@miroslavhavrlent1131]

Thank you Lex. One of the best interviews.

[@whiteF0x9091]

enlightening ! Thanks for the great discussion

[@evanwillenson1398]

Lex your understanding of philosophy and biology and neuroscience facilitated a great convo... as ALWAYS

[@patrickmchargue7122]

I appreciate Mr. Hawkins views and this interview. Would love to ask a few questions myself!

[@Pencilbrush1]

Jeff's is Hands down the best Guy out there working on this problem ..

[@DavidBerglund]

What a fun conversation this was! I'm hoping that I'll see Ben Goertzel on this channel sometime.

[@kirstinstrand6292]

This interview answers so many of my questions regarding life, in general and I can extrapolate these ideas into personal experiences. The construct of identification of objects also relates to mental health. I recently isolated myself in order to resolve childhood emotional pain. My theory is that memories get buried on top of each other - separating out OLD feelings and memories seems impossible. The following is how I identified emotional pain from formative years. I had a Lucid Dream of an open flesh wound. One day later, I was cutting vegetables for a salad. I always keep my utility knife razor sharp for ease of use. I accidentally cut across the knuckle of my thumb nearly 1/2 inch. Flesh was exposed, it hurt and bled like hell. It was difficult to stop the bleeding and the pain. It took much longer than one would expect for such a seemingly insignificant wound. I was mystified by the lucid dream and the exposure of flesh. This is only one of many unusual circumstances that I have experienced on my journey of self healing. I eventually moved out of the lower level of my property, back into daylight, where I continue my journey. BTW, CONCIOUSNESS evolves through pain and suffering. However, Consciousness seems to be misunderstood. Clearly, there are different levels.

[@lil----lil]

Excellent Channel! Subscribed. Was looking for something like this but without the TED talk glitz.

[@penguinista]

"You're just deeply enlightened or are completely delusional." 1:52:00 Superb!

[@philoneill9865]

Excellent interview. Well done. I think the idea helps explain humour or comedy. Our brain agrees on the meaning of what the comedian is saying, and then the punch line breaks your brain's pre-agreed understanding. So, a broken reference frame can frighten you, or make you laugh, or even cry. Anyway, that is just an amateur's take on all this.

[@kevinmcbride9277]

Longitude, latitude. What's the word for the third part of this intersection reference of space? Strange how well my mind still can memorize grand theft auto San Andreas especially apache hybrid in vigilante mode! where is the alien fuel we copied with super collider .let me think,I was thinkin I could use me another six or eight cans of that potted meat, if you got any? Um hum

[@musicalfringe]

Good thought, I think you're onto something there

[@iestynne]

A joke doesn't just break a prediction/pattern, it unexpectedly throws you over to a new and wildly different (yet valid) one.

[@user-ol5bj4dm2v]

@@iestynne great point.

[@empathylessons2267]

Neurons communicate with an electrochemical language, consciousness is a conversation. Each neuron telling other neurons what is being told to it, and the world simulated in our brain is the gestalt, the chorus, with each conscious sensation being defined by its relationships to other sensations. Mind blowing.

[@rhondadixon9786]

My first comment ever on any podcast ! Love Lex because he gets the depth and beauty of the tango scene in Scent of a Woman .

[@RooBeeRTooooTReeBooR]

Amazing content, thanks Lex for helping the open innovation and sharing of ideas develop

[@nickking6371]

Lex, i dont know where your confidence level is at in regards to the podcast game, but these are the most 🔥 💰 videos. Your guest choice is doing so much for me. I saw this guy on reddit AMA and was transfixed by him. I spent half the day talking about it on Reddit nd had to put it down for a minute. No doubt in my mind this is where im headed.

[@treeflip7]

Jeff Hawkins is our greatest hope for understanding consciousness

[@alexshneyderman243]

This is an awesome interview!

[@AngadSingh-bv7vn]

this is my favourite lex podcast to date. wow jeff hawkins.

