Support the show to get full episodes and join the Discord community.
Hakwan and I discuss many of the topics in his new book, In Consciousness we Trust: The Cognitive Neuroscience of Subjective Experience. Hakwan describes his perceptual reality monitoring theory of consciousness, which suggests consciousness may act as a systems check between our sensory perceptions and higher cognitive functions. We also discuss his latest thoughts on mental quality space and how it relates to perceptual reality monitoring. Among many other topics, we chat about the many confounds and challenges to empirically studying consciousness, a topic featured heavily in the first half of his book. Hakwan was on a previous episode with Steve Fleming, BI 099 Hakwan Lau and Steve Fleming: Neuro-AI Consciousness.
- Hakwan’s lab: Consciousness and Metacognition Lab.
- Twitter: @hakwanlau.
Hakwan 00:00:03 My graduate advisor, Dick passing him was a great teacher and really trained, taught me everything I needed to know. And in neuroscience, or maybe even in being a good person, he’s really a very influential figure and mentor to me. But one of the constant friction was he didn’t really want me to go into this field and I kind of reluctantly get into theory. And I, and I just say, I have a inclination I hang out with philosophers would be too much. And then I, and I liked thinking about this, but every time I do it, I, I do try to remind myself this is an indulgence. And so I constantly try to force myself to do less theory and more careful experiments.
Paul 00:00:44 What percentage of empirical studies, uh, about consciousness? Do we need to just throw away?
Hakwan 00:00:51 I would think the majority of it and unfortunately, yeah,
Paul 00:00:55 It hurts.
Hakwan 00:00:56 Yeah, it does. Uh,
Speaker 0 00:01:03 This is brain inspired.
Paul 00:01:17 The science of consciousness is a notoriously tricky, slippery, hard endeavor to pursue as is the philosophy of consciousness. Many of us, including myself, I’m Paul. When we enter neuroscience, we’re quickly cured of our desire to pursue it scientifically, but some folks boldly continue. The empirical study of consciousness, Hogwan Lau is one of those folks and he’s just released his new book, all about it. In consciousness. We trust the book, uh, documents, many of the challenges, people like Huck wan face in making progress. It also explains his own recent thinking and work to make progress. Uh, and it’s also part memoir documenting his own path thus far when was on the podcast on episode 99 with Steve Fleming, we talked a little bit about this, but in this episode, we dive deeper and discuss some of what’s covered in his book. So how Kwan is starting his new consciousness lab at the Rican center for brain science in Japan, where he recently took a new position.
Paul 00:02:24 And in the show notes, I link to his book and how to find out more about him. Those are at brain inspired.co/podcast/ 128. Also a quick note, I will very soon, uh, launch the next round of my neuro AI course, which is a series of pretty condensed video lectures on many of the topics related to what we discuss on this podcast and the intersection of neuroscience and AI, plus some other bells and whistles. So if you missed out on the last round or have interest in joining this new round, uh, go to the website to check that out as well, brain inspired.co and in the menu, you can click on the neuro AI course tab. Thanks for listening everyone special, thanks as always to my wonderful Patrion supporters, happy sciencing out there. Hope you enjoy. Welcome back. I, I know that you are a musician and I’m curious how your, uh, how your musicianship has, uh, has been going. Have you been playing lately? What are you and how have you been playing lately?
Hakwan 00:03:27 I played, uh, all sorts of instruments, very poorly. Uh, since the pandemic I have been mostly playing drums, actually I’ve been practicing the rudiments on the pack because it’s somewhat meditative and, and it’s, it’s not really music. It’s more like exercise to me. Oh, nice. And I tried to play piano now and then, but also very poorly. So it doesn’t really,
Paul 00:03:50 So I released a, uh, this neuro AI course that I’d been working on for a long time. And I had promised myself that when I did, I would, uh, buy this electric guitar and a small amp. I haven’t had an electric guitar in a long time because we sold all of our stuff, but that’s for a different story. But, um, I still haven’t pulled it, you know, I, and I released the course and, uh, I still haven’t pulled the trigger because I’m afraid I won’t play with my toys, you know, and that it won’t be justified essentially. So, so I’m glad to hear that. You’re, you’re playing,
Hakwan 00:04:20 Sorry. Wise, I think over the years, I think all guitars in particular, they are very bad at buying new toys. I was guilty of that. Buy a lot of new toys. They’re never using them for real.
