Dan Dennet, D.Phil. - GoldLab Symposium 2018

welcome to the afternoon session of the first day of the symposium it's always exciting to come back it's it's odd I've only been coming three years and this is a huge meeting and yet somehow it still feels like coming to like an extended family or a group of friends so it's just it's always a wonderful experience so it's my pleasure to introduce the afternoon slate of speakers so we'll start off with Dan Dennett oh wait but not right now right right now Oh leave it the break sorry I am really bad at following instructions so I'm glad I have Larry here to assist me so Bob wants everyone to know that he's changed his talk to be less scientifically intimidating and so if you were thinking of cutting out because it sounded boring please stay yes I got two hands raised okay sorry about that so back to the the introductions so I'm excited to be able to introduce Dan Dennett so Dan is a co-director of the Center for cognitive studies at Tufts University and also the Austin B Fletcher professor of philosophy I have to keep this short I was told so I'm only going to mention in addition his most recent book from bacteria from bacteria to Bach and back the evolution of minds was published last year and on that in 2012 he was actually he was awarded the Erasmus Prize in Amsterdam and I promised I would keep it that short so Dan take it away [Applause] Icahn yes good thank you very much Casey and thank Thank You Larry for inviting me into this this is a wonderful community you're putting together here and as soon as we get Justin here we go with a little [Music] [Music] okay there we go I just want to say something about that we okay that'll do this is the empirical part of my talk because I've been arguing for more than a quarter century there is no Cartesian theater in your head there's no place where the show goes on between your ears there isn't there isn't it seems like there is but there isn't and people say well are you making a conceptual point or an empirical point I said both the empirical point is when we look inside we don't see this that's I think I'd have to agree that's not what the neuroscientist find when they open up the head the conceptual point is it's perfectly possible that could be the reality on some planet and not on ours not with us and the further conceptual point is at some point you're gonna have to go into the head of the homunculus that's in there and do the consciousness of that one and if you're going to avoid an infinite regress you better start early and in fact empirical points on this planet with us with this species it starts right away there is no Cartesian theater there is no little homunculus in the middle pushing the buttons looking at the screens listening to the stereo sound and running your body do I have any disagreement about that because I tell you that message is very hard for many people to accept they keep sliding back into that view and what I'm going to talk about here is some parade cases of this difficulty and I want to try to make it as easy as possible for you to slide back into that view and then I'm going to try to show you how to get out of it so there's going to be some some exercise here okay we're ready to switch now to the PowerPoint I wasn't gonna trust the embedded okay so here we go and sure enough it's my first slide by the way this is also in my book and what you see on the left is an Australian termite castle and on the right is the Sagrada família of the famous Gaudi Antoni gaudí church in Barcelona they are remarkably similar eerily similar not just on the outside on the inside so these are two artifacts made by animals but probably profoundly different R&D and construction methods the termite castle is made by millions of basically clueless termites they don't know what they're doing they don't know why they're doing it they're very competent and they don't understand much of anything and yet this amazing structure arises Gaudi on the other hand it's just the opposite this is small I small D intelligent designer this is a human being who's got blueprints and manifestos and and worked it all out from first principles and he's lording it over his subordinates who were learning it over their subordinates who were learning it over the people at the bottom who themselves know a lot more than those termites do so we have these two very different ways of making an artifact the bottom-up competence without comprehension way and the top-down comprehension comes first way and the puzzle is how do you get to the second way our way the intelligent design way of making things from where we start and the reason for that is that our brain is made of neurons this is a nice this is a few neurons in a petri dish so what have you got you've got 80 some billion by latest count neurons in your brain they are not even as smart as termites they are clueless myopic they don't know what they're doing or why and somehow you can get a human mind a gaudi mind a Larry mind any of your minds we can get one of those minds out of the coordination of 86 billion clueless little idiots neurons how is it done and I'm going to tell you thanks to the lovely talk by David Clemen this morning I can give you a tiny couple of sentences that will you'll know what this what the basic my basic answer is to that question which if you want to know the details many nice turns in those details you have to read the book he showed you the fabulous Tree of Life the trees of life and the great moment lynn margulis really gets a lot of the credit for it it was marish coughs key before that of endosymbiosis the fact that you had a