What do you think?
Rate this book
The Computer and the Brain
John von Neumann is generally regarded as one of the greatest mathematicians in modern history.
His last work „The Computer and the Brain“ was begun shortly before his death in 1957 and compares the human brain with the computing machines of his time. Von Neumann discusses analogies and differences with unparalleled clarity and precision, making his thoughts still relevant today.
His last work „The Computer and the Brain“ was begun shortly before his death in 1957 and compares the human brain with the computing machines of his time. Von Neumann discusses analogies and differences with unparalleled clarity and precision, making his thoughts still relevant today.
95 pages, Kindle Edition
First published January 1, 1958
167 people are currently reading
4045 people want to read
About the author
John von Neumann
78 books366 followersJohn von Neumann (Hungarian: margittai Neumann János Lajos) was a Hungarian American[1] mathematician who made major contributions to a vast range of fields,[2] including set theory, functional analysis, quantum mechanics, ergodic theory, continuous geometry, economics and game theory, computer science, numerical analysis, hydrodynamics (of explosions), and statistics, as well as many other mathematical fields. He is generally regarded as one of the foremost mathematicians of the 20th century. The mathematician Jean Dieudonné called von Neumann "the last of the great mathematicians." Even in Budapest, in the time that produced Szilárd (1898), Wigner (1902), and Teller (1908) his brilliance stood out. Most notably, von Neumann was a pioneer of the application of operator theory to quantum mechanics, a principal member of the Manhattan Project and the Institute for Advanced Study in Princeton (as one of the few originally appointed), and a key figure in the development of game theory and the concepts of cellular automata and the universal constructor. Along with Edward Teller and Stanislaw Ulam, von Neumann worked out key steps in the nuclear physics involved in thermonuclear reactions and the hydrogen bomb.
Ratings & Reviews
Friends & Following
Create a free account to discover what your friends think of this book!
Community Reviews
Displaying 1 - 30 of 101 reviews
February 6, 2020
Picked this up to read with a friend who said they were reading it so I'm still waiting for them to finish it. I think it was really cool reading for understanding the advent of computer architecture and where it was at this point and seeing computers compared to the brain. This book combined all the things I'm really interested in lately and what I'd love to see would be an updated look at the ideas approached in the book. Like computers have gotten way smaller and faster and able to hold much more memory than at the time this was written so would it be fair to still make the argument that the brain computationally is more powerful? Also I thought it was interesting to think about the memory capacities and the different ways in which memory may be stored in the brain. I think the role of the Hippocampus was realized a little later, like the 70s, but it does help encode/consolidate memories and it was just interesting to see some parallels there with the stored memory in computers. Also really want to know more about the math specially the differential analyzer and the difference in the ways we understand/do math and the other possibilities which was kind of brushed upon at the end.
April 9, 2015
A short series of lecture notes written by the great mathematician and physicist John von Neumann while on his deathbed. The book is split into two sections, the first of which details early computing devices and their functioning. Essentially, logical "organs" are orchestrated in sequential patterns to perform intensive arithmetic and memory-related procedures. Here we find some of the first mentions of random access memory (RAM), short code (high-level programming), and other elements that are ubiquitous in modern computing.
The second portion of the book presents a balanced assessment of the computational nature of neurological systems. Much to my liking, von Neumann exhibits a healthy dose of skepticism while drawing comparisons between constructed and natural systems. After initially hazarding a hypothesis about the digital nature of neuron-axon-synapse transmissions, he proceeds to point out that these are, in actuality, complex systems that exhibit analog/continuous characteristics at the molecular level in which the mechanical, electric, and chemical realms begin to blur together. Though they possess considerable disadvantages in "clock speed" and precision, natural systems make up for their lack of logical depth by being massively parallel.
Despite vast improvements in our understanding of computer science and neuroscience, von Neumann's ideas hold up remarkably well.
4.0/5.0
The second portion of the book presents a balanced assessment of the computational nature of neurological systems. Much to my liking, von Neumann exhibits a healthy dose of skepticism while drawing comparisons between constructed and natural systems. After initially hazarding a hypothesis about the digital nature of neuron-axon-synapse transmissions, he proceeds to point out that these are, in actuality, complex systems that exhibit analog/continuous characteristics at the molecular level in which the mechanical, electric, and chemical realms begin to blur together. Though they possess considerable disadvantages in "clock speed" and precision, natural systems make up for their lack of logical depth by being massively parallel.
Despite vast improvements in our understanding of computer science and neuroscience, von Neumann's ideas hold up remarkably well.
4.0/5.0
December 27, 2020
Ho scoperto questo libro per caso, curiosando nella libreria di mia sorella.
Conosco l'autore, John von Neumann, per la teoria dei giochi e sono stata incuriosita dal titolo del libro.
E' un manoscritto incompleto in quanto Neumann morì prima di completarlo e in realtà lo scopo non era scrivere un libro ma tenere 7 lezioni sulla correlazione "Computer-Cervello".
