Consciousness Live! with Steve Fleming (S6 E5)

0:00 just uh before we actually start let me click the buttons and then see if we're live but I'm clicking the buttons now to see if it works I've been burned one time you know oh with Rachel actually um just recently I thought we were live for a half hour we weren't really live oh no oh for a half hour dude we were like chatting she was talking about her work and I was like listening and I was gradually Dawning on me that maybe this wasn't live um and then I had to like talk about meta cognition I had to like uh I had to decide in the moment whether I

0:34 should interrupt her and be like we're not live right now but I was also listening to what she was saying so I was like well should I let her finish and then I'll tell her afterwards and then I was like you could see the panic in my eyes like I'm like I don't know what to do I waited and then I was like we're not live sorry but we are now so hopefully it's it's not quite that bad yeah okay nice nice okay so I'm just checking and it says one we have one con uh one viewer um so if you could check and see if you could hear us that would be great put a lot of responsibility on our one

1:06 [Laughter] viewer um this one's earlier than most because you're coming from where are you exactly right now Steve I'm in Zagreb right now um so yeah it's I always have London time on my computer but it's yeah 37 p.m. um here so yeah wow okay so it's 9:00 a.m. over here where I'm at but that's okay because the kids have got me up since uh well today 4:30 yeah oh man that I mean before five

1:41 is just it's just a bit brutal yes I agree um it feels different before five it just is like there's a a qualitatively different feel to it so I it looks like we're live Steve uh okay I'm going to go ahead and Trust our gut instinct here so uh joining me today a very special guest Steve could you say a little bit about yourself and what you do for those who might not know you yeah sure glad to be here um and thanks for having me on so I'm Steve Fleming I'm a

2:15 professor of cognitive neuroscience at University College London and I'm based in the experimental psychology department there and also at the center for human neuroimaging in Queens Square in London so I run a research lab that focuses on meta cognition um and I've always always had a long-standing interest in Consciousness and the the relationship um between empirical work on metacognition and uh the science of

2:50 Consciousness very cool stuff and we were supposed to talk about your book last two a year and a half almost two years ago I think um when it was uh first coming out the science of self-awareness um and various things came up so that we had to juggle it around so long long foot off but uh first of all I wonder are you sick of talking about this book yet no not at all I mean I think it's um you know really is

3:23 the I I felt it was a chance to put the empirical science of metacognition into a format for a broader audience and I love talking about this stuff and um I love explaining to people that there is a science of it out there we can we can do empirical work on a component of what we call in everyday life self-awareness um and we can start shedding light on things like why self-awareness fails um how it plays a role in our everyday

3:58 life in terms of learning decision making collaboration with others and so on so I think there's just it's a rich topic it's something that I work on every day and that our lab works on um and the book was a chance to yeah kind of broadcast that to a broad broader Orin so I I I'm thoroughly enjoying talking about it very cool okay and so it was because it was published by basic books right which I is like it's meant for a broader audience it's not technically like a academic book it's a what they call it a trade book or something

4:30 yeah that's right that's right okay so but you still cover quite a a lot of Science in there and there's a lot of interesting detail but before we go into all that stuff I want to ask you about what you just said actually which was one of the questions that was in the back of my mind as I was reading this was that um you know I kept forgetting the title of the book because the book I thought was about metacognition but the title so it's not called the science of metacognition it's called the science of self-awareness and you just said that in common sense terminology this is what we would call um self-awareness so I wanted to but it does seem to be to me that you

5:06 cover a lot more than what you might find in your typical like metacognition section of a a general psychology class or something like that so I was wondering if you could say a little bit about how you think the the technical notion maybe of metacognition and then the common sense notion of self-awareness how you see them relating and why why you think uh you approached it this way yeah yeah and it's a good question I mean I think I guess one thing to say is that we debated of over whether to have meta

5:39 cognition front and center in the title um or whether to use a term like self-awareness which does have a more colloquial meaning and I think because it was a trade book the decision ended up falling on the side of using the more accessible term um but I also recognize that to people who work in philosophy work in cognitive science the term self-awareness can also take on other meanings that go beyond metacognition so

6:12 people often when they talk about self-awareness have in mind the bodily self inter reception kind of the notion that we are a um you know we have this awareness of ourselves as a body moving through space um and that's distinct to what I was focusing on the book where really what I was focused on was the more mental aspect of this broader notion of self-awareness and I think the the mental aspect of

6:46 self-awareness does coales quite nicely with um the more technical terms of metacognition and introspection so the idea that we have a class of mechanisms that allow us was to form beliefs about and introspect on other mental operations so these beliefs that's more the more monitoring aspects of metac cognition can then be harnessed for self-regulation for what people call metacognitive control and also for communicating those assessments to other people right so if I say to

7:21 you oh man I think my memory is not as good as it used to be then that's an aspect that in colloquial terms we would call I'm being self-aware of the fact I might not have a good as good a memory but that's a kind of um canonical example of Master cognition that you would find in a psychology textbook so I think there is interesting overlap there but I also appreciate that um you know the the term self-awareness is much of a bigger umbrella than the term metac cognition yeah and but I but I also

7:55 think that you do a good job I mean the part of I guess what you the main one of the main goals of the book maybe if you would put it that way is to a convince people that there's a science of this thing but also how important it is um and then the ways in which it can be affected by what what you do and stuff like that um but a kind of naturalizing of this idea because I think as you point out that we have a lot of Science of perception and stuff but we don't we haven't really until recently spent too much time talking about our awareness of that um from this

8:29 metacognitive standpoint uh so when you've been talking about this it made me wonder like how in the non-academic realm how widely is some of this stuff known or not known or what have you sort of encountered in that aspect of the discussion yeah I I I mean the regular response I have when I give talks on this to a general audience or whether I when I talk to people who have read the book or read articles that have been based on the book people often say say that they didn't know this term existed

9:04 metacognition but they recognize it immediately they kind of didn't realize there was an area of psychology that dealt with this I guess is the is the common response I have um and because I often go back and forth on this myself within the lab I often feel like like like other areas of psychology when you put a label on something it's useful but it can also potentially be a bit dangerous because then you Usher in a science of the label right so then you generate a whole

9:39 industry that for instance is pursuing the science of working memory because that's a label that was used conceptually to denote a particular mental operation back in the 80s and then we've essentially had like this Neuroscience of that phenomenon since then right and that's told us a lot but there's a worry that you're kind of putting your blinkers on and and just looking at it through a fairly narrow lens and I do worry about that somewhat with metal cognition actually in the lab but when you kind of adopt a more public facing perspective I think it's hugely useful

10:12 because then you in a way like Channel people towards a particular literature or area of study that they might not have thought was a science before yeah interesting so do you see um someone's asking in the in the chat which uh sort of to a question that I was thinking of as well but I wonder do you see any relation to the ideas of like meta cognition and the notion of the super ego or other Freudian Notions or are these separate completely because I do think sometimes these questions

10:45 come up especially uh to people who are interested in this from a different maybe not so cognitive neuroscience perspective or or so I'm just wondering what your view about the relation of these Concepts might be yeah I I mean I think there's a loose connection I think in terms of um you could think of matter cognition is certainly being squarely within this broader set of what what psychologists call executive functions that are doing something in higher cognition to control

11:18 and regulate other aspects of the mind and and and the body and our Behavior so in that sense there's a there's a hierarchical structure that you can layer on to that so you can say that there's like processes that take as input other processes and quotes reflect on them and we need to be careful here because we need to make sure when we use terms like reflect on we don't think of a little homunculus doing the reflection we need

11:52 to put that in put that itself in computational terms and so I think that's where it differs from from thing from um the Freudian or the psychoanalytic notion of ego because ego then does have a kind of little man in the head um kind of connotation um but when you start fleshing out those ideas in cognitive or computational terms then you probably end up somewhere reasonably close to the broader modern notion of executive function and we would I I

12:26 would see metha cognition as being part and parcel that although I think it's also important to point out that it's conceptually and empirically distinct from what psychologists would call fluid intelligence so it's it's possible and indeed pretty common that we metacognitively evaluate the operation of classical executive functions so I might reflect on whether a solution to a logic puzzle was appropriate for instance or think to myself oh no I've just made the wrong response on a quiz or even on an IQ test

13:01 you can imagine meta cognition going on while we fill out an IQ test so there's kind of this um notion of monitoring that's distinct to um to logical reasoning yeah I'm guess so part of the reason why I was curious about this was by thinking about um well actually you know I talked with Heather Berlin which I don't know if you know her or not but she's a yeah I know Heather yeah okay so and she was you know suggesting that maybe some of work on priming could save some of the

13:34 Notions uh in a way could save a notion of uh defense mechanisms or or something like that if there was kind of an an unconscious part of you that was you know because there was one particular study where if you if you show images to one side and they're congruent or not with the sexual orientation of the person then they're quicker or not to answer questions on that side of the screen suggesting perhaps they sort of or unconsciously aware what the image was and it either attracted or repelled them in in as by their sort of orientation and that that was sort of um

14:09 maybe an interesting Avenue towards thinking that there was some you know some kind of Freudian esque mechanism here that was keeping the person away from the thing that they didn't want to in a sense and when I was reading your book I mean you didn't talk about this but I was just wondering if you that's what brought this question up in my mind is do you see I mean you talk a lot about the ways in which this is shaped and motivated and stuff but do you see that there could be a story in terms of the psychoanalytic processes or a background or something that would determine these kind of magn cognitive

14:44 Properties or yeah so anyway it's kind of a half big question but since I saw in the chat I figured I'd bring it up yeah I'd have to think more about that I mean I think that I guess a more meta comment is that you Freud obviously has a bad rap when it comes to Modern psychology um and I think that yeah you know it could be time to reconsider that to some degree so I think there's and the reason I'm saying that aside from this this interesting

15:17 conversation about how it might relate to Modern Notions of metacognition is that the the science of Consciousness has right rightly so in my View kind of tended to focus on a very reductionist view of the contrast between conscious and unconscious processes so you have the contrastive method let's create stimula near threshold let's see say whether people saw something or not but the debates there then often become about things like um the Criterion problem like do

15:52 are you really not are you really unconscious of the thing I'm showing you right um and I wonder whether part of the issue here is that we're essentially trying to aim at the wrong type of unconscious that that that weak type of unconscious that you get in a perceptual psychophysics experiment that generates so much of this debate on whether perception is Rich or Spar and whether things are below threshold above threshold and so on is is they're super interesting and great for getting the science off the ground but I wonder whether a much stronger type of contrast is by

16:26 revisiting the broader Freudian notion of the unconscious um with modern experimental techniques because the the critique that's rightly leveled at Freud is that it's a just so story there's no real purchase on data um that that gets you but if you kind of took that more heavyduty notion of the unconscious which may or may not exist you know we'd have to start pursuing that empirically right um interesting and that could give us a stronger contrast between conscious and unconscious that we're looking for huh I don't want to get too distracted

16:59 by this but just to make sure I understand what you're saying Steve so is it that you're saying that so as far as I understand the Freudian notion of the unconscious it's it's that it's like a another realm or something of the Mind where there's like active processes guarding the door to keep stuff down there so it's a much more active sense than oh you're just happen to have some process that isn't in conscious that you're not aware of or something is that the idea that that's the more robust notion or did you mean something metaphysical or so no so was staying on the

17:32 psychological um level yeah and I I wasn't so I think the Notions of repression and so on may end up being what people have pointed out are Just So Stories and you know kind of post talk explanations of why someone might be the way they are um but I was more thinking about the notion of kind of of and I realize we're getting very off track here but no no it's good it's good um it

18:08 actually part of the reason I said what I said was because this is exactly what I've been thinking about very recently in the lab about how to connect up the science of Consciousness that we often pursue with minimal perceptual paradigms and how to connect that up with the science of decision- making which is doing amazing things in terms of explaining how we choose how we plan especially and so planning is a super interesting area of

18:41 computer science of reinforcement learning um you know how you essentially decide um what to do next based on many different possible options there's been a couple of false starts in Psychology suggesting that like that might be the function of the unconscious if you like so there's this classic kind of dster house paper suggesting that you make better decisions when you're unconscious and that turns out not to replicate that particular result but I think the general idea is that what you're talking

19:15 about yeah I think there was yeah I think I can't remember the exact task but it was a science paper 2004 I think okay but so that particular result doesn't replicate and that's I think created a bit of freeze on the um this idea but I think it could be time to revisit it because a lot of these just in intuitively and this is I'm saying this with like no data total speculation but intuitively it feels as though the kind of computations that people are discovering support complex planning and you know how do I decide how do I figure out how

19:51 to get to the bathroom in my house I I I kind of don't think about it that's the point right is it feels very unconscious um and yet Neuroscience is showing us there's pretty sophisticated stuff going on there so I don't know this is very handwavy I'm just suggesting that that could create a he more heavyduty contrast between conscious and unconscious States um or conscious and unconscious computation that might end up going back to the older notion of the the unconscious as being something that

20:24 drives our Behavior but that we don't have access to interesting yeah so I mean this is kind of related to the point to bring it sort of back to your book um that you make in the book and you made just now about fluid intell intelligence being different because you point to some evidence empirical evidence that suggest that uh well not only is metacognitive abilities varied amongst individuals and species but also that it may be absent in some species that are pretty well off otherwise um intelligent in other words so I was wondering could

20:57 you tell us a little bit about about like these particular types of monkeys like the two the ones that do and don't in your opinion have the the metacognitive abilities and what the reasons for thinking the difference may be yeah so I guess um the to to get to an answer to that and need to just um step back one step which is just to say about how people have um gone about quantifying metth cognition um in these experiments so the general approach here is to

21:33 measure some kind of behavior quantify some kind of Behavioral performance on a task um whether that's a memory task or a task involving some kind of perceptual judgment um and then also introduce an extra measure which measures um which is designed to tap into the metacognitive evaluation so that often has the form of a confidence estimate or a non-verbal analog of a confidence estimate if you're um looking at metac cognition in animals or babies um and two broad lines of work

22:10 have emerged in in The Psychology and Neuroscience of these measures over the past few decades so the first is in humans look looking to catalog these individual differences um in me cognition look at their structure across different tasks and how they might be affected by say naturally ing brain damage or disorder and the second has has been kind of more focused in on looking at what is the computational and neural basis of the formation of the Judgment itself the the how does the mass