Paul 00:04:32 Yeah. Yeah. Just, you know, I need the discipline anyway. Well, I’m glad to hear that you’re playing. So, uh, in consciousness we trust, um, this book. Well, first of all, let me, can I ask you just to explain the title? I was curious where the title came from
Hakwan 00:04:47 Actually public came from music. Uh, I think a while ago when I was in the, um, the house of blues I saw in, in blues, we trust and that that slogan left the first drone pressure. And I think, I think sight is just a riff on that. And it really has kind of two indirect meaning one is that, that one is the scientific content part. The, the book, ultimately one of the theoretical idea is that consciousness is the, is the content from perception to cognition. So it’s basically that consciousness is the interface for which your, your perceptual content directly impinge on your cognition. So whatever you’re consciously experiencing you, you tend to trust it. It’s very hard to ignore your conscious content. So in that sense, it’s the kind of the foundation of everything we believe in trust.
Paul 00:05:42 Oh, I see. So it’s like wrapped up in the whole perceptual reality monitoring theory. I hadn’t put that together. That’s interesting.
Hakwan 00:05:48 Sort of, yeah. I didn’t really explain it explicitly either in the book. Um, I just thought it’s okay if people can configure it out. And of course there’s a second kind of sense that it is a bit like a manifesto, uh, like a political agenda almost. And I also like the ring of it. And again, I never explained that directly, how it’s connected, but I just left it there.
Paul 00:06:13 Well, uh, so re reading the book, especially the first half, which goes through all of the, uh, thorny issues of empirically studying consciousness, it gave me nightmarish flashbacks to when I was starting in graduate school, thinking about how, how to apply, you know, scientific practice to study something like consciousness. Um, but you really, uh, go through so well here, forgive the pun, but consciousness is a confounding thing to study, right? Because there are so many confounds, uh, to deal with and you go through them in tremendous detail, um, and, and sort of sets the stage for, you know, all of the different, um, empirical, uh, issues related to it. So thanks for sending me back into cold sweats there, first of all.
Hakwan 00:07:03 Well, thanks for even reading it. I thought that it would already scare a lot of people away just opening up the cans of worms.
Paul 00:07:10 Um, I’m curious, um, if you had an intended audience in mind, because, um, more so than many of the kind of popular science books that are written about consciousness, uh, this really does dive into a lot of the research and, uh, explains a lot of it. So who did you have in mind when writing it?
Hakwan 00:07:28 Yeah, I certainly, um, it, it, it was a very conscious decisions part of the fun, but the, it was definitely written as an, as a monograph, as a book that is really written for people who already have some interests or at least in academic research. So we’re thinking like graduate students to actual full-time graduate students are full-time too, but other, uh, better paid, uh, more, more, more, better paid academics, more job security. And so it’s really written for researchers and I’ve been obviously warned against it and people said, you know, why would you write such a potent? Nobody would read it. You will not make any money out of it. And, and that might have even made me realize that, yeah, that then I definitely need to write this because there are so many good popular books around and they, they do such a great job already. I’m not the kind of person who would write a better popular book than the existing ones, but a few, there, there is maybe, or a room for a book that is aimed for, um, maybe a more than like a niche audience, uh, where people really want to get into the research. But I find that the, the literature is kind of messy. So this is a book trying to hopefully clean things up for some,
Paul 00:08:44 Since it reads, like you are thinking through the ideas yourself in real time. And I don’t know if, if, um, it’s written while you were, you were writing it, if you were conscious of your own trajectory, your own journey in, in studying these things, but that’s the way it reads. Is that accurate?
