On Consciousness with Giulio Tononi, Max Tegmark and David Chalmers

you consciousness is the way is the way to transform is it is what actually transforms syntactical information into semantic information and vice-versa that would be my workable definition of consciousness now in other words it would be like what converts symbols into meaning but now meaning cannot be another symbol meaning is something different than a symbol and as you know computers physical structures and so on are simply transformers of meaning into symbols into other symbols so a computer takes a symbol works it works on it through it a logarithm of whatever transform it into another symbol but the crucial transformation from symbol to Meaney requires consciousness the meaning that I mean is qualia meaning is feelings is comprehension in fact meaning has structure is not just meaning it is a structure in fact comprehension is finding meaning within meaning so it's a recursive structure and therefore comprehension it's like finding long-range correlation between the various meanings that are connected together so artificial intelligence claims that some artificial intelligence workers claim that it will be possible to make a conscious computer and I claim that that is not possible based on the definition that I have given of consciousness because a computer simply trust transformed symbols into other symbols in fact max max would be right to say that physical reality is mathematics if in fact within of consciousness because mathematics is a simp simple manipulation and physics will be described simply by symbol manipulation and there will be nobody conscious to actually feel anything or have any experience in that world so to me ontology resides within the meaning aspect the conscious aspect of reality now I'm working on a beginning framework to explain more about this what I can speak now because I only have a few minutes it's called CIP framework and I believe that we have a very good theory of the cognitive observer in other words a conscious observer in how this conscious observer in fact relates with physical reality in a way that implies an interaction between two spaces the space of consciousness in the space of information which is the space of symbols thank you thank you so um I'm tempted by already get a chance to talk about consciousness the other day I'm tempted to say something about IIT but maybe it'll come up in the discussion instead I thought maybe just say something general about why physicists and cosmologists ought to care about consciousness I mean I see attendance is slightly down at this at this session so there's a variety of opinions on that on this today I'm giving a very nice talk this morning on why physicists don't need to worry so much about the hard problem of consciousness they try and think there's something to battling well if you're very ambitious physicist who wants to see physicist physics is giving a theory of everything then you better have something to say about the hard problem how all this will explain consciousness or if you've got the strong kind of view we're observers is gonna play a causal role in physical dynamics then you better have something to say but I think other than that then maybe it's right that physicist and cosmologist don't get to get too caught up in the philosophical arguments about the problem of consciousness but I think it would be a mistake to conclude from that that physicists and cosmologists shouldn't care about consciousness in the sense of subjective experience at all I think at least the kinds of physicists and cosmologists who are interested in the role of the observer which is something that came up a lot at this meeting I think people who are interested in that really do have to think about consciousness in the sense of subjective experience and the basic reason is that well I mean the central role for the observer that almost everybody agreed on here with at least plays an epistemological role we've got some evidence and our physics has to at least be able to to explain and be consistent with our evidence and our fundamental evidence is the evidence of subjective experience that's the one thing that we really we have to save the data save the direct evidence it's subjective experience that we have to say so this comes out I think in a number of ways in the debates at the conference say the debate over Boltzmann brains you know the kind of Boltzmann brain that we worried about the kind of Boltzmann like improbable scenario that we worry about is the kind of scenario that reproduces our subjective experience you know a Boltzmann neuron is presumably not gonna be a problem for for our theories but merely predicts the existence of Boltzmann neurons why because we don't think Boltzmann neurons gonna have the kinds of subjective experience that we do Boltzmann brains are a problem for our theory precisely to the extent that we think that those Boltzmann brains will have subjective experience if your theory of of experience says that you can have brains of that form without experience and it's not a problem for theory so in the debate the other day about whether you need Boltzmann solar systems or Boltzmann brains I think Alan's at all.i we only need the brain because that's what we need to save the evidence that's relevant precisely because it's going to reproduce subjective experience now in that case it's like okay so that needs certain claims about the connection between physical processes and experience like a brain but pretty plausibly duplicate our experience and maybe that's not a strong claim but it's a claim about the connection between the two very relevant to assessing the role of observers in our theory likewise in quantum mechanics where a lot of people see is one of the major problems for many interpretations of quantum mechanics how to reproduce the standard observations of the physical world and once again what really needs to be needs to be