22:43 cognitive judgment get formed in a particular task like how is what what is the brain process what are the brain processes involved in forming confidence in my performance um and the work in animals is tended to focus on that so tend to ask well um if I design a non-verbal confidence task for use with an animal then I can ask a behavioral question about that which is does my confidence tend to track my fluctuations in performance and if it does then I can say that species seems to exhibit some

23:17 metac cognition about its performance um and what's interesting there is that in general there's been quite widespread findings of metacognitive capacity defined in that way across lots of different species so Maat monkeys rodents Dolphins David Smith and Michael Baron have done lots of cool work on this um then there's some isolated sorry even found in David [Laughter]

23:51 Smith sorry keep going sorry I did no no um but then there was some isolated um exceptions um and there was this famous paper on capuchin monkeys where they found that if they train the monkeys to perform an initial task and then learn to opt out of that task so declined to participate on Trials where they didn't feel as confident

24:23 then Maat monkeys can do that no problem they decline the test when they're likely to be wrong on the test so that suggests some kind of form of meta cognition um whereas the caput chains they couldn't get the hang of that and whereas when you reframe that task rather than it being an opt out judgment but now you simply has have to hit a third key on a two alternative Force choice test you have to hit a third key when the stimulus is somewhere in the middle of the range so formally that's

24:57 equivalent to opting out because you should opt out when it's in the middle and you're not very sure about it right they do fine so there's something about this framing of the task as a metacognitive judgment that the Capuchin in that experiment didn't do so well on so just to be clear though what the the Capuchin could do was it could tell when the stimulus was in between the other two but it couldn't use that information uh as a as a way to say well gee I can't tell which one it actually is then on the basis of that or something yeah yeah exactly um now I think

25:34 there's you know I think there's the animal method cognition literature is still pretty young I mean the human metth cognition is also literature is also quite young um as a science so I think we're still trying to figure out what do these isolated failures mean um and I think what's become increasingly clear is that and it goes back to my comment before about the danger of creating a label and studying the label that what we mean by metac cognition is really a collection of algorithms or internal processes that we use in

26:08 different context to solve this interesting challenge of evaluating our own performance um and so it that the the comput like fails that particular test um but whether it would pass other tests for instance that have a similar flavor but a set up different we don't know I don't think um yeah is the idea that we we can't be sure if these if metacognition is a kind of psychological kind um that there may be many different like computational

26:44 mechanisms performing this function of monitoring so that you can't kind of judge from a failure in one case that there might not be some other mechanism doing a similar job is that the idea uh that's interesting I I would so that wasn't the idea no no I guess I guess what the I'm just trying to map that onto the way that we kind of think about the way it works in humans so I think we have a in terms of the psychological dissection of this I think we have the best

27:16 understanding in humans it's hard to do the animal work because you have to create these clever non-verbal animal analoges of the the human metacognitive judgment TS and I think the way the way that we currently think Ma cognition Works in humans is that there's a number of different component computations um that that come together to support um M cognitive judgment on these simple tasks and so the and this is also what I sketch out in the book

27:48 the first half of the book you can think of it as like unpacking those different components um oh I am I uncertain am I judging my own actions correctly those elements yeah yeah exactly and and I I I recently actually wrote this up for a as an academic for an academic audience in a annual reviews of psychology paper um trying to kind of take those themes from the book but now um you know putting it uh in firmly in the scientific context so trying to

28:24 unpack this General framework if you like for for thinking about how meth cognitive judgments work and there what's and the reason it's I think helpful for contextualizing the competion example is that one component of that that seems important is switching from a world directed encoding of uncertainty so the idea is that we within all our perceptual system

28:57 systems memory systems and so on there's some uncertainty associated with the representations there and that uncertainty might be important for self- evaluation because if I'm not sure I know something then that might be connect to the fact that the first order representation is more uncertain of that thing um and in perceptual cases people have studied this a lot in the case of vision for instance like how populations of neurons in early visual cortex represent that uncertainty literally is the distribution of firing

29:31 rates for so you can think of it in very much basian terms there's like representations of that uncertainty that's what that's that's one strong hypothesis about um how uncertainty gets in the game here and I guess what I point out at various points in the book and also in this recent paper is that that seems really useful for doing metacognition all that all those uncertainties that are kind of spread through the system seem really useful but to we need to then transform that into a frame of reference that refers to my own

30:04 judgments and actions whether these are overt or whether these are hypothetical so that's what I call a self-centered or self-directed um frame of reference and in a classical human metacognition task that's made explicit because I get you to do something and then I ask you did you think you did that well or how confident right whereas in a lot of the animal tasks and this has been Peter K's point for a number of years a lot of the animal tasks are actually ambiguous about whether they're tapping into World

30:37 directed or self-directed frames of reference um including that computition task so you know there's there's kind of there's certain things you can solve just with first order representation there's certain things that seem to need this extra level of coding things in a self-directed frame of reference referring to my own or judgment and then there's the certain tasks that seem ambiguous they kind of sit in the middle and you could solve them um the metacognitive way and you could solve them the first order way and these tasks

31:09 that we're talking about with the capu kind of sit in the in the middle here yeah interesting so on your so it's an empirical question whether these uh creatures do or don't have metac cognition but on your view conceptually could there be a creature as sophisticated as an a mammal um that lacked metac cognition of any kind or is it like you said there's some things that's necessary for it but some things you can do without it so how how far can you get up without without uh without it do you

31:44 think can you be a frog yeah I mean I that's that you know it's really interesting I haven't thought about it in that that way before I I think that I don't think you would get very far um which is also why I think the kind of a a form of meta cognition and when I say a form of meth cognition I'm distinguishing this

32:16 from fully-fledged self-awareness in humans which may go um where as essentially the there could be extra components and this is what I the kind of position I defend in the book that in humans at least and possibly is also the case for some other animal species in humans at least it seems that what is part of our metacognition is a broader self model where we have background beliefs and um we have knowledge of the context that

32:54 affects any individual judgment so if I give you an example so if I um want to evaluate in the moment how well I'm hitting a tennis ball for instance then that judgment might be affected by my background knowledge of the fact that I've practiced tennis for a number of weeks before that point and the knowledge of like my um skill and ability in other areas and so on so there's kind of and

33:28 you know there's a whole number of experiments in the lab that show this that are human method cognitive judgment seem very um modelbased in the sense that we bring to bear a lot of other information about ourselves when we're evaluating um our performance you're saying that could be missing so you could have levels of meta cognition without that right exactly and what's interesting about that is that it also seems to that aspect that more model based aspect seem in humans seems to develop quite late around the age of four or five it comes it seems to come

34:03 along with a richer capacity for thinking about other minds and so the the position I defend in the book is inspired by the K's position that there's a aspect of self-directed Master cognition that co-ops resources for thinking about other Minds that we do it in a very similar way we represent those things about ourselves in a similar way to which we represent um others now that doesn't mean that self-awareness is just theory of Mind turned on ourselves as the kind of um tagline for that position often gets sold I think that it needs to

34:39 then interface with all this Rich uh internal monitoring of uncertainty and so on and how that interface happens how the kind of bottom up and top down stuff comes together is super interesting I don't think we know the ansers to that but but so yes I think there are plenty of animals um many have been studied and many have not yet been studied who I would predict have this model fre metacognition that have this kind of monitoring of confidence and uncertainty and error in a lot of um settings and essentially that's being

35:14 used as a proxy for feedback there's no reward per se but you're quotes recognizing you're off track and so you correct and I think those vicarious trial and error type phenomen have been seen across the animal kingdom you see that a lot so I think that level of monitoring control based on internal error signals is I would predict pretty ubiquitous so it's almost like this is the recurrent processing or predictive processing model um and then conjoined with the metacognition stuff you get this idea

35:49 that gee metacognition is everywhere in the brain because there's predictive processing in the brain and that is a kind of meta error error reduction C or whatever well so so yeah I guess it also depends then it becomes somewhat um semantic about where you draw the lines there but I I would view the so I I'm very much on board with the broader predictive processing um framework I think it's very powerful for thinking about how perception and cognition work but in terms of how metacognition fits in there I would view it

36:22 as helpful to draw the line at um when you get into this self-directed reference frame so when you're essentially dealing with monitoring your own actions right um and they could be hypothetical actions so I might think to myself um yeah I think Island won the rugby yesterday right and then I think no actually they didn't they did actually they W on Friday night but um you're always talking about these Sports I don't know anything I have never heard of but that could be a that could wholly be an internal monologue um so I

36:57 wouldn't have to do anything avert but it's in a self-directed reference frame because I'm thinking it's my own action internal action that I'm judging so so I think that yes that could well be based on and I think it probably is based on the same kind of prediction error Machinery that has been Advanced to think about the develop the the um refinement of first order representation but now you're dealing with something that refers to your own um yeah you refers refers to yourself

37:32 interesting okay I mean yeah there's there's a lot that I want to ask you about but um I I wanted there's believe it or not this all start because I just want to get clar some big picture and stuff before he talks about the details but uh uh yeah anyway but one thing that comes up a lot in the book especially in the second half is the notion of autonomy and control and even Frankfurt shows up but I couldn't help but notice that a certain free will like the the phrase free will or discussion is not put in terms of free

38:07 will but it does seem to me that that's a discussion of your view on Free Will and that you defend a kind of compatibilism and a two level view of you know but but but I want to check with you and see is that how you saw your discussion of autonomy and and Frankfurt I mean first you could tell us a little bit about that cuz I didn't really set it up but uh um and then explicitly related to the philosoph philosophical discussion but then I was like curious about your bigger picture ideas about Free Will and why they didn't show up in the book sure yeah I mean I guess

38:43 um to take the last part of the question first I I think the reason Free Will didn't show up in the book is I I don't have a strong position on this I I personally I I feel like I've always been somewhat puzzled about the debate if I'm honest I I um and perhaps that's because it does put me on the compatibl side of things um I don't really see a tension between acknowledging that things can be deterministic or random at

39:17 a a micro level of physics and and thinking about autonomy and freedom at a psychological level I think we obviously don't know the connections between all those levels um because Neuroscience is still a very young project but in terms of the broader philosophy of how those things all might interact I I yeah I don't really see the pull of the debate there I know other people do and and you know I've seen with Kevin Mitchell's book coming out and Robert spolsky's book there's obviously all these debates

39:50 going on um and it's interesting but I guess like like I I you know I'm very much on Kevin's side here like I think there's a there's a naturalist explanation that once you get up to a psychological level um I think you do have free will in psychological terms and I think what the point the more the smaller narrower point I was making in the book about autonomy there was that there seems to be a role for metha

40:24 cognition in um whether to ourselves it feels as though we're free I think that's that's that's maybe the distinction that I want to draw is that it's not trying to engage with the metaphysics of free will do we have it or not it's more saying if we acknowledge that some things feel more free than others then when don't we feel free we off a a a case that the the classic Frankfurt case is we might not feel free if at a m cognitive level we want to do something that at a first order level we feel unable to resist um the whereas if

41:00 we're yeah sorry I said the attic not the attic the the the the person with an addiction is the classic example exactly exactly yeah and so and then yeah just um so it was a pretty short section in the book but it was more trying to point out that we've done a little bit of work on um met cognition that goes beyond the more classical perceptual or memory based task setup where we've looked at how people evaluate subjective choices

41:34 um and the subjective choices we used there in the lab were choices between different food items so these are choices that don't have a objectively correct decision anymore because sounded like a fun stud to be a member of a part of by the way yeah it it was it was fasting well yes yeah yeah yeah so this was a collaboration with a really good friend of mine benedetto de Martino who's also a professor at UCL he's a behavioral Economist neuro does Neuroscience of um economics and so he's really interested in how people make

42:06 these subjective uh decisions and in behavioral economics they want everything to be incentive compatible so that's why we had to get people to fast because we wanted to the choices that they made about the foods we wanted them to be meaningful so if you tell someone you're going to be in the lab for the next four hours and the only thing you're going to get to eat is you chose on this computer screen then they tend to care about it um and how many food did you have because you said had hundreds of combinations yeah so it wasn't very many items I think it was maybe under 20

42:39 items but we paired every item with every other one in the set okay um so the idea is to get ratings of every like individual preference between pairwise basically right yeah exactly so we we get we asked them to um people to State how much they're willing to pay for each item in the set and again this is incentive compatible because we deduct that money that they're willing to pay from their earnings in the experiment um and so that gives us a um a a measure of how much you care about

43:13 the Snickers bar compared to the kettle chips and and so on so we have all these items in the set then we get you to choose between Pairs and the important and then the the novel thing we did was just ask you very simply a metacognitive judgment how confident do you feel that you made the right decision about that pair and when we originally planning this study I remember lots of conversations with benetto about whether this even made sense and I guess this kind of hits up against the philosophy a little bit like does it make sense to ask you to evaluate a totally free choice

43:48 because what's the notion of error there like can you be aware of making a error when it was up to you in the first place which one to choose right um and if you thought you should have chosen B rather than a then why didn't you choose B in the first place right so that was kind of we went back and forth on this a lot but what we found was that people found this a very natural and their confidence judgment showed a lot of the same properties that we see in other tasks that do have an objectively right answer

44:21 so um you tend to be quicker when you're more confident about the judgment your the judgments are confident predicted whether you have chosen the thing that later on you would say is more valuable to you um and also those confidence judgments tend to engage the same um neural networks in the prefrontal cortex that we've seen engage confidence in other tasks that have an objectively right answer and so what we essentially concluded from that study is that first

44:54 of all metacognition about subjective choice always seems alive and well and second of all that when you make a judgment of high confidence that you've made the right choice in a setting like that what you're essentially doing is self- endorsing it you're saying that yeah if I was to face with that opportunity again I would make the same Choice um and people like Pascal mamasan and others have actually very recently suggested that we should broaden our notion of conf of confidence in the literature to not just refer to a

45:26 probability of a judgment being correct but to refer to the probability of a judgment being self-consistent over time and I think then you actually do get to quite a Frank 1an level of autonomy from these kind of um investigations yeah yeah very interesting so you know I I I was going to ask something about the franker stuff but just really quickly about the choices so how does this interact with the choice blindness stuff that you you discussed in your book buts Al just more more generally