Quando capì che non sarebbe riuscito a sostenere il viaggio e la fatica si dedicò alla scrittura dei concetti base ma - come anticipato - non riuscì a completare l'opera.
Il testo si compone di circa 140 pagine; nelle prime 50 pagine troviamo la prefazione e l'introduzione. Prefazione - alla seconda e alla terza edizione - molto interessante perché è stata scritta negli scorsi anni ed è quindi attuale.
Neumann nelle prime righe scrive subito che lui è un matematico e quindi l'analisi del cervello che farà non sarà dal punto di vista psicologico/neurologico.
Essenzialmente ci sono due capitoli dove prima esamina la composizione e il funzionamento del computer e poi confronta, a grandi linee, i risultati ottenuti con il funzionamento del cervello.
Pur essendo stato scritto quasi 70 (!) anni fa i temi sono attuali, interessanti e spiegati bene.
Personalmente avrei apprezzato un maggiore approfondimento delle singole sezioni - che invece sono molto brevi, anche mezza pagina - ma capisco che lo scopo era dare uno sguardo d'insieme e non una visione al microscopio.
Lettura scorrevole, si legge in una/due giornate.
Conosco l'autore, John von Neumann, per la teoria dei giochi e sono stata incuriosita dal titolo del libro.
E' un manoscritto incompleto in quanto Neumann morì prima di completarlo e in realtà lo scopo non era scrivere un libro ma tenere 7 lezioni sulla correlazione "Computer-Cervello".
Quando capì che non sarebbe riuscito a sostenere il viaggio e la fatica si dedicò alla scrittura dei concetti base ma - come anticipato - non riuscì a completare l'opera.
Il testo si compone di circa 140 pagine; nelle prime 50 pagine troviamo la prefazione e l'introduzione. Prefazione - alla seconda e alla terza edizione - molto interessante perché è stata scritta negli scorsi anni ed è quindi attuale.
Neumann nelle prime righe scrive subito che lui è un matematico e quindi l'analisi del cervello che farà non sarà dal punto di vista psicologico/neurologico.
Essenzialmente ci sono due capitoli dove prima esamina la composizione e il funzionamento del computer e poi confronta, a grandi linee, i risultati ottenuti con il funzionamento del cervello.
Pur essendo stato scritto quasi 70 (!) anni fa i temi sono attuali, interessanti e spiegati bene.
Personalmente avrei apprezzato un maggiore approfondimento delle singole sezioni - che invece sono molto brevi, anche mezza pagina - ma capisco che lo scopo era dare uno sguardo d'insieme e non una visione al microscopio.
Lettura scorrevole, si legge in una/due giornate.
June 11, 2017
Perhaps more so than Alan Turing, (If we accept the intro by Kirtzweil, a man with serious credentials), John von Neuman was one of the most important figures in developing the basic architecture of the modern digital computer. The two did work together, but von Neuman was the senior and I propose had a better grasp of the juncture of math and machine.
The Computer and the Brain is the last published work by von Neuman and was an attempt to bring together what was known about the machine qualities of the brain and what the machines of 1957 might one day be able to accomplish.
As such almost everything is dated and new discoveries in neuroscience more so than in computers place limits on the absolute value of his comments.
That said, there was a head line in recent science news that there is a prototype computer in testing that combines both digital and large scale parallel computing in the manner von Neuman suggests as the model for how the brain may work.
The Computer and the Brain is the printed from of a lecture. He was too ill (mortally) for the series he was offered to conduct. So great was the respect for the man that he was allowed to present only these papers, sufficient for one lecture and about 3 hours reading. I do not think he read the paper; he had just the strength to write it.
Besides having been a vitally important mathematician, he was active in the cause of scientific ethics and as the man who drafted the letter, signed by Einstein credited with America committing to atomic research he is therefore a originator of the atomic age. He was a man of great thought and influence.
Reading this book is a chance to listen to a great mind. I make no claim to have understood all of it. I suspect that no one should read it in an effort to be at the leading edge of math, computers or neurology. It is a hard, but worth it read, and a glass into our recent history.
The Computer and the Brain is the last published work by von Neuman and was an attempt to bring together what was known about the machine qualities of the brain and what the machines of 1957 might one day be able to accomplish.
As such almost everything is dated and new discoveries in neuroscience more so than in computers place limits on the absolute value of his comments.
That said, there was a head line in recent science news that there is a prototype computer in testing that combines both digital and large scale parallel computing in the manner von Neuman suggests as the model for how the brain may work.
The Computer and the Brain is the printed from of a lecture. He was too ill (mortally) for the series he was offered to conduct. So great was the respect for the man that he was allowed to present only these papers, sufficient for one lecture and about 3 hours reading. I do not think he read the paper; he had just the strength to write it.
