7 ms·
How to Create a Mind - Book Review
- reader5000 14y agoI think trying to understand "the mind" by way of the brain is like trying to understand Windows OS by way of the CPU. It's just not the way to do it. Having said that it seems like the book reviewer's central complaint was To create a mind one needs at a minimum to create consciousness.... I'm quite skeptical of this claim simply because "consciousness" is such a sketchy concept: http://www.skeptic.com/eskeptic/08-02-27/ http://www.skeptic.com/eskeptic/08-02-27/
- mtrimpe 14y agoI couldn't agree more. It seems that whenever computers become able to do something that we previously used to define consciousness, we just create a new definition of what it means to be human/conscious that doesn't include that new capability. Essentially we're continuously redefining consciousness as "that which humans can that machines can't." My guess is that we'll keep expanding that until human consciousness is just "the version that operates in flesh and blood rather than silicon".
- seiji 14y agoThat's one of those clever things people seem to bring up -- "oh, we used to think playing checkers required consciousness, but it doesn't." At no point did someone think a computer would wake up, sing Hallelujah, and start composing public policy when it finished playing checkers. Consciousness is clearly: the ability to have directed agency in the world (i.e. "wanting" to do something), following an action path to create a desirable outcome, and forming opinions/judgements/actions based on prior experience ("memories"). Fun thought experiment: when does a baby become conscious and not just a bundle of predefined cry-poop-eat behaviors? Every animal walking around in the world is consciousness (if you can "turn off agency" in the form of sedation during surgery, the thing was consciousness), but human-level intelligence consciousness only forms in our squishy brains at this point in our world line.
- mtrimpe 14y agoMy point is more that computers are already conscious, but that we're constantly redefining our definition of consciousness to exclude it. From a philosophical perspective it's saying that consciousness is a property of reality rather than of humans / animals and that we're merely choosing to define an essentially arbitrary subset of it that's sufficiently similar to our own as 'true consciousness'.
- seiji 14y agoMy point is more that computers are already conscious Saywhatnow? consciousness is a property of reality Again, saywhatnow? Perhaps your usage of 'reality' is throwing me off. Consciousness is just a very high dimensional aggregation of "stuff" (e.g. synaptic connections) with a processing framework (e.g. neural architecture) for recall/storage/action based on outward perception and internal state. We're both agreed that consciousness doesn't require squishy brains. I'm just saying human level consciousness, at this point in time, only exists in our squishy brains. We'll figure out different ways of creating computational consciousness in the future, but we aren't there yet. Our tiny recommender systems, search engines, face detectors/recognizers, and "my first backprop neural net 101" aren't conscious.
- mtrimpe 14y agoI can tell that the full gist of what I'm trying to explain will be hard to do in an HN comment, but for simplicity's sake let's just say that I'm defining consciousness as a scale which includes "my first backprop neural net 101" (and your table, but that's the saywahtnow part of a would-be philosophical discussion ;) Given that assumption, all I'm saying is that as computers progress along that scale we are currently continually moving the point on that scale at which something becomes "human level" to specifically exclude computers.
- omarchowdhury 14y agoAny time I see someone that thinks they can put conciousness into words, and they start off by saying "Consciousness is just..." I know that person has no clue.
- AJ007 14y agoDid you read the book? Because that is one of the points Kurzweil makes. The book review in question is crap. I'm happy to read Kurzweil critics but claiming something is too complex to understand to does not make for legitimate criticism, in any field. Pattern recognition is a given, how to do it is not. That is really what the book is about (hint: hierarchical hidden markov models.) As for the philosophical stuff, I read The Age of Spiritual Machines many years ago when it came out (and I was barely in high school), so there was nothing in How to Create A Mind that challenged views I've had over a decade to think about. The whole issue of consciousness I think is minor. Which, I personally would explain consciousness as "living" awareness ( in the book Kurzweil says even the smartest people start to sound stupid when trying to describe this.) The good news is our industry is not going to get dull anytime soon.
- mtrimpe 14y agoJust to be clear, I didn't read the book. I even just skimmed the original post. I was just responding to the parent comment and it's claim about the sketchiness of (the definition of) consciousness. Perhaps I should read Kurzweil once even if just to see how much our views are aligned..
- pbw 14y agoYou absolutely could understand Windows by studying the CPU, you latch on to some of the traces and record all the signals going by. That's what electrophysiology is. Oh, there's KERNEL32.DLL, I wonder if that's important, let's look at the contents. We constantly study signals within the brain, not just the dry physical connectivity.
