A Thousand Brains: A New Theory of Intelligence
- Downloads：2457
- Type：Epub+TxT+PDF+Mobi
- Create Date：2022-01-16 07:16:08
- Update Date：2025-09-23
- Status：finish
- Author：Jeff Hawkins
- ISBN：B09QB56D18
- Environment：PC/Android/iPhone/iPad/Kindle
- Download
Summary
An author， neuroscientist， and computer engineer unveils a theory of intelligence， of understanding the brain and the future of AI。 For all of neuroscience's advances， we've made little progress on its biggest question: How do simple cells in the brain create intelligence？ Jeff Hawkins and his team discovered that the brain uses maplike structures to build a model of the world-not just one model， but hundreds of thousands of models of everything we know。 This discovery allows Hawkins to answer important questions about how we perceive the world， why we have a sense of self， and the origin of high-level thought。
Reviews
Delson Roche
，
I love reading stuff which is completely new to me， and this book was exactly that。 I enjoyed the first part， which explains how the human neocortex works， remembers and recalls。The second part was about how AI system works and how can it be made similar to the human mind from its current structure。The third part is about the future of AI。 The first part is what I connected to the most and was a wonderful learning experience。
B
，
Jeff Hawkins is an entrepreneur who wants to explain the human brain， "the most complex thing in the whole universe"， as he describes it in his introduction。 Even though there are "thousands of papers written every year"， he claims to have finally found the theory that connects all the dots。。。I think there is a high bar in intelligence and humility when it comes to amateurs wanting to revolutionise a field。 As so many other Silicon Valley kind entrepreneurs， Jeff Hawkins seems to try to make up Jeff Hawkins is an entrepreneur who wants to explain the human brain， "the most complex thing in the whole universe"， as he describes it in his introduction。 Even though there are "thousands of papers written every year"， he claims to have finally found the theory that connects all the dots。。。I think there is a high bar in intelligence and humility when it comes to amateurs wanting to revolutionise a field。 As so many other Silicon Valley kind entrepreneurs， Jeff Hawkins seems to try to make up his lack of the latter with the former plus a pinch of bravado。 I don't buy it。My problems are not with how Hawkins present his own persona。 He his certainly successful and a certain braggishness is a character flaw that is anoying but not really important in a book。 Yet， his theory seems underdeveloped， relying on flashy， yet overly simplistic metaphors instead of reasoning， leaving wide gaps between the layers of abstraction he aims to connect （the actual challenge in cognitive science）。 This choice is strange， as he throws away the second half of the book for wild speculation and awkward narratives。 What is even more frustrating is that he doesn't give the reader any threads to pull at。 Hawkins seems to genuinly believe he was the first to think the way he does， that all "his" ideas are uniquely his。 It is possible that he drew many of the connections himself， but this should have suggested to him even more that others did so likewise， that his ideas are part of the vague common knowledge in the field he is working in， and that the real contribution （especially of a pop sci book like the one he is writing） lies in illuminating the important works of and debates between others and drawing the lines for those who are only visitors in his world。 He has nothing of this。 Somehow Dawkins ended up writing the introduction， but appart from that it almost seems as if Hawkins is the sole hero working in the field （except of his mentor， of course， who died before showing him the full power of the force。。。 eh。。。 the full depths of his insights）。 But I get it， hero's journey-kind stories are what sells these days， and are probably what you get when A/B-testing your pitch TED-style。 。。。more
Sivakumar Thulasimani
，
If your expectation is to learn about the author and how he has a possible breakthrough then you will be fine with book。If your expectation was to learn about empirical information about brains and intelligence it’s scarce here and no experiments or evidence for anything presented here worth the time 。 Dropped it after part 1 of this book。
Jodie Madden
，
I thoroughly enjoyable book， providing many concepts that are new to me and considerations around their application in the real world in the near or distant future。
Alessandro Berardi
，
The first part， which presents the Thousand Brains theory， is fascinating and holds a lot of potential for neuroscience and AI。 The rest of the book， however， tends to present a lot of opinionated speculation， which， while interesting and thought-provoking， is less valuable。
Nayan Barua
，
nice
Ricardo Acuña
，