Besides having been a vitally important mathematician, he was active in the cause of scientific ethics and as the man who drafted the letter, signed by Einstein credited with America committing to atomic research he is therefore a originator of the atomic age. He was a man of great thought and influence.
Reading this book is a chance to listen to a great mind. I make no claim to have understood all of it. I suspect that no one should read it in an effort to be at the leading edge of math, computers or neurology. It is a hard, but worth it read, and a glass into our recent history.
May 13, 2008
Too short to really bring anything new to the table (save some tantalizing references to analog computing I need chase down in more detail), but also too short to piss you off or demand more than a half hour. More like a pamphlet than a real book, I was worried I might put this in my pocket and send it through the wash.
February 19, 2024
The midpoint of this short manuscript is the literary equivalent of the beat switch in Nights by Frank Ocean
April 23, 2019
An unbelievably insightful and tantalising look at the similarities and differences between computers and the human nervous system. At just over a hundred pages it can easily be read in a day, though some background knowledge of computing and neuroscience would help. Although von Neumann was writing in 1958, his thinking is just as relevant today as it was then. By today's standards the section on digital computers may seem quaint, but it is a succinct and interesting summary of the state of affairs then, and the explanations are easily extended to today's technology. The section on analog computers, by contrast, is fascinating, because it's easy to forget that they were such an important predecessor to digital machines.
The fundamental question of the paper is: "To what extent are we digital, and to what extent analog, and what might our understanding of computers tell us about the way our brains work?" The "winner-takes-all" nature of neurons firing and the binary nature of genetics imply that a large part must be digital, and yet the way neurons use time summation/frequency and several other mysteries about the statistical nature of our nervous systems mean we must be at least in part analog.
The massive increase in computing power in the meantime makes updating von Neumann's figures as an implicit exercise for the reader. Obviously computers have come a long way since then, while the brain hasn't, and yet virtually everything von Neumann writes still applies. The brain has far more neurons than any single computer has transistors, and neurons interact with so many other neurons in such complex ways (not just electrically but chemically, mechanically, spatially, temporally) that to treat them as equivalent to any single "active organ" (i.e. transistor) is a mistake. The questions of how memory works in the nervous system and the limitations of mathematical precision in the nervous system capable are equally fascinating, as is the question of parallel versus serial processing and their respective memory requirements.
In some ways, this book is interesting precisely because of its simplicity in comparing the two systems. Today, most people would dismiss the question by saying that the brain is too complex to compare to a computer. But why and how is this the case? That his writing is cogent and coherent for non-experts also makes this a great read. Highly recommended.
The fundamental question of the paper is: "To what extent are we digital, and to what extent analog, and what might our understanding of computers tell us about the way our brains work?" The "winner-takes-all" nature of neurons firing and the binary nature of genetics imply that a large part must be digital, and yet the way neurons use time summation/frequency and several other mysteries about the statistical nature of our nervous systems mean we must be at least in part analog.
The massive increase in computing power in the meantime makes updating von Neumann's figures as an implicit exercise for the reader. Obviously computers have come a long way since then, while the brain hasn't, and yet virtually everything von Neumann writes still applies. The brain has far more neurons than any single computer has transistors, and neurons interact with so many other neurons in such complex ways (not just electrically but chemically, mechanically, spatially, temporally) that to treat them as equivalent to any single "active organ" (i.e. transistor) is a mistake. The questions of how memory works in the nervous system and the limitations of mathematical precision in the nervous system capable are equally fascinating, as is the question of parallel versus serial processing and their respective memory requirements.
In some ways, this book is interesting precisely because of its simplicity in comparing the two systems. Today, most people would dismiss the question by saying that the brain is too complex to compare to a computer. But why and how is this the case? That his writing is cogent and coherent for non-experts also makes this a great read. Highly recommended.
May 12, 2014
A short and accessible historical gem that explores, well before its time, points of convergence and divergence between natural and artificial intelligence systems. It's particularly interesting as a meter stick of progress between the 1950's and the current day: whereas the computer side of the question is in some ways quaintly outdated, von Neumann's outline of the human nervous system and the functioning of individual neurons remains near the limit of what we understand even today. Not quite exactly at the limit, but certainly not as dreadfully short of it as 1950's computer science is to our current level of technological understanding. Psychologists often point to the youth of our field as if it were an apology for the slow progress-- still waiting for our Newton or our Einstein--, but the contrasts laid out in this book, as a modern reader, question the validity of the excuse. Computer science is younger still than psychology, yet the speed of knowledge acquisition dramatically outpaces that of psychology.
Nevertheless, as von Neumann outlines presciently, the points of divergence may overwhelm the points of convergence, rendering the comparison moot. The essay stands as a reminder of the limits of artificial intelligence when compared to biological intelligence: we may never be able to build real brains from artificial components running current computer architectures. I was especially piqued by his closing remarks that the mathematics of the nervous system may reflect different mathematics than the ones we currently understand. He compares our known mathematics to assembly language and the hypothesized neural mathematics to higher-level programming languages, suggesting that we may simply not have found the right method for interpreting the latter.
