What do you think?
Rate this book
The Emotion Machine: Commonsense Thinking, Artificial Intelligence, and the Future of the Human Mind
A leading contributor to artificial intelligence offers insight into the numerous ways in which the mind works to demonstrate how emotions and feelings are just different ways of thinking, in an account that poses controversial ideas about the potential for designing machines that are capable of thinking like humans. By the author of The Society of Mind.
400 pages, Hardcover
First published January 1, 2006
207 people are currently reading
3220 people want to read
About the author
Marvin Minsky
41 books318 followersMarvin Lee Minsky (born August 9, 1927) was an American cognitive scientist in the field of artificial intelligence (AI), co-founder of Massachusetts Institute of Technology's AI laboratory, and author of several texts on AI and philosophy. Marvin Lee Minsky was born in New York City to an eye surgeon and a Jewish activist, where he attended The Fieldston School and the Bronx High School of Science. He later attended Phillips Academy in Andover, Massachusetts. He served in the US Navy from 1944 to 1945. He holds a BA in Mathematics from Harvard (1950) and a PhD in mathematics from Princeton (1954). He has been on the MIT faculty since 1958. In 1959 he and John McCarthy founded what is now known as the MIT Computer Science and Artificial Intelligence Laboratory. He is currently the Toshiba Professor of Media Arts and Sciences, and Professor of electrical engineering and computer science.
Ratings & Reviews
Friends & Following
Create a free account to discover what your friends think of this book!
Community Reviews
Displaying 1 - 30 of 59 reviews
May 26, 2010
Oh Marvy. After the three chapters, I was ready to stop reading -- you simply weren't saying anything important. But then you whipped out a well-placed discussion and your take on the whole consciousness debate; it pulled me back in. Alas, you slipped again as you barely held on to me as I waded (and progressively sped up out of disinterest) through the next chapters.
I get it, you're a computer scientist and attack the understanding of the brain as if it is hierarchically organized in the same (or similar) ways to our programmatic abstractions. Of course, I'm biased to think like you (a classically trained computer scientist), but unfortunately your arguments and ideas take form in the same abstract space you're trying to define. That is, there is hardly any substance, evidence, or research backing up your claims. You casually brush this off by claiming that any such research would simply be out-dated in the future, as if this book will be read in the generations to come. I agree, it's hard to take a theoretical stance on science that is constantly in revision, but sometimes you need to back up what you say. That being said, I do admire and agree with some of your ideas. These gems just end up being spaced on average to about every 30-40 pages.
As for your writing, my biggest problem is your style. Instead of wasting time in the current chapter hinting about what you'll talk about later, try introducing something now! I got the feeling that the entire book was alluding to evidence, anecdotes, or ideas which will always be coming in later chapters -- and many of them never came.
The short of it: this book is worth skimming through, picking up from the library, or reading on the toilet. It has some clever ideas, but is poorly executed.
I get it, you're a computer scientist and attack the understanding of the brain as if it is hierarchically organized in the same (or similar) ways to our programmatic abstractions. Of course, I'm biased to think like you (a classically trained computer scientist), but unfortunately your arguments and ideas take form in the same abstract space you're trying to define. That is, there is hardly any substance, evidence, or research backing up your claims. You casually brush this off by claiming that any such research would simply be out-dated in the future, as if this book will be read in the generations to come. I agree, it's hard to take a theoretical stance on science that is constantly in revision, but sometimes you need to back up what you say. That being said, I do admire and agree with some of your ideas. These gems just end up being spaced on average to about every 30-40 pages.
As for your writing, my biggest problem is your style. Instead of wasting time in the current chapter hinting about what you'll talk about later, try introducing something now! I got the feeling that the entire book was alluding to evidence, anecdotes, or ideas which will always be coming in later chapters -- and many of them never came.
The short of it: this book is worth skimming through, picking up from the library, or reading on the toilet. It has some clever ideas, but is poorly executed.
April 24, 2015
For certain books I find myself reading the Goodreads reviews to see if my responses jibe with those of others, especially in terrain foreign to me, as AI is. I have been impressed with the reviews as mostly intelligent perspectives on the book and author.
I was delighted to be in the hands of a clearly brilliant thinker, and was wowed by the breadth of his knowledge of a variety of disciplines and individual great minds, past and present. But I too found the book less than exciting and coherent, although Minsky constantly makes reference to his discussion in a specific sub-chapter read previously or in an upcoming chapter. I read only a few pages every day, so it took quite a while to get through the book. Nevertheless I am grateful for the exposure to what seems to be a divergent way of regarding thinking, consciousness and the self. Concurrently I have been reading "The Most Human Human" by Brian Christian about the Loebner/Turing competition, and am clearly in the camp of wanting to understand how the human "mind" works; not how to design the most human computer.
I was delighted to be in the hands of a clearly brilliant thinker, and was wowed by the breadth of his knowledge of a variety of disciplines and individual great minds, past and present. But I too found the book less than exciting and coherent, although Minsky constantly makes reference to his discussion in a specific sub-chapter read previously or in an upcoming chapter. I read only a few pages every day, so it took quite a while to get through the book. Nevertheless I am grateful for the exposure to what seems to be a divergent way of regarding thinking, consciousness and the self. Concurrently I have been reading "The Most Human Human" by Brian Christian about the Loebner/Turing competition, and am clearly in the camp of wanting to understand how the human "mind" works; not how to design the most human computer.
July 25, 2008
Not that Minsky isn't an intelligent guy, he just isn't on the right track. His model of the human brain is complete shit. In fact its an awful stretch to even call it a model of the human brain, because Minsky has never studied the brain. As far as I could tell he just sat around and thought about thinking and then he wrote a book about what he came up with. There are a lot of crazy models we could come up with just from introspection, only one is right, Its not this one. If you are trying to actually build intelligent machines, take this book as a lesson on what not to do.
September 19, 2017
It took me a long time to finish The Emotion Machine. I read it on and off for almost two years, starting and finishing other books in between. Although some parts are remarkable, some other parts are boring as hell.