45:59 um uh where people are presented a choice they pick one then they're presented with the other unchosen option asked to say what they liked about their first choice and they often make stuff up yeah so how much of how much skepticism should we take from there to these findings about these choices and confidence ratings here where we might think gee isn't this all just made up and I wonder what would happen if you did a kind of choice blindness confidence mashup yeah what you would do

46:33 yeah I think that's a great idea I I've always thought that would could be fun to do um I so the choice blindness studies are fascinating and the way they tend to work is they're they they use slate of hand they use a magic trick essentially to show you the wrong the the wrong thing right so you you know you the way it works is that you're you choose between two things um the original study had which face do you prefer is more attractive and then the experimentor slides over

47:07 the one that you didn't choose and asks you to say a bit more about that like why do you think that she's more attractive um and then the result is that a people do that they go on to justify choices they didn't make but be a decent proportion of people I can't remember the exact percentage are are surprised that they've been fooled in that way yeah and they didn't notice the manipulation so I think that what that shows is I I do feel that those justifications are an aspect of method

47:41 cognition about your choice you're trying to introspect about you're trying to and the kind of um the split brain style narrative Constructor is like just doing its job you you've been asked a question at that metacognitive level and so you try and answer it um and it confabulates because we know people are quite good at making up stories about why they've done things but I think it's a very interesting idea to ask would there be telltale signs of that construction in the confidence

48:16 estimate so in something more systematic because that that metacognition there studying there is very free form like it's just say whatever you like um but if you did something more atic and using the tools of psychophysics to look at like how does confidence map on to the Judgment then I agree it could be very interesting and this is actually exactly it comes back to the conversation we're having earlier about the Freudian unconscious like is it does it really exist like can you can you kind of try and investigate this what appears to be quite yeah is is there is there more of

48:49 a um how permeable is that um I think it'd be very interesting yeah so the you of metac cognition that you come to in the book what would you what and that you've also in your own work come to what would you predict or what would it have any predictions like what would you expect that uh because it sort of seems like aren't people doing that even in the case that you presented I mean they seem yeah um so yeah yeah so I think I think one one so one thing there is that um I I would in that case

49:26 where they they're using the Slate of hand to um it would I think it like a lot of these things it would the devil would be in the details it would be about how the Paradigm Works set up yeah and and one thing we found in our confidence studies is that postdecisional processing plays a strong role in the construction of the eventual metacognitive judgment you make so not just the positive evidence just um the fact that if I ask you to make a

50:01 choice based on some say some very briefly flashed information then if then I um go on to study the properties of what happens in the brain after you've made that choice then there's good evidence that you continue to accumulate evidence about the original thing to figure out whether you might have made an error or how confident you feel in it and then you also have these yeah these additional biases that come out in confidence estimates that might be partly due to this

50:33 postdecisional phase and so what's very interesting and I'm just realizing this talking out loud with the choice blind and stuff is that essentially they're giving you a postdecisional phase that strongly pushes one option to you right so because when you originally see these two things you're probably fairly indifferent so you might be like okay yeah a and you get a brief glance and then I slide you B and you might feel a bit like really but then as long as there's

51:08 the the trust there with the experiment a decent of people will will push through it and accumulate evidence in favor of that thing they've chosen um so I yeah I I mean in general this all sounds very interesting and I wonder whether there is quite a lot to be done I mean the interesting one that always struck me the most was the one they did in the supermarket I mean who knows if these ones Will Survive replication but the one where they switched out the jams um so that's just a brief glass a brief glance you know that's how you have seems a little more direct but yeah so that's very absolutely they also did

51:43 one with um this is petty Hansen um who's run these studies they also did one with political attitudes so they got people to fill in a survey on the street and then they you know You' literally be there ticking all these boxes about how much you support a particular position on climate change and then they'll just give you the opposite if people would do it really I mean I I struggle I I'd like to think I would not be taken in by that but you know well some it's a bit of it you wonder like at that point like if you

52:15 need a different Theory like are they cooperating with the experiment or are they yeah you know uh doing some kind of mesque C something else is happing at that point yeah there's clearly big Dem characteristics here there's trust in the experiment there's yeah absolutely yeah interesting but also it's part of your the way you think about meta cognition that it's not always right um and that there's a distribution amongst the populace uh kind

52:48 of just naturally I guess or I guess you environmental factors lots of tough influences it you even say we can practice mindfulness and improve it over time so it's not just a set I think that's an important point of your too it's not a set quantity like you're not born with x amount of metacognition that's it you can there are things you can do to harm it or help it yeah but uh it is a naturally varying quantity in the in the population yeah that was like a interesting main kind of discovery of the science of this stuff that you might not have

53:19 expected yeah absolutely so I think you know when you Define metacognitive ability as the mapping between your metac cognitive judgments and your performance over time so you look at how those two things can vary how good someone is at realizing they've made errors or being aware they've made a good judgment and so on then we can give we can just compute a single number for that there's various ways of doing that and part of the innovation in this field has been

53:52 developing ever more sophisticated tools for studying the statistical properties of those mappings when you do that and you have a a number that corresponds to that person's metacognitive ability then what you can do is look at individual differences in that and how it generalizes across different tasks and so one interesting finding is that there seems to be some domain General component to ma cognition so if you have good metacognition on one task then you'll tend to have good metacognition on another the reasons for

54:24 why that is are still um under investigation um and that individual difference um in a lot of the work we've been involved in is related to features of um how the brain works so the structure and organization and function of the prefrontal cortex but as you say that doesn't mean that that it's just set in stone so there's a whole other line of work looking at

54:57 interventions what might change metacognition so there's people have done work on meditation we've done a bit of work on training and I think that's still um you know a a a really cool area that's only just started getting going in terms of like how much can you learn to be more metacognitively able we don't know I think is the answer we we did one study that suggested there was an effect of train to become better at your confidence judgments but then there was a worry that that might be due

55:31 to that might all be an artifact of how people are using the confidence scale that we give them to report their confidence on So within the context of the task they might be getting better but does that mean they have improved their metacognition beyond the task and so Natan favor group um ran another experiment suggesting that actually that artifact explanation of our original study might be the might be the correct one there but that doesn't mean there you know that that perhaps perhaps says that yeah

56:05 exactly if anything it just makes us more interested in thinking well okay that was a bit of a shot in the dark that turned out to be an artifact but is there are there any other Roots there's so many other Roots you could think of of to try and systematically investigate what experience does to metacognitive evaluation what we're talking about is like non-invasive ways because there also invasive ways yeah sure but these would be like things you could do at home to improve your metacognitive ability yeah and just to be clear you think this is an important thing that

56:38 should be improved if it could be right yeah well so yes I think broadly yes um I mean I remember there's a well there's a there's a there's a there's a line in um so John Fel who was a psychologist at Stanford in the 60s and 70s and he really kind of kicked off the field of the modern field of MTH cognition research he wrote an influential article uh in 1979 laying out meth cognition as a new

57:11 field of inquiry and his Focus there was on children he was interested in how kids learn and he pointed out that if kids had better metal cognition they'd learn even better because they could know what they don't know they could guide their own study and so on and so on the applied side there has been this um ground swell of interest from educa um in educational settings on what might how might we go about improving method cognition and what's the

57:47 the the there's a bit of a disconnect there between the cognitive science and Method cognition which tends to try and system matize things and and put numbers on things and the educational interventions which are more qualitative but seem to have big benefits into when they actually measure the impact on entainment so I think there's a fascinating distinction there we were actually just talking about this in a meeting a few weeks ago where the Ed the the teachers seem to have tapped into the secret but yeah the the scientists are kind of

58:22 stumbling around the dark we might have you know more systematized tools for measuring it but we don't have the kind of way of developing the intervention that the teachers seem to have kind of figured out so I'm actually involved in a project with educational psychologists at the moment to try and learn more about this and and see whether we can cross fertilize these these literatures but to go back to Fel he he has a line in that original paper which says something like I'm fly firmly of the belief that we need more rather less met cognition in the world which is a pretty

58:56 big statement right why is that a big statement well I mean you might say according to you just we need more self-awareness you know no I mean I think sure exactly and you know you can take this stuff back hundreds if not thousands of years like in terms of people being urged to have more self- knowledge to to be more wise in their in their um judgments with respect to what they can and can't do and so on um but I think you know if you imagine

59:29 other areas of psychology where you just treated things like an unalloyed good people might get a little worried right so think of like a memory research saying all we need more of in the world is people with even better memory um and there are cases of people with like these super memories and they're not very happy people because they remember too much and you know these classic Laura cases so I do I do worry slightly that I I kind of agree with Fel the sentiment of it but I do worry slightly that we haven't really considered potential downsides um and the obvious

1:00:03 downsides would be something like if people become more introspective and More cleari in their self judgments then that could create a danger for things like depression and anxiety I mean this is just spe Illusions and realize that they suck yeah pretty much right I mean that's that's I think the reason worry I mean there's actually an active um there's an active discussion on this in um the research on metacognition in dementia because there there's a very

1:00:37 interesting idea that if you improve metacognition or you give people with memory problems the tools to know that their memory is not so good as it used to be then they can more effectively offload the environment they can put up notes write lists and so on but there is a worry there that if you suddenly make someone aware that their memory is shot to Pieces then that obviously has emotional consequences and perhaps they're better off believing that it's actually okay you know so there's fascinating conversations about what is the right um

1:01:11 paternalistic kind of medical approach to this stuff um I think we're obviously a long way away from needing to properly engage with this because I don't actually think we have the science at the level of doing proper interventions yet but in a decade's Time might do and then there will be a I think a serious conversation to be had yeah interesting um so I mean because a part there's also the worry about like do we want bad people having good metacognition you know so if this is generally enhancing

1:01:45 your overall abilities then you know the Psychopaths and um yeah other sort of dangerous types out there there's a question about whether uh across the board this is a good thing to have um but I I I was going to ask about something else but since you brought this up it's it's kind of related to I think the discussion at the very end of your book where you talk about um two different ways forward so to speak by thinking about how like to incorporate um machines actually uh into meta cognition uh one by either building

1:02:19 metacognition self-aware machines but others by building them sort of in to us so to speak so that we medic cognate the machines and I thought that was a very interesting diff I haven't really seen anyone talk about this so I know it's kind of you know not exactly related to what we're just talking about but I I think in some sense it is because it it it's a question of how would you maximize uh metacognition the most useful way and one way to do that you might say is by just so if you just you

1:02:52 know Skip Machin one way to do that is by enhancing our metacognitive ability another is just to enhance our lower level abilities and our ability to survey that um so I I wonder first of all I know it's kind of you know talking about machine Consciousness some people love it some people hate it but your your view seems to be committed to it so I wonder if you could talk a little bit about this uh aspect yeah no I mean I think there's lots of interesting things to say here and I think it it you know the end of the book is a bit sci-fi it is a bit um kind of thinking

1:03:25 quite far in the future about and it it builds on this notion of autonomy that if we if in some sense metac cognition is important for this feeling of freedom and having autonomy over um what we think is a good choice and so on and I think it there is um a case to be made for that then we it might impact on how we think about Outsourcing certain decision- making capabilities to the ever

1:03:58 increasing um uh the the the more and more sophisticated um developments in AI so I think there are two questions here one is a more pragmatic one about does AI need meta cognition um and I think there's a lot of consensus on that that yes it would be useful to have it in some form um presumably the lower level more ubiquitous animal metacognition form that we talked about earlier essentially having the ability

1:04:31 to estimate in a precise way its confidence in various courses of action and on a on some level know that it does not know the answer so that it can um appropriately refrain from engaging in um areas of endeavor that it is not trained to do and I think we've seen a lot of the Falls of AI systems not having meth cognition in the all the debate over chat GPT and confabulation and and you know hallucination and I

1:05:04 think yeah hallucinations and so on so and so for instance we have a project with an Oxford the the a robotics lab in Oxford um that are trying to think about like what kind of implicit signs would you want in a robot to communicate its confidence externally and how would that benefit human machine collaboration um we're also doing some work on looking at how do humans intuitively assess the metacognition of the machine um so one thing we did there which we have a pre-print out on this

1:05:39 recently is look at if I get you to interact with a human giving you the answers to something and interact with chat gbt giving you the answers to something then how much confidence do you ascribe to those answers and it turns out that people think think chat GPT is more reliable and confident than the equivalent human which is kind of worrying right because so wor then yeah and so so there's there's a two-way thing here there's there's both the Met cognition within the machine and then there's the um human machine interaction

1:06:13 component which is how do we interact with a machine and share metacognitive estimates um in that interaction and then there's the final more sci-fi thing which you alluded to which is is perhaps the um end game of all this is either we outsourced everything to the machines and we are just kind of you know kind of walking around having Leisure Time Machine just became angry Steve you froze for a second just in the mid in mid sentence you're back now though sorry go ahead okay okay good um yeah so you know

1:06:48 if we drive to do more and more Outsourcing to machines without much metha cognition then we might be ending up in a somewhat dystopian scenario where we lose metacognitive contact with all the first order of stuff that we you we that may have an important role to play in our feelings of autonomy whereas an alternative is like perhaps it would be nicer as humans once this AI becomes part of the fabric of Our Lives I mean it already is in a lot of ways but would it be better to essentially like be like

1:07:23 the tennis player or the Formula One driver where their technology is essentially part of the fabric of who they are in that space of time so you know they those guys talk about like feeling the car they they literally feel like part of it and the same with the tennis player they don't really think about the racket they just think about there's this notion of extension of tool tool use to Encompass the space that you're using that tool in and I think there's a very interesting analogy to be made by with all this work on BC's brain computer interfaces um which by the way

1:07:59 that's been going on way longer than Elon Musk oh yeah but clearly becoming very it's hitting I me you know the capucha night was it a capucha or a Mak the monkey that they well I mean it's kind of mean but they amputated its arm and then um Tau him computer to interpret its neural signal it controll the robotic arm um so so that that kind of research was done long before musk was doing right anything that was the only point I was making BC yeah yeah yeah no exactly I

1:08:33 mean I think you know it is interesting like a lot of these things it is interesting when the private sector starts getting involved because then things can accelerate much more rapidly than it would have done within Academia um but I think there's you know there's in that whole technology raises I think very interesting questions for Consciousness science for met cognition um for philosophy about like what kind of lives do we want to be leading and I think that um there's a lot of work that just hasn't been done about what is it like to have one of