- worfly 14y agoThe CPU, RAM and non-volatile storage architecture do tell you a lot about what a program can and cannot do. Furthermore, if your Windows OS does something magical on the screen over the course of a minute, you can be confident that there is a definable limit to the amount and type of code that could have been executed in order to achieve that "magical" result. A more accurate question might be, "can you recreate the running code of the Windows OS, given an installed desktop running Windows OS". Oh, and we also have some copies of the source code--if that helps (but it is hand-written in Mandarin). :)
- andrewtbham 14y agoI don't understand why the author is adamantly opposed to the book. The author, Colin McGinn, states his only credentials as a psychology student in the 70s. I believe Kurzweil is likely correct. Pattern recognition is the essence of human intelligence.
- andrewtbham 14y ago"I am thinking about London while in Miami I am not recognizing any presented stimulus as London—since I am not perceiving London with my senses. There is no perceptual recognition going on at all in thinking about an absent object " You could see Miami and you could recognize it as similar to London. That is pattern recognition.
- sp332 14y agoHumans are good at finding (or projecting) patterns, but there's more to a human brain than that. Memory is one of those things which is not just pattern-recognition.
- andrewtbham 14y agoI suppose I don't see a distinction between memory and pattern recognition. To me, memory is implied in the definition of pattern recognition. The definition of recognition is "the identification of something as having been previously seen, heard, known, etc." i.e. memory
- fmoralesc 14y agoNot necessarily. If you have some stymulus like > rrrrrrrtrrrrrryrrrrrurrrrirrrorrpr you can recognize patterns in it even if you have never seen, heard, or known them before. EDIT: of course this example is maybe too simple.
- andrewtbham 14y agoi completely agree with that. that is what i am intending to convey. the point of others and the author is that pattern recognition doesn't explain the mind because it doesn't account for memory. my point is that the term recognition obviously implies a mechanism for memory.
- stcredzero 14y ago> There is no perceptual recognition going on at all in thinking about an absent object. So pattern recognition cannot be the essential nature of thought. This point seems totally obvious and quite devastating, yet Kurzweil has nothing to say about it, not even acknowledging the problem. Is it really the case that my brain is not recognizing patterns when I remember something? I don't think this is the case, otherwise I would never realize something about a remembered object after the fact. > He does in one place speak of dreaming as a “sequence of patterns” and he might try to say the same about thinking. But this faces obvious objections. First, even if that is true, there is no pattern recognition involved when I dream, or when I think about London and my friends and relatives there. Seems to me the brain is taking random stimuli and piecing it together into a narrative when one dreams. Sounds like there could be some pattern recognition involved there. Also, the author has never spoken of absent friends and said something like, "Doesn't Bob look sad?" Is he remembering the recognition of sadness, or is he recognizing sadness from the memory, and how could one be sure of the difference? Methinks the author either knows something left unexplained, or isn't being nearly as clever as he thinks.
- seiji 14y agoI think the review author means physical in-your-face "perceptual recognition." Thought remembering is a weaker form of physical recognition, but you're right -- it works mostly the same as far as recall is concerned. With physical recognition (touch, temperature, smell, taste, sound, sight), all the details are instantly available to our perception and recall/memory access is easier. It's difficult to fully remember the feeling of a childhood holiday only from your memory. You get a much richer recall experience if you go back to the place/time of year you lived it. Little things will trigger your memory you didn't even know you remembered. The review author would be less confused if "pattern recognition" was replaced with "memories linked to what you're thinking about." Saying "pattern recognition" of London doesn't make sense to non-tech people. Saying "You remember things attached to London in your mind" helps out more, right? I like that we are kinda strange.
- seiji 14y agoWe can all agree Kurzweil is past his intellectual expiration date. The saddest part here is he wasted such a good title, "How to Create a Mind," repeating his historical non-usable drivel (not to mention it sounds like a Culture series book). My current "create a consciousness" track is to get something with pattern completion completion working (NB: aspie nerds often fail this step interpersonally ("Oh, Hello!" "blank stare"), but not intellectually/professionally ("What does this thingamajigger do?" "Oh, well, that's there because historically we used a 3.2 bit kernel to operate the Arm of God and... [30 minutes later]... Why are you falling asleep?"). The thing inside your head stringing your actions together into a coherent after-the-fact story isn't pattern recognition or even pattern completion, but pattern completion completion. You have to take the context of the new world generated by your prior pattern completion and run a new internal world simulation based on that, then take action. This does leave out intrinsic or emotional motivations pulling you hither or thither, but that's a vague writeup for another time. Though, there are plenty of smart people running around in the world with seemingly no emotions. Emotional feeling doesn't seem to be a prerequisite for consciousness (or for being usable to be exploited by employers to make them 10x-1000x richer while you get 0.5x richer). It's fun to think about at least.