coming together of two unicellular entities which had the only independent histories of rd for millions and millions of years and I come together and you get this tremendous technology transfer boom just like that and they travel together thereafter the difference between a termite colony and a human brain is that well a colleague of mine has a nice expression he once said you can't do much carpentry with your bare hands and you can't do much thinking with your bare brain a termite colony is a bear brain no thinking tools our brains on the other hand are hugely equipped with thinking tools where do we get them we didn't have to build them all ourselves you got calculus and reading and writing and all the rest of your thinking tools from your culture this is the second great endosymbiotic revolution after the after the eukaryotes what we are where you primate we're primates with infected brains our brains have been colonized by hundreds of thousands of culturally evolved thinking tools and toys and junk and that's the secret of our success we don't have bear brains we have elaborately equipped brains with many thousands of tools and tools for making tools and tools for making tools and so forth so that's the short answer and I'm not going to say anything more about that millions of termites eighty billion neurons trillions of cells there's the question how can an aggregation of trillions of cells make a person that can love and notice and wonder a termite colony is competent but not conscious now you know we're Paris's you remember your college years you understand the word promise your neurons don't they don't now if you listen to Christof Koch you can talk about your jennifer aniston neuron met some interesting research but even if we allow for the oversimplification of his account its many neurons not just one the Jennifer Aniston neuron doesn't know Jennifer Aniston from anything it may be excited whenever Jennifer Aniston is the topic but it doesn't know that it's not conscious it's clueless these little agents signal without understanding and yet their activity somehow enable you to understand and notice and love and promise so one problem a very big problem that we're now working on is how do you organize systems of neurons and their many many helpers ten times more astrocytes how do you organize them into something that is a you that has a that has a self that that has memories and purposes and integrity plans and makes promises and loves and hates and all the rest of that but I want to start with something simpler and one that's more this is a great distractor for many people little experiment here now just yeah I'm pretty sure it will work here I want you just to look fixate on the white cross for a few more seconds for a few more seconds I'm pretty sure that's long enough what do you see alright everybody see the flag okay here's a question the lowest short red stripe didn't intersect the black cross do well let's go back we'll try it one more time really quickly ready ready ready ready ready ready ready okay how many say it did all right now what are you talking about there was no Red Cross on the screen no red there was no red stripe on the screen there was no red stripe in your brain there was no red stripe in your retina there was no red stripe anywhere anywhere it just seemed like there was if there was a Cartesian theater that'd be a red stripe on a screen looked at by the homunculus but there isn't a Cartesian theater so there's no red stripe one more example I will see if this one will work just stare one of those little black dots in the center works for me how many of you see rotating motion yeah it's pretty good surprise surprise nothing is rotating on the screen and nothing is rotating in your brain nothing nothing is rotating in your brain there is no rotary motion anywhere what there is is representation of rotary motion but it isn't done by turning something it's done in some other system of representation so now we're ready for two obvious facts first there's as I say no second transduction the light streams in your eyes the photons hit your rods and cones bla bla bla it's transduced by the by the by the rhodopsin in your in your in your cones and rods in two spike trains your neural spike trains those are not transduce back into something else like color its my trains all the way that's what we learn when we look inside there's it's the whole job is done with Spike trains that means that there aren't any qualia to be presented because there's no place where the qualia appreciator resides to appreciate the qualia they're being presented all that work has to be broken up and done by lesser functionary in the brain that even don't know what qualia are a DVD it can represent motion sounds colors shapes no microscope you can literally where you're going to see those motions those colors those shapes because those are representations in a different coding system for all of those wonderful properties compare the DVD with good old-fashioned film celluloid notice that on film on each frame of film there's actually color you're using color to represent color that simple fact has implications I think that are hard for people to come to grips with they think okay my eyes my rotten cones they transduce everything into spike trains okay now we're sort of in DVD mode we've got these pulsing signals of some sort when do they get transduce back into color when are they shown through the rejector they aren't never ever they don't have to be because the job of the appreciator the recognizer of red and blue and motion and all of us those that those are jobs are broken up into a