1:09:07 those devices I mean you know there's there's so far rightly so the focus has been on let's solve the first order problem let's get you to walk again or let's get you to control this you know this this device right but then once you've learned to use that device what does that feel like on a more metacognitive level I think is a fascinating question exactly and that hasn't been really addressed yet I mean so I'm I'm all for the you know the fut the Smart House of the future like I imagine being able to play Call of Duty on the couch

1:09:40 and having my robotic arm installed on the sink in the kitchen and then I hook to a Wi-Fi with my brain fi and then scrub the dishes while I'm playing the uh while I'm playing the video game so when I tell students about this they become very they're like what this is I think gee this would be amazing but uh so I do think that there is um we haven't really got to the point where there's widespread acceptance of the idea that like the machines could become part of us so to speak right uh but that might be the next step we need

1:10:13 to take because I think that's where we're going if unless we want like you said to Outsource a complete to them um which seems kind of immoral to make conscious machines or self-aware machines or something yeah I mean I think I guess yeah I guess that's kind of the the gentle point I was trying to push back on in the book was that there's often I think the notion of kind of Minds melding with machine is often portrayed as a dystopian scenario whereas I guess I was trying to suggest that that could actually be a better scenario for

1:10:46 retaining autonomy because we then naturally interface with all the higher the lovely stuff we have for monitoring our own actions and our own behavior already and so if that becomes part of us as it were then that might actually be a more natural ironically Endo than Outsourcing things exactly very interesting and and it might enhance our autonomy as you're saying but uh by giving us more control um yeah and and and thereby making us more to you know the kind of Tom you were talking about is making us

1:11:19 more able to make sure that the things that we want are the the things we want to want and that there's an alignment between these right exactly exactly yeah yeah exactly so unfortunately I mean one one thing that doesn't come up a lot uh in the book is kind of the the realities of the real world distribution of this so if if there are people who are you know let's say a career criminal who's making bad choices over and over again that's partially due to their metacognitive abilities that they just

1:11:51 don't realize that they're making poor choices over and over again would you think that that's that's part of the explanation or is it or do these kind of social political issues um are they entirely separate or are they going to come into play here because you do sort of talk about like you know the false memory stuff and in a little bit of that which is directly related to eyewitness testimony and all that stuff but I just wonder like how much of this do you think is going to play out in this discussion about um well levels of autonomy in the population differing and therefore individuals having different

1:12:23 levels of autonomy I mean there is a you know the Eugenics movement stuff like that lurks in the background so there is a danger in that along that Dimension as well too don't you think 100% yeah and I think this is where um you know I don't want to this is this is I guess we Circle back to the point we were discussing about free will the point about autonomy there that I was making was a fairly narrow point about like how meta cognition contributes to that I wasn't

1:12:56 trying to argue for a um a sufficiency case um no I didn't mean to suggest that no I I agree but I was just saying the flip side of saying if you improve meta cognition then you improve autonomy um the flip side of that is well if you have less autonomy then you have less metac cognition and if there just is a kind of well I guess I guess I was um I would challenge that like because I don't want to draw a one toone correspondence there I think

1:13:28 there's plenty of different ways that you could have and this also then comes down to what we're defining here as autonomy which is also you know broad as a concept so I think there are there are many different ways you could have autonomy I would just pointing to this more specific way this specific aspect of if you have a mismatch between your metacognitive assessments and your first order desires actions and so on then that would seem to be a case where your

1:14:01 autonomy is reduced because you are not able to pursue the thing you want at the higher order level whereas when things are aligned then it seems to be increased but that not so I wouldn't want to say that therefore you need the metacognitive element to have this broader notion of autonomy I'm just saying that like one contributing component to having having yeah exactly okay there wouldn't be like a you don't Invision people reading your book and then becoming convinced that

1:14:34 they need to go and like it's a moral issue that we improve the a the metacognition of like people across the board you know and we should be aiming to do that sort of you know well I so I so so I think that there are cases so people might um draw some moral implications but not from that narrow autonomy point the just the broader point that um and this again goes back you know the very old notion of

1:15:07 why um having self- knowledge is something to strive for that um and I talk about these experiments in the book that we've we've done some work looking across larger populations about what are the features of the way people form beliefs about real world issues how does that track with metacognitive ability as measured in these tasks and we have found that people who form more rigid and dogmatic beliefs and so are essentially less sensitive to evidence

1:15:40 against their position or that they think they're right and everyone else is wrong and we've looked at this both for political beliefs also for climate change beliefs those are the people who also have lower metac cognition in these laboratory tasks and that makes total sense if you think that there are some computational components involved in essentially checking whether your first order beliefs are on track or not and so I think it all makes sense and there's one it also goes hand inand there is I don't think it's a very

1:16:14 controversial statement to make that part of the um reason that educators are interested in improving metacognition in educational settings is ESS to create Learners who are not like that they're not saying you know what okay you've given me the answer I'm done um see you in a couple of months for the exam you know they they want to be like n okay this is nuanced this is complex I'm going to revisit it I'm gonna so I think that I could see where your question is going a bit in terms of like is this dangerous are we kind of

1:16:47 creating an imperative to improve mental cognition but I think if we view it through the lens of why do we educate our kids then I don't think it's that dangerous really it's just part and parcel of creating more open-minded Learners yeah cool interesting um uh I want to switch gears a little bit if I can and harass you about Consciousness since allegedly this is supposed to be we're supposed to be talking about Consciousness I don't take the title of this Consciousness life business too serious uh because I think about much more about the mind but anyway um you do

1:17:20 have views about Consciousness and they're very interesting uh so in the book you you try to relate metacognition to Consciousness and maybe we'll talk about that more generally but one of the thing that kind of one thing that kind of struck me in the discussion was your uh approach to phenomenological overflow um so roughly the idea that there are experiences that we you know uh that are outside of our ability to report or attend to or notice or something like that depending on who you talk to they'll say different things about it um it it sort of felt to me

1:17:54 like what your view was that gee this isn't an empirical question so I guess if you like that notion of Consciousness that's great but it doesn't really do any interesting work in the things I'm thinking about so that was the impression that I got from reading what you wrote it and I'm wondering if that was the wrong impression of your view about like unaccessed Consciousness if it was more like you sort of felt like it was a terminological issue or it wasn't something important I hinged on it so I was just curious if that was the wrong impression or if that was indeed uh sort of the way you you feel about this

1:18:26 debate so maybe just say briefly again what you took away from that so what what was the impression so in your discussion of it you say okay so you know some people think that there's you know Consciousness um in the absence of metacognition basically this kind of self awareness and other people tend not to say that and and then you kind of just say um yeah it sort of seems like you could say

1:19:00 either one because I don't find nothing that's literally of a contradictory nature in that but I don't also think it plays any interesting role in like a theory of or what we think of as interestingly human or so that was where I kind of lost track of what I thought what were you trying to say exactly like are you saying it's not a real thing it's not or it's a methological question that we can't address or so I just kind of that's fair no and I think that um you know

1:19:34 that I think I was if I that that chapter that kind of briefly detours into the connection to Consciousness is looking back the least developed um in the book and I think it something that I'm thinking much more about now and um you know trying in fact a lot of the lab's work has switched on to that track is to try and rather we are still doing empirical work on metacognition for its own sake

1:20:09 um a lot of the kind of things we've been talking about so far but a large chunk of the lab's work is now devoted to investigating in computational terms like using the methods of um cognitive science and cognitive Neuroscience to think about like how does metha cognition broadly construed contribute to primary experience to phenomenal experience um now in terms so that's all the preamble to say that I

1:20:41 think there's whatever was exactly said there is um underdeveloped and I think that my current view on overflow and the richness debate is that I do wonder whether it's methodologically an impossible um Conant to break out of that that it often does de does turn on the contrast between different types of measure and someone can always turn

1:21:16 around and say well you know that that it's possible that under that measure they were actually conscious before you measured it and so on you know the whole back and forth with um and with the fleeting nature of of of the experience and and that's what's often used to bolster the rich position um and I think that that that so the way I see it is that it's hard to see how empirical evidence can be used to adjudicate that

1:21:50 particular debate and I wonder whether we're actually investing our efforts in the wrong place there because really what we should be doing is investing our efforts into understanding why um people have these beliefs that they do about their experience meta problem is that what you're talking about yeah okay so yes yeah so are you an Illusionist now Steve what's going on here so I I don't yeah okay so yes

1:22:25 basically but I don't like um I I'm not a fan of the term okay because then everyone turns on you and accuses you of not being conscious yeah I'm about to but you know I do I do feel as though um you know there's there's been an interesting shift and this is where it connects to master cognition and now I'm talking just very general terms there's been an interesting shift in the world of cognitive science from kind of Bernie bars in the 80s talking about availability of

1:23:00 information being shuttled around different processes to considering more and more about the importance of um a computational explanation of belief so what makes a system think it has this kind of experience and and there I think that you then end up encroaching on The Illusionist position because you are then saying um you know that's the thing empirically we want to try and

1:23:34 explain now the the critique that then comes back is well you're just explaining away you're you're you're not engaging with the experience itself you're just engaging it with the what people think about their experience like and and I think that I like Andy Clark's terminology here where he calls it revisionist illusionism where what what he wants to do is say no it's the system that is you know the perceptual system has these features that give it this um sense of external reality that just has

1:24:07 overwhelming force that we which is the reason we show up at Consciousness conferen is because we find it fascinating but that doesn't mean that we can't explain it in the terms of cognitive science which refer to information and beliefs right so that's the I guess the general thrust of the way I think is empirically productive here but you know I I recognize that this is a very contested um view yeah so I mean I hear way maybe two positions in what you're saying and

1:24:42 maybe waffling between them I'm not sure um but so uh one is what I would call methodological illusionism or whatever which might say that you know given the tools of science that we have what we can explain is why people make the judgments that they do why they hold the beliefs that they do um and but that methodological claim is a claim about what we can study and what science can achieve but it isn't a claim about what's there so maybe there is phenomenal Consciousness and that's what

1:25:15 you know part part of you in the book you sort of say maybe maybe okay so so maybe on this methodological interpretation look science can't address that issue science addresses this other issue that you know is more friendly to illusionism but it's still you know in reality that thing may be there but the more stronger of view the one that the illusionists often take and the one that gives them push back is not methological it's like ontological or something it's like in reality there is only these beliefs and the other thing isn't there so do you come down on a

1:25:48 side in that uh distinction or do you think that's not a meaningful way to distinguish things yeah no that I think that's really helpful um so I think I'm definitely on board with the methodological position um and I and then I and I think because if you are on board with that position then you're I guess naturally sympathetic to The ontological Illusionist position but I mean I I'm not an authority on on top as in like I

1:26:21 don't know what's real like that that's the point of doing these this science and then there might be let's see I guess I guess my feeling about this is that let's see how far how much progress we make yeah with the an explanation on the psychological level and I think that you know actually the tools and the the methods for doing this kind of work you know looking at um you know taking metacognitive processes

1:26:55 seriously taking confidence belief formation subjective inflation how that interacts with um qualitative judgment I I think there's loads of cool things to be done in in that and then if we end up at a position where we kind of can explain all your judgments about your experience in psychological terms that are rich enough to capture those judgments and can also then kind of show why you might have this overwhelming force that makes you think

1:27:29 there must be more going on I think that's the really interesting meta problem aspect to it then I think that you know maybe we'll convince a few people that there's not much more to be studied but obviously you know clearly there could be more there could be more um I just feel like then you do start hitting up on the buffers of what cognitive science can can do um yeah I mean yeah so it's like you left you left New York and went out into the you left the phenomenal light of New York and are now in the cold of London

1:28:02 Town I can see why you'd be so depressed St but come back into the light my man just kidding but uh yeah but so this really I think gets to sort of a heart the the heart of the matter in a sense because there really are two approaches here um what what you start out wanting to explain so one side the side like me says well you start with Consciousness that's where it all you always you can't start anywhere else you start with like here's me in the world and it's like this um and so that's like

1:28:39 a datum that there's something that is like that there's a phenomenal aspect and you've got to have that in the explanation at the end it explain them or whatever I don't know Latin well enough to know that I'm making a silly mistake I think I know I'm making a mistake M but I don't know it's explanans whatever I do know anyway um so uh you want it there and the other side says okay no that doesn't go into the data what goes into the data or what people say about that but not that and that's like the den heterophenomenology

1:29:10 versus like reality I would say but uh versus the sort of first person approach um so I do think well so just so just I mean just so I wonder whether we've been I I I think um I I very much hear what you're saying that this is a kind of contrasting fundamental difference in starting point yeah um and you know the the The Great Divide the Charmers in the Charmers 1996

1:29:44 book right the kind of is is is like a fundamental Fisher in in the field I feel um that and I do wonder whether this is where adopting more and more naturalistic cognitive science approaches will help bridge this it it it might not convince people who think that this is already beyond what psychological science can do and again I'm not saying that those that that

1:30:16 position is wrong I just feel like it's beyond what psychological science can do so you might end up with a metaphysical position picture that you think is the correct one that is outside of what you wouldn't need to run a cognitive science experiment but I'm just within this um this debate about starting point imagine a case where you can put on a VR headset and turn all the dials until you match exactly the properties of the thing you started with this is the kind of thing

1:30:48 that anal s group are working on um and other so you know the methods of Cai tend to be like butt and presses and language right and and that's where dennet when he was writing about H phenomenology was starting from because that was the tools that were surrounding him at the time I just wonder whether when you get tools that essentially exhaust it in interesting ways I mean all the Mike Cohen richness stuff is really cool because essentially that's what he's doing is getting people to like match their experiences right in a way where they're surprised that it was

1:31:25 not like what they thought it was and then you can use those data as a basis for starting to develop what processes might support that particular experience right so I think yeah I I just wonder whether some of this fundamental Fisher can be healed with methodological development in psychological science yeah it's interesting and also so not only does it sort of is it a fish or in like how how you start but it's also a Fisher in like how you think science should proceed um

1:31:58 because you know for so for me I'm very influenced by the kind of Ned block approach uh I'm influenced by a lot of things but part of me is influenced by the by the by blocks approach where the idea is that you're not really going to get an explanation in terms of cognitive science of of Consciousness because it's sort of you know just a a a biological thing that is going to you be used to explain other things so that if you want to explain why someone you know is not wanting to grab the knife by the sharp end part of the answer is that in the