- andremendes 14y agoI would suggest the lecture of Raja Yoga by Swami Vivekananda. I remember in the beggining of the book, it touches this topic in a way that I can't describe using my English habilities.
- carwithcookies 14y agoI'm glad McGinn criticizes Kurzweil's use of anthropomorphic language. I'm only in CH 3 and have been waiting for him to translate his friendly, talking recognizers back into neuroscience. Also bothersome is Kurzweil's repeated but unacknowledged use of Chomsky's ideas about recursion in language.
- tehwalrus 14y agoWhile the reviewer is mostly correct in their objections, probably, I have to disagree that "red" or "the smell of a rose" are not patterns. The brain perceives by receiving electrical signals from neurons, whether from the optic nerve (which, incidentally, encodes both colour and geometry) or the nerves from the nasal cells. These are going to arrive in the brain as identical chemical 1s and 0s for the brain's sensory pattern recognition circuits to deal with (or whatever.) Independent thought, recursive concept matching (smell -> memory of childhood, memory of childhood -> fear, adrenal response) and all the rest are a strong point of the criticism, though, since these are obviously not based on pattern recognition.
- pbw 14y agoMcGinn says that "thinking about London while in Miami" involves no pattern recognition, because no senses are involved. But Kurzweil's claim is pattern recognition is key to memory, and certainly memory is involved when recollecting London. It's fine to disagree with Kurzweil's claim about memory, but to imply he's discovered a big loophole ("this point seems totally obvious and quite devastating") and suggest Kurzweil thinks patterns only apply to our senses, is completely silly. McGinn's tirade against anthropomorphic language is also silly. It's obviously something he's trotted out from previous writings, which is okay if it were insightful. Can't he mentally translate "a neuron shouts when it sees a pattern" into "a neuron fires when a certain combination of it's input terminals are excited" or perhaps instead of "fire" which is potentially anthropomorphic itself, Kurzweil should write a script to replace "fire" with "the voltage-gated ion channels embedded in the plasma membrane will open allowing an inward flow of sodium ions, etc. etc". Please, where does it stop. Look beyond the words to what the person is actually saying. It's similar to his criticism of Kurzweil's "law" of accelerating change. Sure, it should be called a theory not a law, big deal. Kurzweil is a huckster, he is selling books, he's going to overstate things. Just make the mental translation law->theory in your mind and move on with your life. McGinn meanwhile has no opinion on what is actually going to happen. He says accelerating change is "just a fortunate historical fact about the twentieth century". So is he saying it has stopped now, 13 years in to the 21st century? I don't believe it has. Or that it will stop? When will it stop McGinn, in 2015, 2029, 2045, 2387? Because this matters a lot in terms of how much computer power we end up with. He has no opinion at all what is going to happen, he only cares what we call things.
- fmoralesc 14y agoYes, McGinn's step from pattern recognition to perceptual pattern recognition is argumentally bad, and kinda jarring in the article. I think McGinn's point is that human cognition involves creative elements (imagination for example) that don't seem well modelled on pattern recognition, but that doesn't come across that well. McGinn's tirade is not again antropomorphic language per se, it is against homunculism. The idea that a human mind is composed by agents (homunculi) that exhibit mental properties leaves the mental properties of those agents unexplained. That's the reason he proceeds to argue agains the use of "information processing", because he seems to find that also explanatorily lacking. > It's similar to his criticism of Kurzweil's "law" of accelerating change. Sure, it should be called a theory not a law, big deal. It should be called a generalization, or maybe an hypothesis, it isn't even a theory. McGin's question is: is this generalization valid? Can we say? If he sees fault in Kurzweil's evidence for it, why not say it? Why would he have to provide evidence as to when Kurzweil's prediction will fail? The burden of proof is not on him.
- goldfeld 14y agoI was really interested in what a Mind-Book would be, and how one would go about about reviewing one such book. I was disappointed.
- dschiptsov 14y ago"Kurzweil's new book on the mind is magnificent, timely, and solidly argued! His best so far!" ~Marvin Minsky
- worfly 14y agoI think this question is central to his disagreement with Kurzweil, "Are we seriously to suppose that dreams and thoughts have geometrical shape?" Which to my ear, belies his fundamental apprehension of geometry.