lot of neural specialists that are themselves unconscious and they just do the work aha I want to drive this my life I learned I have to people really have to come to grips with one little fact here suppose you made an autonomous weaponized drone I don't heard you two and suppose you decided out of a sense of conscience or something or because the dark that told you to that it needed to recognize red cross vehicle so it's not to not to drop its munitions on red cross vehicles so you'd need to make a red cross recognizer okay state of the art it's possible right now you get a color video camera and you write a red cross identifier it's just the same range of the sorts of problems where you have programs that can recognize handwritten numerals and so forth so this is a solved problem notice what you wouldn't do you wouldn't build into the drone a little TV screen that showed the Red Cross in color and then have a little television camera looking at that TV screen no you you cut that all out and you go directly from the bit streams coming from your digital camera into the algorithms that are going to identify the Red Cross and your home you got it you got it you've got a Red Cross recognizer well but the drone isn't conscious right in what sense well although it notices the Red Cross it can't notice that it's noticing a Red Cross that's another layer and one of the amazing things about our minds is that we don't just notice things we notice that we notice things and we notice that we notice that we noticed things and it is this recursion if I may use that word it's this reflective echoing principle which makes possible our brains and only our brains to be the creative intelligences way beyond the beaver building its dam way beyond the bird building its nest that we have between our ears what you do is you wouldn't transduce the colors bring the colors back as you would if you had a DVD and you popped in your DVD player and then looked at the colors on the screen you wouldn't transduce you just translate you would translate you would have algorithms for translating the format that the the coatings coming direct from the camera you'd be massaging those coatings in various way so you've got them in two user-friendly versions for the next part of the system and what that does is it creates a sort of user friendly user illusion of one part of the system for another and that's basically what my claim is about human consciousness it's your human consciousness is the user illusion that the brain has of itself and it's what makes it so easy for you to do all the things you do you don't have to worry about all the rubble complexities of what's going on inside your brain because the brain has evolved systematic ways of boring over those details and giving you translations and not transduction translations into formats from which the information can be derived information about whether there's a red cross on the top of that vehicle for instance there's no transduction into a show in your head there is no inner audience that just seems to be it spike trains all the way no how does that work this is what I call the hard question not the hard problem that's David this is the hard question I asked this in 91 and a book back then my hard question is and then what happens and I have an example here well first I said then one of the occupational hazards of neuroscience seems to be the tendency to think of consciousness as the end of the line then I we've reached the finish line end of research here for instance is a hypothesis hazarded by Francis Crick and Christof Koch we have suggested that one of the functions of consciousness is to present the result of various underlying computations okay and that this involves the attentional mechanisms to temporarily bind the relevant neurons together by synchronizing their spikes in 40 Hertz oscillations they never went on to ask and what does that accomplish right you've got the forty Hertz of synchronization and then what happens they just stopped it they didn't have any more to say on the subject so a function of consciousness is to present the results of underlying computations but to whom the Queen Creek and Coke do not go on to ask themselves the hard question and then what happens well that was many years ago I was 91 a couple of days ago tackling I was on my way out here when I got this Scientific American in the June issue Christophe has a piece on consciousness and it says the origin and nature of these experiences sometimes refer to as qualia have been a mystery from the earliest days of antiquity right up to the present scientists are beginning to unravel a mystery that has long vexed philosophers many modern analytic philosophers most probably perhaps Daniel Dennett of Tufts University find the existence of consciousness such an intolerable affront to what they believe should be a meaningless universe of matter and the void that they declare it to be an illusion ouch well Christophe and I have a long history but he's not alone just a few weeks earlier a non-scientist Christophe sir scientist he's the director of the alain lat Allen Institute in Seattle one of the top neuroscience research outfits in the world Christoph's not alone few months earlier Galen Strasser who's a well-known philosopher Strawson wrote in the New York Review of Books about the silliest claim ever made by me yeah I'm honored and he calls it the denial this is the great silliness we must hope that doesn't spread outside the Academy or convince some future information technologists or robot assist who has great power over our lives that's you folks so be careful if you listen let's