Hakwan 00:09:03 Yeah. I think there’s a small element of, of it being, um, somewhat like a memoir. Um, it is something that I’ve done, as I said in the book to basically, maybe not so explicitly, but I basically, it is something I’ve done my whole adult life. So ever since I, I achieved not, not the consciousness in the sense of, um, having subjective experience, but basically having an adult level, rationality, that that’s something that I’ve always been doing. And this book kind of goes back to the very beginning and it’s not, it doesn’t reflect thinking on the fly because all the things I’ve thought through, I think the past two decades, and that’s why I wrote it, but I try to basically recall the whole journey and, and especially something in some places went into mistakes I’ve clearly made and think about the logical errors and struggles I’ve had.
Paul 00:09:56 Yeah. Well, you’re very honest throughout the book in documenting your own rethinking of your positions, uh, as you’ve gone along. So I enjoyed that as well, because, uh, I think often when you read, you know, when you read an account, someone doesn’t take the, uh, self-reflective time and energy to, um, Hmm. I don’t know if confess is the right word, because that’s the way science is supposed to work, but, uh, but I enjoyed that aspect of the, of the book.
Hakwan 00:10:24 Thank you. Yeah. I think, I think that’s, uh, in part, has to do with the, the, the, um, the memoir, like aspect of it. I, I, I wanted it to be on record to remind myself how, how easy to make those mistakes, um, how tempting those mistakes were. But, but also maybe that’s part of the, the rhetorical, um, intention too, is to really try to say, like, these are not easy problems. And, and I think we in particular is about to put the positioning of the book is related, right? So there are a lot of popular books out there. They, they gave very slick and exciting stories and I think they are great. I mean, they’re important for bringing in new audience, but I think for researchers within the field, we have to be honest that we have not thought through these issues, some of the very basic logical issues, uh, we have not thought of through
Paul 00:11:16 The other thing that I, um, had, you know, realized through reading the book is it may have been like two months ago, but this has been on my mind how valuable it would be to have a sort of, um, well, it was a project that I propose to someone and we haven’t moved forward on it. Um, but maybe now we don’t need to, to have like a repository of all of the research studies about consciousness, like with the paradigms, the tasks, the confound, and your book kind of encapsulates, uh, all of that. So, uh, maybe I don’t need to make that project, that project anymore.
Hakwan 00:11:48 Maybe you should, because that’s just my, you know, my version of it, but that’s the at least that’s they attempt. And I, in a way I hope more people are doing this instead of going for the more exciting, cute ideas, or maybe not cute, even profound ideas. I think, I think we have too many ideas and not enough people doing this organization, the work and try to say like, okay, how do we, how do we even evaluate these ideas beyond saying that this idea is sounds cooler than the other? Um, so I think that that’s part of the
Paul 00:12:20 Well that hearkens to, you know, a lot of people in neuroscience, not necessarily the study of consciousness, but, um, talk about how we are bereft of theory, right? We have all this data, uh, and we, but we’re what we’re lacking is theory, but are you saying that in consciousness, there are too many theories or, or just ideas?
Hakwan 00:12:39 Definitely. I think, um, both theories and ideas and we, we, we have too many of them, I think to the extent that you think neuroscience has too many of them, we have a problem that is much bigger in magnitude. And the idea is, tend to be maybe, maybe also the nature of the ideas, a bit worse. So ideas, often people put a lot of personal prescription and stick into the ideas like, like this is my greatest idea. Before I, before, before I win my second Nobel prize, I should really, you know, air out these ideas on something like that. There’s a kind of set sense of personal indulgence involved. And then the people who test those ideas also tend to have a kind of sense of almost like idolizing the people, giving those ideas. And I think that that’s just not very healthy.