1:32:32 past they felt pain and that feeling of pain is acting as a a reason for them not to uh want to have it again and well how can you explain that well if pain is a certain bi feature of the brain then that is part of the explanation but you're not going to explain why pain is that brain State because it's like explaining why water is H2O at a certain level it just is that and that's the end of the discussion and then you go on to explain other things using that and so the block style of of argument then goes okay so would we ever have evidence that

1:33:07 these things are there and that's where the whole mesh argument comes in you look at what Neuroscience has discovered what psychophysics has discovered you kind of overlap them um and what you get is kind of it looks like a picture that emerges is that maybe there's more to what people experience and what they can metac cognize and so that you don't get a regular neuro a regular cognitive Neuroscience explanation you get this other thing that emerges from doing all the regulative cognitive science like a

1:33:38 an argument to the best explanation or anyway but uh yeah yeah yeah so it's just a different approach that's what I was trying to contrast not to say which one is right or wrong but that it reflects more than just like I like Consciousness it all also reflect like how you think the science of Consciousness should proceed I think that's really the fundamental issue 100% no I I think there's lots of other yeah and I I think because I've done meth cognition research and I'm I've got you know papers on higher order Theory and prefrontal cortex and

1:34:13 stuff I I think I'm often held up to someone who thinks you know it's it's only what you can metac cognize and it has to happen in the PFC and like it's and I actually I'm very sympathetic to the block kind of let's get to a place where we're as consistent as possible between what the behavioral data um right as M can we exhaust as much as possible the first person experience using third person tools and see how that lines up with the

1:34:47 neuroscience and you know that that's I think definitely been influencing me in terms of like thinking about this distinction between um you know maybe the all the first order representation provides the content that might be sufficient like that's that's a kind of back of the brain style of theorizing um I think there are additional things that are needed to um get you to a position where you know something becomes conscious in a in

1:35:25 a in a psychological in a way that can be studied in a psychological experiment but yeah I I I definitely feel that we're only just scratching the surface about what is the right way of characterizing the psychological processes in neuronal terms I think that's like that project basically hasn't really started I feel exactly um yeah and but then that I I I guess I would and it's really interesting we're having this discussion

1:35:58 now because I've been thinking a lot about so there's this um I'm involved in this Symposium at ASC in July with Ned and annel Seth and Le mudri organized by bu on yeah on um yeah on uh functionalism and because that is it here because not a functionalist about about mental quality right right exactly and and I guess like and I'm sure that the you know the the non-f functionalist would say the

1:36:33 functionalists are being way too premature and thinking everything's functional but I would kind of say the non-functional of being a bit too premature and saying everything's biological or it it feels as though like let's give the functional explanation a chance to even work um before we take that off the table I mean it was mocked in the 80s before they even could build it like oh Chinese room China Nation but no one had even built anything yet they were like mocking it conceptually so I I take your point they haven't really had the chance to even get off the ground empirically yeah and I and I also definitely do think that like

1:37:06 um the the kind of metal cognition we were talking about in the first part of this discussion is the the the explicit Judgment of a decision or a person that's not what I you know I've I have in mind here when I'm thinking about confidence in belief or I think that is much more allied with the Notions of precision in predictive coding architectures that are for the system um rather than something that's like postdecisional or goes beyond it um and yeah so I I I do feel as though this is an exciting opportunity

1:37:41 to try and pursue those kind of explanations um in psychological terms it's interesting too because I remember one because I like the biological View it's been my my oldest probably Prejudice in in philosophy actually that uh there's something biological about us humans um like innate distinctly biological uh so you know when I find people that are smarter than me that also think that I think okay let's see what they why they think this um but uh one thing that struck me was I remember

1:38:16 I was because anyway my point was I've had these views for a long time I haven't been able to give rid of them although I sort of becoming more friendly to functionalism as I get older um I'm sure I'll be fully friendly to functionalism by the time I have my first artificial hippocampus implant but I remember well there you go yeah there there it really does uh it really does connect to the last part of the discussion once it becomes real but I sort of remember one time I was talking with hawan or maybe yelling at him drunkenly you know it's hard to tell the difference sometimes uh but uh I was talking about this stuff and he was like

1:38:50 look what do you think biology is it's just function that's what biological means it's just this what one thing we've learned is that what it means to be biological is to like to to be doing functions basically so what else is there to biology besides I like I don't know the calcium man yeah no absolutely yeah that's a more fundamental point is like when people say well can what shouldn't it be biological then it really is open question what do we mean by that yeah and I I mean I I I'm always um I always

1:39:25 think back to the think I think it's kinds of Minds that denn it has this passage about if you really cash out the functional story in all its Glory then you might end up with something that basically is indistinguishable to biology you we don't know yet like whether that's yeah if you really want contact with all the kind of Rich sensory input that we have and the way we interact with it then you might end up with a robot that's basically a a biological replica it feels that goes to Ned's you know block has this argument where he says look do

1:39:57 we when we create artificial Minds do we have to create them lack ours and his example is like you know uh people with dental fillings who go up in airplanes above a certain level experience a certain pain in their head um do we need to build a robot that does that in order to say it has Consciousness like ours like I I don't know like so the questions become murky once you realize that maybe what we think of as conscious experience exists on a spectrum or within a space and that we might occupy narrow region of that space actually um so it does yeah this is why I'm a bit worried about this Symposium because n

1:40:30 Ned always has an answer right so it's usually very comp it's usually very compelling and bummed that I can't be there for that uh I have to watch May hopefully it'll be on video but that'll be a very discuss hopefully yeah it'll be fun it'll be really fun so is this what are you going to talk about qual quality spaces and um things of that nature well so yeah so I think um you know I think aside from just trying to articulate the broader position briefly um and the broader promise of of This research

1:41:04 program um you know I I'm going to talk about psychophysical investigations of reality monitoring as a way into this so essentially that is asking when do people think or believe an experience is real and when do they think it's imagined and we've been doing quite a lot of empirical work on this recently um you know we can show that people have surprisingly poor um monitoring of this that they have a kind of and I think that's again an interesting thing that people can be surprised about um on a intellectual level that their kind of

1:41:41 the force of their experience is actually more murky than the source of that is more murky than you other would otherwise believe I mean I think graciano has an nice um Michael gra has a nice example of this where he talks about the um hallucinations as a as a kind of test case of this like if you um if you're suffering from a delusion that you have something lodged in your head that's not actually there then you know we might be able to figure

1:42:14 out and people are working on this kind of thing exactly how the those representations come into Force like literally where are they located in the brain how are they coded um how do you assign a confidence or Precision estimate to those those um representations and that might end up being a kind of sufficient explanation for why that person is absolutely convinced that they have this thing that's lodged in their head that's not actually there and so I feel like you know there's an opportunity for

1:42:47 similar explanations of why people think there's definitely a grating on the screen even though we never presented them with one and and so on so I think you know that's the kind of thing that empirically I'm going to talk about obviously this is like very minimal and constrained um but the idea is that if we can learn something about the principles that attach confidence to minimal forms of perceptual experience then we could start to generalize that to richer forms of experience interesting Steve you're doing it wrong you have to talk about Mary in her black and white room

1:43:21 oh yeah yeah I know exactly you're GNA get sucked into a discussion about that you know that no I know um but then the the other the other thing that I think is really important for all that is that like if if you are going to be a functionalist about this then you also then need to think about a a you need to start talking in terms of functional role and like what why why would the system be doing that like that that I think is if you could start making progress on the and David Mah has this you know this whole idea that if you make progress at

1:43:55 the Y level at the the why is the system set up like that then you can make a lot of progress at the lower levels in terms of How It's implemented and and so on so you know he the the people often use the examples of the bird the feathers on a bird if you didn't know they were for flying then you would have a hard time figuring out why the bird have wings um whereas if you know that this kind of weird what um seems to happen that we kind of if on our like our quasi Illusionist story is is on the right

1:44:28 lines then there must be some functional role to why we have this like um grabbing force of experience like why are we why are we um convinced that there's you know something special about phenomenal Consciousness and um and I think one another angle we're trying to pursue here is what you know what does this do for the system and and this is inspired partly by Haan LA's ideas on what does the aceto force of the um reality tag do for you in terms of

1:45:03 helping you then base rational action and planning and decision- making on a belief about what the world is like right now and that's distinguished from what I'm doing internally so I need some way of kind of fixing if you like the external World while I then act and plan and so on a bit like it would be annoying if I was like sitting there at a chess board trying to make a chess move and someone came along and moved all the pieces around it's like hang on just fix the board for me for a second while I plan and that's the

1:45:36 general idea is that like we're going through this process of like perceptual fixing and then we're yeah and and other people yeah this is not novel I'm just trying to pull together a few strands like other people have talked about this a lot before yakob how's account in his predict M book is quite similar yeah yeah but that that's interesting because it's like similar in a video game you know uh first person games that you can pause and look at your menu they have there's two different kinds there's ones where the world outside doesn't pause while you do that so you have to be looking at your supplies while everyone's attacking you running around

1:46:10 and then there's other games where the outside world freezes and let you sort through everything and make a decision before getting right back to it so what you're saying is that second model is sort of more efficient for a model of how we do things um so we take like a snapshot and use that to to reason about yeah exactly and and I think that um I mean so it gets quite quite fraud about snapshots because the the wrinkle here is that the snapshot is probably itself Dynamic um you know includes motion and things that are moving in the

1:46:43 world but I mean a lot of the um here I've been quite um influenced by this really beautiful work from Michael herzog's group on temporal Windows of integration um where there's this surprising predictive and postdictive influence on how you essentially become conscious of a moment in time it it still phenomenally it still appears Dynamic but the idea is that that's somehow being built and then served up into conscious experience

1:47:16 and then you know if the if the windows a little longer than might might appear useful for the immediate control of action that then starts you thinking well what is what is it for and I I still I don't think we know I think those experiments are again just scratching the surface on what it means to kind of build this slightly longer window of integration um interesting I mean yeah so so I think but there's lots of interesting work on the temporal stability of experience and the oscillation inherit the perception

1:47:49 action cycle um there's there's cool stuff in fact Doris s had this talk at ASC just gone on these oscillations that seem to be related to um flips between integrating information and then creating the the conscious experience and but this was all happening like at a theer cycle 250 milliseconds or so yeah which would somewhat fit with the the hertog DAT as well so yeah I think there's lots of Stu to think about uh well okay so I have there's at least

1:48:23 two more things I want to ask you about big picture- wise and I can't keep you forever so I know you said you had open schedule but I know that's there's still got to be some uh within reason some cut but but before um I mean we were just talking about Consciousness science and I someone want to ask you about that but also I want us go to the frontal regions of the brain before we end our discussion because You' become quite an expert on the uh um effect of damage to the prefrontal cortex onc conscious experience and you

1:48:56 kind of you say in like almost in passing in in the book that you know you could damage the prontal cortex and you do get some effects on experience but this of course has been a huge controversy especially in the last few years about people thinking that you know you could basically remove the entire prefrontal cortex and and not affect conscious experience and you're often cited as someone who's work has been um suggesting the opposite of that and you mentioned the Paradox of the prefrontal so I I wonder if you could just talk a little bit about what your

1:49:29 view is about these kinds of claims um with respect to the prefrontal cortex and and Consciousness and meta cognition how they all relate yeah yeah no I mean I think I mean what we've shown is that if you have a prefrontal legion that encompasses the frontal polar cortex so robman area 10 right towards the front of the brain then you have problems with perceptual matter cognition so you have problems with assigning confidence to particular percepts to particular perceptual decisions um

1:50:04 and that is in the absence of any other change in or any other difference in neuros pychology in IQ we have age match controls who with the same intelligence neuropsychology and so on who don't show that effect and so that's I think um I said paper in brain in 2014 and that's um a paper that's often referred to as you say to claim that there are subtle but systematic differences in something perceptual because it was specific to perception that's another

1:50:37 thing to point out that when we gave them a memory metacognition test there was no difference to controls and the effect size is pretty huge as well so there are about 50% impaired comp there's a 50% drop in the metacognitive efficiency matric compared to the control groups um so and you know these that was a naturalistic um sample of patients with lesions of varying size due to tumor resection we did have a lesion control group of patients with temporal lob lesions of

1:51:11 similar sizes but still it's the lesions are hetrogeneous messy um but there's been recent work from kentaro mamoto where he's used reversible in activation of frontal polar cortex in maax to show something pretty similar in the context of a um a perceptual um uh task where you have to then they they have this non-verbal confidence estimate and they have a big effect size drop in metacognitive efficiency when you in inactivate the

1:51:43 frontal protocol set so I think there's pretty good causal evidence that these anterior regions of the pref CeX are doing something important for perceptual Master cognition now how that then relates to Consciousness I don't know is the short answer good um right because we're dealing there with a metric that is um going to draw upon all the things we were talking about in the first half of the call um these you know we're dealing with an explicit metacognitive

1:52:17 judgment that's going to draw on all these other cues that people have to their confidence yeah the biases and so on so so I think that I'm I'm also sympathetic to the case haam makes that if you're tapping into an explicit metacognitive judgment and you show that is screwed up then that could be evidence that more implicit forms of perceptual method cognition that are relevant for these models of Consciousness might also be impaired so I'm sympathetic to that view but I don't think we have the data to show that that's the case in those patients so but do you think that we

1:52:51 have data to show that you can have prefrontal cortex largely absence but no disturbance of uh I mean but so I mean one of the critiques that I've seen of this line of argument which is kind of what I'm s to hint at is that the S the effects in your work from 2014 are very subtle um but uh wouldn't you expect wouldn't you expect like much larger effects if you had um like in the case of patient a you know has that very probably not all of the prefontal cortex

1:53:24 but most of it so and there's lots of damage I mean I mean you know so the the kind of philosoph uh the philosophical case from Bene who I know you know um uh from Benji from who sorry uh Benji Kik oh yeah yeah sorry um uh that he relies on is mostly surveying the empirical literature and saying look in all these cases of of um damage to prefontal Cortex you have all this preserved ability um and you would expect some deficit in like if