I have a mixed feeling about this book。 The first part was interesting and informative for me， which talks about how the neocortex works from the point of view of the theory of a thousand brains。 However， the second and third parts are rather speculative since Hawkins writes about futuristic scenarios of artificial intelligence and human intelligence。 （Even though Hawkins admit that he doesn´t like science fiction） For me only the first part is worth it。 The second and third parts do not convinc I have a mixed feeling about this book。 The first part was interesting and informative for me， which talks about how the neocortex works from the point of view of the theory of a thousand brains。 However， the second and third parts are rather speculative since Hawkins writes about futuristic scenarios of artificial intelligence and human intelligence。 （Even though Hawkins admit that he doesn´t like science fiction） For me only the first part is worth it。 The second and third parts do not convince me and there are other authors who write and develop much better the subjects of artificial intelligence and its future。 I was expecting much more from this book， considering that I liked Hawkins's other book: "On Intelligence。"。 I think that "A Thousand Brains" is more about a popular science book intended for the general audience to publicize the research work that is carried out in Numenta research labs。 。。。more
Christian Adriano
，
Even if do not agree with them all， it is a good introduction to Hawkins ideas and how they related to current topics on AI。 I mostly focused on chapter 11 （existential risks of machine intelligence） and 14 （merging brains and machines）。
Stephen Smith
，
I think this is a case of a tech billionaire surrounding himself with yes men to tell him constantly how brilliant he is。 There is no one to tell him when he has a bad idea。 The book is annoying to read as every second paragraph is about how someone told him he is really smart， or right。 He presents a theory of intelligence in the neocortex which I'm sure has some merit and is part of the puzzle， but he then claims it does everything and to prove it he defines intelligence as something the uses I think this is a case of a tech billionaire surrounding himself with yes men to tell him constantly how brilliant he is。 There is no one to tell him when he has a bad idea。 The book is annoying to read as every second paragraph is about how someone told him he is really smart， or right。 He presents a theory of intelligence in the neocortex which I'm sure has some merit and is part of the puzzle， but he then claims it does everything and to prove it he defines intelligence as something the uses his algorithm and then disparages other theories as non-intelligent because they don't use his algorithm。 He is adamant that intelligence is limited to mammals because they have a neo-cortex， disregarding the intelligence of cephalopods and many birds such as crows and parrots。 He seems ignorant of modern computational complexity theory， chaos theory and emergent behaviors。 Then the third part of the book goes off the rails， I guess Hawkins is worried about his legacy， especially if the human race goes extinct， so we need to send this book to nearby aliens or something so they can see how brilliant Hawkins is。 Anyway， I don't recommend this book。 。。。more
Yudhishtir
，
The book takes a road downhill when it takes an exit from the science theory to the personal half baked theorisations of what intelligent men and ordinary men want different always classifying his self as the intelligent kind。
Ven Sio
，
This book is divided into three parts。Part 1 is a well-written explanation of his theory。 Part 2 is an application of his theory on AI development。 I like his idea that AGI can be built using the same framework he theorizes in Part 1。 An interesting leap for me that he thinks that building it like our own brain will result in the creation of true machine consciousness。 Part 3 is a bit of rant on what he considers to be False Beliefs and the existential threats it presents。 I prefer Yuval Noah Ha This book is divided into three parts。Part 1 is a well-written explanation of his theory。 Part 2 is an application of his theory on AI development。 I like his idea that AGI can be built using the same framework he theorizes in Part 1。 An interesting leap for me that he thinks that building it like our own brain will result in the creation of true machine consciousness。 Part 3 is a bit of rant on what he considers to be False Beliefs and the existential threats it presents。 I prefer Yuval Noah Harari's take on this and the whole body of Jordan B。 Peterson's work in psychology。 Then the rest of the chapters before the end is a speculation on how we can preserve humanity/genes/knowledge。 There's some interesting takes on it， but it does wander away from the A Thousand Brains topic and acts more as his proposal for the future。 All in all， a stimulating read。 Find a good buddy to talk about it with afterwards。 。。。more
Daniel Gorin
，
The most important book ever written。
José Miguel
，
A really good book。。。 It covers a variety of topics， from grounds of neuroscience to good insights into the future of AI。 I think it is worth the time。。。
Dmitriy Nevzorov
，
The first part of the book is definitely worth reading。 I wouldn’t spend time on the second part though。
Stoyan Minchev
，
The first chapter very she real theory。I like the first chapter。 This is the big breakthrough。 Afterwards are just thoughts and future predictions。 Two totally different subjects