The book attempts to provide a model for how humans think. Arguing that countless explainable processes are hidden behind many "suitcase-words" that we use everyday and we cannot explicitly explain them, such as consciousness and feelings, Minsky severely criticizes many theories developed about human mind in fields like psychology and philosophy.
The book is supposed to be non-technical but it is not. Therefore, I do not recommend it for non-academics since it is going to be confusing and super boring. However, a glimpse at the breathtaking list of the names Minsky has acknowledged in the end due to their fruitful conversations and ideas suggests how profound the notions described in the book can be. Whatever your field is, you'll find familiar big names. Leaving alone that Minsky himself is considered as the father of Artificial Intelligence.
The book attempts to provide a model for how humans think. Arguing that countless explainable processes are hidden behind many "suitcase-words" that we use everyday and we cannot explicitly explain them, such as consciousness and feelings, Minsky severely criticizes many theories developed about human mind in fields like psychology and philosophy.
The book is supposed to be non-technical but it is not. Therefore, I do not recommend it for non-academics since it is going to be confusing and super boring. However, a glimpse at the breathtaking list of the names Minsky has acknowledged in the end due to their fruitful conversations and ideas suggests how profound the notions described in the book can be. Whatever your field is, you'll find familiar big names. Leaving alone that Minsky himself is considered as the father of Artificial Intelligence.
September 11, 2018
It was hard to track what Minsky is doing with this book. “The Emotion Machine” seems mistitled. He sees emotions as a form of thinking. We have multiple ways to think and he says that an “emotional state” is one example of this. Definitions of emotion are all over the place, but Minsky is in another ballpark. While we may attach words and arguments to them, anger and love are quite different than thought. Minsky might mean that this way of thinking is emotion-infused in Hume’s sense, but then Hume didn’t see emotion-based thought as one among “multiple ways to think” as Minsky argues. Hume saw that all thought has an underlying emotional component (thought serves the “passions”). (1) It’s also not clear why Minsky's title refers to “machine.”
Minsky has several layers of thinking, beginning with instinctive “reactions” and proceeds up to the highest level of thought, self-conscious reflection. Each rung in this hierarchy involves a goal (instinctive initially, then proceeding to “ideals embedded in conscience, ideals, or moral codes”) that he sees in human thinking. That we have such high-level goals is clear, but the question is always why do we have this goal/ideal/code and not another? Many, or most, might attribute their origin to the “environment” or “nurture” – to upbringing and culture (“The infant absorbs the goals and ideals of parents and friends”), but a Freud or evolutionary psychologist would move the origin of high-level goals back farther. For example, the infant, then the child, absorbs goals and ideals of its social group because that is evolution’s ways of merging the self with the group to ensure survival. And this, in turn, brings in a full suite of psychological, emotional and social dynamics such as the role of the superego, conformity at the expense of the self, loyalty regardless of right and wrong, etc. Pretty soon, the self is in deep, dark territory -- and far from consciousness -- regarding the explanation for its behavior. (2)
Whatever else Minsky believes, it’s clear he does not give much weight to biology. He’s a mind person. The mind is in charge. (3) It sets the agenda and the body follows, but that model only takes us so far. At the highest level, for example, Kant can tell us to respect others as ends and never solely as a means whereas a healthy skeptic can always ask in return: “Why respect others as an end? And an answer to that question can take one back to evolution’s self-interest and Hume’s passions: We respect others because we are attached to them (a feeling) or because we understand at some level that it’s in our best interests (another feeling, understood cognitively) to do so.
Minsky’s other point is that the mind is a collection of subsystems, functioning more or less as autonomous units, depending on the circumstance. That makes sense, though Minsky goes on to argue that there’s no centralized self that is in control of all of this. “When you say ‘Self,’” he writes, “you are referring not to a single representation but to an extensive network of different models representing different aspects of yourself.” Later, quoting Herman Hesse, Minksky states that there’s no single human being but, rather, “multiple subpersonalities.” The counter-argument is that if life is seen as a vehicle to house a replicating vehicle (a “machine” of sorts) as Dawkins argues, then the body needs to survive to replicate, and the mind needs to support this task, including everything that goes with survival – work, status, love, creature comforts, family, friends, and group, etc. there is a singularity of some sort that pulls all of this together in a more or less coherent way that can be viewed as a "self." (4)
Given his view of mind-emotion as a machine, it’s interesting that Minsky says nothing about how, chemically, behavior is transformed into representational thought.
(1) Later in the book, Minsky treats emotions in a more or less traditional way when he states that emotions are (a) classes of mental conditions, (b) that produce external signs, and (c) that make our behavior more predictable to others. That is the role of the primary emotions – anger, fear, sadness, disgust, surprise, curiosity, and joy. Why these are primary emotions are not clear; if they are from Darwin’s Descent, they are physical manifestations, “expressions of emotions.” While Minsky says we have these because they signal to others, it could be we have these emotions because we are mad, we are afraid, sad, disgusted, surprised, curious and joyful regardless of the presence of others. In other words, an inner state expresses itself outward.
(2) In his six-layers of thought model, Minsky says we create new emotions. He illustrates the highest stage, self-reflective consciousness, by saying that if a woman is late for a meeting, she wonders what will others think of her. With the conflict between being late and being concerned about what others think, a new emotional reality is created. A goal that resolves the conflict is created, and a change in behavior follows. Her reality is transformed by an “ought” but that just begs the question, “Why is it an ‘ought’ with mandatory power? The deeper issue is why does she care about what others think about her in the first place, and isn’t it this that guides her choice? If she fears ostracism, she conforms to group expectations. If she values her own independence, then she doesn’t care what they think. In a book of photographs on the building of the Union Pacific to Promontory Point, there is a picture of one official looking “stern and somber, a perfect picture of how a trusty nineteenth century railroad builder should appear.” This is another example of Minsky’s “ought” as created by an emotion. The question of course is why the gentleman would care to present himself in this way, a question that evolutionary psychology, say, would have an opinion about.
(3) When Minsky writes, “Our species surpasses all the rest in our flair for inventing new ways to think. We fabricate weapons, garments, and dwellings,” he conflates ways of thinking with the products of thinking. Ways of thinking remain the same – we solve new problems, but the problems and the solutions vary across time. He also uses Hawkins and Reeves as case studies to illustrate the importance of the brain. Of course, there are numerous examples of demented people who have healthy bodies.