reproduced they sure have pointed mechanics could give rise to subjective experiences like like this one how in the quantum wave function for example as a challenge say for the Everett view how is it and this Everett like quantum wave front you get experiences like this one one way to do that would be by showing you get a brain like this one in an environment like this one everything else in the world but the minimal thing that theory needs to do is to get experiences like this one I think at that point again thinking about the connection between physical processes and subjective experiences really is really vital in fact in a book I wrote about 20 years ago I tried to defend the Everett view by arguing that in Everett's our wave function you'd implement every computation that would be implemented in a branch of the wave function by itself and then combined with a separate claim that reproducing the right computation will give you the right subjective experience that was meant to help solve the problem for the other of you how can you make it consistent with our observations and maybe that was right maybe that was wrong but that's just a general instance of the fact that if - in order to argue that these theories are going to what you need to do to save the phenomena is to make sure these theories are going to be consistent or maybe even predict our subjective experiences - think about that you at least need to think about the connection between physical processes and experiences that doesn't mean solving the hard problem but it doesn't mean at least giving some serious thought to consciousness well that that beautifully segues in some of our remarks later but let's thank David for those so the meeting organizers have always asked the moderators to let sparks fly and in these panel discussions and I'm finding that a little tricky here because it's such a AIT heavy contingent so I'm gonna rely on the audience when we get to audience Q&A in a bit to really pound on these theories and ask your top questions yes I think there's been a function that's weighted in a certain direction but there are two lines or questions I'd like to that's developed before I do open it up for general comment and the second box would concern some of your remarks on the connection to the other themes of the conference but first I'm noticing a lot of people are coming to IIT really maybe even for the first time today or it's an unfamiliar theory for them so there's maybe some some IOT 101 that'd be fun to to to get into and and one questions I've always had about the theory and this is really for anyone or everyone who's who's involved and done work on the IIT or thought about it how does it connect how does this higher level of abstraction and this is very much in the spirit of what Jen was saying connect to the dynamics so what is happening when I experience the world what is actually going on in my brain to bring about the sensations and when I become conscious of something how does that how does that occur according to according to the theory I think Julie overall looking toward you so I had one slide showing a network with units of connections among them units functional and at the fixed point so that the network will stay there like that indefinitely so in some sense nothing was happening and yet IIT predicts the network would be conscious so nothing needs to happen in any sense what matters is what exists so it's being versus happening and this connects a little bit to one question that was raised Carlo and not answer which is one of the difficult things to understand about IOT so IIT tells you that you take a system so a network of three simple binary and an elements connected in a certain way it would be an example and you want to know what actually exists there you need to unfold its full cause-effect power that network in that state at t0 so to speak and of course you know carlo came up with a question which is a natural question to ask so you are telling me if I paraphrase you correctly you tell me you're telling me that something is what it is by how it constrains so to speak the past and the future its own past and future or rather it's inputs and outputs I would say and that what is doing right now doesn't matter okay now it's not exactly it doesn't matter it matters normally has to be in a particular state but what something is is defined but it's causal power not by what is happening or what is doing and one very simple way to at least think about what that means is take again two simple logic gates take an or gate which is on and that end gate which is on just like that okay and then ask yourself what makes something as a being an end gate versus an or gate they're both on there both gates with two inputs in one output okay to find out what it actually is it's not enough to look at what's happening when they're both on it's not enough to look at what might happen later okay what you need to do is figure out the cause effect power and when you perturb them in all possible ways and look at what they do you can see then that one is an angei the other one is an or gate define based on its constraints the constraints it goes onto itself so the state of the system matters in the sense that he provides further constraint in addition to the connections it says this set of connections in this state constrains the past in the future in the following way in that you need to do to figure out actually exist instead a comment on this with my physics hat on which is I think we physicists need to be a bit more humble when we interpret observations of things because we usually naively think of us the experience or seeing something is actually experiencing what's out there which is a clearly nonsense because I've had some very interesting dreams where I experienced being in a room and seeing people even though there was absolutely no sensory input at the time so so we know and Julie likes to emphasize this that even that there are