Harry
，
an important book for knowing how we knowI felt the book's importance and scientific impact was most prevalent in Section 1， less os in Section 2， and mostly speculative in Section 3。 Personally I enjoyed part 1 the most。 It contains the information and ideas that Hawkins is most qualified to answer。 All of it is highly readable to anyone with the interest and doesn't require a background in science， math， or biology。 an important book for knowing how we knowI felt the book's importance and scientific impact was most prevalent in Section 1， less os in Section 2， and mostly speculative in Section 3。 Personally I enjoyed part 1 the most。 It contains the information and ideas that Hawkins is most qualified to answer。 All of it is highly readable to anyone with the interest and doesn't require a background in science， math， or biology。 。。。more
Phoebe Phuong
，
Enthralling book that talks more than about how our brain works。 Jeff is truly is inspirational in his argument into a future where human intelligence will be more important for our survival。 Highly recommend。
Ravi
，
Reads like a white paperDescribes how neocortex is organized with fundamental building blocks and how it uses reference frames is the meat of the book。 A decent start in the first few chapters but loses focus in the later parts of the book
Josh
，
About 10 years ago I read the first book by Jeff Hawkins， On Intelligence。 It blew my mind。 Now， 10 years later I had the chance to reread his first book and then read his new book。 My impressions of On Intelligence were more mixed this time around。 Overall， I found it inspiring， but I saw more areas where he seemed to be too disconnected from other researchers to be taking he brought enough set of point of views into account。 Well， his next book A Thousand Brains completely resolved that。 More About 10 years ago I read the first book by Jeff Hawkins， On Intelligence。 It blew my mind。 Now， 10 years later I had the chance to reread his first book and then read his new book。 My impressions of On Intelligence were more mixed this time around。 Overall， I found it inspiring， but I saw more areas where he seemed to be too disconnected from other researchers to be taking he brought enough set of point of views into account。 Well， his next book A Thousand Brains completely resolved that。 More than ever， Jeff Hawkins tied his theories into the latest research on deep learning and neuroscience， and he also situated his hypotheses in the cultural agita around artificial intelligence。 This book was exciting and it will certainly leave an imprint on my own thinking。 I can't wait to reread it， soon！ 。。。more
Frederik
，
Part 1 claims to explain intelligence* Focus on the neocortex， labeled the new brain （vs the 30% old reptilian brain）* Basic units are cortical columns which act as universal predictors and are trained through feedback loops* 1。000 columns predict and vote in parallel* Prediction/understanding requires "reference frames" aka contextFascinating topic， provocative ideas， very interesting read。 BUT: the author seems over-confident / self-inflating， relying on very few sources besides himself - this Part 1 claims to explain intelligence* Focus on the neocortex， labeled the new brain （vs the 30% old reptilian brain）* Basic units are cortical columns which act as universal predictors and are trained through feedback loops* 1。000 columns predict and vote in parallel* Prediction/understanding requires "reference frames" aka contextFascinating topic， provocative ideas， very interesting read。 BUT: the author seems over-confident / self-inflating， relying on very few sources besides himself - this triggered my cross-check impulse。 After some research （see eg Robert Thibadeau） i think this "theory of intelligence" is rather an additional glimmer on the topic than ground-breaking understanding。 And probably many presented ideas are not new in the computational cognitive neuroscience community。 Nevertheless intriguing to me！Part 2 and 3 on AI and human future are a collection of personal viewpoints which i only skimmed through quickly。 。。。more
Hao Hu
，
As a former biologist who has been working in academia for almost a decade and a current AI scientist， I found some of the ideas stimulating。 However， for anyone with a formal training in science， here's my suggestion: lower your expectation if you want to enjoy this book。 It's not a science paper， a survey of papers， or even a solid theory。 The book is more like a collection of essays， which presents some half-baked ideas on neurology and intelligence。 If you take it too seriously， you'll likel As a former biologist who has been working in academia for almost a decade and a current AI scientist， I found some of the ideas stimulating。 However， for anyone with a formal training in science， here's my suggestion: lower your expectation if you want to enjoy this book。 It's not a science paper， a survey of papers， or even a solid theory。 The book is more like a collection of essays， which presents some half-baked ideas on neurology and intelligence。 If you take it too seriously， you'll likely be very disappointed。 