I really enjoyed the quick read and recommend it to anyone who works with computers, brains, or-- most especially-- both, as an important historical document.
Nevertheless, as von Neumann outlines presciently, the points of divergence may overwhelm the points of convergence, rendering the comparison moot. The essay stands as a reminder of the limits of artificial intelligence when compared to biological intelligence: we may never be able to build real brains from artificial components running current computer architectures. I was especially piqued by his closing remarks that the mathematics of the nervous system may reflect different mathematics than the ones we currently understand. He compares our known mathematics to assembly language and the hypothesized neural mathematics to higher-level programming languages, suggesting that we may simply not have found the right method for interpreting the latter.
I really enjoyed the quick read and recommend it to anyone who works with computers, brains, or-- most especially-- both, as an important historical document.
January 10, 2020
John von Neumann applied his exceptional rationality to everything: from computers to economics and to the brain.
I have a goal in my life, which I am aware of how unachievable it is, which is to describe the world mathematically. I am not sure what that means yet, but it is mainly about finding out whether the precision we have in physics, chemistry and biology, and try to apply that in the humanities. Many geniuses have tried that, in one way or another, including Gottfried Leibniz and Bertrand Russell.
Another name to include in that list is von Neumann. Applying rationality, science and mathematics to the brain is just beautiful. However, I think he did it in the wrong way. Not that he was stupid (or that I am ANY near his or anyone’s intelligence), but it was probably due to his times being different.
My issue with his approach is that he takes the time-sensitive parts of the computer (time to compute, time to complete a circuit, recover time...) and of the brain (synapsis duration, quantity...) and uses THAT as the means of comparison. Trying to understand the brain through such metrics (which are now 100% different for the computer, and will be much different in the future) seems useless to me. It tries to grasp the similarities between the two machines from a practical and mechanical point of view. I believe the beauty of trying to find the similarities between the computational machines we built and the computational machines God built (or we have, if you are an atheist) is at the abstract level. The same level as mathematics and logic. Not at the application (which varies, changes and renews year after year) level.
It is not the scale of speed/memory that matters when comparing the brain and Turing machines. It is the functioning of each system at the most abstract level possible.
Anyways, I love the topic and if you disagree or if you have books to recommend on the subject, I am all ears.
I have a goal in my life, which I am aware of how unachievable it is, which is to describe the world mathematically. I am not sure what that means yet, but it is mainly about finding out whether the precision we have in physics, chemistry and biology, and try to apply that in the humanities. Many geniuses have tried that, in one way or another, including Gottfried Leibniz and Bertrand Russell.
Another name to include in that list is von Neumann. Applying rationality, science and mathematics to the brain is just beautiful. However, I think he did it in the wrong way. Not that he was stupid (or that I am ANY near his or anyone’s intelligence), but it was probably due to his times being different.
My issue with his approach is that he takes the time-sensitive parts of the computer (time to compute, time to complete a circuit, recover time...) and of the brain (synapsis duration, quantity...) and uses THAT as the means of comparison. Trying to understand the brain through such metrics (which are now 100% different for the computer, and will be much different in the future) seems useless to me. It tries to grasp the similarities between the two machines from a practical and mechanical point of view. I believe the beauty of trying to find the similarities between the computational machines we built and the computational machines God built (or we have, if you are an atheist) is at the abstract level. The same level as mathematics and logic. Not at the application (which varies, changes and renews year after year) level.
It is not the scale of speed/memory that matters when comparing the brain and Turing machines. It is the functioning of each system at the most abstract level possible.
Anyways, I love the topic and if you disagree or if you have books to recommend on the subject, I am all ears.
January 30, 2015
It's really sad that this is not completed. It's like you just finish an appetizer and get thrown out of the restaurant. Anyhow, I love his writing and recommend it as an important text. He carefully analyze the computer-brain analogy in terms of memory, logic, arithmetic, capacity, precision and processing time etc. I wish all neuroscientists could do the same job before making certain assertions.
There are a few discussions on the functioning of the brain are especially interesting, e.g. (1) work of the brain relies on back and forth digital-analog conversions, which, I think, suggest current modeling efforts to look beyond complete digital simulations; (2) language of the brain is not the language of logic and mathematics as how we use it, otherwise nothing would work with its low arithmetic precision; (3) memory relies on the activities neurons and their genetics, but not sure where it's stored. - Probably just a few things I like to keep in mind as a student of the subject.
He spent substantial passages on memory, which also fascinated many greatest minds in the past. But as a random nerd, I find the analogs (not just by von Neumann) between computer memory and human memory the funniest of all time. We objectify some of our memories as physical markers in the computer, but are these markers also memories of the computers as they would call it? Maybe there is some asymmetry.