certain states of information processing in our brain you know whether IIT is right or wrong that field subjectively like there's an outside world so as physicists it's pretty obvious whether what's really going on is our sensory organs and so on they are constantly taking in input and then using that to keep updating this internal world model so it doesn't just kind of hallucinate on its own like when you're dreaming but it tries to correlate with what's actually happening out there but what we're fundamentally experiencing when we say I observe the dial and my measurements thingy to be in position seven is something in our head maybe I can bring in muscle for me and larysa on this as well I mean let me try to be more precise suppose I see a rose or how do i how does that how did the qualities of the rose leave an imprint in my brain according to this theory has it changed my life streaming of consciousness so your your system that is forms the major complex which is what we call the the substrate of consciousness is not isolated it has inputs and when we click it thinks we take them as background conditions so in the calculation they're fixed but of course they can dynamically influence the system so what would happen is I show you arose it activates your retinal neurons it activates some neurons in your collar areas and everywhere and so the reason that you experience a rose is not that I showed it to you because that all the right neurons changed their state to the state of seeing a rose and that state of this neurons constrains where they probably came from so what the probable inputs are were for this neurons and it constrains what probable effects they will have and all these constraints form this conceptual structure where part of that structure corresponds to the rose and the rest of the structure corresponds to the rest of your experience so that would be the conceptual answer so and likewise Masafumi when I am conscious of being conscious how does that state of the network how was it communicated to the network how did can the network respond to it itself in that way yeah it's kind of a higher-order consciousness it's not just an external stimulus that enters the network it's the property of the network that has to enter the network how does that happen to be conscious actually that the system has to actually itself and that part questions that system kind of determines a cause and also the system after observing itself system has to do something that power is a common effect responding to some calls and both need to be neatly dix to be conscious that is there ìit so I'll Frederico please what I find difficult to understand is whether Phi is a measure of consciousness or simply a measure of a complete complexity holistic complexity of a physical system they could support consciousness in those two concepts are very different so so III don't see how we could call fie consciousness or qualia should I take this all right so this is something that happens many times to many people especially if you give a 20 minutes presentation but even in an hour and even sometimes after reading the papers so it boils down to our here is a measure this measure of Phi which is interesting mathematically maybe and in a mock show that is actually not completely mistaken maybe and so on and it's some kind of measure of complexity and then there is this giant claim made that that should be the quantity of consciousness okay but first of all let me say there is more than that as I said this is a conceptual structure which is an infinitely more complicated things than Phi is not a single number it's a giant structure that is supposed to correspond to the quality a particular experience you are having but even augmented by the quality it seems like there's this big claim being made why would that be the case what is typically forgotten is what I spent roughly 30 seconds about which is the axioms of the theory so it's a key thing you know in fact it is the fundamental thing at first the first step is always the most important step you don't start from the physical world develop some kind of measure of complexity and say oh maybe that's consciousness ok would be nice everywhere you start from consciousness itself that is the data that david says you cannot throw away your own consciousness and you say what are its essential properties the ones that are true of every experience those are the axioms those accident happen to be those 5 I think after very long consideration and then you say well the physical world which is a postulate from within consciousness let's remember that a very good posture about certain hypotheses that we make from within consciousness has to support those properties that consciousness has to begin with so it's not a measure that one then develops and claims they might have something to do with consciousness but it is consciousness that leads directly to the relation of that measure so I very much admire this this feature of IIT that starts with consciousness and kind of reads off kind of deduces I can us have some axioms of consciousness and let's just figure out the mathematical properties that a physical system we just have to have in order to have those features of consciousness but I think this actually has to do a lot of work for the theory given that empirically it's so hard to measure Phi and measure consciousness I think the empirical support is you know fairly sketchy and so this I think is really is the big pillar of support for the theory so it doesn't mean I think we need to pay a lot of attention to those those motions early in the papers and the talks where Giulio goes from the axioms to the postulates to the to the mathematics because that's very much playing a load-bearing role and I guess one point where I'm just a little bit skeptical I like the axioms but ago at a certain point these axioms get turned into postulates and they get turned into mathematics and so quite transparent all