go back to Christophe talking about spike trains they've got a new technique for studying their their dynamic evolution as they unfold it over time these traces recordings of neural activity and what he calls the hot zone each corresponding to a specific location in the brain below the skull then he says this yielded a movie no it didn't think about that drone no movie in that drone there's no second transduction now this is I've discovered a real problem because we have people like Jack Gallants lab at Berkeley there they're doing work where they get a lot of data and then they've got these these rendering algorithms which make movies that you can look at and they do not go out of their way to say that's not what's going on in the brain this is a sort of poetic license this is a a dressed-up version of what's actually going on and so these people are actually feeding on our love of that cartesian theater and the idea that we've got to still have it there now so coaxial asks the wrong question then people ask me you know what's philosophy good for and I said well philosophy is what you're doing whatever you're doing when you're not sure what the right questions to ask are and what we philosophers try to do is help you find better questions to ask and even scientists sometimes ask the wrong question and when they make breakthroughs it's usually because they've been doing a little philosophy really they don't call it philosophy and they may do it to the seat of their pants but they are thinking outside the box they are reassessing their very starting points and some of the assumptions but look what what Christophe has to say the abiding mystery is how and why any highly organized piece of active matter gives rise to conscious sensation and he goes on to put it this way what is it about the biophysics of a chunk of highly excitable brain matter that turns gray goo into the glorious surround-sound and Technicolor that is the fabric of everyday experience well if that's the question you ask then you're going to be gobsmacked now and forever and that's of course David Chalmers hard problem and the solution to it is to say don't worry about the hard problem worry about the hard questions and then what happens worry about actually figuring out how the machinery works how it enables people to have all the convictions they have all the reactions they have all the all the memory updating all the adjustment of priorities and so forth if you can solve all those programs there won't be anything left for that is that Technicolor movie in your head recently this illusion that I've been talking about has been nicely diagnosed by a man named Richard power whom I just met last fall here is Richard power some of you may know him he's a retired AI researcher in England and we met at a meeting in Edinburgh and here's his he just sent this to me out of the blue and he well he says we understand the concept of representation from extra representations such as pictures or verbal descriptions for these representations we can have direct experience of both a representor the portrait painting and a representing the person painted call these the medium at the content well I found this on the web does a nice example it's not of a person it's a landscape and you can see the landscape painting in the landscape and notice you can and we all have done this on here you can tell look from what the other right there you got them both in the same perspective and you can mash up the blobs of pigment here with the color you see out there right that's that's our model of representation in short we conceptualize the medium of our internal representations by abstracting some features from the content and attributing them to some kind of spiritual or ghostly substance that's the best we can do since actually we cannot experience the medium at all and have to look for analogies in the external world this is the conceptual scheme we're that we bring to internal representations because it's the only one we have but there's a huge difference for external representation we can experience both medium and content oil-on-canvas as well as people trees or whatever but for internal representations we do not experience the medium at all the idea that the medium is some state of the brain seems intuitively absurd so powerful is the illusion that we're dealing with an iconic representation in the medium of spirit so using this image is our crutch we imagine that there's an internal picture of the world in some sort of like colors but they're not there's no light in there after all but we tend to attribute properties to the medium which are in fact properties represented by the medium that's the distinction we have to keep in mind now I mentioned Galen Strawson and his remarks about the silliest claim ever made here's what he says we know something fundamental about the essential nature of conscious experience just in having it no we don't Descartes was just wrong let me just give you on the on the inbound path here are things that we don't know from the inside first of all we just don't know how vision works from the inside you have to remember not so long ago in Aristotle's day they thought that vision worked by projecting something out from the eyeballs honor the front surfaces of objects it's not obvious from the inside how it works it's taken two thousand years of of optics to figure out what vision how vision actually works most of us are surprised where we learn about cicadas those little jumps that your eye makes you're not aware of them at all but you couldn't see without them then there's the all the retina topic maps that have