Customer Reviews

The Emotion Machine: Commonsense Thinking, Artificial Intelligence, and the Future of the Human Mind

5 star:		(11)
4 star:		(6)
3 star:		(1)
2 star:		(4)
1 star:		(2)

 

 

Minsky is well known in the field of cognitive research (AI) and his earlier book was very interesting. However, his latest was a great disappointment to me. Part of this is the fact that I have high expectations for him and the book just didn't meet them; part of this feeling is simply that the book is lacking a lot.

It would seem to me if you're going to write about emotions that you would start by trying to understand the biological basis for them. That is, try to resolve the question of their utility - if they evolved as "higher" functions then they should have a major utility.

So the best place to start would be with the biology, medical and neurologists who have studied them. LeDoux's "The Emotional Brain" is the foundation for this area of research. Oddly enough, LeDoux references Minsky's earlier book; however, Minsky does NOT reference LeDoux. This is very odd since LeDoux's work is the de facto standard.

For anyone who has read LeDoux, a number of Minsky's hypotheses and conclusions are erroneous. If you intend to contadict the best theory for actual working "emotional based systems" then you had better have very strong arguments for why this is so. Such arguments are not within Minsky's book.

Instead, we have more vague "thought experiments" and hand waving about agent-based emotional subroutines. Sorry, this is why AI has not developed anything resembling even the intelligence of a wasp or ant in over 30 years...

Go and buy LeDoux's "The Emotional Brain" if you really want to learn something.

I agree with the reviewer who noted how odd it was that a book titled "The Emotion Machine" does not discuss Joseph LeDoux, even if only to refute him. But I think that the problem is with the title, not the book. I found many of Minsky's insights very helpful - it is a very good book about how machines think. And if you are not a dualist, then those insights apply to people too. The book is very well organized and clearly written, and helps you think about thinking. I especially enjoyed his discussion of qualia (although he does not use the term), and why he thinks it is not quite the problem that so many philosophers want to make it.

Minsky's main take on emotions is that emotional states are not fundamentally different from other types of thinking, and that the entire dicotomy of rationality v. emotion is misleading. He prefers to view them all as different ways of thinking - of utilizing various mental resources at one's disposal, some conscious and some not. He organizes his discussion of difficult material very well, but I wish there was more grounding in the underlying neural anatomy of human emotion.

Progress in the design and creation of intelligent machines has been steady for the last four decades and at times has exhibited sharp peaks in both advances and applications. This progress has gone relatively unnoticed, or has been trivialized by the very individuals who have been responsible for it. The field of artificial intelligence has been peculiar in that regard: every advance is hailed as major at the time of its inception, but after a very short time it is delegated to the archives as being "trivial" or "not truly intelligent." It is unknown why this pattern always occurs, but it might be due to the willingness of researchers to engage in philosophical debate on the nature of mind and the possibility, or impossibility, of thinking machines. By indulging in such debates, researchers waste precious time that is better used dealing with the actual building of these machines or the development of algorithms or reasoning patterns by which these machines can solve problems of both theoretical and practical interest. Also, philosophical musings on artificial intelligence, due to the huge conceptual spaces in which they wander aimlessly, are usually of no help in pointing to the right direction for researchers to follow. What researchers need is a "director" or "set of directors" that are familiar with the subject matter, have both applied and theoretical experience in the field of artificial intelligence, and that eschew philosophical armchair speculation in favor of realistic dialog about the nature and functioning of intelligent machines.

The author of this book has been one of these "directors" throughout his professional career, and even though some of his writings have a speculative air about them, many others have been very useful as guidance to those working in the trenches of artificial intelligence. One can point to the author's writings as both inspiration and as a source of perspiration, the latter arising because of the difficulty in bringing some of his ideas to fruition. It would be incorrect to state that the author's ideas have played a predominant role in the field of artificial intelligence, but his influence has been real, if sometimes even in the negative, such as his commentary on the role of perceptrons.

There are intelligent machines today, and they have wide application in business and finance, but their intelligence is restricted (but highly effective) to certain domains of applicability. There are machines for example that can play superb chess and backgammon, being competitive with the best human players in this regard, but these machines, and the reasoning patterns they use in chess and backgammon cannot without major modification indulge themselves in performing financial prediction or proving difficult theorems in mathematics. The building of intelligent machines that can think in multiple domains is at present one of the most difficult outstanding problems in artificial intelligence. Some progress is being made, but it has been stymied again by overindulgence in philosophical speculation and rancorous debates on the nature of mind and whether or not machines can have true emotions.