the steps from the axioms to the postulates to the mathematics you get the sense you know something is being slipped in you know some extra highly specific content is being for example slipped into the mathematics along the way so let's say it's given this support for each other it's a kind of transcendental deduction of how the great Pro philosophy how the physical world would have to be to support consciousness but yeah you want to pay very very careful to how very very very careful attention how those steps in the deduction go and yeah you wouldn't we don't surprised to find that another another researcher to Nonis star it ended up in a somewhat in a very different place from very similar starting point but David can you be more specific about what this kind of explanatory gap the theory has between the axioms and the other levels so they have something that they can respond to it comes up in a few places one player I think it really comes up is in the principle of exclusion sure how do you state the axiom of exclusion media's consciousness is contains what he contains not less not more and he runs a display of temporal grain it runs not less not subscribe FAC that's a principle that as a given subject has a state of consciousness which is one very definite state of consciousness it's got boundaries it doesn't contain other things somehow that gets translated into a principle about within a given physical system if there is a subject with a certain degree of something to reify in another sorry now the system with a certain degree of Phi and another system with a lower degree of fire that only the first system is conscious of systems right and the and the subsystem is not to someone we've got a principle that subjects having different states of consciousness now to a principle that system supporting distinct subjects you might have thought of okay well in principle you could have had two different subjects they're both with definite states of consciousness without doing anything to contravene the exclusion axiom that subjects always have a a definite state so we've moved from this single subject principle to a cross subject principle in a way that looks I think to many people as if something new is being slipped into the theory not really it's a good point but since you know you bring in transcendental deduction meaning can't I'm going to bring in the principle of sufficient reason therefore lively it's okay and we'll see who wins between liveness and cont but but having say that so exclusion is in fact perhaps the most difficult axiom of the theory but it is yes based on a single subject you know so but what does it say essentially it says countenance is definite so definite means it has borders things are in or out and it could be interpreted as if it were saying there is only one experience that is being had but it only says really my experience is definite it has borders when you move to the posture in the translation and believe me there is a lot of work in how to do that and it is indeed a fundamental issue to make sure that that work is not cheating okay when you move there the idea is so if existence is cause-effect power constant exists physical substrate must have cause effect power if intrinsic existence it cause effect power on yourself if composition is the parts of course effect power if information is a specific course of problem not generic if integration is its cause effect power of one entity not two and so on exclusion means thus cause effect power must have borders must specify a cause effect structure with borders how do you find the border so it must have them okay and since existence is cause effect power this physical existence there one then the only way for the principle sufficient reason if you wish to find the borders is to ask what exists the most and that is the only existence based way of getting the borders there are no other ways so while there is still work to be done to sort of prove the uniqueness of the derivation of the postulates from the axioms there is it's not really proven I think you know much of the effort has precisely being that mathematize in those and making sure that we have just exactly the right set of axioms and just exactly the right set of postulates that uniquely translate those axiom but the key here is prolific power given that I'm the only no supporter of the IIT in this at least I have questions you know in other words I mean I appreciate the theory and I really think that it's a major step forward in understanding the nature of consciousness don't misunderstand me but I have issues with it and I would like those issues to be clarified only peeps specific what is the issue is the one that I mentioned early basically if you postulate the existence of consciousness since postulated in your axioms then how can you claim that consciousness emerges out of it has a part of a physical system is already postulated there so so all the can emerge is what this physical system allows to emerge of something that already exists so consciousness has to be a primary aspect of nature isn't it yes and it's actually our starting point that that consciousness is the one thing that we know and the one thing that we know exists and maybe the way to to understand this derivation is that the primary thing is consciousness and through consciousness because it has structure we postulate that there is a physical word out there that that is the reason for the structure in our experience so then we know that there is a physical or we inferred that there is a physical world now we we take all these properties for from consciousness and we basically project them onto this inferred physical world and we pick those parts of the physical world that have the Bertie's - to correspond to those characteristics and so we're not saying that so you can't make you can't then go and say oh all that we're base our theory on is that we identified some things and they give rise to