(4) Minsky says the will-to-survive and pleasure-pain are “simplistic. Simple yes, simplistic no. They are simple because they reflect our evolutionary essence – survival rests on species-specific needs that motivate much of our behavior. We can trace what we do back to their evolutionary impetus: work and nurture to sustain ourselves, group life for security, fear and anger to protect ourselves, and of course sex as replication. We are the tip of the spear this way, but the motive forces go way back. In Schopenhauer’s sense, need is pain (and about objects related to pain), and pleasure is the satisfaction of need. Need and pain are embedded in each of the emotions, and so is their satisfaction-pleasure.
Minsky has several layers of thinking, beginning with instinctive “reactions” and proceeds up to the highest level of thought, self-conscious reflection. Each rung in this hierarchy involves a goal (instinctive initially, then proceeding to “ideals embedded in conscience, ideals, or moral codes”) that he sees in human thinking. That we have such high-level goals is clear, but the question is always why do we have this goal/ideal/code and not another? Many, or most, might attribute their origin to the “environment” or “nurture” – to upbringing and culture (“The infant absorbs the goals and ideals of parents and friends”), but a Freud or evolutionary psychologist would move the origin of high-level goals back farther. For example, the infant, then the child, absorbs goals and ideals of its social group because that is evolution’s ways of merging the self with the group to ensure survival. And this, in turn, brings in a full suite of psychological, emotional and social dynamics such as the role of the superego, conformity at the expense of the self, loyalty regardless of right and wrong, etc. Pretty soon, the self is in deep, dark territory -- and far from consciousness -- regarding the explanation for its behavior. (2)
Whatever else Minsky believes, it’s clear he does not give much weight to biology. He’s a mind person. The mind is in charge. (3) It sets the agenda and the body follows, but that model only takes us so far. At the highest level, for example, Kant can tell us to respect others as ends and never solely as a means whereas a healthy skeptic can always ask in return: “Why respect others as an end? And an answer to that question can take one back to evolution’s self-interest and Hume’s passions: We respect others because we are attached to them (a feeling) or because we understand at some level that it’s in our best interests (another feeling, understood cognitively) to do so.
Minsky’s other point is that the mind is a collection of subsystems, functioning more or less as autonomous units, depending on the circumstance. That makes sense, though Minsky goes on to argue that there’s no centralized self that is in control of all of this. “When you say ‘Self,’” he writes, “you are referring not to a single representation but to an extensive network of different models representing different aspects of yourself.” Later, quoting Herman Hesse, Minksky states that there’s no single human being but, rather, “multiple subpersonalities.” The counter-argument is that if life is seen as a vehicle to house a replicating vehicle (a “machine” of sorts) as Dawkins argues, then the body needs to survive to replicate, and the mind needs to support this task, including everything that goes with survival – work, status, love, creature comforts, family, friends, and group, etc. there is a singularity of some sort that pulls all of this together in a more or less coherent way that can be viewed as a "self." (4)
Given his view of mind-emotion as a machine, it’s interesting that Minsky says nothing about how, chemically, behavior is transformed into representational thought.
(1) Later in the book, Minsky treats emotions in a more or less traditional way when he states that emotions are (a) classes of mental conditions, (b) that produce external signs, and (c) that make our behavior more predictable to others. That is the role of the primary emotions – anger, fear, sadness, disgust, surprise, curiosity, and joy. Why these are primary emotions are not clear; if they are from Darwin’s Descent, they are physical manifestations, “expressions of emotions.” While Minsky says we have these because they signal to others, it could be we have these emotions because we are mad, we are afraid, sad, disgusted, surprised, curious and joyful regardless of the presence of others. In other words, an inner state expresses itself outward.
(2) In his six-layers of thought model, Minsky says we create new emotions. He illustrates the highest stage, self-reflective consciousness, by saying that if a woman is late for a meeting, she wonders what will others think of her. With the conflict between being late and being concerned about what others think, a new emotional reality is created. A goal that resolves the conflict is created, and a change in behavior follows. Her reality is transformed by an “ought” but that just begs the question, “Why is it an ‘ought’ with mandatory power? The deeper issue is why does she care about what others think about her in the first place, and isn’t it this that guides her choice? If she fears ostracism, she conforms to group expectations. If she values her own independence, then she doesn’t care what they think. In a book of photographs on the building of the Union Pacific to Promontory Point, there is a picture of one official looking “stern and somber, a perfect picture of how a trusty nineteenth century railroad builder should appear.” This is another example of Minsky’s “ought” as created by an emotion. The question of course is why the gentleman would care to present himself in this way, a question that evolutionary psychology, say, would have an opinion about.
(3) When Minsky writes, “Our species surpasses all the rest in our flair for inventing new ways to think. We fabricate weapons, garments, and dwellings,” he conflates ways of thinking with the products of thinking. Ways of thinking remain the same – we solve new problems, but the problems and the solutions vary across time. He also uses Hawkins and Reeves as case studies to illustrate the importance of the brain. Of course, there are numerous examples of demented people who have healthy bodies.
(4) Minsky says the will-to-survive and pleasure-pain are “simplistic. Simple yes, simplistic no. They are simple because they reflect our evolutionary essence – survival rests on species-specific needs that motivate much of our behavior. We can trace what we do back to their evolutionary impetus: work and nurture to sustain ourselves, group life for security, fear and anger to protect ourselves, and of course sex as replication. We are the tip of the spear this way, but the motive forces go way back. In Schopenhauer’s sense, need is pain (and about objects related to pain), and pleasure is the satisfaction of need. Need and pain are embedded in each of the emotions, and so is their satisfaction-pleasure.
June 13, 2016
I started reading this book to know what Marvin Minsky's thoughts are about thinking machines and AI, especially how he thought a truly intelligent agent can be created. While in the middle of the book I realized that the ideas discussed in this book are helping me to understand myself, leave aside the creation of intelligent machines. This is not just an AI book, it also has a psychology and cognitive perspective to help understand how we naturally intelligent beings think and behave.