There are a few discussions on the functioning of the brain are especially interesting, e.g. (1) work of the brain relies on back and forth digital-analog conversions, which, I think, suggest current modeling efforts to look beyond complete digital simulations; (2) language of the brain is not the language of logic and mathematics as how we use it, otherwise nothing would work with its low arithmetic precision; (3) memory relies on the activities neurons and their genetics, but not sure where it's stored. - Probably just a few things I like to keep in mind as a student of the subject.
He spent substantial passages on memory, which also fascinated many greatest minds in the past. But as a random nerd, I find the analogs (not just by von Neumann) between computer memory and human memory the funniest of all time. We objectify some of our memories as physical markers in the computer, but are these markers also memories of the computers as they would call it? Maybe there is some asymmetry.
March 28, 2023
The books is the last writings of the great John von Neumann, which was going to be a lecture series in a week. The book is incomplete since he dies before finishing the notes. So a small incomplete book from Neumann... I bought it to read his last work.
The book is in two parts, hence the name. The Computer, which we start to gain knowledge about digital and analoge computers, how they work and what are needed for each. And The Brain, where we learn about the actions happening in brain for a computation.
Since the Turing paper applies for any physical system, Neumann tries to connect the two in this book, and show the alikeness and difference between them.
The book is in two parts, hence the name. The Computer, which we start to gain knowledge about digital and analoge computers, how they work and what are needed for each. And The Brain, where we learn about the actions happening in brain for a computation.
Since the Turing paper applies for any physical system, Neumann tries to connect the two in this book, and show the alikeness and difference between them.
June 23, 2018
This book is the manuscript of a lecture that Von Neumann planned to give at Princeton in the 50's. Owing to that, it is terse in a way, and that makes it a bit dry to read.
That said, it goes over the state of the art in computing in the middle of the twentieth century, and tries to draw parallels between the computer logic processing and memory hardware of the time, with the then nascent understanding how neurons worked.
It was a good history lesson of computers, though not comprehensive by any measure. Moreover, there were a few back of the napkin calculations that compared brain hardware with computer hardware which were fun to ponder.
The book is unfinished, because of the author's death before completing the book, so it does end abruptly. Towards the end of the book, he seemed to be going in an interesting direction, comparing software programming with information that is coded and translated in the mesh of neural connections in the brain; their similarities and differences as then understood. Would have been an interesting analysis had he gotten the chance to complete it.
That said, it goes over the state of the art in computing in the middle of the twentieth century, and tries to draw parallels between the computer logic processing and memory hardware of the time, with the then nascent understanding how neurons worked.
It was a good history lesson of computers, though not comprehensive by any measure. Moreover, there were a few back of the napkin calculations that compared brain hardware with computer hardware which were fun to ponder.
The book is unfinished, because of the author's death before completing the book, so it does end abruptly. Towards the end of the book, he seemed to be going in an interesting direction, comparing software programming with information that is coded and translated in the mesh of neural connections in the brain; their similarities and differences as then understood. Would have been an interesting analysis had he gotten the chance to complete it.
July 1, 2020
With all my respect for John von Neumann and his achievements in computer science, I need to admit that I am quite disappointed with 'The Computer and the Brain.' Against all my thoughts about the content of the book, it rather discussed existing models, a small crash course into hove computers work. Meanwhile, I considered to read about ideas that von Neumann would offer to the future of computing so-called manifesto or memorandum.
However, we should also note precise and accurate introduction to the world of computers for students who are new in computer science, whom I recommend to read it. Moreover, ideas and models that are described in the book are still used to build modern computers, which are, in fact, many times powerful comparing to 1950s computers.
However, we should also note precise and accurate introduction to the world of computers for students who are new in computer science, whom I recommend to read it. Moreover, ideas and models that are described in the book are still used to build modern computers, which are, in fact, many times powerful comparing to 1950s computers.
September 14, 2017
The book is very well-written; however lack of illustrative figures makes it hard to digest the content unless you already know about it. This is the only reason that I rate it one star less. The book might represent state of the art knowledge at 1956 however our understanding of operations of the brain has significantly changed. This book serves as a good read for historical comparison of our understanding of the brain and its similarity to modern computing machines.
July 13, 2017
John von Neumann's ideas was so influential that now a days they sound intuitive and plain. Makes you remember once again that even the ideas you take for granted needed to be comprehended and relished by someone(-where)(-how)(-time-ago). No need to read this.
September 25, 2018
Can't wait until I can understand this book.
November 30, 2019
Erfreulich einfach und lesbar. Die Hauptverständnisarbeit besteht im Übersetzen der Beschreibungen in heutige Begriffe. Es geht dabei meistens gar nicht um konkrete Technik, aber schon die Umschreibung der Funktion von elementaren Computerbestandteilen ist halt eine andere als heute. Bei manchen Dingen ("short codes", hä?) bin ich mir am Ende immer noch nicht sicher, was gemeint ist, aber ich bin zuversichtlich, dass sich das durch einen schnellen Blick in die Sekundärliteratur klären lässt.