consciousness no we had consciousness and we project down and this is the set of physical substrates that we identified to have the potential to give rise to consciousness the first thing until I attend is completely consistent and with Federico's claim that consciousness is a fundamental aspect of the universe through that lens would just be that the fighter consciousness principle would be regarded as something like a fundamental law that connects consciousness to other principles I take it it's also consistent with a reductionist view we're all there is to consciousness is integrated information but as far as I could tell the theory is consistent with both views has been patiently waiting I just wanted to make an analogy inject both some caution and some optimism into this so I want to make an analogy between two different subjects which both start with CEO both of which I've worked a lot on in my career and both of which started out being extremely controversial and branded as flaky by many I'm referring to cosmology and consciousness so in the checkered history of cosmology it used to be largely derided as philosophy only sorry physicists often use your wonderful field as an insult and and not much progress was made for for thousands of years cosmologists were called by a famous physicists as people who were seldom and were often wrong but never in doubt and a lot of very compelling sounding arguments based on principles and axioms and so on were made for example that ultimately you had conclusions that were completely wrong such as that our universe should be static you'll never not be changed even Einstein fell into that trap what was it that changed cosmology into a more respectable science now it was ultimately data get back to that and and so in other words a lot of these variant were a lot of them very nice sounding predictions but they were testable and most of them now are in the dustbin of history and what survived we take very seriously the reason I'm like I T so much is not because I am I'm convinced that I T is correct no I'm not convinced of that at all maybe it's correct maybe it's not what I love about IT so much is it actually makes testable predictions and after just like cosmology suffered the Dark Age very very long period of time when there's basically no data you could drink beer and speculate but there was no measurements we got all this wonderful data and now it's more scientific I feel we're on the cusp of exactly the same thing happening with consciousness there's this avalanche of high-quality data about what's happening in the brain from outside that you can compare whether your subjective reports and begin to test these things so I'm gonna take you very hard noise physics is approach to this and only be convinced not just because someone has eloquent arguments but because testable predictions are tested and found to be correct might say that it's a beautiful property of the theory that it's testable in principle but actually extremely hard to test in in practice right the course of it so hard to measure Phi and it was easily testable it might have been tested and refuted or refined right but as I was arguing in my talk I think we're actually on the cusp of being able to start making some real tests so let me start with a quick poll because I guess every panel has to do that the speed round will a theory of consciousness help us with fundamental physics yes or no you're not even allowed to say it's complicated yes or no okay definitely yes yes yes yes okay actually I'm noticing a queue there that I've been completely ignoring so maybe we'll go to that and I've got some other questions we can come back to if the queue is exhausted so please I'm sure it seems to me like my impression of this development of IIT it seems like there's a impression that I should be some sort of unique measure but if I want to talk about the fluffiness of a sheep it's not really clear to me that there should just be one measure of the fluffiness of a sheep so especially in the arguments about what whether or not there's a derivation of five from the axioms there's not been in my mind an argument that there should even be a single formula that we should be able to derive from nexium so maybe you could explain to me if there should be or if there's not why I'll say one quick thing about this essentially everything we've tried to do for many years is based on the notion that consciousness is the way it is and therefore it cannot be arbitrary how it is instantiated and therefore measured and it cannot even be done in two or three similar ways like the fluffiness of the Sheep so for instance this means that when you actually choose the distance is it cool but Lila is it earthmovers distance as we are doing okay the whole exercise was actually in finding what is the right distance measure it cannot just be some distant measure that more or less approximates what consciousness is consciousness is the way it is and we need to find the correct answer there is only one answer in my mind so how how would the theory be falsified if you put a cat in an MRI scan machine and see a positive five would that falsify the theory or how would you do that now if you see zero fire in a cat I will be worried if I see a good amount of fire wouldn't be so worried what if it's Schrodinger's cat it can't be both ways but whatever the point the point is though is maybe David made a comment that you know saying the evidence is sketchy etc in as usual in theories there is a predictive aspect and we saw some explanatory aspects there are many things that IHC explains there are basic things why this piece of brain doesn't do it and this piece does it or maybe not even prefrontal cortex does it even it's a piece of cortex all these are very basic facts ok and IIT has a principal explanation of them but basically you find nowhere else there are many other things like that and there are also very specific predictions which we won't get