There are too many citations and occasionally some quotes that did not make any sense to me. I guess i have to read it again! Some of the claims are without proofs and experimental evaluations, we gotta take them as author's opinions.
As a grad student interested to seek the truth of AI, I happened take graduate level A.I. course (this book was actually mentioned by my A.I. professor at USC as one of his favorite book). I am also trying my best with recent advancements in machine learning. I wasn't and am not convinced that what most people call AI these days will ever yield human like intelligent. In my opinion, the recent development in machine learning is merely mathematical tricks (in other words, a mechanism to achieve the goals in terms of probability and stats summary, sometimes with logic and data). This book offers a different and broader perspective. This book defines 'what', 'why' and 'how' questions about thinking, consciousness, self and why the modern human society is the way it is. Often mentions so many questions we don't have answers yet. I am surprised to see how little we know about ourselves!
Yes it was worth the read!
There are too many citations and occasionally some quotes that did not make any sense to me. I guess i have to read it again! Some of the claims are without proofs and experimental evaluations, we gotta take them as author's opinions.
As a grad student interested to seek the truth of AI, I happened take graduate level A.I. course (this book was actually mentioned by my A.I. professor at USC as one of his favorite book). I am also trying my best with recent advancements in machine learning. I wasn't and am not convinced that what most people call AI these days will ever yield human like intelligent. In my opinion, the recent development in machine learning is merely mathematical tricks (in other words, a mechanism to achieve the goals in terms of probability and stats summary, sometimes with logic and data). This book offers a different and broader perspective. This book defines 'what', 'why' and 'how' questions about thinking, consciousness, self and why the modern human society is the way it is. Often mentions so many questions we don't have answers yet. I am surprised to see how little we know about ourselves!
Yes it was worth the read!
January 5, 2009
I really wanted to like this book more, and it does have a number of fascinating ideas, but overall I found it rather incoherent and undirected, and ultimately unsatisfying. It was hard to make myself get through it all, and I can't point to any particular insight that will change the way I think about AI or intelligence in general. Minsky is a great thinker, but I'd give this book a pass.
September 29, 2015
So, how does one go about understanding how the human brain works? One way would be to introspect, to discuss it with others, to speculate, etc. The other would be to try to reverse engineer it. In a lot of ways, it turns out the best way to understand how anything works is to build one yourself, and this is (perhaps surprisingly) no less true of the human brain than with anything else.
Fortunately, even if you fail, you learn a lot.
Marvin Minsky is an MIT professor and researcher who has been trying to build a brain since the 60's. Full disclosure: in a previous decade, I got a Master's degree in neural networks. In that field, Minsky is known as the guy who did a hatchet job on the whole field in the 70's. Basically wrote an article that says, "not only doesn't this work now, this approach will never work, and here's the mathematical proof of why". Whether it was true or not, it was convincing enough to cause most funding for research to disappear almost overnight.
Besides killing previously thriving fields of research, what has Minsky been doing in almost half a century of thinking about the requirements of artificial intelligence? To judge from this book, a lot of diagrams. It's one of the book's strong points, actually, that he has very clean and simple diagrams for a lot of the concepts which he discusses.
In a very real sense, what Minsky believes he has discovered is that emotions are a requirement for any rationally functioning mind. The central problem of artificial intelligence is that, while any approach has certain problems it can solve well, there are always others which that approach is fundamentally inappropriate for. To deal with this, we need to be able to think in different ways, at different times.
What would it feel like, to change how you think? What would you call that switch from one way of thinking, to another, that happens when your mind decides, "this isn't a situation for Way of Thinking A, this is a situation where Way of Thinking B would be more appropriate"?
According to Minsky, it would feel like going from happy, to angry. Or bored, to excited. Or depressed, to being in love. Parts of your brain, that were firing away moments before, shut down. Other parts, fire up. Some parts are held in common, of course. Why is it hard to think rationally when you're afraid? Why is it hard to be suspicious when you're in love? Why is it hard to think ahead and plan for the future when you're depressed?
It's an intriguing theory, and Minsky gives us about 350 easy-to-read pages on what it would mean if his theory were right. Essentially, it would mean that you're not You, you're Them, a host of mini-yous who are more like an orchestra than a computer. At any given time, some of the mini-yous are playing, and others are silent. The result, makes the piece sound angry or melancholy, bright or dark.
But is it true? Who knows? Until Minsky (or some latter-day Minsky of a future generation) succeeds in reverse engineering a new model of the emotion machine, we'll each be entitle to our own opinion. But it's a fun read, and thought provoking to boot.
Fortunately, even if you fail, you learn a lot.
Marvin Minsky is an MIT professor and researcher who has been trying to build a brain since the 60's. Full disclosure: in a previous decade, I got a Master's degree in neural networks. In that field, Minsky is known as the guy who did a hatchet job on the whole field in the 70's. Basically wrote an article that says, "not only doesn't this work now, this approach will never work, and here's the mathematical proof of why". Whether it was true or not, it was convincing enough to cause most funding for research to disappear almost overnight.
Besides killing previously thriving fields of research, what has Minsky been doing in almost half a century of thinking about the requirements of artificial intelligence? To judge from this book, a lot of diagrams. It's one of the book's strong points, actually, that he has very clean and simple diagrams for a lot of the concepts which he discusses.
In a very real sense, what Minsky believes he has discovered is that emotions are a requirement for any rationally functioning mind. The central problem of artificial intelligence is that, while any approach has certain problems it can solve well, there are always others which that approach is fundamentally inappropriate for. To deal with this, we need to be able to think in different ways, at different times.
What would it feel like, to change how you think? What would you call that switch from one way of thinking, to another, that happens when your mind decides, "this isn't a situation for Way of Thinking A, this is a situation where Way of Thinking B would be more appropriate"?
According to Minsky, it would feel like going from happy, to angry. Or bored, to excited. Or depressed, to being in love. Parts of your brain, that were firing away moments before, shut down. Other parts, fire up. Some parts are held in common, of course. Why is it hard to think rationally when you're afraid? Why is it hard to be suspicious when you're in love? Why is it hard to think ahead and plan for the future when you're depressed?