However， I treat my reading experience as reading an internet forum。 Most stuff are probably known or wild speculations， but occasionally there are something sparkling。 However， I do want to point out a few key flaws about the book which may upset some science readers。1。 The so-call "thousand-brain" theory is a basically three ideas which has only some very preliminary supporting evidence， and they are by-no-means the "secrete-ingredient" of machine intelligence source that Jeff claimed them to be。 For example， a） the idea of many cortex columns working together to improve decision making， is not a novel idea at all。 In machine learning field， people call this "ensemble-model"。 And it has been very widely used for the design of model algorithms。 b） I failed to see the relevance of so-called "reference frame"。 Isn't that just a fancy name for "memory"？ And doesn't everyone already know our brain recognize the world in the context of existing memory？ c） The author claims that our brain learns by "moving"， but defined "moving" so vaguely that it can be anything from moving your fingers， changing your body position and making a mathematically deduction。 This makes me question how concrete Jeff's theory really is。2。 Sometimes （more often than it should） Jeff makes logical deduction that are from naive to laughable。 In one chapter， he gave a lot criticism over existing definition of "consciousness" in animals or machine。 Then he made his own definition: being able to create experience and later play back the experience。 By this definition， my cam recorder or even paper notebook would be qualified as "conscious being"。3。 His treatment of the "existential threat of AI" chapters seems to careless。 First of all， the part （basically part 3） is only very loosely related to the so called "thousand brain theory"。 Second， he claimed that we should not worry about machines because 1） they cannot replicate themselves and 2） they don't have an "old-brain" counterpart of human to make them greedy or fear。 The discussion ends here， which seems superficial。 I'd like to see why he thinks about the possibility of some people granting an "old-brain" to AI。My apology for the harsh review。 However， I do feel this book has some interesting points but overall unsatisfactory for a scientific audience like me。 。。。more
Alina Mitruţ
，
Definitely a complex read， and， must certainly， a MUST read。 It's remarkable how the author instructs the reader about the differences between the old， and the new brain， and the repercussions we would face depending on which of the two is in charge。The reference frames theory， through which - apparently - we learn models of the world， even though is mind-boggling， it made it so much clearer to me how the brain actually works。 Checkpoint。 ;）The author also redefines what I thought machine inte Definitely a complex read， and， must certainly， a MUST read。 It's remarkable how the author instructs the reader about the differences between the old， and the new brain， and the repercussions we would face depending on which of the two is in charge。The reference frames theory， through which - apparently - we learn models of the world， even though is mind-boggling， it made it so much clearer to me how the brain actually works。 Checkpoint。 ;）The author also redefines what I thought machine intelligence means， and I found very appealing both his perspective on downloading our brains into machines， and his take on focusing more on the survival of our intelligence than to our species， though， ideally， we could do both。 I always love a book full of promises， that leaves you wondering， and this one is immensely rich in provocative ideas！ 。。。more
Anibal Diaz
，
Perhaps is the most interesting scientific book I have read in the last ten years。 Challenging intelligence， genes， and biology， to preserve knowledge as the most valuable asset of the civilization。
Bhuwan Chand
，
A wonderful read， opens up mind to so many possibilities that exist for the future of humanity。 Which path we would take - the one lead of the thought of saving our Genes or the one which leads to saving the Knowledge humanity is accumulating continously。 Thats a mystery for the future。
Kaveh
，
This was the third book I read about AI （although this book is not exclusively about AI） and the first one that's written by a Neuroscientist and an actual AI expert。It definitely has a more optimistic tone about the future of AI。 The other books I read about this were "Superintelligence" by a phosphor and "Life 3。0" by an astrophysicist。There is night and day difference between the way their projections。 I think this debate is essential regardless of who is right。 We need to be prepared and hav This was the third book I read about AI （although this book is not exclusively about AI） and the first one that's written by a Neuroscientist and an actual AI expert。It definitely has a more optimistic tone about the future of AI。 The other books I read about this were "Superintelligence" by a phosphor and "Life 3。0" by an astrophysicist。There is night and day difference between the way their projections。 I think this debate is essential regardless of who is right。 We need to be prepared and have some ground rules。 I think by addition of more technical knowledge about how intelligence actually works Hawking's argument is more agreeable and I hope he's right。I hope to see more books （and read more books） written by AI experts and others。The rest of this book （religion， etc。） although still good didn't really interest me much。 。。。more