Humans can of course think in multiple domains. Indeed, a good human chess player can also be a good mathematician or a good chef. The ability to think in multiple domains has been christened as "commonsense" by many psychologists and professional educators, and those skeptical of the possibility of machine intelligence. It is thought by many that in order for a machine to be considered as truly intelligent, or even indeed to possess any intelligence at all, it must possess "commonsense", in spite of the vague manner in which this concept is frequently presented in both the popular and scientific literature.

The nature of "commonsense" is explored in an atypical manner in this book, and in this regard the author again shows his ability to think outside of the box and phrase issues in a new light. This is not to say that advice on how to implement these ideas in real machines is included in the book, as it is not. But the ideas do seem plausible as well as practical, particularly the concept of a "panalogy", which is the author's contraction of the two words "parallel analogy". A panalogy allows a machine (human or otherwise) to give multiple meanings to an object, event, or situation, and thus be able to discern whether a particular interpretation of an event is inappropriate. A machine good in the game of chess could possibly then give multiple interpretations to its moves, some of which may happen to be similar to the interpretations given to a musical composition for example. The machine could thus use its expertise in chess to write musical compositions, and therefore be able to think in multiple domains. On the other hand, the machine may realize that there are no such analogies between chess and musical composition, and thus refrain from attempting to gain expertise in the latter. Another role for pananalogies, which may be a fruitful one, is that they can be used to measure to what degree interpretations are "entangled" with each other. Intepretations, which are the results of thinking, algorithmic processing, or reasoning patterns as it were, could be entangled in the sense that they always refer to objects, events, or situations in multiple domains. A panalogy, being a collection of interpretations in one domain, could be entangled with another in a different domain. The machine could thus switch between these with great ease, and thus be effective in both domains. It remains of course to construct explicit examples of panalogies that can be implemented in a real machine. The author does not direct the reader on how to do this, unfortunately.

The author also discusses a few other topics that have been hotly debated in artificial intelligence, throughout its five-decade long history, namely the possibility of a conscious machine or one that displays (and feels!) genuine emotions. The nature of consciousness, even in the human case, is poorly understood, so any discussion of its implementation in machines must wait further clarification and elucidation. Contemporary research in neuroscience is giving assistance in this regard. The author though takes another view of consciousness, which departs from the "folk psychology" that this concept is typically embedded in. His view of consciousness is more process-oriented, in that consciousness is the result of more than twenty processes going on in the human brain. An entire chapter is spent elaborating on this view, which is highly interesting to read but of course needs to be connected with what is known in cognitive neuroscience.

It remains to be seen whether the ideas in this book can be implemented in real machines. If the author's views on emotions, commonsense, and consciousness are correct, as detailed throughout the book, it seems more plausible that machines will arise in the next few years that have these characteristics. If not, then perhaps machine intelligence should be viewed as something that is completely different from the human case. The fact that hundreds of tasks are now being done by machines that used to be thought of as the sole province of humans says a lot about the degree to which machine intelligence has progressed. Whenever the first machines are constructed to operate and reason in many in different domains, it seems likely that they will have their own ideas about how to direct further progress. Their understanding of ideas and issues may perhaps be very different than what humans is, and they may in fact serve as directors for further human advancement in different fields and contexts, much like the author has done throughout a major portion of his life.

Early efforts to model human-like thinking with machines using rules were interesting but failed in a number of ways to capture even simple ways that humans think. Marvin Minsky, AI pioneer at MIT, insists that we understand the mistakes and can begin to appreciate how the mind actually works in functional terms from the lessons we have learned. Learninig from our past mistakes, what a novel idea.

To put this into perspective, the question of whether a machine model can adequately describe a brain has long been considered in terms of either strong AI or weak AI. Most people find weak AI plausible: computers can solve certain kinds of problems better than humans. We mostly balk at strong AI however: machines can literally think like humans and solve the same kinds of problems just as well.

In The Emotion Machine, Marvin Minsky presents a very machine-like architecture that he claims actually represents the way real minds probably work in fundamental respects. That sounds pretty much like strong AI. So a lot of people will reject the concept of this book out of hand. I think that would be a mistake. Minsky has done a very good job identifying plausible specifics of why AI programs have failed to deliver on, where they have actually managed to deliver, and speculates on how we can fill in the gaps.