It's an intriguing theory, and Minsky gives us about 350 easy-to-read pages on what it would mean if his theory were right. Essentially, it would mean that you're not You, you're Them, a host of mini-yous who are more like an orchestra than a computer. At any given time, some of the mini-yous are playing, and others are silent. The result, makes the piece sound angry or melancholy, bright or dark.
But is it true? Who knows? Until Minsky (or some latter-day Minsky of a future generation) succeeds in reverse engineering a new model of the emotion machine, we'll each be entitle to our own opinion. But it's a fun read, and thought provoking to boot.
October 19, 2014
This is a book to set your mind on fire. Why we want what we want? We choose among the things we want, but can we choose *what* we want? Is there a self behind the wheel managing all of our often contradictory drives? The answer is as simple as surprising. This book is sort of a sequel to Society of Mind, which I'm yet to read.
June 11, 2016
Very interesting and unique perspective from the pop-psychology slant. It examines how the mind uses different parts of the brain to work together, and how emotions are central to how the mind works. I love his quotations from Shakespeare.
March 29, 2019
The book is basically Minsky’s theory about how the mind works (in this case, brain ≅ mind). I can’t help but speak about what I didn’t like first…
When I started reading it, I couldn’t shake the feeling that I’m reading pseudo-science book (which it could be, if you take everything for granted, but Minsky repeatedly clarifies that it’s just his theories). I might feel this way because I recently read a very good (popular) science book If the Universe Is Teeming with Aliens ... WHERE IS EVERYBODY?: Seventy-Five Solutions to the Fermi Paradox and the Problem of Extraterrestrial Life, and in stark contrast to that, this one had many fewer references. It got better as I read on though, but have in mind that this book is ⅘ hypotheses, ⅕ science. Minsky’s kind of reverse engineering the brain. But in theory.
Another thing I disliked was the style... The examples he uses to illustrate ideas are quite dull, and he keeps coming back to them over and over... But the worst is constant references to the upcoming (and previous) chapters. I mean, you don’t need to keep reassuring us that you will expand on that... Unless he intended this book to be a manual for how to develop his AI theory (which it isn’t), I don’t see the point... Also, he is often citing himself...
Apart from these (not so) minor drawbacks it was a good book. I didn’t find it to be revolutionary at all (but then, I did study AI), but he did invent (?) many concepts like “selectors”, “critics”, “ways to think”, “panalogies”, etc. These I found of interest, especially panalogies and his take on the way we retrieve our memories. In principle, all these theories could be tested by just making an AI that would work the way he describes and seeing if it’s human-like… Of course that’s ridiculous, because he barely scratches the surface of AGI with this book, but I do feel like he could be onto something, and some theories could be very spot-on. Research ideas.
I wonder what do psychologists think about this book, and then people who have neither psychology nor AI/CS background. I think it could be mind-boggling to some readers if they aren’t deterred by mediocre writing.
⅗
When I started reading it, I couldn’t shake the feeling that I’m reading pseudo-science book (which it could be, if you take everything for granted, but Minsky repeatedly clarifies that it’s just his theories). I might feel this way because I recently read a very good (popular) science book If the Universe Is Teeming with Aliens ... WHERE IS EVERYBODY?: Seventy-Five Solutions to the Fermi Paradox and the Problem of Extraterrestrial Life, and in stark contrast to that, this one had many fewer references. It got better as I read on though, but have in mind that this book is ⅘ hypotheses, ⅕ science. Minsky’s kind of reverse engineering the brain. But in theory.
Another thing I disliked was the style... The examples he uses to illustrate ideas are quite dull, and he keeps coming back to them over and over... But the worst is constant references to the upcoming (and previous) chapters. I mean, you don’t need to keep reassuring us that you will expand on that... Unless he intended this book to be a manual for how to develop his AI theory (which it isn’t), I don’t see the point... Also, he is often citing himself...
Apart from these (not so) minor drawbacks it was a good book. I didn’t find it to be revolutionary at all (but then, I did study AI), but he did invent (?) many concepts like “selectors”, “critics”, “ways to think”, “panalogies”, etc. These I found of interest, especially panalogies and his take on the way we retrieve our memories. In principle, all these theories could be tested by just making an AI that would work the way he describes and seeing if it’s human-like… Of course that’s ridiculous, because he barely scratches the surface of AGI with this book, but I do feel like he could be onto something, and some theories could be very spot-on. Research ideas.
I wonder what do psychologists think about this book, and then people who have neither psychology nor AI/CS background. I think it could be mind-boggling to some readers if they aren’t deterred by mediocre writing.
⅗
November 22, 2020
The central thesis in the emotion machine is that current approaches to neuroscience are in limited in what they can discover. They are limited because they follow the dominant paradigm of scientific theories: explain as much as possible using as little as possible. Occam's Razor. Minsky thinks our fetish for simple laws that describe the world is rooted in physics. In physics we have had much success distilling the laws of nature into small equations with few variables. And as a result, we have called the process of finding ever smaller or fewer equations with fewer variables "beautiful". Naturally all the other sciences followed suit because physics very much sets the standards of science for all the other sciences.
There's an alternative hypothesis that I fancy. It's that human brains are intrinsically limited in its computation to understand complex theories. It's commonly known that humans can only hold 6 items in their short term memory. I don't think it's a stretch to think there are other striking limitations. That is to say, the main scientific paradigm cannot help but favor Occam's Razor. Complex theories may just be beyond human understanding. There are other patterns to how humans do science. For example, we tend to favor thinking in paradigm of atoms. Our sciences usually consist of describing the smallest possible component of the system (the atom) and build complexity by composing atoms together or describing their interactions otherwise. It's very possible that other paradigms of scientific theory are equal or better at describing nature. The most cliche example is of variational calculus which describes nature without the arrows of a time, which seems very strange to our conventional scientific thinking where we like to think about things as they evolve through time. This illustrates human's prior or bias in thinking about science as we experience our everyday lives: an evolution forward in time. Of course, this is inevitable as all human thought is analogies (or panalogies according to Minsky).
