Most Helpful First | Newest First
5.0 out of 5 stars a fascinating way of looking at intelligence,
This review is from: On Intelligence (Kindle Edition)
This is a wonderful book. A way of looking at the brain not as a computer at all but as a memory system. Unlike other books which emphasise the Swiss arm knife view of the brain (lots of different bits doing their stuff) the author argues that the whole neocortex basically works by running the SAME algorithm. One that uses stored invariant memories to make predictions. That is, unlike a computer the brain solves problems not by computing the solution step by step but by recalling stored memories of the experience. The author gives the example of catching a ball. Whereas a robot will have to try to calculate moment by moment the trajectory (which is so complicated not robot can at the moment - which should be a good clue that this is probably not what the brain is even attempting to do, since neurons are orders of magnitude slower than even a laptop), the brain simply recalls stored patterns of muscle commands activated by the sight of the ball. The author argues that these memories are "invariant" in form. That is, as the details of the experience get passed up the neural hierarchy (he explains this in clear and fascinating detail) the details are slowly left behind and a general sort of "overall" memory is left (my words, not his - he is far clearer!). We don't store all the details of our experiences, rather the relationships between the stimuli. Another example of his is our memory of music. We remember a pitch invariant version of a song. By went confronted by actual input, a note of the song or the sight of the ball, we predict what is going to happen next by using the stored relational template. The cortex runs an algorithm which constantly uses these stored invariant memories to identify patterns in our input and make suitable predictions. It is a simple theory or model. I loved the book because it is the first book I have ever read about the brain which has given a plausible theory of what the brain is actually doing. As the author says again and again, it is a first draft of a theory and may turn out to be wrong but I can't help feeling that the answer, whatever it is, will look something. Like this.
5.0 out of 5 stars Excellent and thought provoking,
I love the way that the Hierarchical Temporal Memory of the neo-cortex is reverse engineered, and how it matches up so well will observations in behavioural economics made many years later in books such as Daniel Kanehman's Thinking:Fast and Slow.
Had a look at the respective open source libraries, but this is still an early stage technology.
Loved the bit later on on what Jeff thinks consciousness is.
5.0 out of 5 stars Best book on theory of the mind/brain,
This is simply put the best book on AI, the mind and approaches to replicating intelligent machines that I have ever read. It is a seminal piece, and will no doubt have left its mark on the AI landscape in years to come.
5.0 out of 5 stars This book blew my mind,
Amazing book. The main argument of the book is that neuroscience is too focused on collecting data and not focused enough on making a theory of the mind.
4.0 out of 5 stars A most likely theory about the neocortex,
Not only does Jeff Hawkins put forward a very attractive model of the workings of the cortex, he puts his theory in language suitable for general reading. In step with the approach of the engineer/scientist he is he offers various ways to test his theories. I warmly recommend this informative and well written book.
5.0 out of 5 stars How and why the brain works,
I read this book when it was first published (in 2004) and recently re-read it while preparing for an interview of one of countless thought leaders who have acknowledged their great debt to Jeff Hawkins for what they have learned from him and, especially, for what they learned from this book. Written with Sandra Blakeslee, this book provides a superb discussion of topics that include
o Artificial intelligence
o Neural networks
o The structure and functions of the human brain
o A "new framework of intelligence" (more about that later)
o How the cortex works
o Consciousness and creativity
o Hawkins' thoughts about the future of intelligence
As Hawkins explains, his goal "is to explain [his] new theory of intelligence and how the brain works in a way that anybody will understand." However, I hasten to add, this is not a book written for dummies and idiots who wish to "fool" people into thinking they know and understand more than in fact they do.
Early on, Hawkins acknowledges his skepticism about artificial intelligence (AI) for reasons that are best explained within his narrative, in context. However, it can be said now that after extensive research, Hawkins concluded that three separate but related components are essential to understanding the brain: "My first criterion was the inclusion of time in brain function...The second criterion was the inclusion of feedback...The third criterion was that any theory or model of the brain should account for the physical architecture of the brain." AI capabilities, Hawkins notes, are severely limited in terms of (a) creating programs that replicate what the human mind can do, (b) must be perfect to work at all, and (c) AI "might lead to useful products, but it isn't going to build truly intelligence machines."
The material in Chapter 7, "Consciousness and Creativity," is of special interest to me as I continue to read recently published books that offer breakthrough insights on creativity, innovation, and the processes by which to develop them. (The authors of many of those books, to borrow from a 12th century French monk, Bernard of Chartres, are standing on Dawkins' "shoulders." It must be getting crowded up there.) Hawkins asserts that creativity does not require high intelligence and giftedness, and defines creativity as "making predictions by analogy, something that occurs everywhere in cortex and something you do continually while awake. Creativity occurs along a continuum...At a fundamental level, everyday acts of perception are similar to rare flights of brilliance. It's just that the everyday acts are so common we don't notice them." I call this phenomenon "the invisibility of the obvious."
I am among those who are curious to know the answers to questions such as "Why are some people more creative than others?" ""Can you train yourself to be more creative?" "What is consciousness?" and "What is imagination?" Hawkins has formulated answers to these and other questions and shares them in this chapter. He concludes the book with eleven predictions and #8 caught my eye: "Sudden understanding should result in a precise cascading of predictive activity that flows down the cortical hierarchy." In other words, revelations (whatever their nature and scope) help us, not only to connect dots but to connect those that are most important.