No, he doesn't spend time arguing against Searle's Chinese Room or other conundrums of AI, he just presents his case and gives examples in a clear, simple, accessible way. And I am persuaded that he probably gets a lot right. Probably more than he gets wrong. And that's a lot better than a lot of critics will give him credit for because it goes against both the mainstream disdain for strong AI and the mainstream love of flashy neuroscience images.

Minsky skips right on past the issue of connectionist networks vs. semantic networks and simply posits that we had to evolve semantic representations at some point. How is left as an exercise for neuroscientists. There is a lot of "details to be filled in later" sort of thinking here, so don't look to this book as a detailed physical model of the brain. This is a high level functional model of the mind and I like it.

So I claim that this is an important book that seems to promise a 21st century reboot of scientific naturalism as our guiding philosophy for the future. Minsky takes on nothing less than an overall architectural model for the mind in natural terms. It is brilliant. Too brilliant to be appreciated in its time because Minsky makes complex ideas so accessible that the biggest challenge for this book is that people will not appreciate its power. It reads like a simple AI model of a mind, but it is much deeper than that because of the amount of deep thought that has gone into it and the consideration of the weaknesses as well as strengths of previous AI programs.

We are currently in the grip of a widespread fascination with poorly understood pop neuroscience, and most readers will be deeply disappointed that this book does not attempt to wrestle with brain science at all. I think that's a strength because it means Minsky is not falling into the weird metaphysical spins that we too often see in pop neuroscience books, especially those by non-researchers and over-enthusiastic under-trained journalists.

What Minsky is doing here is simply coming up with a logical model of what a mind has to be able to do to provide the capabilities that we observe real human minds to possess. Sounds simple, right? No, not at all. The reason Minsky has accomplished something special here is that he recognizes many of the powerful fallacies we usually fall into when we introspect about thinking and rely on traditional models. We tend to think of emotions and reasoning as separate kinds of things, and then we talk about how they are both needed and how they interact. But as Minsky points out, both neuroscience and psychology seem to provide us evidence that these are points on a continuum, not different kinds of things. Minsky takes that seriously and builds on it.

The result is something amazing that looks like a simplistic mechanical model of the mind but captures some deep insights into how minds really work.

The central implication of Minsky's model is an epistemological stance that resourcefulness in human thinking is a matter of switching between different kinds of representations, each used in a different way of thinking, each of which captures something essential about specific things in our world while neccessarily leaving out other details. A mind can't comprehend everything at once. Some decisions simply don't have an optimal answer because they look different from different angles.

The key concept underlying Minsky's model is that minds as we think of them had to start with simple rules for recognizing and responding to cues, had to be able to incorporate goals in some form in those rules as well, and then eventually had to be able to recognize kinds of problem and activate appropriate ways of thinking. It makes sense to think of this in terms of logical levels of recognizers and responders, and importantly, what Minsky calls "critics" and "selectors," where each new level provides some way to resolve conflicts that arise in the level below it.

So conflicts in our instincts can be resolved by learned rules, conflicts in learned rules can be resolved by deliberation strategies, and in turn levels with different kinds of representations of the problem and eventually the problem solver and their own ways of thinking. Once the problem solver can represent themselves and their own thinking, we have the power to shape our own thinking in meaningful ways.

I'm really not doing justice to this book in this review, because it's power is in the details of his examples and how they illustrate the architecture at work. Suffice to say that I think if you find a functional architecture of the mind of interest, I highly recommend this book. I think it gives a much more fundamental understanding of how minds most probably work than any amount of flashy recent brain scans, and certainly more than untestable holistic and quantum mechanical theories will ever tell us until we better understand the functional design. Neuroscience in the future will, I believe, be filling in the details of a framework very much like this one.

The ''Emotion Machine'' by Marvin Minsky is an introduction to how our minds work.
An endeavor to understand mind (thinking, intellect) in terms of its design (how it is
built, how it works).
Sure, the inner workings of the mind sometimes appear to be impervious to any kind of
scientific approach. Routine stuff like making mental models of the world, design plans,pursue goals and feel desires aren't all that routine, when you think about it. And the way it all plays together - from simple, instinctive kinds of thought to more complex forms, such as consciousness or self awareness - is obviously very complex.

It would be very easy to get stuck in too much detail or to be too superficial. But somehow Minsky finds the right balance. So overall, the Emotion Machine is a brilliant introduction to how our minds work.