One might imagine a distant future where humanity meets alien societies. They might each have their own very different approaches to understanding or describing nature. Will our atomic, conventional science be primitive compared to their sciences?
Some might argue all this philosophical musing (or meta-science if you will) is pointless. If there are alternative approaches to science that cannot be understood by humans then we would never know about them or be able to validate them. It's an unverifiable hypothesis. That is true but I think as it's often the case, it's better to be cognizant of your ignorance than blissfully stupid. Furthermore, there may actually be approaches to science that are understandable but are not yet part of mainstream science. Other approaches, again calculus of variations (is our argument resting on a single example?!), and perhaps a theory of science that is not atomic, may be discovered some time in the future or has already been discovered in the past. However, I predict it would be very difficult to overcome the inertia of mainstream science and develop these alternative approaches or theories to their full potential. From a very young age, our education system teaches to think in the way of mainstream science. Our academic institutions reward research that looks like mainstream science. How would we ever get a chance to break out of the loop?
Coming back to the emotion machine, Minsky audaciously proposes his pet theory of how the brain works. The theory is based off of mainly introspection rather than scientific evidence and it is designed to be anti-conventional science. It is complex, it is messy, and it is redundant. Minsky argues these are all necessary characteristics of a full theory of the mind. Here I agree with him. The reason I suspect this is true is because the mind is unlike any system we usually study in the sciences. Most sciences study simple systems that fit into the cookie cutter framework of atomic theories. True there is biology where we study organisms made up of billions of different parts but usually the solvable problems in biology are those that tend to simple (in the sense of simplifying to a few ideas) to begin with. Think genetics where all the chemical, cell, tissue, organ interactions are abstracted away by "are your eyes green or brown". Systems which are complex or emerging such as protein folding are a lot more unwieldy. We usually need to resort to simulations and numerical methods rather than elegant, simple equations. The brain is probably one of the most complex, emergent systens for the reason the emergent phenomena of the system is more intelligent than those of any other system. Therefore, we expect science of the mind to look more like protein folding and less than genetics. I don't necessarily think the specifics of Minsky's theory are correct. I do however, agree with his high-level approach.
By its very definition, a theory that is counter-conventional science will piss people off. The emotion machine didn't receive the most generous ratings on Goodreads, mostly because a lot of people are so turned off by his "unscientific theories". I would argue these people are missing the forest for the trees. It's quite clear we've hit a wall with our current approaches to neuroscience and understanding the mind. Actually we can generalize this to our scientific understanding of any complex, emergent system but this is another story. Any endeavor into the land of unconventional science will be long and nonlinear. There is no foundation to stand on. There are no prior works you can cite on Arxiv. You will have to set footholds but you will inevitably make some bad ones. But this doesn't mean you shouldn't try. People who wanted a fully coherent theory of mind from Minsky's book (I know I did at first) are expecting nothing less than omnipotence from Minsky. As well accomplished as he is, Minsky is human. And like the rest of us humans, he has a hard time searching the space of possible theories when there is little empirical experimentation to ground the search in.
I'll be the first to admit parts of the book were hard to understand. Maybe they could have done a better job editing. And I believe Minsky's theory is actually ambiguous, redundant and trivial at times. However, this is definitely a book I will read again. There is so much to discover between the lines. Peeks into the many conversations Minsky had with his colleagues, all bright minds in their own right. Minsky's capacity for introspection is undeniable. Maybe I'm just someone who likes this stuff - but his insights to how humans think and his ability to construct the right sequence of english symbols to describe these phenoma is inspiring. As a reader of machine learning papers, I can say authors often attempt to motivate their work with insights from psychology and introspection. These can often feel cheap or shallow. Emotion machine is a refreshing divergence from that.
There's an alternative hypothesis that I fancy. It's that human brains are intrinsically limited in its computation to understand complex theories. It's commonly known that humans can only hold 6 items in their short term memory. I don't think it's a stretch to think there are other striking limitations. That is to say, the main scientific paradigm cannot help but favor Occam's Razor. Complex theories may just be beyond human understanding. There are other patterns to how humans do science. For example, we tend to favor thinking in paradigm of atoms. Our sciences usually consist of describing the smallest possible component of the system (the atom) and build complexity by composing atoms together or describing their interactions otherwise. It's very possible that other paradigms of scientific theory are equal or better at describing nature. The most cliche example is of variational calculus which describes nature without the arrows of a time, which seems very strange to our conventional scientific thinking where we like to think about things as they evolve through time. This illustrates human's prior or bias in thinking about science as we experience our everyday lives: an evolution forward in time. Of course, this is inevitable as all human thought is analogies (or panalogies according to Minsky).
One might imagine a distant future where humanity meets alien societies. They might each have their own very different approaches to understanding or describing nature. Will our atomic, conventional science be primitive compared to their sciences?
Some might argue all this philosophical musing (or meta-science if you will) is pointless. If there are alternative approaches to science that cannot be understood by humans then we would never know about them or be able to validate them. It's an unverifiable hypothesis. That is true but I think as it's often the case, it's better to be cognizant of your ignorance than blissfully stupid. Furthermore, there may actually be approaches to science that are understandable but are not yet part of mainstream science. Other approaches, again calculus of variations (is our argument resting on a single example?!), and perhaps a theory of science that is not atomic, may be discovered some time in the future or has already been discovered in the past. However, I predict it would be very difficult to overcome the inertia of mainstream science and develop these alternative approaches or theories to their full potential. From a very young age, our education system teaches to think in the way of mainstream science. Our academic institutions reward research that looks like mainstream science. How would we ever get a chance to break out of the loop?
