From the Bookshelf of Science and Inquiry…

A Thousand Brains: A New Theory of Intelligence

Jeff Hawkins,

Richard Dawkins (Foreword)

Start date: July 1, 2022

Finish date: July 31, 2022

Discussion: Book Club 2022

Find A Copy At

Libraries

Group Discussions About This Book

Showing 2 of 25 topics — 446 comments total

Book Club 2022

November 2022 - Rise of the Mammals
By Betsy , co-mod · 15 posts · 98 views

last updated May 05, 2024 12:39AM

December 2022 - Immense World
By Betsy , co-mod · 29 posts · 122 views

last updated Jan 28, 2023 09:23PM

showing 1 of 1 topics view all »

What Members Thought

Dec 15, 2022 Petra X marked it as 1-tbr-owned-but-not-yet-read

Intelligence, to me, is the development of a reaction, probably chemical to start with. If a single celled creature, say an amoeba, is able to turn away from obvious danger, there must be a time where it is faced with two threats and no where else to go. Which one is the least dangerous? All creatures must be able to decide that. And to me that is where intelligence begins.

The book is more about the neurological structures in the brain that give rise to intelligence. I will get round to it one d ...more

68 likes · Like · see review

Aug 11, 2022 aPriL does feral sometimes rated it really liked it · review of another edition

Shelves: science, academic-notations, mostly-conversations-and-ideas, non-fiction

‘A Thousand Brains: a New Theory of Intelligence’ by Jeff Hawkins is a good overview of a lot of the questions many of us are asking and some theories by Hawkins to answer those questions. The questions I am talking about are: why do people believe in fake news and religions despite a lot of facts that proves their beliefs are incorrect? Why do most people, whether they believe in global warming or not, environmental concerns or not, continue to fill up the world with trash and pollution, using ...more

15 likes · Like · see review

Jul 19, 2022 David Rubenstein rated it it was ok

Shelves: evolution, science, technology

What a mishmash! The book starts out with Hawkins' hypothesis about the organization of the neocortex. Here, he even gets the subtitle of his book wrong; it is not a "new theory", but a "new hypothesis". For his idea to be a theory, it must be supported by an overwhelming preponderance of data. This simply isn't supported in the book.

Hawkins has an interesting biography. He founded Palm Computing, and designed the Palm Pilot and Treo. In 2002 he founded the Redwood Neuroscience Institute, now at What a mishmash! The book starts out with Hawkins' hypothesis about the organization of the neocortex. Here, he even gets the subtitle of his book wrong; it is not a "new theory", but a "new hypothesis". For his idea to be a theory, it must be supported by an overwhelming preponderance of data. This simply isn't supported in the book.

Hawkins has an interesting biography. He founded Palm Computing, and designed the Palm Pilot and Treo. In 2002 he founded the Redwood Neuroscience Institute, now at Berkeley. Then he founded Numenta, and independent research company. The goals of his company are twofold: 1) to understand how the neocortex works, and 2) to apply how the brain works to machine intelligence.

The first half of the book is ok, even fascinating at times. The focus is on how the brain makes many parallel predictions about the environment. The neocortex learns multiple models of the world, makes predictions based on those models. In order to learn about the world, to formulate models, you have to move. Something has to change in order to see, feel, or hear it from different aspects. Then there is a "vote" as to which of the models is most accurate.

The neocortex is a sheet about 2.5 mm thick. It is composed of about 150,000 columns through the sheet. There is very little physical difference between the various areas of the sheet, and very little difference between the columns. Each cortical column generates a model of physical objects. It generates a reference frame for each object and then makes sensory-motor predicitons. Cortical columns model reference frames not only for physical objects, but for abstract objects like math, politics, and language. A key aspect is that reference frames can refer to other reference frames, in a recursion. Language is full of recursions, as well as physical objects.

In 1978, Vernon Mountcastle proposed that a common algorithm unerlies all perception and cognition. The author claims that the common algorithm formulates the reference frames. The "Thousand Brains Theory" states that knowledge of any particular item is distributed among thousands of complementary models. A perception becomes stable, because the columns hold a vote.

So far, so good. I think that some reference to Bayesian reasoning would have helped Hawkins' case for his "theory". Bayesian reasoning and calculations could form the basis for holding multiple models in memory, and taking a vote, by combining memories with incoming sensory information. I hold this as a significant lack of insight on the author's part. (Hawkins; you should be talking to some physicists and mathematicians--I am trying to help you!)