January 12, 2025
A prescient mathematical examination of the early development of computers and parallel comparison of the natural computing structures of the human brain. Far ahead of its time, and a worthwhile read for anyone interested in understanding computation or neurology and how each can inform the other.
October 13, 2022
A very nice read on early history of electronic computing from one of its founding father.
January 16, 2023
Simply extraordinary to read in 2023 how much of what is written in this written in the 50s still holds today... What a visionary!
July 8, 2024
It was very hard to read, but it had interesting ideas to think about.
August 3, 2018
My 3-star rating has nothing to do with my awe at John von Neumann as a towering genius. I'm also not rating the three introductions to the book. Instead, I'm using strictly as a rating of my enjoyment of this unfinished lecture.
However, I would like to talk about the introductions to this Third Edition published in 2012. My favorite, by far, is Ray Kurzweil's. It is the most recent and the most fun to read. I was particularly thankful for his summaries of where our knowledge of computing and of the physiology of the brain have come since 1957. I also found his explanations of Claude Shannon's Mathematical Theory of Communication to be incredibly lucid. I have a couple of Kurzweil's books on my shelves and based on the strength of this introduction, I'll be bumping those up on the reading queue!
The second introduction was written in 2000 by the Churchlands. It is also helpful, covering similar ground to Kurzweil. It's short, but I found it quite dry.
The third introduction was written by Klara von Neumann and is an excellent summary of how the unfinished book came to be. Klara was John's wife and the introduction is particularly heartbreaking when you realize that it was written in September, 1957. John had finally succumbed to cancer in the February of that year.
In all, there are 51 pages of introduction and then 80 pages of the lecture material. Though von Neumann never lived to give the lecture, or even finish the written work, it still stands as a testament to his ability to absorb and understand vast amounts of material.
For the most part, this book is nothing more than a time capsule describing the very state of the art in computer architecture and brain physiology at that point in time. I'm not sure if we even know where he was heading with this material - would he have proposed avenues for simulating human brains with computers?
One thing that really stuck out immediately in the actual lecture material were the terms that von Neumann used to describe how computing architecture is organized: he refers to the different parts of a computer as "organs" such as the "memory organ", "input" and "output" organs. In fact, it's very interesting to consider the need, at the time, to describe in analogy to biological systems, the parts of computing that we now take for granted. Nowadays, we're probably just as likely to use computing terms as analogies to simplify the function of biological systems!
The von Neumann architecture of the 1950s is completely recognizable in today's computers. This struck me, in particular, when reading the section titled Modus Operandi of the Memory-Stored Control. In this section, he describes what we would now call the storage of programs in memory:
He then goes on to describe in-memory self-modifying code, which was in vogue in "the old days" (however you choose to define that) when programs were written for specific machines in assembly (or even bytecode!) and whenever there was a need for extracting extreme performance from limited hardware. It's a heady concept, but generally frowned upon for maintainability and "graspability" reasons.
At the halfway point, the computer lessons turn into biology lessons as we switch from the architecture of the computer to the architecture of the brain. It's all very carefully written, admitting gaps in our knowledge (then and now!) about how, exactly, certain parts of the brain actually function.
For me, the highlight of the brain portion is the end, starting with a section snappily titled "Arithmetical Precision or Logical Reliability, Alternatives."
The book ends with a general statement that "the nervous system appears to be using a radically different system of notation" than that of mathematics (and traditional computing) and then concluding that the "language" of the brain may be in layers (which remind me of the layers of network protocols we enjoy today: physical voltage and timing, data link, routing, error correcting, and ending at the application level.)
I'm not sure modern readers will "learn" from this book. But it's certainly worth a visit to read a description of modern computing at the moment time of its birth straight from one of its creators.
However, I would like to talk about the introductions to this Third Edition published in 2012. My favorite, by far, is Ray Kurzweil's. It is the most recent and the most fun to read. I was particularly thankful for his summaries of where our knowledge of computing and of the physiology of the brain have come since 1957. I also found his explanations of Claude Shannon's Mathematical Theory of Communication to be incredibly lucid. I have a couple of Kurzweil's books on my shelves and based on the strength of this introduction, I'll be bumping those up on the reading queue!
The second introduction was written in 2000 by the Churchlands. It is also helpful, covering similar ground to Kurzweil. It's short, but I found it quite dry.
The third introduction was written by Klara von Neumann and is an excellent summary of how the unfinished book came to be. Klara was John's wife and the introduction is particularly heartbreaking when you realize that it was written in September, 1957. John had finally succumbed to cancer in the February of that year.
In all, there are 51 pages of introduction and then 80 pages of the lecture material. Though von Neumann never lived to give the lecture, or even finish the written work, it still stands as a testament to his ability to absorb and understand vast amounts of material.