[@danbreeden1801]

Fantastic I really enjoy this series and the interviewer and the fascinating minds he has on his program

[@jyreHeffron]

THIS guy is such an awesome communicator... great interview - deepest dive yet on most of the topics concerning brain function... just az¸A

[@earlewischmeier3128]

I mentioned Ingo Swan in one of your videos yesterday and goodness gracious we're talking about dimensional information trying to take it to the physical. Consciousness reality what a adventure I would have loved to see you guys interview him. But you can see his last talks on KZbin befor he died. It would be in interesting discussions to incorporate his part of the information into A I. Earl

[@jimfraser9563]

"you need the winter to refresh the palette" excellent Lex.

[@DalgaSar]

Dear Lex, thank you for these talks. Could you talk to David Deutsch and Frank Tipler? Is there a way to arrange such a talk?

[@wts97m]

Dang this one makes so much sense to me now that I think about it. I do voting logic in some of my work when trying to decide what a person will do with data, and I can see how a blind person would almost seem to have supernatural powers when it comes to sound because they have trained more "votes" to weigh in on a map that is sound based vs visual and can therefor create a better map than we can if we close our eyes. His description of waking up to go pee but keeping his eyes closed really helped me understand that he is creating a map, and only using the visual cues hinders your map, since the only "votes" that are able to chime in are visual. The same thing works when you stick your hand in a box and feel an object. Your brain is working out a map it can recall from touch only (no visual votes allowed), and the more your fingers feel the object the more of the map it creates, all the while having many different maps being recalled from memory that resemble what you are feeling "voting" on what the object is, but if you had also previously listened to the sound of what the object makes because you are blind, you would have more "votes" in your mind of what the object actually is, much like if you could see what you are touching and your sight could also "vote" on what it is. I would love to hear more about how those maps are stored in memory, and when you are touching something, how much of it would be essentially ram vs rom memory.

[@jbaker91585]

Good job so far. The eye contact and q card problem is handled well. Your insight from the 10000ft view is wonderful! Im on board keep it flowin

[@yobeenocreative6984]

I really enjoy listening to your podcasts.

[@dadman9799]

I started this video thinking “im probably not going to understand this very well” BUT I was wrong thanks to the way the guest was able to break things down. 👍👍

[@ivanr7725]

Found it very interesting for myself. Looking forward they build proper neural network.

[@dapdizzy]

Ivan R they won’t, but we can try. I thought that timing of firing is crucial just from the top of my head, may be because that’s so obvious when you even observe how things happen for yourself. But I don’t think timing as a technique will be any different than just assigning he weights. I’ve heard of bigger (higher level) constructs called cognitoms. Those are bigger parts of the cortex working together as a system. One researcher told things like mood, thoughts arise from cognitoms working together. There’s something magical in how smaller parts tend to work together as one bigger unit and I don’t think we understand this generally enough. Like a flock of birds flying together. One said there’s a similar mechanism in play when a group of people heather in the smoking room, so, I believe, it’s working for ourselves even right now and we are up to the challenge of finding out how it really works.

[@pebre79]

Please do an interview with his former partner Dileep George and Scott Phoenix - both currently work at Vicarious

[@ryanstrozeski976]

Sound writing music is like making models of sound. That actually makes a lot of sense to me.

[@lnc-to4ku]

What an incredible privilege it was to listen to 2 brilliant minds discussing my favorite topic- the human brain and all of its deep complexities! This guest was fascinating on so many levels, I hope we get to see him here again. And thank you Lex for being such an exceptional human being! Most people who have the intelligence and thinking power that you have, would feel the need to continually interrupt their guests throughout the conversation to prove how smart they are. But not you. It proves how many different levels of intelligence you possess!

[@murcoutinho9988]

This is probably the best conversation I have ever listen to, just amazing

[@delatroy]

Been listening to Jeff before he started Numenta. Interesting to see how his thought process on the subject evolve. Like watching a very long, slow paced football game

[@jaivigneshvenugopal]

you guys inspire so much you have no idea

[@marirsg]

Consciousness = the "C" word . Lol, well played Lex.