Hawkins lists what he calls the 4 attributes of AGI (Artificial General Intelligence). They are:
1) Learn continuously
2) Learning via movement
3) Many models
4) Use reference frames to store knowledge
While these attributes may be necessary for intelligence, they are certainly not enough. I personally have written software systems to have all these attributes--including multi-level recursions--but I certainly would never claim that the software has "general intelligence".

Then the next portions of the book chase some ideas about intelligence, artificial intelligence, the search for other intelligences in the universe, and leaving our knowledge in permanent structures for posterity. It's all fascinating, but I got whiplash from all these ideas coming out of nowhere, without an attempt to adhere to an overall theme. What a mishmash! ...more

3 likes · Like · see review

Mar 27, 2021 Terri (BooklyMatters) rated it it was amazing

I love reading about the brain; about neuroscience, consciousness, and emotions, and there are a few books (non-fiction) I have read in this category that have absolutely changed my life. (At the top of this list is “The Brain That Changes Itself”, by Norman Doidge). And I now have another to add to this list - “A Thousand Brains: A New Theory of Intelligence” by Jeff Hawkins.

While I definitely did not agree with all of the authors suggestions, and found several of them downright terrifying (par I love reading about the brain; about neuroscience, consciousness, and emotions, and there are a few books (non-fiction) I have read in this category that have absolutely changed my life. (At the top of this list is “The Brain That Changes Itself”, by Norman Doidge). And I now have another to add to this list - “A Thousand Brains: A New Theory of Intelligence” by Jeff Hawkins.

While I definitely did not agree with all of the authors suggestions, and found several of them downright terrifying (particularly in Part Two and Three) this book was mind-blowing in it’s scope, opening my eyes (and my thousand brains) to a whole new way of thinking about intelligence - how it is constructed, modeled and stored in the human brain - and what this new thinking means for future machine intelligence (AI) and its potential and ultimate place in the long term survival of the human race. Heady stuff. (Pun intended!)

Written in a way that is much more accessible to the lay-person than a typical book on such subjects, the first part of the book describes a new theory on how intelligence is constructed, based on the neocortex (the wrinkled wrapper around the brain, a 2.5 mm layer that is the newest layer of the brain, evolutionally speaking). In this theory, Hawkins provides a fascinating look at neurons, the brains nerve cells, and how they are aligned along the neocortex in rows, but more importantly, as far as intelligence is concerned, in columns. These cortical columns, of which there are about 150,000, are each identically structured with general-purpose cells capable of modeling and manipulating objects or concepts, learning as they go, and using this learning to continuously form predictions. This is achieved through the use of reference frames, conceptual maps in essence, that are cross-tagged dynamically with thousands of objects as we learn and connect them to form our ongoing mental representation of reality.

Simulations of these cortical-column-based modelers can form the basis for new and evolving forms of machine intelligence, providing the much-needed impetus, Hawkins believes, to fulfill the promise of the exploding field of AI. This is explored in Part two of this astounding book in some detail, along with Hawkins thoughts on what truly intelligent machines, constructed in this way, (that is, self-learning and independent), can be used to achieve along with what risks to humankind may be involved in the wide-scale adoption of these super-robotic intelligent beings. This section is more speculative, less grounded on science and facts, and definitively more controversial, both in its content and its treatment of ethical questions. There were many areas here that made me cringe, ethically, particularly in the treatment of “consciousness”, including both what it is and what it implies. For example, - Hawkins determines there is no ethical quandary in turning off a “conscious” robot (assuming such a robot could be constructed) as they would have no feelings, (fear or survival instinct). In my mind, extending this premise to other conscious-but-not-feeling creatures leads to some pretty scary ethical decisions.

Hawkins closes with a discussion of the old-brain vs new conflict, the long term survival of the human race, human or robotic colonization of other planets, intergalactic communication, and our obligation to preserve knowledge as the legacy of the human race.

Phew.

This is a book to devour, think about, sift through the ideas, and take away some nuggets to be pulled out, again and again, to reconsider. In my case at least, a great many of those nuggets are mind-altering.

A big thank you to NetGalley, the publisher, and the author for an advance review copy of this book. All thoughts presented here are my own. ...more

Like · see review