1:54:00 you had found a central area that was important for Consciousness uh now you know so I know you have a response to that but I'm just curious if you could tell us about it yeah I mean I think so you know having worked with patients with brain lesions either due to stroke or due to um tumor res C things are very hetrogeneous and and you almost you very R see clean bilateral damage like the you know that cuts across two hemispheres um and so you

1:54:35 mentioned that RF effects are subtle I actually think that RFS are surprisingly big given that there is a lot of sparing of of tissue we you know we're talking about a 50% drop in efficiency that's not ablating the whole of the prefrontal cortex right far far from it right even not ablating the whole of the frontal polar cortex in most patients um so I think that you know the you need to take the results in context of what's common in the prefrontal legion literature I mean if you had a total as

1:55:08 an adult if you had a total excision of your of your frontal Lo then you'd basically be unable to do anything you'd be unable to move and speak and and so on right so and so I think that we're dealing here with subtle differences the prefrontal cause is a big territory of tissue um I I but I kind of know I I I'm sympathetic to the thrust of the critique which is I think there's been a notion that if Consciousness is quotes in the PFC then you take it out and then

1:55:43 you should become unconscious and I just feel like I I don't think that's true either I think the I think there's overwhelming evidence that in perceptual tasks the primary and and secondary sensory areas going back to like the late 1800s are absolutely Central to qualitative features of perception you know and and perceptual processing the the question is like can you get from these Legion cases insight

1:56:15 into whether some aspects of those responses now [Music] become more on the leaning towards what we would perhaps classify as unconscious in say a primary experiment or you know something to where we have lots of control over the stimulus and we can measure it properly so I I don't know is the question is is the answer I think I'm just trying to say is that like I don't feel like it's simplistic as either Camp makes out right and that's why you're attracted to your own like your higher order by the way I saw the discussion with you and

1:56:48 Victor at the last assc before last and he says what's this hos what the S stands for and you say states that was a classic moment but anyway so that's why you're you're drawn to the like kind of uh the split level or joint determination split level as I call it sometimes um kind of view where the metacognitive stuff in so far as is important for Consciousness is is not like providing the its red part it's it's providing a kind of you know how bright the red is or how Vivid it is or

1:57:21 something or right but the redness of the red on your model is encoded by the first order States um or at least that's the way you would like it to be yeah I mean I think there's I say it wrong well so I yes a little so I think the the the model is just a model of you know a a um I I definitely wouldn't go so far as to think of it as like a

1:57:54 fully is is far from being a fully worked out theory of like Consciousness r large I would say this is a model of the kind of cognitive process that might account for some of the features of um how the system can make inferences about these two important components of conscious judgment which is both their content and also the level of awareness you have of them so if you take that as

1:58:27 your starting point there's more the qualitative psychological side of things and then you say well how can a a system that's we think is doing inferences at various levels of abstraction essentially how can we kind of extend the regular predictive coding view of perception out to include this more more abstract notion of being aware of lots of different types of possible content and so there I and and there it is kind of tilting at this [Music]

1:59:00 um this psychological notion of awareness that we can study in experiments by asking people to report on what they see and hear and so on um so there is still then this open question that comes back to the Overflow debate about like if you take that level away what are you missing and so the in the paper to articulate weak and strong views the strong the weak view would be you take that level away you're missing reportability but we're just going to be like agnostic about what else you're missing right and the strong view is that that extra level is like playing the same role that a higher order

1:59:33 representation plays in higher order theories that it's making you phenomenally conscious the first order representation right and so you you kind of haven't I mean the paper when it first came out was like this is just a model of like how people make inferences and report and and then later like oh maybe it's a theory of Consciousness and now you're saying well actually the first one it was a theory of Consciousness I just didn't want to say it because that's what Consciousness is it's just making these judgments and then issuing the reports it's what the model ultimately turns out

2:00:05 to be saying yeah this we're gonna get you onism one way or the others D there a yeah I know there's an interesting way of characterizing the journey it all seemed a bit more joined up than that but you might be right no I mean I think that like um yeah I I think that you know I I do worry about and this more gets to the kind of um metascience level of like you know if we were in the the

2:00:42 memory field like and I suggested a model of how a particular feature of memory works then I wouldn't be asked to say is this a theory of memory is this actually a theory of all memory like you know you got everything in it and and you know I feel like there's quite a push towards universalism when it comes to theories of Consciousness and I can understand totally get why that is um interesting but I do worry slightly that like and I think part of the um you know my this model together with um yourself

2:01:19 and others is part of this adversarial collaboration on higher order theories and I my reading of that is that like there's this umbrella term which we call high order Theory and we now have like kind of different models of component processes within this broader umbrella term that were were contrasting and trying to test against each other and that to me feels a bit more like Cog cognitive science in other domains whereas there's you know other um versions of those adversarial collaborations which are really kind of pitting Theory r large against each other and I think that a different sort

2:01:52 of that's a different sort of endeavor and and I I guess like the the ambition of hos is not to say we've got the ANS to it terms like this is a fully fleshed out model it's to say this is one particular competition that I think we should take seriously um as potentially playing a role in the explanation here just just really quickly on the universalism stuff do you not get that with memory so we do kind of have a standard picture of memory at the psychological level at least that you might think of is a um

2:02:27 starting with the working memory Bandera kind of style updated with what we've learned about automatic encoding and and um classical conditioning and stuff like that so different ways to encode but there is kind of a bigger like a system of memory um which then you're right has different components like there's the episodic memory semantic memory procedural memory but they're all part of this larger system which does get explained generally in the same sort of way one might think so is that I mean I

2:03:01 get what you're saying but isn't there isn't that the end goal to end up with something like that for Consciousness too so I I would say yes but I I I um I guess I would question and this is saying this obviously is a outside it's to the memory field um but I would question whether there is a unifying force at work there that where aside from the fact they're all studying aspects of those computations that are relevant for what folks psychologically we call

2:03:33 memory um I feel like in Consciousness the push is much harder much stronger to say ultimately you need to be aiming at this one unifying notion of phenomenal Consciousness and if you're not tackling that one unifying notion then you're not really doing conscious science and I feel like that's not that unifying um force is less prevalent in other areas of psychology yeah um I mean there is a the sense that when you teach

2:04:06 a class on this uh on memory for example cogni neural science of it by the end students are like where's the memory because it's you get in there you say well there's a thing called memory they're like actually no there isn't there's these four different things that each have their different systems that each kind of work in a different way um so in that sense so I I so I think sorry to say I think that's where we're going to be when we're teaching our classes on Consciousness in okay yeah interesting that's what I was wondering but I mean I'm kind of saying

2:04:40 it tongue and cheek a bit because I know I'm getting back to your uh favorite topic of illusionism but like essentially that's where you know that's that's I feel like there's the yeah there there is an the the adopting the perspective of kind of component computations or models of different interacting parts does mean that you end up with to some extent explaining away the original folk psychological thing you started with um and I think we've become more used to

2:05:13 that in other areas of psychological science yeah and I do wonder whether we'll end up getting more used to it for Consciousness eventually or maybe we won't maybe we'll just you know still consider this as like an outlier in in the field I'm not sure where it will go yeah interesting I mean I would say that the thing that's left over in all those cases is phenomenal Consciousness in that the thing that we accept is kind of in the other terms is the part that we can say okay like the part of memory that's okay to uh explain away is not

2:05:46 the consciousness of my memories or what it's like to experience them but there's other stuff about storage formation updating blah blah blah they're related in some interesting way but I do I would be on the other side I don't think we're gonna get to that point but you know who knows I I'm happy no I appreciate that but like but also when you talk about memory in those terms you know you've been kind of we've we've both grown up in a world of psychological science a kind of cultural M where that is just the way of talking about it especially

2:06:20 with the computer metaphor and so on so yeah I and conscious science in terms of the trying to do essentially kind of do um yeah empirical science that tries and tilts at the phenomenology um that does have a long history but it's a bit more checkered yeah and in terms of a systematic science that's trying to connect down to the computational ultimately hopefully the neural level I do get a sense that's quite young so I

2:06:53 wonder whether someone growing up in a world where that's just the standard way of pursuing very different view yeah would might feel differently I yeah that's interesting I mean we'll have to wait and see um for me I would you know I have a kind of death first attitude towards illusionism so uh yeah well yeah yeah that's true okay but because I do sort of think like loss of consciousness is almost worse than death like I do find like if I were to find out I was a zombie that like yeah gee Life Would Suck um anyway

2:07:28 so that well that's that's why if you if you come over to The Illusionist Camp then you can't you can't have that right so then you're all right illusionists are zombies I mean that a world where I have I have this little jab in my book where I say a world where illusion of is true let's call it diluted Earth um instead of PL Earth in a place like diluted Earth where the theory of illusionism is true there show zombies I I can't live in that place I don't live I don't exist there um I don't think I mean I feel that as like that back to that thing we were talking about before

2:07:59 about a fundamental difference in starting point like there's no possible way that I'm an an Illusionist World um but anyway I I mean you know as I get older we'll see check back in 35 years because I probably would have said the same thing about most of the things I believe now if you had to ask me you know when I was younger so that's Works um but but I before I I let you off the hook and I know you know I can't again T keep you forever but it's turning into forever it only feels like forever Steve no it's been fun it's been fun but uh um I want to ask you you brought up the

2:08:33 adversarial collaboration and you are in an interesting position with respect to that maybe a unique position because on the oneand you're a leader of uh as you said one or suer or a co-leader I don't know how you want to phrase it yeah yes you're up there in charg this of and adversarial collaboration but also you are the lead author or at least the first author on a critique an open that open letter which critiques IIT but also the adversarial collaboration I would

2:09:06 say um between IIT and Global workspace um so it's sort of interesting that you find yourself in a position of being like a leader of one of these things but also critiquing how it was done so that's not obviously contradictory or anything like that it's just an interesting position to be in so I wonder if I could ask you about that interesting position so how did you come to feel well to be involved in in the Templeton project but it also feel compelled to be involved with this other letter yeah yeah yeah no I mean I've been reflecting on this a lot recently

2:09:41 um I mean I think the second half of last year was um in on a personal level very I think stressful because I felt in this position of tension as you articulate um I usually don't like conflict and I think the letter was um in hindsight you know made it was incendiary and um you created a lot of a lot of conflict um and I think people

2:10:14 you know too like you know anel and guys are around neighborhood yeah I mean I think so so I mean I've talked to a lot of people on both sides of this since then and I feel that there are lessons to learn on both sides which hopefully that is a good thing that comes out of the of of the SEC the past few months I mean I think on the letter side I so you mentioned me as first author I'm first of 10 so like I was part of an initial group of joint first if you like um as

2:10:49 at all everywhere I've seen it well that's because I'm it's alphabetical um in that first 10 and there is a note in the preprint that explains that um but obviously that gets a little uh missed but yeah so I mean I wouldn't say I'm you atically first on that letter I am I am um yeah absolutely you know I'm part of you were part of like it was written by a core 10 people of which you EXA absolutely no exactly so so um I was part of an initial group um

2:11:24 that edited that um initial the initial letter um that provided comments and suggestions on the initial draft um I you know I the one of the major regrets is that I didn't realize at the time how incend the title would turn out to be so I think that you know a lot of people as as happens and I do it myself like when you see papers come past you might just glance at the title and and look at that or you know read the first

2:11:58 few lines and I think that actually the you the content is a lot less inflammatory than the title itself um and I think there's and you might think they don't doesn't really deliver on the title no exactly so I think that there is a technical use of the term pseudo science which is a worry about not making predictions about observable data and the diagnosticity of of those predictions and I think that this is a fair challenge for IIT and especially the adversar collaboration the cogitate

2:12:30 project and I've written about that elsewhere I had a blog post about how you know the reason the main reason for me that I got involved in this was that I was really worried about the lack of those diagnostic predictions that were being made by the IIT side the experiment and the project was fantastic in terms of just ushering in this um new way of doing conscious science like AV aeral collaboration I think it's fantastic it's partly why I've also agreed to co-lead the the higher order side um and the experiment was done with

2:13:05 state-of-the-art imaging the data quality is amazing and so on but the experimental design was similar to a lot of those that are already out there in the perception and vision literature where you show task Rel task irrelevant features and you look at how these are represented in multivar neural activity and the duration of those the those um the the extent to which you can decode those features in different regions of the brain and I think that because I am also part of the community in my kind of other life outside of

2:13:40 Consciousness science I go to you know a lot of conferences on Vision on perception on compensational Neuroscience in those in those fields and my real concern was that like this was being put forward as the kind of like the flagship project of recent years in Consciousness science but those predictions that there would be decoding these particular brain areas for the were not diagnostic of IIT so nowhere was the theory used to make predictions in terms of calculating fi or simulating the capacity of different architectures

2:14:14 different sub regions to um to have that Capac so I you know I I think that like the technical term that was that was used there is is a fair challenge but then there's obviously this more colloquial meaning which has a much more accusatory tone to it suggests like that you're being Mal you're being malign in your actions and yeah you know so I think exactly I mean that's what I always said is like this these are like fighting words you're accusing someone of being a bad person basically when you

2:14:46 say this very care no exactly exactly and I think that um you know there is now this longer version that's that's um in the works that we'll hopefully do a much better job of unpacking those worries um I I also think that there's you know there were special circumstances here because these points including the use of the word pseudoscience by the way have been made by people in the community in the past like there's I mean I would just just sorry to interrupt you but just really quick I was just watching this video I

2:15:20 mentioned it of you at the assc um on that panel discussion and I think I heard lead or Lucia one of them say the word pseudo science about IIT she said Victor said that you could consider this as pseudo science and I was like whoa so that was in the assc before last when obviously someone had said that what people were talking about this already so I I think your point is right you guys are not the first person or that letter is not the first instance I mean Michael graan is in print on it and Jake Hansen has a talk that's got a lot of views on

2:15:52 YouTube where he's worried about the demarcation problem when it comes to IIT and so on there's also the the the nature human behavior piece that did that I think were you a co-author on that the 54 yes exactly we didn't use we didn't say suici no we didn't use the term suici but there is a paragraph in that and that say Bas same thing exactly right um the and also people who signed that were also involved in the Cog project so you know there's been worries in the community and I think from I I was kind of like personally I like you

2:16:27 know IIT is this metap metaphysical position as a philosophical mathematical project seems totally fine to me it's like interesting right but the worry I had was that like there was this kind of the the the the cogitate project meant that there was this like sudden jump over into mainstream cognitive Neuroscience without articulating any of those bridging principles into it really right and that's what really worried me because