For the most part, this book is nothing more than a time capsule describing the very state of the art in computer architecture and brain physiology at that point in time. I'm not sure if we even know where he was heading with this material - would he have proposed avenues for simulating human brains with computers?
One thing that really stuck out immediately in the actual lecture material were the terms that von Neumann used to describe how computing architecture is organized: he refers to the different parts of a computer as "organs" such as the "memory organ", "input" and "output" organs. In fact, it's very interesting to consider the need, at the time, to describe in analogy to biological systems, the parts of computing that we now take for granted. Nowadays, we're probably just as likely to use computing terms as analogies to simplify the function of biological systems!
The von Neumann architecture of the 1950s is completely recognizable in today's computers. This struck me, in particular, when reading the section titled Modus Operandi of the Memory-Stored Control. In this section, he describes what we would now call the storage of programs in memory:
"In this case, since the orders that exercise the entire control are in the memory, a higher degree of flexibility is achieved than in any previous mode of control."
He then goes on to describe in-memory self-modifying code, which was in vogue in "the old days" (however you choose to define that) when programs were written for specific machines in assembly (or even bytecode!) and whenever there was a need for extracting extreme performance from limited hardware. It's a heady concept, but generally frowned upon for maintainability and "graspability" reasons.
At the halfway point, the computer lessons turn into biology lessons as we switch from the architecture of the computer to the architecture of the brain. It's all very carefully written, admitting gaps in our knowledge (then and now!) about how, exactly, certain parts of the brain actually function.
For me, the highlight of the brain portion is the end, starting with a section snappily titled "Arithmetical Precision or Logical Reliability, Alternatives."
"It should be noted that the message-system used in the nervous system...is of an essentially statistical character. In other words, what matters are not the precise positions of definite markers, digits, but the statistical characteristics of their occurrence...nearly periodic pulse-trains, etc."
The book ends with a general statement that "the nervous system appears to be using a radically different system of notation" than that of mathematics (and traditional computing) and then concluding that the "language" of the brain may be in layers (which remind me of the layers of network protocols we enjoy today: physical voltage and timing, data link, routing, error correcting, and ending at the application level.)
I'm not sure modern readers will "learn" from this book. But it's certainly worth a visit to read a description of modern computing at the moment time of its birth straight from one of its creators.
December 9, 2021
Nøye og kompakt (dog utdatert ved flere punkter) sammenlikning mellom hjernen og maskinen. Hadde vært gøy med en 2021-versjon (mulig det eksisterer).
Disse drøyt 80 sidene har tatt lang tid, hver side har inspirert til tankesprang i alle retninger
Disse drøyt 80 sidene har tatt lang tid, hver side har inspirert til tankesprang i alle retninger
January 26, 2025
What makes this work so valuable is that John von Neumann was trying finish it during his last days, while struggling with cancer. These lecture notes were meant to be read by him at Yale university, but he was not able to finish them.
April 17, 2021
Good mostly for historical value; if you’re into math and logic, you may enjoy the details, but you can get most of it out of reading a summary. Still interesting that his main thesis (that the nervous system is neither fully analog or digital but some unique combination that incorporates elements of both) has seemingly not been explored much further by more modern writers on consciousness or neurobiology.
October 31, 2017
This 1956 monograph is a series of notes for a projected series of lectures in the"The Silliman Memorial Lecture Series" at Yale. Part of the guidelines for these lectures was that they be appropriate for a general audience. Unfortunately, manuscript remained unfinished and the lectures were not delivered, because von Neumann died while he was working on it.
The premise of the book was to explain how analog and digital computers process information, then compare that with the way the brain processes information. In the days of ferro-magnetic core memories, cards and tapes, before Moore had a chance to observe his law, the terminology used to describe how computers work is a bit alien to the modern reader. Von Neumann seems concerned about issues, like error correction and reliability, which are second nature to us 60 years later. And although he surveys many computing devices from history, the focus is on the computers that use the "von Neumann" architecture, which includes most modern computers.
I wanted to read this book to sample von Neumann's style, the way I read Freud on psychology. I was impressed with his method not only to offer theories, but to quantify them with estimated calculations. The brain might consume 10 watts of energy. The computers of those times would consume much more. I worked on a minicomputer in the 1970's that required a 600 volt power supply to operate the ferrite core memories.
And as primitive as neuroscience was in 1956, to his credit the author deduced that the brain was conducting massively parallel processing with billions of slow firing neurons, while computers were using much faster electronic switching to solve math problems a step at a time.
The quality of the prose trails off a bit toward the end, because of the state of his declining health. I enjoyed the glimpse of greatness and a nostalgic return to the 1950's.
The premise of the book was to explain how analog and digital computers process information, then compare that with the way the brain processes information. In the days of ferro-magnetic core memories, cards and tapes, before Moore had a chance to observe his law, the terminology used to describe how computers work is a bit alien to the modern reader. Von Neumann seems concerned about issues, like error correction and reliability, which are second nature to us 60 years later. And although he surveys many computing devices from history, the focus is on the computers that use the "von Neumann" architecture, which includes most modern computers.