5.0 out of 5 stars Read this book,
If you think computers will one day become intelligent you should read this book. If you think computers won't one day become intelligent you should read this book. If you are interested in just how the brain takes sensory input and creates memories and actions you should read this book.
On Intelligence is such a good read for anyone interested in computer intelligence. Hawkins is a computer man talking about biology, analysing how the brain makes sense of the world. To what extent his theory about the neo-cortex's role in intelligence is accurate I cannot judge and only time will tell. But for this part-time programmer the operation of the brain now seems considerably less mysterious and the path to artifial intelligence seems a lot clearer.
One of the best books I have read for a long time.
5.0 out of 5 stars Excellent book,
This book is easy to understand (not too technical, but enough) and very convincing, as the presented theory makes a lot of senses explaining how the brain works. It was written in 2004, so now Jeff Hawkins has more to show (great talks available on his website).
5.0 out of 5 stars On On Intelligence,
It was the enthusiasm of other Amazon reviewers that caused me wonder to if this book might have some genuinely novel insights into the tangled mystery of the brain on offer. I actually took time out from my slow, careful plodding through a more hefty and substantial neuroscience text, Rodney Cotterill's Enchanted Looms: Conscious Networks in Brains and Computers, to indulge myself in a quick bit of light relief. How glad I am that I did. Given the subject matter, it is a marvellously easy little book to read, albeit there is a small section in which Hawkin's outlines his theory of the operation of the mammalian cortex that does require a more careful reading. He states from the start that most of the ideas in the book have been around for a while in various forms, and that what he provides is a synthesis that sets those ideas into a new framework.
The broad essence of his argument is based on the observation of Vernon Montcastle that the mammalian cortex has a uniform global and microscopic structure. The cortex is the crinkly sheet that we see when looking at the brain from above and the sides, and that is wrapped around the more evolutionarily primitive inner components. A possible corollary of this observation is that `cortex is cortex', and that it is all implementing the same highly generalised processing algorithm. This is a rather counter-intuitive proposition as it would seem reasonable that the brain is doing a diversity of things and is therefore using a diversity of mechanisms to accomplish them. A vertical section through any part of the cortex reveals it to be comprised of six layers, each with a distinct composition of types and densities of neurones, and synaptic interconnections. Closer examination shows that these neurones are organised into a semi-astronomical number of transversely arranged microcolumns, with many interconnecting vertical synapses between the constituent neurones working to make each microcolumn into a tiny processing unit. Microcolumns operate together to make the functional areas that neuroscientists have been mapping in ever greater detail over the last century or so. These areas or regions are interconnected in a complex, but highly organised way, to establish a hierarchy in which areas connected to sensory inputs are at the bottom, and areas of increasingly abstract association are towards the top. The puzzling fact that there are more backward connections flowing down this hierarchy of areas, than there are forward/upward connections, has been known for a long while, but has arguably been largely ignored. This connectivity can be understood however in the light of Hawkin's proposed 'memory-prediction framework'. According to this model the brain's operation, and the essence of intelligence, consists of higher cortical areas constantly seeking to predict what patterns will be encountered next in the lower areas to which they are connected. These predictions are based on comparisons between memory, that is the cumulative analysis of previous patterns, as extracted by blind and simple algorithms, and the patterns of current input. Hawkin's thus argues that each area of the brain is constantly trying to anticipate its future inputs from its lower areas. Where such prediction fails we have the experience of surprise or novelty, and attention on behalf of areas further up the hierarchy is required in order to subsume that input under existing patterns, or to derive new patterns. Such new patterns will cause changes to flow up and down the hierarchy, this process being learning. He even argues that movement, as a result of activations in the motor cortex, is implemented in the same terms. Thus we actually move by anticipating the sensory inputs from our bodies, including the vestibular (balance), proprioceptive (disposition of the body in space), etc. that will arise as a result of issuing motor signals, and that it is these predictions themselves that drive the motor areas. He goes on to propose a reasonably detailed description of how this pattern-predicting model might be implemented down at the level of microcolumns and the synaptic connections between the neurones in the six layers.
For such an easy to read little book this is quite an extraordinary hypothesis that, at a stroke, makes a great deal of sense out of a mountain of baffling detail. If Hawkin's has achieved nothing else it is to demonstrate ways of thinking and writing about neural architecture that are more transparent and intuitive than has arguably been accomplished thus far. I am going to have to spend a while thinking about his theory, and considering whether his model really does capture everything that the cortex, and the generalised intelligence that gives us knowledge, skills, reasoning, language and so on, does for us. I have returned now to the Cotterill book and already I am finding myself thinking about what I am reading in a rather new and different way. Time will tell whether Hawkin's theory will turn out to be a master key that will bring some overarching sense to the mass of messy detail that my current knowledge of the brain presents me with. Time will also tell how his predictions about intelligent machines and the social revolution they could engender will transpire. That such machines are possible, and will be built I have no doubt. How long it will take is rather trickier. However, when they finally arrive it may be that we come to look back on this little book, which is as much a pamphlet or manifesto, as a milestone in intellectual history.
5.0 out of 5 stars The most fascinating book I've read about the brain to date,
If you're curious about the brain then this book gives a very unique insight into a theory for how it works, one that really makes sense and can explain a lot of unanswered questions. It really is one of the best books I have ever read. I've read several books on the topic and this one is by far the most thought provoking.
Most Helpful First | Newest First
On Intelligence by Sandra Blakeslee