Coming back to the emotion machine, Minsky audaciously proposes his pet theory of how the brain works. The theory is based off of mainly introspection rather than scientific evidence and it is designed to be anti-conventional science. It is complex, it is messy, and it is redundant. Minsky argues these are all necessary characteristics of a full theory of the mind. Here I agree with him. The reason I suspect this is true is because the mind is unlike any system we usually study in the sciences. Most sciences study simple systems that fit into the cookie cutter framework of atomic theories. True there is biology where we study organisms made up of billions of different parts but usually the solvable problems in biology are those that tend to simple (in the sense of simplifying to a few ideas) to begin with. Think genetics where all the chemical, cell, tissue, organ interactions are abstracted away by "are your eyes green or brown". Systems which are complex or emerging such as protein folding are a lot more unwieldy. We usually need to resort to simulations and numerical methods rather than elegant, simple equations. The brain is probably one of the most complex, emergent systens for the reason the emergent phenomena of the system is more intelligent than those of any other system. Therefore, we expect science of the mind to look more like protein folding and less than genetics. I don't necessarily think the specifics of Minsky's theory are correct. I do however, agree with his high-level approach.
By its very definition, a theory that is counter-conventional science will piss people off. The emotion machine didn't receive the most generous ratings on Goodreads, mostly because a lot of people are so turned off by his "unscientific theories". I would argue these people are missing the forest for the trees. It's quite clear we've hit a wall with our current approaches to neuroscience and understanding the mind. Actually we can generalize this to our scientific understanding of any complex, emergent system but this is another story. Any endeavor into the land of unconventional science will be long and nonlinear. There is no foundation to stand on. There are no prior works you can cite on Arxiv. You will have to set footholds but you will inevitably make some bad ones. But this doesn't mean you shouldn't try. People who wanted a fully coherent theory of mind from Minsky's book (I know I did at first) are expecting nothing less than omnipotence from Minsky. As well accomplished as he is, Minsky is human. And like the rest of us humans, he has a hard time searching the space of possible theories when there is little empirical experimentation to ground the search in.
I'll be the first to admit parts of the book were hard to understand. Maybe they could have done a better job editing. And I believe Minsky's theory is actually ambiguous, redundant and trivial at times. However, this is definitely a book I will read again. There is so much to discover between the lines. Peeks into the many conversations Minsky had with his colleagues, all bright minds in their own right. Minsky's capacity for introspection is undeniable. Maybe I'm just someone who likes this stuff - but his insights to how humans think and his ability to construct the right sequence of english symbols to describe these phenoma is inspiring. As a reader of machine learning papers, I can say authors often attempt to motivate their work with insights from psychology and introspection. These can often feel cheap or shallow. Emotion machine is a refreshing divergence from that.
September 10, 2018
Skimmed and consulted, rather than rigorously read.
I'm still making up my mind about the value of this book for me, as some of the conclusions seem too rushed (ie, we have multiple images of self, therefore an unified self doesn't exist") or failing to even attempt to provide an explanation of the phenomena they claim to explain (ie, on the subjective experience it only points to a multitude of stimuli, the challenge of processing them, etc... and it simply stops there).
I do, have to give him credit for a thorough attempt of creating models for each of the examples.
Chapter 5 proposes six levels of processes
“each layer is built on the previous one—until they extend to processes that involve our highest ideals and personal goals”
P146 – levels of mental activities
Self-conscious reflection
Self-reflective thinking
Reflective thinking
Deliberative thinking
Learned Reaction
Instinctive reaction
P2. “Consciousness refers to more than twenty […] processes”
P111. Quotion Aaron Sloman (1992) ‘Developing concepts of consciousness’ “The phrase ‘human consciousness’ typically corresponds to such large cluster of features and capabilities (many of which we don’t understand or know about) that its set of possible subsets is astronomical.
P292-293, Hierarchy of representations. “The sections above have briefly described several kinds of structures that we could use to represent various types of knowledge. However, each of those representation types has its own virtues and deficiencies -so each of them may need other connections through which to exploit some other types of representations. This suggests that our brain need some larger-scale organization for interconnecting our multiple ways to represent knowledge. Perhaps the simple arrangement would be a hierarchical one. [….]”
He suggests -from top to bottom: micronemes, neural networks, k-lines and k-trees, semantic networks, frames, trans-frames and narrative stories (top layer)
I'm still making up my mind about the value of this book for me, as some of the conclusions seem too rushed (ie, we have multiple images of self, therefore an unified self doesn't exist") or failing to even attempt to provide an explanation of the phenomena they claim to explain (ie, on the subjective experience it only points to a multitude of stimuli, the challenge of processing them, etc... and it simply stops there).
I do, have to give him credit for a thorough attempt of creating models for each of the examples.
Chapter 5 proposes six levels of processes
“each layer is built on the previous one—until they extend to processes that involve our highest ideals and personal goals”
P146 – levels of mental activities
Self-conscious reflection
Self-reflective thinking
Reflective thinking
Deliberative thinking
Learned Reaction
Instinctive reaction
P2. “Consciousness refers to more than twenty […] processes”
P111. Quotion Aaron Sloman (1992) ‘Developing concepts of consciousness’ “The phrase ‘human consciousness’ typically corresponds to such large cluster of features and capabilities (many of which we don’t understand or know about) that its set of possible subsets is astronomical.
P292-293, Hierarchy of representations. “The sections above have briefly described several kinds of structures that we could use to represent various types of knowledge. However, each of those representation types has its own virtues and deficiencies -so each of them may need other connections through which to exploit some other types of representations. This suggests that our brain need some larger-scale organization for interconnecting our multiple ways to represent knowledge. Perhaps the simple arrangement would be a hierarchical one. [….]”
He suggests -from top to bottom: micronemes, neural networks, k-lines and k-trees, semantic networks, frames, trans-frames and narrative stories (top layer)
December 23, 2020
کتاب تلفیقی از روان شناسی فلسفه علوم شناختی ریاضی هوش مصنوعیه. مطالب کتاب البته سنگینه و نیاز به فکر و تحلیل داره. ایده کتاب اینه که ذهن از بخش های مختلف تشکیل شده که هر کدوم با بخش های دیگه معمولا تداخل پیدا می کنه و ما ذهن واحد و خود واحد نداریم. خود مطلق توهمی بیش نیست. ما بین مدل های مختلفی که از خودمون ساختیم در نوسان هستیم. و اینکه با شناخت ذهن می توان کامپیوترهایی ساخت که شبیه انسان فکر کنند. از نظر تکامل یک خود واحد نمی تونه مسئول تمام رفتارهای ما باشه، چون تصمیمات این خود مطلق و تسلط کاملش مثل غذا نخوردن یا تولید مثل نکردن نسل بشر رو نابود می کنه بنابرین بخش های متضاد به گونه ای تکامل پیدا کرده که بتونه بقیه بخش ها رو محدود کنه. به همین دلیل تضادهای ذهنی یک فرآیند طبیعیه که همه درگیرش هستند.از این نظر هدف یعنی نادیده گرفتن اهداف دیگه. ما مجبوریم هدف هایی انتخاب کنیم تا بتونیم دوام بیاریم و رقابت کنیم.