Often he will give us an idea about what must be going in computational terms, and then shield us from the usual deluge of ''neuro-technical'' terms. Because (in his own words) ''research on the (actual nitty gitty of the) brain is advancing so quickly that any conclusion one might make today could be outdated in just a few weeks.'' In one reviewers words: ''The Emotion Machine rewards careful reading. You'll learn a lot about how your mind works, even if you won't be all that much wiser about what is actually going on within your brain.'' :-)

Nevertheless, the book is obviously based on the latest advances in computer science, psychology, neuroscience, engineering etc. And certainly, his book is not a bad place to start if one wants to be a little wiser on what goes on inside our heads.
A wonderful book.

-Simon

I am currently reading various neuroscience, emotion, psychology, and even anthropology books. This one has, as of yet, and without a reasonable doubt been excellent, and I highly recommend to anyone even considering the thought of what we as humans hold. It's quite easy to read, so even a child would understand it, but the information herein is absolutely beautiful to know. It is definitely a favorite, and I must say that Prof. Dr. Minsky is one of the brightest minds this generation may have available right now.

A good book with interesting ideas. However, there was a fair ammount of rehash from his other book (Society of Mind), there was alot of conjecture with not much expereimentally backed up theory, and of course no sourcecode!

Marvin's last major book The Society of Mind was chock full of creative new concepts, e.g. the brain is not a monolithic whole but a collection of cognitive subsystems with quirky interactions. The long awaited sequel, portions which have appeared on his website since the mid-1990s, starts out with an interesting title- implying emotion is an interesting and essential component of cognition. Several other articifial intelligence researchers agree- that language and robotic interfaces wont be effective unless they capture emotional nuance. However, only a couple of chapters address emotion, while the rest deal with general topics in cognition and artificial intelligence. Unlike Society of Mind I found few new ideas in this survey.

The other major problem with the book is its choppy flow. It reminds of a power point presentation in outline form. Many, probably too many lists, diagrams , and quotations break up the sparse prose making it slow reading. Its like a set of lecture notes not entirely fleshed out. I've been spoiled by the calibre of recent science writing tending toward high journalistic standards.

One of the great pioneers and researchers in Artificial Intelligence world Marvin Minsky takes on this work the deconstructing of our emotional Mind, Consciousness, Self in the hope of providing a kind of algorithm for artificially creating the Artificial Feeling Intelligent Being Machine of the future. Or that's in any case how I interpret what he is doing.
He looks at various mind- functions such as goal-making, resourcefulness, the way pain becomes suffering, and engages in explorations of the subjects aimed at providing understanding. He in doing this uses students' queries and replies, and also quotations from the best minds he can find who address the problems. The result is amusing if difficult chapters, interesting forays into providing better insight into how the old machine inside us, or upstairs from us ( just joking) happens to swoon when it hears 'Moonlight Sonata' or cry when it is told that it has dirtied the floor with crumbs after the better- half has just spent two hours washing it.
I take what Minsky has done as a kind of scientific post- modernism, a reconstruction and breaking down into various narratives what obviously mysteriously operates as some kind of whole.
To give a bit of flavor of what he is doing I take a piece from his website in which he summarizes chapter one of the book.

"Summary: This chapter addressed some questions about how people acquire the goals they pursue. Some of these are instincts that come with our genetic inheritance, but others are subgoals that we construct to achieve other goals that we already have. I also conjectured that some of our highest-level goals are produced by special machinery that lead each person to try to adopt the values of other persons who become what I call that person's "Imprimers."
Imprimers are parents, friends, or acquaintances to whom a person becomes 'attached,' because they respond actively to one's needs--and they then can induce special feelings in us, such as guiltiness, shame, and pride. At first, those Imprimers must be actually present, but older children form `mental models' of them, and can use these to evaluate goals when those imprimers no longer are on the scene. Eventually, these models later develop into what we call by names like conscience, values, ideals, and ethics."

If you like the sample you should go for the whole book. It may not give the Answer but it will certainly provide a great deal of insight and information into how the old 'thinking reed' wakes and shakes as it lives and dies.

Marvin Minksy's follow-up to Society of Mind (SOM) is a more legato paraphrasing of some of the ideas in SOM. As a whole, it may not be the most incredible read, but each sentence or idea in itself is profound. After reading this book, many other books seem flowery, fluffy, and self-important. True (as others have noted) the title is a little misleading, but Marvin's ideas and concepts like panalogies, levels of thinking, imprimers, etc. are solid platforms for the things that humans experience but often don't know how to think about. Moreover, the concepts are general enough to apply to all phenomena of human action, behavior and thought, but succinct enough to be considered rules and logic that will hopefully serve AI coders well in the future...