Peter Molander
，
Great insights， and also a very thoughtful summary at the end。 One example of the insights， is the fact about the neocortex being not one area but rather 150，000 mini columns with almost identical structure/ content。
Gede Manggala
，
Great book！ Bring you to wander and imagineGreat book！ As a layman in this field I enjoy the story and wonder shared by the author。 The clarity and the flow from paragraph to paragraph， chapter to chapter is very easy to follow。 A lot of great insights new to me and also very helpful in some specific definition， e。g。 what is intelligence
Shantanu Kumar
，
A Thousand Brains will definitely keep you up at night， but not always for the right reasons。 The human mind is a complex “machine”， with trillions of connections between cells and so much more going on。 The book is an excellent introduction to the field of neuroscience， and it will definitely inspire you to learn more about the fascinating subject of how the human brain works。 The sheer cogs and wheels involved in even the most simple of tasks will blow your mind， and it will make you stop and A Thousand Brains will definitely keep you up at night， but not always for the right reasons。 The human mind is a complex “machine”， with trillions of connections between cells and so much more going on。 The book is an excellent introduction to the field of neuroscience， and it will definitely inspire you to learn more about the fascinating subject of how the human brain works。 The sheer cogs and wheels involved in even the most simple of tasks will blow your mind， and it will make you stop and think about just how amazing our brains really are。 The book is divided into three distinct parts， all united by a single theme: **old brain， new brain。**According to Hawkins， the old brain is what harbours our desires cultivated through years of natural selection， more primal in nature。 The new brain is the thinking/reasoning-brain， that houses our learned knowledge of the world。**Part 1** of the book lays the foundation of his new theory of intelligence， updating an age-old theory of hierarchical learning with the Thousand Brains theory and summarising a lot of past work。 In classic Hawkins’ fashion of oversimplification， I’d say the theory replaces the idea of “one part of the brain doing all the processing” to “each cortical column is a learning-machine in itself”。 The language is easy to follow if you’re new to the fields of neuroscience and AI/Computer science， therefore accessible， but missing and incorrect on a few technical details。 On further reading， I found that practically most of the ideas presented in the book are before Hawkins’ time， which often makes it like he’s taking credit for something that isn't his。**Parts 2 and 3** are an absolute drag wherein he changes gears to turn philosophical about AI， the risks to human-kind， and what goals should be important to conserve our race。 The number of over-reaching assumptions and random musings present in this section make me believe that this was just written to “sell” the other parts of the book and latch on to the wave of what’s trending and what’s not。 With little to no link to the previous section of the book， Hawkins is convinced that the only route to intelligent AI is through reverse-engineering of the brain， and that AI *cannot* be dangerous till the time that happens。 > Once again， intelligent machines will not have human-like emotions and drives unless we purposely put them there。 Desires， goals， and aggression do not magically appear when something is intelligent。He doesn’t spend enough time addressing the concerns that other great thinkers have put out re: the risks of AI （eg。 Bostrom）， and does not clearly explain why programming tasks or goals for a machine may not generate desires （mostly because he posits that’s how the human brain works， and that’s how machines HAVE to work）。 We don’t even need intelligent or strong AI for it to be dangerous， but Hawkins dismisses these possibilities。## Future + Connected Reads1。 [**Superintelligence: Paths， Dangers， Strategies**] by Nick Bostrom2。 [**Thinking， Fast and Slow**] by Daniel Kahneman3。 [**I， Robot**] by Isaac Asimov4。 [**The Unabomber: Industrial Society and Its Future**] by Theodore Kaczynski。5。 [**Final Thoughts on Free Will**] by Sam Harris6。 [**Chaos: Making a New Science**] by James Gleick。 7。 [**The Consciousness Instinct: Unraveling the Mystery of How the Brain Makes the Mind**] by Michael Gazzaniga。8。 **[Homo Deus: A Brief History of Tomorrow]** by Yuval Noah Harari。 。。。more
Barbara
，
A very interesting walk through the science of how intelligence arises from the brain。 The author is a neuroscientist who has contributed substantially to the field。 He comes across as somewhat arrogant but also driven by deep curiosity。
Satish Antony
，