into but I was briefly mentioning to Sean before that for instance one has to do with space one very precise prediction ability which are testing is if you change the strength of a connection in this back part of your brain of just some connections between neurons it will deform the phenomenal perception of space it will contract it even though the activity of the neurons the state of minerals doesn't change at all that's a very country into it prediction to most neuroscientists the results actually seem to support that psychophysical results and another prediction is the one I showed you before that an inactive but functional brain would be conscious and that's a game to the country in truly the most people so I think these are in the tradition of counterintuitive predictions in physics something that you can actually test with AIT and we artists before we go Federico let me let me ask you and bring you into this is there something you would like to apply as a test a falsifiable test no I'm not at that level in my development I've been working for about three years and this in terms of you know real real work so no I don't hear very good please a quick yes/no question for the panel so does consciousness exist at a more fundamental level than the space-time matter level and there could be a more simpler level so max would probably say it's mathematical the basic level but the question is does consciousness exist at a more fundamental level great what's this matter yes or no yes no and no evasion this time I have an equally fundamental level not at the moment okay fair enough I do more fundamental level it's complicated no no yes yes its intrinsic existence as opposed to for somebody else we've all been behaving as if the number of consciousness is in this room is equal to the number of people but that's a very individualistic way of looking at things it's often said that if you have a lot of people together in a room the number of opinions is greater than the number of people runs less and that the whatever we have that distinguishes us from inanimate matter may be more like what Jim Hartle said the other day when he says that the whole scientific establishment or the whole of human civilization is just one I guess not a lot of different conscious consciousness is and perhaps other less individualistic philosophies would look would pose the problem of consciousness that way now my other thing is I guess it's not a question either but that Scott Aaronson has criticized IIT is producing a lot of false positives things that you wouldn't think are conscious that would have high values of it so I guess that's something people should consider let me just he'll Don the first part of your remark in particular I'll direct this question to you David is IIT boner pill to a dancing clearly a type of argument that you could switch back and forth between levels and consciousness with fade in and out well here is actually one place where I think it might be might be vulnerable to that style of argument Julio's claim that a computer simulation of for example a brain on a one Normand computer will have no consciousness because it has zero fight whereas say a neuromorphic version would have would have hi-fi and so it would have consciousness I think we could you could imagine a gradual transition from a euro morphic version to a 111 version where we kind of take the direct connections between neurons and route them through a CPU with the basic reason why they why the monoamine version has zero fires because there's a one momen CPU through which everything passes and somehow that makes that brings fire down to zero but I think that we could have a gradual transition where we replace all these parallel connections by one element connections gradually the functioning of the system will main remain completely the same throughout five will at least by Giulio's account go from something very high down to zero my consciousness will go from something very high down to zero and you know the question is what happens to my consciousness on the way I'm gonna do lean either gonna have to win count at some point well that won't happen on Julia's view it'll gradually go down despite the fact that my my behavior is is the same throughout it's really hard to make sense of those intermediate stage these faded consciousness states where yes I'm here behaving and even thinking normally that my consciousness is fragmented down to down to just bits without me realizing and we probably could do dancing quality which is going back and forth between the neuromorphic modes and the bonhomie inversion and consciousness which is you know now without ever showing up with my functioning I find all those predictions extremely counterintuitive and best avoided if they can so I'd much prefer a version of the theory that aside hi-fi to the one women simulation of a the brain but maybe I can some say something in direct response to the series so I mean a lot of people really like this idea of a group of consciousness but um phenomenologically we know about our own and that some has has limits itit has borders and and so the way people think about group consciousness is typically oh I want all of the people here to have a consciousness and then maybe I want 50 people together to have a consciousness but if you relax this exclusion this exclusion postulate or the axiom that that only Maxima of off consciousness exists it's not just me and the group of 50 people it's also me mine one urine - two neurons - three neurons - five neurons all these would exist it's me - a molecule it's me - and so it's not just like oh I excluded this one other possible consciousness I excluded millions of ridiculous multiplications of my own consciousness thank you yeah so a fairly radical suggestion that's been floated once or twice in the conference is that conscious creatures don't follow the laws of dynamics as we understand them in other words you have to take account of