[@lunakid12]

:) Yeah. Actually Max Tegmark used that phrase in an earlier interview with Lex (~2 years ago).

[@MeanGeneHacks]

Read Hawkin's first book "On Intelligence" and am excited for his new book! I also loved my Palm Pilot! Thanks Jeff!

[@MarkoTManninen]

I can tell you this: He is the man of inspiration and enthusiasm. Jeff got me interested in AI for the second time after 90' by his book On Intelligence, and now he is definitely getting me to read their recent papers. Kind of third awakening call for me. But, please don't underestimate feelings, intuition and belief systems, influence of media in human understanding and communication. It is intelligent to account them even they themself may lack epistemological intelligence. True progress of the AI system can be martyred only if it is resistant to bias, solves the mass opinion voting problem and still maintains possibility to interact constructively with the environment. I'm still not sure if his thousand brains system can overcome this. I mean, teached wrong, but correct itself by introspection. @Lex have you considered interviewing Timo Honkela? He might give one more interesting view to the already fabulous serie of AI interviews with his Machine of Peace proposal.

[@tosvarsan5727]

very very great interview, it makes me want to look more about this guy

[@cjansenATL]

An interview with John Vervaeke might be worthwhile. See his Escaping the Meaning Crisis series.

[@msulemanf]

My favourite guest/podcast so far

[@romanarutyunyan8709]

Amazing podcast. Jeff is a very interesting person. I'd like to see him at jre or mindscape too.

[@Niki007hound]

Great interview. Exciting to be alive at times like this, thanks to Jeff and his collaborators.

[@kas4223]

thanks Lex this was so good, crazy enjoyed it! even got some Yoko Taro feels at the end 'to build something that will outlast humans' niceee

[@autotechpintura8775]

Love your program

[@thepoliticalcrass5665]

i like his method of approaching 'thinking'

[@rysw19]

This is the best one I’ve seen bar far

[@funduk89]

Jeff is very inspiring. Can you maybe talk to him once again? Maybe other topics such as his entrepreneurship?

[@mackie-themadscientist6057]

What an incredible conversation, one of my favourites. Totslly agree the C word is overrated. Might have to take Jeff up on understanding the math behind General Relativity ! When Jeff went into detail there, it really opened up some synapses in brain that haven't been opened before...

[@sixtynine999]

Great interview and great close to it

[@MyTimeTravels]

Sounds like lucid dreams I had back when I use to mess with 3d modeling as a hobby. In the dream I would analyze all points and textures of sand I had grabbed from a beach. Made me think how is it possible that the brain can remember so much info.

[@goldenbeers88]

This guy's great! (Jeff) and Lex you're great too.

[@OHOHOHCOME]

This was the only podcast that I listened to completion. Jeff is like the only guy in the AI community that is onto a concrete architecture of a neural net that can solve problems intelligently. There is a small chance that you can engineer an intelligent machine without understanding neuroscience, but I can FOR SURE engineer an intelligent machine by understanding neuroscience! EDIT: Did anybody else cry at the end? Because what Jeff said at the end about human species was so beautiful. What an inspiration.

[@jikkh2x]

I thought this was good until the end, when he decided to project his knowledge fetishism onto everyone. Lots of people do care about genes, for aesthetics or other reasons. He kind of came across as a wimpy nerd finding refuge in intelligence.

[@OHOHOHCOME]

@@jikkh2x Do you at least agree that our legacy as a species is our knowledge, not our gene? That is literally all he is trying to say, he is not refuting peoples interest in their genetics.

[@adeline7980]

Please does anyone know the name of the experiment he mentions at 46:01!!! the one about the reference frames and the birds. I really need to try and find that

[@venice__waves]

I had to pause now at 18:20 and reflect, because of the beauty and the complexity of the statements that were made. Now I will return much later to the video in order to take it all in :))

[@makanani1014]

I always thought of emotions as coordinates within our brain maps. One could see how emotions have many more permutations and can be tied to specific memories. Ptsd is like a heavily entangled memory to a strong emotional response. Just food for thought.