I wanted to read this book to sample von Neumann's style, the way I read Freud on psychology. I was impressed with his method not only to offer theories, but to quantify them with estimated calculations. The brain might consume 10 watts of energy. The computers of those times would consume much more. I worked on a minicomputer in the 1970's that required a 600 volt power supply to operate the ferrite core memories.
And as primitive as neuroscience was in 1956, to his credit the author deduced that the brain was conducting massively parallel processing with billions of slow firing neurons, while computers were using much faster electronic switching to solve math problems a step at a time.
The quality of the prose trails off a bit toward the end, because of the state of his declining health. I enjoyed the glimpse of greatness and a nostalgic return to the 1950's.
November 24, 2018
I've read this book in an hour or less. It's difficulty is of medium challenge, although the content of the book could be understood more easily if it was accompanied by some illustrations or technical representations of the composing parts of the book which contain explicit representations of how the computer works in a similar way to a brain, considering mainly their hardware/biological function. A thing which Neumann cannot explain in this book is what the "mind" is as he clearly makes a distinction between the mind and the brain. Yet the mind being more of a storage unit for the logical process rather than an epistemological necessity to be held in the same way as a philosopher.
The book is divided in two main parts:
1. The Computer
2. The Brain
The computer part basically describes the classic Von Neumann architecture which is basically in the same form more or less: Input Device > Central Processing Unit (Control Unit/Logical Unit/Registers), Memory Unit > Output Device.
There is a somewhat archaic almost comical representation of the type of technology used in the control process which contained electrical plugged connections. The lines of command for how electric stimuli would be released would be through punctured paper or tapes. This seems quite a primitive method by now.
On the other hand a most interesting view of the brain is that the neurons that form a dense connection among themselves through axons are physically ordered by means of logical operations like an integrated circuitry which perform by the logical operations of "and" and "or".
The book is not finished as it requires a final chapter which possibly holds a more synthetic view of both the computer and the brain and why not, a cybernetic merge between these two if possible.
The book is divided in two main parts:
1. The Computer
2. The Brain
The computer part basically describes the classic Von Neumann architecture which is basically in the same form more or less: Input Device > Central Processing Unit (Control Unit/Logical Unit/Registers), Memory Unit > Output Device.
There is a somewhat archaic almost comical representation of the type of technology used in the control process which contained electrical plugged connections. The lines of command for how electric stimuli would be released would be through punctured paper or tapes. This seems quite a primitive method by now.
On the other hand a most interesting view of the brain is that the neurons that form a dense connection among themselves through axons are physically ordered by means of logical operations like an integrated circuitry which perform by the logical operations of "and" and "or".
The book is not finished as it requires a final chapter which possibly holds a more synthetic view of both the computer and the brain and why not, a cybernetic merge between these two if possible.
July 27, 2020
The book isn't really relevant anymore, unfortunately: the first half is a fairly basic description of how computers work, but you can get much better explanations of this in countless computer science books today. The second half about the brain was interesting only in that it made me realize that no real progress has been made in reverse engineering how the brain works in the last 65 years! The speculative musings about what the logical model of the neuron could be (no, it's not just threshold(a*w+b)), how memory might be stored, or whether frequency coding is a feature or just a technical necessity, etc. - these are the exact same things people are still theorizing about today, constantly reshuffling the same ideas that were already presented here. That is a remarkable case of scientific stagnation.
It's a real pity that this is the only book for a lay audience that the remarkable von Neumann left behind.
It's a real pity that this is the only book for a lay audience that the remarkable von Neumann left behind.
October 28, 2020
A quick read but packed with insight. Von Neumann describes the basic architecture of a computer (now called Von Neumann architecture, is still used today) and then proceeds to analyzing neurons and the nervous system. He concludes in awe of the nervous system - it is imprecise in the technical sense but quite adept in other ways; for example, it struggles to sum 3-digit numbers and yet can perform image recognition in an instant. Of course, this contrasts with our artificial computers, even of today, which would find the former task trivial and the latter exceptionally difficult. What is the reason for this discrepancy? Von Neumann can only conjecture that a totally different mathematical system is used in the brain. Overall, a book of incredible depth and foresight.
May 3, 2015
Here we have the very computer architect comparing the 50's machines to the human nervous system. He starts describing the 'current' (1956) digital/analog computers. If you are not into machines, the first chapters will bore you down (but hold on and wait for it). Else, you are going to enjoy this pre integrated circuit/microchip description. Then, he goes about making ingenious comparisons between the natural and artificial machineries. You may have lots of fun updating his figures to recent computers. Hey, note that the human nervous system is pretty much the same (sic), but the computers evolved a bit last ~60 years.
Displaying 1 - 30 of 101 reviews