their consciousness in order to predict exactly what they're going to do I just wondered we could pull the panel on whether they think that's like what what their credence is well they take the chance of that main trails and if they could say in a sentence why I mean it's at least a possibility I take seriously given the I think is a very good argument for consciousness being fundamental in nature once it's fundamental in nature it'd be great if we could play a role very natural disposed it plays a role in the dynamic something the one place we're supposed obvious to look for to play a role in the dynamics is of course quantum mechanics where traditional views suggest some people go left or right well we've developed interpretations of quantum mechanics that don't give a role to the observer well that's great and I like a lot of those views but hey well there are interpretations of quantum mechanics also that do give a role to the observer and consciousness I think we should at least be trying to developers in analogous detail 20% if you consider the a loop of observation in action then you have a understanding which is quality quality and that means that you can down generally symbol that interact with other symbols in physical reality okay then you can get the information the transformation or reality done by physical reality - the symbol that you inserted into physical reality you receive that symbol you convert it into a different meaning and that means that you can now have a loop of action through the physical world they effects the physical world through the meaning the increased meaning that you had through this cognitive loop that's the way I look at it but it still it's very hard to say very course at this point man it needs to be it needs to be refined much more but that's the way I look at an observational process is not an event is actually a process and is there loops they repeat and then increase comprehension max did you want to comment on the question as well yeah my guess is that we is maybe I'm a small minority on the panel I think my guess is we don't need anything beyond standard physics protons and all that stuff and the reason for that is by an analogy I think we have many times in physics made the mistake of blaming or failures done on in vote by info on invoking the need for some additional magic fairy dust ingredient I've before Newton people a lot of people thought that the reason the moon wasn't falling down even though rocks did was that somehow heaven objects had different properties and earth objects and his genius was to do away with that extra fairy dust and realized that it was all just the same physics similarly not that many hundred years ago people thought the difference between a living bug in a dead bug was some extra life sauce fairy dust wears now you just think of a dead bug is broken same physics and I'm by analogy I I think it's very tempting to and to say that we need something extra to get the subjective experience but I would love to try that hardest first to see if we can explain it with just what we have and it's exactly the same minimalistic principle of trying to avoid fairy dust it's led me to my most controversial view of all which is this this guess that them there's no difference between mathematical existence and physical existence because you can write down the standard model equations and a bunch of other equations and say that there's some magic extra physical existence fairy dust that you pour on one particular mathematical structure to make it smell pop into being and I think it's much more elegant if you don't need that either you can just say well there is nothing extra and all the different mathematical two things are equivalent that that's my guess the melamine analyst yes but then if you are consistent you should not accept the postulate that consciousness exists before anything else which is the postulate or existing IIT for me it's not a possibility for me it's just the direct observation that my subjective experience exists and that's more primary than any of my other observations the reason I believe that you exist and the reason I believe that atoms exist is the more indirect inference from from my subjective experience very good thanks okay okay no follow-ups just now because we have a line bling so this is something that David already began to talk about and I guess it's addressed to those who take IIT seriously even if they're not committed to it and and that is this issue of the claim that of on Norman computer could not be yeah could not have a high value fine so I could for those who think that well first of all the people could say whether they agree with that or don't agree with that and for those who do agree with that where they explain why the the arrangement of things in three-dimensional space that you know we seem to be living in should be so critical to to something which seems as your more fundamental than that consciousness so David clearly didn't agree with that the claim Oh still your series so he can sequence together how how how five works but I couldn't say I would have thought there's at least a version pretty close to there's a theory pretty close to IRT that allows a von Lohmann computer to have to be hi-fi because basically my understanding of how the theory works is you can take a system you can call screener in all kinds of ways to figure out here are the units and here are the connections and I would have thought there'd be a theory that would allow a kind of coarse graining so that all the you know the points in the array outside the CPU are the units and the CPU itself is just part of the causal background between the units and on that kind of coarse graining would have exactly the same fight as then you're off occasionally don't for some reason on Julia's version that's intuitively reasonable to me but that's why I want to find out yeah well so once again a remark that we start from the axioms not postulate