2:16:58 I'm kind of part of that field that they're jumping into and I feel like if we want to nurture this connection between conscious science and cognitive Neuroscience then we want to avoid kind of yeah making that leap without articulating the bridging principles and so I I actually think right so I've been thinking about recently that a lot of this would have been avoided this breakdown if the I theorists had said you know what we can't operationalize IIT proper so we're going to create a weaker version

2:17:30 called neuronal IIT and try and test this we'll use the connectivity metrics inspired by fi but applied to inter Regional connectivity using the kind of stuff that people like Adam Barrett have been developing that's all cool stuff really cool stuff right and I totally believe that you would find there's an important role to play for information integration in the human brain measured using those techniques um but that's not a test of IIT that's a test of this new Theory you've developed it's a bit like

2:18:03 the difference between Global workpace Theory and Global neuronal workpace Theory but we don't really talk about yeah so the the neuronal version of gwc is saying let's call the front paral Network the workspace or the way that the workspace gets built and then let's map all our predictions onto that and we can get sophisticated about that and you know it has got sophisticated but it's still testing a much weaker um view than than like the idea that you can build a little workpace and you know connect up a few nodes and so

2:18:36 so I think there's like a kind of um if you know and and it's almost like the stakes are a little higher for it I feel because it is as an Seth pointed out in his noralist piece the theory is a bit Bonkers in ascribing Consciousness to systems that folks psychologically we would think do not have Consciousness like sets of inactive logic gates and tol Boos and all you know all the kind of examples that have been used and in that piece you know he defends the right to outlandish thinking and I think he's right to say that

2:19:10 because I you know we need people to articulate different metaphysical positions and that's you know many of those positions May well one day be on to something that we just can't conceive that it could be the the you know I IIT core what I'm saying is it could be right totally it's it's a it's fascinating I honestly when I read those papers I feel like it's like an amazing thing that has been created here yeah but it it's not right so far

2:19:46 because of Neuroscience science so I think when you things get really worrying when you start to turn it around and say like we're now going to go from the Neuroscience back to this highly unusual what anal calls Bonkers um I mean and some people have been saying things like this and I I I wouldn't say something like this but I've seen people we know say things like well G you can't really use empirical work to test whether or not inactive lodic gates are conscious and I I guess what the criticism there is that like the theory

2:20:20 is kind of so Bonkers that no empirical work could ever establish it would you go that far do you think or do you think it's uh I mean not the weak version because I by the way just before you answered this question I 100% agree I know hawman has a paper on this too with Mattis I think on the dangers of yeah yeah well I mean there's also the yeah yeah yeah absolutely and then there's also the the tiex paper from annel Seth and um Adam Barrett and others right on weak versus Pedro mediano on weak versus strong IIT now I didn't know that yeah

2:20:56 and and they essentially lay out the the thing is that I do I worry that like by calling it weak IIT what they've essentially done is like say you can use you can pursue those metrics and gain evidence for the strong version of the theory I think what you need to do is do a global workpace move and say new yeah yeah yeah by the way Bernie was mad about that was he he thinks they I don't know you know Bernie well yeah he's like

2:21:28 that's my theory how's it different like you can't just put a he thinks that is his theory it's not importantly different but I think it is anyway so I I think you're going to disagree with that in Stan and those guys have a way different interpretation of how it's impl in the brain and I think Bernie did anyway so I think there's a lot of interesting discussions around this um around this question because it does seem so this is what I want to get at so first of all I want to ask you like how you respond to these kind of common objections that I've been hearing go around but also first like what is it that you thought was like warranting the use of that term because it's not just

2:22:01 in the title by the way it's in the in the main body of the letter too where you say you know we indeed do think that this pseudo science label is is relevant or whatever however you phrase it so it's in the title it's in the text so what in your mind was the thing which made you think okay gee this isn't such a bad a bad thing to do um I mean not not like you know use the word at all but what about IIT made you think that it Rose to that level and what I heard you say I think but I want to check was

2:22:33 that well they're saying that it's empirically confirmed when it isn't um so that's why my previous question which I want now loop back to it so how much like so okay is it that you never you think it could be like they shouldn't even be saying that they they can't it's like the pany saying I want to empirically confirm my theory like like Philip Goff trying to ask money from the NSF to to prove pan that electrons are conscious and you're like no you can't that doesn't even make sense or is it this other weaker claim that what they have is like evidence for

2:23:07 a different Theory not for their other Theory whether or not you could have it for that yeah so I mean I think I mean I've I've written about this on the blog post I mean I'll put a link by the way so people can read that okay great yeah that'd be great I mean so I think that um for me it's both it's both of those things um so the the way and and this is something that I theorist freely admit that like

2:23:40 it's essentially impossible to compute f for a a network the size of the human brain and probably network is much much smaller than the human brain but we need to start in humans um right otherwise we can't go beyond humans to much smaller systems right where it does start to become tractable and so I think there's a there's just an empiric you just an empirical brick wall there and then in terms of like the um the the specific um critique of the recent

2:24:18 experiment it's the diagnosticity of the prediction that like okay you might then say well let's have a weaker form where we just say in general there's some features of the brain that make it um that that make our Theory more likely to be right where you find decoding for instance but they had those predictions had no diagnosticity and I my bet would be that if you did actually articulate predictions that did have diagnosticity for a variant of the theory you would end up with something that's much more like this weak neuronal version we've

2:24:50 been talking about interesting but they didn't no one's actually done that and you know but if may you know that that seems like a really important thing to do um and I remember having this debate with people saying well isn't it just the same for Global neuronal workspace like they don't they have no bridging principles either and I was like well no because that's been part of the theory for quite a long time that like this it has been articulated in print that that is what they're calling the work it's pretty vague but then any bridging principles to the human brain at this

2:25:22 stage is going to be vague because we just don't have very Advanced Neuroscience so I think My worry is that you're essentially trying to like take the current state of of Neuroscience and say we can gain information from those kinds of experiments to support our um theoretical position and that's doing science that's that's essentially um advancing a theory is being supported but it's masquerading as science it's not actual science because

2:25:54 you're you you can't make that link based on those data so so it has nothing to do with pan psychism then the way you're saying it no has nothing to do with pan psychism or starting with Consciousness first or the axioms or any of those things but it's just rather this um well more basic issue of is the thing you're testing a prediction of the theory like that's the more the more basic issue like is the data even relevant to the theory at all yeah yeah I I think for me that um yeah it goes back to can you actually

2:26:29 get off the ground with the human brain here because when I saw this the conversation I keep mentioning at ASC but Melanie was there and one of the things that came up was the kind of PCI test so the zip and zap stuff I guess and someone was saying well gee doesn't U you know Global workspace make the same sort of prediction and Melanie pushed back pretty hard and said no this is a prediction of IIT it was it was it was predicted by IIT we did the test that way so this counts as an empirical test of that theory even if you can't I think this is her way of saying you could test it even if you can't compute fi um but no one really followed up on

2:27:02 it so I'm I mean I don't want to you can definitely test no the P the PCI stuff is really cool and you know potentially even might have clinical importance like all the stuff from marello mamini and and colleagues and Melanie others and it's really cool but I do feel like if you're going to use that evidence to claim support for a theory you're in the weak territory you're in the you're in the variant of the theory which is then not what you started with basically which is developing mathematical postulates from

2:27:35 axium that give you these equations that you can only apply to very small systems right right so that that's the so you get this impression that they started with the axioms god equations and then went to the brain but the equation stop at like the thre part logic gaug basically yeah anything more that is an extrapolation but even if it didn't stop at the three-part logic gate I think you'd still be in trouble because modern Neuroscience isn't of the level where you can even start building that bridge right so

2:28:07 yeah you know so you're in the realm and you certainly can't build that bridge from non-invasive data in humans at the moment and actually cogitate had invasive data right it had um inal recordings um so so yeah so I think that um yeah I I think it's it in a way like it it's kind of made me um I do I do Wonder like is this productive has this been productive at

2:28:42 all I think that like I the the one thing for sure is that if it hadn't been for things like the adversarial collaboration we would not be having these kind of conversations right I mean Leah mudrick has this lovely paper where she shows that everyone before this point in time was just looking under the street light at their own Theory and trying to confirm it right so it definitely has brought clashes to the four um but I I do worry that like um

2:29:20 what we're seeing here is a clash between just two totally different styles of trying to introduce Theory and the they and this is partly what I point pointed out in the blog post and Jonathan bur had a nice post on this as well where he talks about like the the Overton window of Science and and and so on like I feel like yeah it's really cool and essentially like my kind of nagging worry is that what makes it so difficult to

2:29:52 get theories like IIT and Global workspace even in the same arena is that Global numer workspaces operating in I mean now we're coming back to functionalism right it's opening a kind of functionalist um mainstream Neuroscience tradition like you go to Society for neuroscience and if you watch someone giving a post and said do are you a functionalist they probably wouldn't know what to respond to you but they are ex yeah

2:30:25 so James or something William James or something like yeah yeah as in like it's so embedded in the way yeah the the the mainstream is working that it's not really questioned now I think it is really interesting to have discussion about whether that itself should be then questioned like what do we mean there's a whole other area of debate here about like what do we mean by causal explanations in neuroscience and psychology what's the goal of modeling there and so on but in a way with Consciousness science

2:30:58 we're at a younger stage we're a much more nent stage with empirical Consciousness science and and my worry was that like we were essentially jumping the gun massively by saying we had those bridging principles in place to bridge over to the mainstream Neuroscience work that's going on that's not trying to like glorify mainstream Neuroscience I think it's got a whole host of problems itself but for better or worse that is kind of the Paradigm we're operating in at the moment and if we want to bring Consciousness science into it then I think we're going to need to do it step

2:31:29 by step rather than just say let's take evidence from this brand IM experiment and use it to support this um you know pretty extreme with respect to that view metaphysical position it's like the the you know the the the quote extraordinary claims require ire extraordinary evidence is it was it Carl Sean some some I think famous yeah exactly and is like absolutely if you had extraordinary evidence then I'd sit up and listen but if you're going to show me decoding from

2:32:02 the visual cortex and say that that supports this extraordinary claim then I think we've got real real problems I guess that's the that's the point yeah I mean I missed out on of this because I was writing my book at the time and I didn't even go to this big event that everyone was mad about um I get the feeling that maybe I was supposed to organize it but then I didn't because they were saying everyone was saying oh a qualia Fest would be great would line up with this and I was like I don't have time to organize a music event for the assc or actually what I said was uh that's a great idea thanks for

2:32:34 volunteering um to do it Dave and then I think that that that result was uh was kind of what happened there was Joe in the big event or whatever but um so I wasn't there I didn't attend it but not because I out of principles or because I you know people think I was like involved in someh in this Cur fuckle but I was just like locked up here basically thinking about my book and I only heard about it about it secondhand um but I but they did send me the letter to ask me to sign it and I was like guys dude uh no because and some of this has already come out but it's like so why a

2:33:07 letter you can't do what you need to do the title is inflammatory you've already admitted in retrospect maybe the unpacking of it would be better in a more paper sized and one kind of criticism that you guys have received is that this should have been in a paper it should have been peer- reviewed and it should have went through the normal channels rather than kind of on the psych archive parentheses is there new change and moderation because of this have you seen that there now instituting some new policy I was wondering if this is because of you guys

2:33:40 by the way but uh that's partially a joke but also not um so anyway I I did wonder like why why did you think like a letter as opposed to like something maybe even you know letter to the editor like put this in nature put this in ticks well why on psych archive where it's not reviewed at all it kind of comes across as like punching down in a way um by a bunch of famous people which is not what I'm saying you guys are doing but I'm just saying that I've heard people say that and it does why you chose this method yeah yeah no I

2:34:13 think a lot of the things you're saying that I in retrospect I think a very fair point so the first of all it it was designed as an expression of concern the kind of thing you get as a letter to the editor right um and it was submitted to a journal for review by as a less to the editor and it didn't get um reviewed there and and I think there was basically just the passage of time had elapsed and sign signatures have been

2:34:46 acquired and then there was like are we just going to let that um sit I see where unseen and and so I think there's and S you know there is now a culture of everything that gets submitted to a journal gets pre-printed now I think that here I do worry that pre-prints were not the right Avenue for this and I also worry that perhaps the cogitate paper the preprint wasn't a great hav't you for that either because right the stakes were so high on both of these

2:35:19 sides that it would have been good to have a bit of peer review in the process to essentially encourage private revisions on both sides and so I do worry that preprints have kind of overreached a bit here that they their original purpose of like facilitating communication has backfired somewhat um I also regret there somewhat haphazard and ad hoc way the signatories were collected because I it would have been much better to been more transparent and systematic about this um and that and to your point about open letters yeah I I think that a petition way of doing things is just it's not a very

2:35:53 productive way of doing things and I wouldn't myself have been motivated to get involved with this if I didn't think that something needed to be said like essentially like how do you then get that voice out if it if it had been a um academic paper with a few people saying what's already been said a lot of times I just don't think it would have been heard very much to be quite honest um now there's a question whether was it worth the total uproar that

2:36:26 happened I think ideally it would have been somewhere in the middle where it wouldn't have been total chaos and it would have been okay yeah we hear you now like you've got quite a few people saying a similar thing and now we need to take these issues seriously and I feel like they are getting taken seriously now at least by some people um where they might not have been worried about before in the same way so I I definitely have misgivings about the way it was done and I I think there are definitely lessons to be learned on both sides of this I think hopefully you know things are calming down now this longer

2:36:59 piece is getting hopefully peer reviewed right and you know there's a bit more exchange rather than one side says this one side says that kind of um Dynamic um but yeah I mean like my a few weeks after this whole thing went down there was some um at my daughter's Nursery there was some petition getting organized because they didn't like lock the gate properly or something and my wife was like are we going to sign this open letter to the NTI I was like no I'm not signing any more open letter

2:37:31 yeah like yeah yeah it is funny like you said we the open letter that we originally wrote didn't receive this attention we didn't use that inflammatory term so may I think no one sort of predicted that you guys would have this reaction that it would be this like everybody was talking about overnight sort of thing um so yeah it was a bit surprising I mean I could tell as soon as a letter came to my inbox I was like holy Fu uh but but mostly not because I disagree but because I think the case needs to be made a lot more seriously uh and if you're you know it's like if