been discovered in in in cortex who know that we're going to be retinotopic map so we certainly had no clue about that then there's the fact that para phobia Lee your vision is very coarse-grained and you don't have any color vision out there either you can prove that to yourself I have a simple test to do take a deck of cards look in the mirror pick a card hold it up here facing you but look straight at yourself in the mirror and see how far up you have to have it right about here before you can tell whether it's a face card or what it's number is whether it's either red and black its suspending demonstration of the coarseness of peripheral vision well if we're that clueless about the inbound path what makes us think we're any better on about what happens next we don't but we think we do and that's why one of the reasons the hard question is an ass lock is that scientists Crick and kokum a classic example of this they tell their wonderful story about the astonishing hypothesis and they stop right at consciousness and people readers say oh thanks science for laying the groundwork but you're now entering the region where we are the authorities and we'll take it over the task of explanation if you don't like the fact is we're not authorities and we shouldn't take that job away from the scientist that's the only way we're gonna have a theory of consciousness all of the comprehension appreciation delight revolts from recognition amusement etc the human beings experience must be somehow composed of the activities of billions of neurons myopic in the extreme but how that's the hard question and it's hard because the general answer is almost anything can happen deep learning systems don't understand yes and this isn't a bug it's a feature because the parts of your brain the corresponding parts of your brain don't understand either the trick now is to see how to take those non comprehending competent fabrics and put them together into an architecture that can achieve understanding that's the next step and current work on turning deep learning fabrics into architectures is going on I don't have time to talk about it I recommend there's quite a bit in my book about this the global neuronal work space developed by de Honda's colleagues in Paris oh it's time for me to stop very quickly I'm almost at the end anyway what are liquid brains it's a new idea it's the idea that your neurons are not locked in to a whole bunch of identical nodes the way the comparable units in a digital computer are there you had 86 billion little individuals in there and they can shift around make new connections and once you start thinking about brains as composed more like termites in fact the the the little units are can be more resourceful they have to fend for themselves they've got agendas which are staying alive and so they're ripe for being rounded up into coalition's where they can get assistance and so we have a whole new range of explorations of computational architectures using these semi mobile units that are very labile not like the sort of the nodes of connectionist networks which usually have two or three degrees of freedom max and the only way to control such a system is by what an engineer would call a virtual governor and I don't have time to explain what that is but probably most of you or many of you know anyway and again some people hate this idea Daniel Dennett is the devil there's no internal witness no central recognizer mean no self others an abstract center of narrative gravity virtual governor which is itself nothing but a convenient fiction pretend it's not the case of the emperor having no clothes it's rather that the clothes have no effort exactly if you think you have a theory of consciousness and you still have an emperor in there you don't have a theory of consciousness you haven't begun to have a theory of consciousness what it's hard for people to realize is that a good theory of consciousness when you go strolling around through the model it's going to be like a deserted factory all this clueless machinery doing all this stuff and there's nobody home if you still have an inner witness in there pushing the buttons you simply postpone your theory of consciousness thank you very much [Applause] so we'll see if we can prioritize questions from the back just to get a bunch of voices do you want to start up here and then we'll work our way Larry if you have a question we'll come to the front I promise sorry we'll defer to multiple Larry's if there are questions thanks dr. Dennett I was a wonderful presentation thank you I wondered if we don't if humans mammals don't corner the market on consciousness then what implications does this have for AI and happy people been bugging about that oh I've been thinking about that for 40 years I have been on record for many many years a it's strong ai ai which is conscious is possible in principle but also it's I think extremely unlikely because I think the difficulty is orders of magnitude greater than the enthusiasts of thought and this is a there's a tradition of over optimism and hype in AI and for all the wonders of deep learning and machine learning those are fabrics they aren't architectures and my current hunch is that if you say whoa okay let's put the fabrics together and they're people who are doing this now and what they started doing is putting the fabrics together by trying to swat them into the slots in good old-fashioned AI architectures going back to soar and things like that from from 30 years ago and those are the wrong architectures - you need to have an architecture for parts that are more open-ended that can develop new architectures