of conscious because as marks a those are the one that are immediately evident okay what everything else is inference from there if the translation is nearly unique which we think is then it follows then a phenomenon computer we sort of recently proved this cannot be conscious as such only these minimal little entities and we did exactly what David is saying so we did all possible black boxes the one that there is a briefly Illustrated for this meaningful normal computer and it doesn't matter how your black boxes are tons of possible black boxing's it never can do that so that's a very very strong consequence of the theory now I will take this opportunity to get back to Charlie Bennett who also brought up one question which was this Scott Aaronson question okay and it relates to your question about the 3d or rather stick to Tootsie for the moment structure of the accurate physical world okay and thus is that in any way important for consciousness so one issue that was always clear in IIT is that grids physical grids not mathematical grids that physical grids because ideas physical in that sense are a good substrate for high Phi and a stack of grids is an especially good substrate so I parameter grid whether you want to comb it and so a Scott Aaronson is a good example if occation of the idea he did he with one term on the matrices and the error correcting code but you had to do it physically and the physical implementation that is a grid a grid has so law algorithmic complexity is such a simple thing that his gut feeling told him he cannot possibly be conscious okay well the evidence very strongly suggest I showed you a couple of slides that we were at least the part of our brain that matter for our conscious our by a large a stack of grid so think about that as evidence for theory not evidence against okay and these are data on top of that the structure of conscious experience also is consistent with what a grid does which is essentially space okay please one more thing about so so it may be the the first way to think about it is that for up for the simulation that variety the implementation matters not just the computation right so and it's hard for me to understand why this is always so hard because if you simulate water the simulation is not wet right but for some reason if you simulate consciousness the simulation is supposed to be conscious right so why is that because people think that pure abstract information is what gives rise to consciousness but IOT is a physical theory and that the information in IIT is physical so only if you have a physical thing that has States those states constrain their past and future and in that sense it's causal information and you're not allowed to leave causal interactions out of your analysis so the only way to get a mapping from a computer to what the brain is doing is by leaving out some of the physical parts of the computer you're not allowed to do that because they're actually interacting right so we started with the actual physical substrate of the computer and and that is the basis for the causal analysis we're so just to follow up quickly and then we go in this entire model that the team has developed about causal efficacy and cause-effect power would do you think that has applications actually - maybe the physics to understanding the hierarchy of levels of description in nature so I mean we have a paper about the cause of complexity and discrete dynamical systems for example where we show that example if you analyze all the elementary cellular automata there is a relation so the understanding the causal structure of a system can help you understand the system and all these high order interactions that are typically ignored when you just have your initial state and then you run the system those high order interactions are important to understand what the what the system is doing so in that sense even if like you take and purely as a measure of cause cause effect power I think it can be valuable yes so something Masafumi said a little earlier got me thinking and and I part of this is my ignorance of IIT and so you can clarify what I wasn't sure if the theory required self awareness or not so my question is is does it and if if it if it does require self awareness I would want to know how do you measure that so for instance if if you're Giulio and you want to find out if object X is conscious how do you how you measure quantity quantitatively the self-awareness if it's not self-aware then I would ask why not because it seems like that could almost be part of the axiom of you know if I take consciousness to be axiomatic because you know and Julian's justification is because and Max said because I know I'm here then that that's a self awareness thing like max said he's more aware of his own existence his own consciousness as then he is anybody else's so what I mean by are saying that the system has to know has to observe itself and that means that the system has to you know know there what is the past state based on on its present state that this kind of self-awareness or it's not necessarily exactly equivalent to the self-awareness but in some sense the system has to know about all its past based on its present so in this sense it is kind of self-awareness but yeah amnesia ah like a make no memory or something not yeah so it does not decay a memory but I declare that it can specify their past based on their own its present so it has to be some mechanism to you know info on its past yeah that that is necessary but that is not necessarily equivalents to the memory function memory may be a function of self-awareness but yeah very good thank was thank the panel for a wonderful discussion [Applause]

Channel: FQXi

Views: 31,139

Rating: 4.8838711 out of 5

Keywords: Giulio Tononi, max tegmark, fqxi, david chalmers, george musser, federico faggin, masafumi oizumi, larissa albantakis

Id: RX-oOIFoY3E

Length: 53min 6sec (3186 seconds)

Published: Mon Oct 30 2017