2:38:03 you're going to use a word that really seems to me more like a more a character judgment about the person who's advocating the view which is not what you guys intended I think but that's the way I not told hear I think a lot of people heard that um and you know just for the record Julia's nice guy super smart extremely charismatic um no he's not mean or he's not a type of person you want to go after really so I don't think that this was a personal attack on him or his Camp although a lot of people interpreted it as like sour grapes and you guys just

2:38:36 being like you know we're pissed off at you or don't like you because you're getting attention but that's clearly not what's going on I would say at the very least yeah not at all and I also you know I know some people like our field is like a big field and I know some people on involved in I research others who others I don't know I don't know Julio personally I I like after all that upro I felt awful I lost a lot of sleep about it actually thinking like this is being perceived as you said a personal attack when it was actually designed or

2:39:10 at least the way I had when I was participating in the initial edits and so on I felt like the ethos of the design there was to express our concern and there was some talk about how you know this needs to be a we need to try and make our voices heard in a sense like there's um try and see if there's other people who are willing to sign on to this um but yeah I think that that for me is the most regretable aspect of it is that it's kind of perceived as a personal attack and I in

2:39:42 hindsight I can understand why that you know that I I've learned a lot about how people treat this term pseudo science and I think that like it's clearly very divisive some some people read it one way some people read it another way I mean I I you know I had one thing where like you know after um it was kind of pre-social media so or pre the like um big um dominance Twitter and so on but like when

2:40:17 I was involved in some of the work on individual differences in brain structure in the late 2000s um you know I had there were a few blog posts out there calling it phenology right so phenology being the kind of study of bumps on the head right and in highsight they're kind of right actually like it was it was a it was a rhetorical device to say what are you learning from this that isn't similar to what what actual pseudo scientists were doing in the like 1800s and I felt like

2:40:53 you know I was like really annoyed by that and thought they were attacking me but actually in hindsight yeah that was not I don't think that's a great way of going about studying individual differences and how the brain works it was popular for a while and I don't think people aren't really doing it now for good reason because it is kind of like neop phenology because you're not looking for bumps on the school but you're looking for small differences in bumps in the brain essentially great mass of volume and so if if you have someone you know I could easily imagine

2:41:26 now if I publish that paper tomorrow in today's social media world and much more interconnected and so on it could blow up someone could write an open letter calling me a phrenologist or our group of phrenologists exactly shouldn't do it anymore and so on and and I like I could imagine feeling that that's that's a terrible way of going about things um because of the blowup not necessarily the thing someone said but because the blowup if you like I could imagine that um and I think something similar happened here like I I

2:41:58 think that the concern is real but the way it was evolved if you like after that concern was then put out there was had unintended damage yeah and I definitely regret that side of things well I know and I know you as reflective and sincere person um outside of uh academic life altogether just as a regular individual so I I never had any doubt that these were uh methological concerns that you were trying to issue and that there was something that was

2:42:31 concerning that you guys were trying to draw attention to and the only criticism I've ever had was that this was not a great way to do it and I think you've you know the way we've been talking addresses that um but do you could you you I mean you've never met Julio personally but when you talked with Melanie and other IIT people um what has been or since this has come out has there been any responses besides this stuff we've been talking about about about the science that have caused you to to rethink your issue a little bit or has they become more or have you been become

2:43:03 more like entrenched in that science uh that the evidence on the science side yeah um yeah I mean I I guess I mean I haven't seen a lot of these people since New York and I'm looking forward to ASC to actually see it's gonna be a great family reunion yeah yeah it's gonna be kind of like Mean Girls High School vibe to it potentially hopefully hopefully not hopefully not well will have home CT Advantage at least right yeah okay yeah um yeah I mean I have talked to a couple

2:43:39 of people about um this weak versus strong and what's the on What's the um the the fact that P PCI and other metrics do well on diagnosing awake versus asleep and potentially even in helping diagnose vegetative state uh patients and so on like what does that tell us about the ontological status of IIT and I think people who are involved in using those metrics have said to me they don't think it has any bearing on the ontological

2:44:12 status of I so and yeah you know I think that's what um I was very keen to make sure we did in the um in the longer form article that I'm not um atically first in no and also not like a a you know corresponding or or lead author on this but I have um provided input and edits and so on and I was Keen to make sure that we it distinguished in a crystal clear way between

2:44:46 the kind of work on complexity metric and connectivity metric that highlight integrates information as a as a potentially important component of what it means to be a conscious brain and then distinguishing that the kind of neuronal or weak complexity view from the core of the theory and I think that's still a really important distinction that needs to get articulated in that in that hopefully in that longer piece so so yeah so I I

2:45:18 think that I've I've been I think there's two well there's probably like two three debates really within this bigger debate one is just on what if anything can science um do to test this Theory and I found surprising degree a degree of agreement between people I I have to talk to who are working on the kind of weak IIT side um not on the strong I side because and I think it' be surprising if they did

2:45:50 agree because they think you know that that is a testable Theory um and then the the second debate is more narrow on like what should adversarial collaborations be doing in you know how how do you articulate bridging principles that can kind of get you a purchase on different theoretical ideas in conscious science and should like adversarial theories be at the same level because that's the thing that a lot of people point out way before this happened was that like

2:46:24 Global workspace theory is obviously a functional Theory IIT is a structural Theory it's not like that's where the unfolding argument all that stuff comes from that people are interested in so people already said it looks weird that these two theories are the ones that are trying to compete or not compete but make competing predictions CU they're just like coming from different viewpoints about about how the mechanisms or even if there are any they're going to work so it seemed weird all from the very GGO one couldn't help but figure think that maybe it's just like these are some

2:46:55 famous people and these guys want them to be the face of the coll you know that they that they these are well-known persons like you know because anyway so I do think that maybe on the part of Templeton there was some reason for having this one be the first and I know they they tried to have high theories in there and hakan you know left for whatever reason not to participate um and that story maybe I think we've talked about it before I mean not you and I here but uh on this podcast I think it's come up um but anyway so it is curious that those two

2:47:29 were the ones that were picked because they come from such different ways maybe that was what they were thinking is like here they're so different like they've got to collaborate but uh good collaboration requires just enough commonalities that the differences can be looked at from the same point of view so to speak yeah yeah I mean I I wasn't involved in the Genesis of of either the arc's or or the cognate projects there are other ones in the in the Templeton stable that are also testing

2:48:02 IIT um there are two just keep talking I gotta this blind over here yeah yeah sure I mean there are two projects um that are testing higher order theories um two two testing I think two testing Global work space so so I think that there it's probably true that broadly speaking IIT is over represented in the arc's I don't know what the reason for that is I I it could just well be organic I mean like you know these

2:48:36 things are not necessarily planned top down in terms of what competitions are going to emerge in in Grant making um but yeah I think yeah well I'm I'm suspicious anyway so however you view about I mean so the kind of more background picture a question that I had in the background while I was thinking about this was what you think can come out of an adversarial collaboration because like for example the one on high Ro of theories roughly they're all at the same level so to speak um maybe

2:49:09 differences psychological versus computational level we can we can quibble about that and stuff like that but they're all sort of within the same ballpark one might think um so your criticisms weren't of adversarial collaboration no not at all right not at all yeah okay yeah yeah and I think I think um well so you know I think that the the as I said before like the achievement of the cogitate project has been to get us comparing ideas I

2:49:45 mean this is also this is all driven the energy behind this is Leah mudri and lucher Maloney who also convene the debate in Amsterdam so the General Drive between come from Le and lucher to stop just existing in our own little silos um and thinking of experiments that confirm our original ideas and instead trying to compare and contrast ideas across theories I think is a great one um and really important but I think that there's

2:50:20 also the and this this intersects in subtle ways I think with the universalism issue we were talking about earlier in Consciousness science that I would feel more comfortable with an arc process that is the an adversar collaboration process that is essentially trying to compare between different variants of models and that then when you find which one best fits your data you revise your theory or your model based on that

2:50:52 outcome right you don't like strike it off the table maybe eventually you do strike things off the table but my worry about applying adversarial collaborations to Consciousness science is essentially you're dealing with a field that already has this um drive towards universalism that you have to have like either it's my theory or your theory and there's no kind of Middle Ground here there's like we've got to kick one off the table so I feel like our higher order project is actually and that's partly why I was happy to get involved and Lead it is actually more in

2:51:25 the former Camp because we're all under this one broader umbrella Theory and we're now looking at different instantiations if you like of that theory now yes there are I think important implications of um the fact that you might find that like um you have evidence for certain form of higher order reputation versus another yeah but at the end of the day you're dealing with comparisons between models um which all have a common flavor to them yeah um so I I

2:52:02 think that's yeah so I think that's um my worry in in conscious science is that like this loadable goal is running up against this um incommensurability between different theoretical approaches interesting yeah and I think we I think we saw that after after the cognit experiment with these two different discussion sections that were written by steren and and Julia ton right yeah um interesting so I agree with you we don't but that is the way it's pitched it's pitched as we got to narrow or wiow down

2:52:36 the field um by pitting these these ideas against each other and what you're saying is maybe what we got to do is not winow but modify enhance incorporate yeah um and see how these fit together in some broader way uh so yeah and and like I think that you know I've always felt as though aside from the um you know I think when you when you try and create Universalist versions of these component

2:53:09 mechanisms then you end up having massive fights um on a you know on what's the ontology of Consciousness in each of these theories but I've always felt as though you know for instance recurrent processing is clearly part of the picture here um and it goes hand inand with some aspects of predictive coding and I've always you know my the work we're doing in my lab on on higher order contributions to awareness is goes hand in hand with PR

2:53:44 of coding even though it's framed as a higher order model so I feel like there's way more commonality yeah between Theory so-called Theory classes than than we've we've um kind of recognized and I know there's people have written about this that you know one there's an opportunity for more integration rather than Division and so I feel like when you set up a competition between two classes then that's worrying because I guess I guess I would have been more

2:54:16 comfortable with different AR different avar collaborations that are all doing like you'd have your Global workspace AV collaboration where you'd have like different variants of global workspace that all come together and do their experiments and then You' you'd make progress you defin you definitely make progress right and then you have all different variants of recurrency local recurrency that like say no it goes as far as this brain region or no it only goes as far as this one or you and I think you'd make make progress that and then eventually you'd have to come back to the table and say okay we now got a

2:54:48 different component um Parts again and and we've made progress on these subcomponents and I think that's often the way cognitive science Works in other fields yeah I mean you're not going to get $3 million for that though it's harder because it's not as flashy and you know yeah no and I I think I think that's right I think I think the part of the issue here is that c science the modern version is a young field Templeton wanted to kind of be

2:55:21 this have this like Flagship initiative which was great but it had to be as you say I think it had to be impactful and and try and contrast these big classes of of theory to and and I in hindsight I worry that that has been the source of a lot of the issues here yeah I mean and but I always felt like behind the scenes everyone kind of knew that that's what you say to get the money but like we were you know it's like we that's not what we really think it's happening here yeah well I mean but these these are not like normal grants right so I

2:55:56 mean I know also that is it's useful to bring up funding actually because you know a lot of the other things that have been said about the letters that is kind of like sour grapes about how much funding is going in and so on and I I always found that somewhat puzzling because like it's not as if you know getting funding to these projects is actually there's a lot of admin and and it's actually like a lot of um effort work to exactly it's a full-time job let's just say it yes it's a and and the funding obviously you know you don't you

2:56:29 don't it doesn't go into your bank account put it that way it definitely goes on less exciting things and um yeah obviously it's important for careers and everything to have these kind of projects but I would actually say that like these projects are unique in that you a lot of scientific Grant making is you come up with an idea and then you refine it as you go along and you probably do something a little different to what you wrote in the grant because your field has moved on and so on now you can do to that to some extent within the templ and framework because you can

2:57:02 refine and update the pre-registration documents but once you've committed to getting a grant between testing for instance these two Theory classes you've got you've got to go ahead and do that so I right I don't think that maybe people did think there was more flexibility I I have no idea I I don't know but it's become clear that you know if you're GNA get a grant to do that contrast between the theories you just have to go and do that contrast between the theories and um yeah interesting okay well Steve I I was going to ask you 400 million more things

2:57:35 but uh in the interest of your health I I'll let you off the hook I do like to ask at the end if uh there's anything that hasn't come up that you wanted to talk about um the answer can be no we've talked about tons of stuff but just in case let me check is there anything that you uh that hasn't come up yet no I mean we've covered a lot of grounds so thank you for um you know thank you for hosting this I think it's been a lot of fun um I did I was glad we got the chance at the end to talk about the letter because it has been something

2:58:08 that's you know weighed on my mind about like just generally what is the Right Way Forward our field um so I think if we can you know I think one of the issues maybe the last thing to end on is just that like any area of Life having Face tace conversations and actually hashing out these issues is the way to do things um and I think that the problem with social media is that things get very incend very quickly and people impute intention when there is none I

2:58:42 think this can go both you know both ways um so I yeah I hope for going forward that there'll be just more of this kind of conversation the kind of conversation we've had today where we properly just unpack everything um exactly these I thank you for saying that I think these kind of conversations are important you know I've invited some IIT supporters that come on so far no takers but uh I I want to get both sides of the story because I think the thing is interesting you I'm not trying to provide a one-sided um no no take on this because I think it's an important

2:59:16 and interesting topic just you know and not that you care about this I'm just saying for anyone that's paying attention I have tried to get other uh more friendly IIT people just you know not everyone is interested in talking about this yeah y fair enough but but I think these conversations are important so I agree with you and um you know I I just want to say that I you bring this back to your to your book I know we've ranged far a field um but just to close on that note to remind people the book The Science of self-awareness uh know thyself it's an excellent book it's it got a lot of information in there but

2:59:50 also a lot of really interesting applications of that information so um if you haven't had a chance to check it out I recommend you check it you know Steve one thing I've always loved about your work is that you're serious about the science and you're serious about the philosophy too and it shows in the book you address philosophers you you interor the empirical work with philosophy it's really um how I think I hope it'll become more common in the future it is becoming you know slightly more common now but it's an example of how that should be done so um I I applaud you for that leading the way there um I was going to say something else but I forget

3:00:25 so on that note stay on the air for one second see I'll stay by but other