on the fly and rather than being being restricted within an architecture I could go on and on about that but leave 2/3 of the audience bored stiff Thanks it was really gone 1991 book that opened my eyes to this consciousness thing but to continue the theme what I think you just said is that there's no philosophical reason why an artificial intelligence system should not have a consciousness of some kind and isn't that what we're all afraid of well yes I think in a way it is and I've also argued that although it's possible in principle it is neither practical nor is it desirable we've got plenty of intelligent designers around this room is full of them and we've got all the more that we can use and what we should instead think about is making not artificial colleagues but smart machines smart machines that don't care smart machines that really don't comprehend that don't have agendas of their own and we can we can treat them like vacuum cleaners we don't and and part of what I would do to encourage that is I would try to reverse a 75 year old attitude which is owed strangely enough and unfairly in a way to touring with touring created the Turing test he created a test which valorized fooling people into thinking you're human and ever since then a lot of people in AI have developed techniques for making machines see more and more human as if this were in itself a good goal I don't think it is and I think in fact we should start penalizing discrediting criticizing the Disneyfication of AI systems make them user friendly yes but don't ever make people think that they're talking to a human agent that understands things that they do until you have one that does and we're not going to see that in our lifetimes then I want you to touch on a topic that you often combined with your philosophy of mind which is the evolutionary perspective we tell us a little bit about how come we have these conscious about feelings and what use they are what selective advantage we get from them Thank You Larry I well I have to now boil down a chapter into a couple of sentences I mentioned degrees of freedom engineering to him drone has old say six seven degrees of freedom and you can control them all with the joystick you know or you can make them autonomous how many degrees are freedom in us millions millions of degrees of freedom and wherever you have a degree of freedom you've got to try to control it and what human consciousness is for is for trying to control all of those degrees of freedom that we get as a result of having our brain filled with all of these thinking tools and we can achieve a level of self control and but only because we're conscious and only because we can notice our noticing the phrase well it seemed like a good idea at the time if you laugh because it's notoriously a sort of doofus thing to say you know it's not a sign of stupidity it's a sign of the greatest intelligence any entity any agent that can say and mean well it seemed like a good idea at the time look what he's got a deficit memory for his own thinking has got to be able to evaluate it has got to be able that's very high-level and it's our ability to use that to debug ourselves to some degree very hard but that's what we're doing and our human consciousness is the arena the global neuronal work space which makes that evaluative reflection possible now maybe one more thanks Stan when I was a kid I was in school over in Thailand at bang cigarette American school and a lot of my classmates were Indian by growing up in the West most of us the change our consciousness you know we drank something we smoked something we laid with somebody so and so forth but I reached a kind of Tiffany my early 20s I can engineer three and so I had a good book on yoga about doing the pranayama and so one day I kind of just I just really held to it and I I wouldn't give up and I was steadfast and and I know this is not going to jive with most people here being scientists but I I went in you know I did the great exhalation and it was it was about midnight and then when I came out of it it was about dawn and service hours had gone by and so theoretically after five or six minutes you're in dead without breathing so I had not been breathing for some time and not quite sure anyways so that was when I guess the term is himself realized after that and so the big thing today that's being pushed is mindfulness and altruism and I was just wondering going through your slide presentation how you might kind of incorporate that into your background thinking tools are apps that you download to your neck top and what you can do depends on what apps you've downloaded and yoga systems like that those are apps too and some of them let you do things you can't do otherwise and so one you get away from the idea that all the all this all the intelligence is somehow for instance just in the genes just in the genome or just in the in the brain in the hardware and you what you have to realize it's only the hardware as configured by the software that gives you the power and every experience you have as that one final point I learned of a nice phrase the other day it's a variation on something Piaget said and I've been thinking about it ever since and I think it's it's as good as a motto for the whole goal symposium intelligence is knowing what to do when you don't know what to do that's a wonderful note to end on thank you thank you [Applause]

Channel: GOLDLABCOLORADO

Views: 5,261

Rating: 4.875 out of 5

Keywords: Medicine, Health, doctor, Dr, PhD, talk, education, science, GoldLab, Symposium

Id: hsz6pqOKwOE

Length: 46min 50sec (2810 seconds)

Published: Sat May 19 2018