دیدگاه نویسنده درباره مغز دیدگاه مکانیکی بدون توسل به متافیزیکه. البته این دیدگاه برای خیلی ها خوش آیند نیست. چون جایگاه شون ممکنه متزلزل بشه.
دیدگاه نویسنده درباره مغز دیدگاه مکانیکی بدون توسل به متافیزیکه. البته این دیدگاه برای خیلی ها خوش آیند نیست. چون جایگاه شون ممکنه متزلزل بشه.
August 10, 2020
I found The Emotion Machine to be very enlightening and refreshing. I am a computer scientist and spend some of my free time researching and experimenting with artificial intelligence. I find it interesting to have such an influential person in the Artificial Intelligence community that challenges how we currently approach things (statistical analysis).
I would call this book a "user manual" to the human brain. The system that Marvin Minsky describes in this book has opened my eyes to certain situations in my daily life at home and with friends. Its very thorough and is clearly a collection of many years of experience in the field!
I would highly recommend this book to any psychologist, cognitive scientist, Parent, Computer Scientist, or AI Enthusiast who is looking for an explanation to "How People Think".
I would call this book a "user manual" to the human brain. The system that Marvin Minsky describes in this book has opened my eyes to certain situations in my daily life at home and with friends. Its very thorough and is clearly a collection of many years of experience in the field!
I would highly recommend this book to any psychologist, cognitive scientist, Parent, Computer Scientist, or AI Enthusiast who is looking for an explanation to "How People Think".
August 1, 2021
Unfortunately this book didn't have a central idea/argument/point/thesis. It was very simply a mix match of various vague ideas, and whenever something started to become interesting, the author claimed "We will get to that later" or "This is unexplainable because the mind is a system specifically characterized by complexity".. I was not satisfied with this reading, if you're looking for an introduction on how the mind works, on some very broad AI and consciousness concep, then it might be okay for you..otherwise, I'd skip it! So many topics are also very commonsense, you could come up with the same ideas through some introspection and self-reflection
July 25, 2024
This is actually a psychology book, in case somebody expects something more technical / computer science based. The ideas are interesting and claims seem justified in my opinion. On few occasions The Emotion Machine triggered deeper considerations on my side - mostly around implementation of these ideas - which I think is a good indicator that this is a worthy read. Be aware though that this is a “high level” AI book trying to reverse engineer human MIND from the top in a functional manner - as opposed to a potential neurobiological analysis focusing on human BRAIN or an effort to engineer a human-like mind from available deep-learning building blocks.
December 15, 2024
Worth a read given the author's stature in the AI community. But then again maybe not. There's not a lot here that wasn't in 'The Society of mind'. Dogmatist to the end. Having almost single handledly killed off pattern recognition approaches - competition for funds and prestige was as vicious and ruthless in the 50s as it is today - he continued railing against it to the end. Just not plausible in the late zeros. His tedious standard behaviorist/functionalist arguments that consciousness is not real and just a confusion of language are painful to listen to.
February 9, 2018
Read this book now three times.
Notably, it's the only book I actually made remarks on the border.
Amazingly, this book is able to inspire wild ideas every time. Always something to pick up in there.
Two remarks:
1. the style is not my preferred style, but this is a book I read because of it's message content, not the delivery
2. When reading this book, be aware WHY wrote it and when (stage and experience in his life). Keep this in mind ALL the time!
Notably, it's the only book I actually made remarks on the border.
Amazingly, this book is able to inspire wild ideas every time. Always something to pick up in there.
Two remarks:
1. the style is not my preferred style, but this is a book I read because of it's message content, not the delivery
2. When reading this book, be aware WHY wrote it and when (stage and experience in his life). Keep this in mind ALL the time!
June 20, 2020
The books is interesting with diverse set of ideas organized around an incomplete model of the mind, and how it works. It is more oriented towards psychological concepts of mind and it's working, and computational implementation of those concepts. Modern approaches to engineering the artificial intelligence are other way around , by building on computational principles and scaling them to generate certain behavior.
January 27, 2025
I learned from this book that the idea of a computer "having a mind" refers to a mechanical mind, and philosophically speaking, it stems from humans operating machines, leading to events that are perceived as the machine having a "mind." This perspective is fascinating and thought-provoking, especially for those envisioning the future of artificial intelligence in the 24th century. Highly recommended for deep thinkers exploring AI and consciousness.
March 24, 2020
A good starting point for those trying to understand the brain and/or venturing into the world of AI. The author constructs some useful models of the brain that are consistent with the everyday observed behavior and even (I feel) with the ideologies of spirituality (based on the little I know from my previous reading)
October 21, 2017
Interesting meta-theory about artificial intelligence as it applies to emotions and higher level and other cognitive functions. Complexity and greater granularity seems to be the way to fill the gaps in knowledge of the mind.
November 5, 2018
Minsky's approach to cognitive psychology is unique, and he is a great thinker. This book doesn't capture his best thinking, and it's arrangement (with "dialogues" or comments from imagined readers/thinkers, for example) is confusing.
September 17, 2018
I will surely read it again...
August 25, 2019
Overall well written, especially good at providing time futuristic thoughts.
This entire review has been hidden because of spoilers.
August 27, 2020
AI is impossible without including emotional thinking
January 29, 2021
Просто гениально просто
June 30, 2021
I really enjoyed The Society of Mind but this book is hard to read. I literally force myself. Decided not to continue after chapter 2.7
August 7, 2021
Probably the most poorly written pop science book I've ever read. You would think that a book about how we think should be organized in a way that doesn't simulate schizophrenia, but
Displaying 1 - 30 of 59 reviews