Paul 00:13:31 Yeah. You’ve written about that as well, uh, elsewhere in other manuscripts as well. Let me, let me read, um, I’m gonna read a couple of quotes from the book here, here. Here’s one toward the beginning of the book, which, uh, speaks to the complications. So all current theories seem to make some rather improbable predictions. Of course, if we know for sure that a theory is correct, we should accept whatever improbable consequences it entails, but I hope that the attentive reader should have been convinced by now that the science of consciousness is just no, such simple matter. It’s fairly depressing. You know, um, we’re going to talk about the contents of the book and, and your ideas and your own research more, but I kind of want to linger on this. Um, the difficulty in studying something like consciousness, have you been like in your own little journey, uh, like you said, your, your entire adult life has been working on this, on these things empirically and thinking about them even, you know, and philosophically as well. I know you interact with a lot of, uh, philosophers, um, given the complexities and the massive amounts of compounds that you have to think about. It seems well, well, for someone like me, uh, it seems likely that I would grind to a halt and just wouldn’t find space to continue. Have you ever felt yourself grinding toward a halt and, and, or been, uh, so put off by all the complexities or the complications that, uh, you’ve thought of moving elsewhere?
Hakwan 00:15:02 Definitely a very many times. And, um, um, or at least I thought, especially, especially about quitting science. Um, so I might still want to think about consciousness and then I might want to write popular books or, or, or just indulge in a bit of unprofessional philosophizing about it, or maybe even being, becoming a professional. At some point, I thought, you know, I have a PhD is not in the discipline, but I publish philosophy papers that are cited in philosophy. I constantly think if I just leave science, pack it in, how, how, how much would it take for me to retrain myself as a philosopher? And I, especially earlier on in my career, I constantly thought about that because doing science in this just seems to be not so realistic at some point, it’s just, it’s not, it’s not that I don’t enjoy the science. Um, at some point the projection became like, I, I thought I hated doing science and every time I was about to quit, I realized, no, actually I really, I would really miss doing it. Um, so it’s not that I don’t like doing science, but doing science in this field is, is really endlessly frustrating. You really few, you’re not making much progress and you’re just trying to put out fires sometimes. And, um, yeah, that, that is very, yeah, that, that actually is still on my mind. Even my reason moved from, from, from one job to another, I thought about this again, all over again,
Paul 00:16:29 But you have interacted with a lot of philosophers and you continue to, obviously I know you’re friends with Richard Brown and, you know, and you have open dialogues with net block and, and on and on. Um, well, this is a terrible question, but I, I don’t know. Can you describe the difference in interacting with other empirical researchers versus interacting with the philosophy side of it? Because I imagine that the philosophy side is also, I don’t know, is it, I don’t know if it’s less frustrating, but maybe you can just answer the question, I suppose.
Hakwan 00:17:03 Yeah. I think they there’s some frustration too, but they’re different. Um, but, but there’s one thing I really like, maybe I say what I not like about interacting with floss. So for first, because it’s less important. So I leave the punchline heavier later. So the, the part is that they, with, with a few exceptions, like people like net block or my co-author Mathias, Michelle, many of them are not really that deep into the empirical literature. So these are people, net and matures are people who sometimes would tell me about papers died. Don’t know. So in the middle of a debate, they would, they, they would, they would pull out empirical evidence that I, I didn’t know that existed. And so that’s really fun, but most of the time is not quite like that. Right? So they, the philosophers who have any time for me quite often do so to talk to me because, because they think that I can, I can share my knowledge in science and they, so their, their level of knowledge is not as good.
Hakwan 00:18:02 And it’s quite often the debates then becomes very purely conceptual, which is interesting, but it’s not, not always so effective in resolving differences. So that’s some mildly frustrating, but the, but the really good part I really like is in some way, philosophy is, is more honest in the sense that you can, at least it used to be more like that. I think it’s becoming a bit different. You can, you really go and attack your opponents positions quite mercilessly, and it would not be considered to be unfriendly or bad form at least most of the time, unless you’re really not polite about it. But if you keep some minimal degree of civility, I think they are, they genuinely appreciate, uh, back and forth arguments because that, that’s how they do things. And they realized the utility of it, which I also really, I also appreciate, but in science, of course, you can’t really do that so freely, if you take the same kind of combative attitude in science, you, well, I tried, you were suffering in your career because there will be retaliation. Then people do re remember things and they, they, they get upset and maybe understandably I’m, you know, you, when you really attack their work. And in my effect, their funding effects how they are the consequences. They, they don’t, they don’t accept the criticisms as, um, as openly as philosophers do.
Paul 00:19:27 It seems like more than other fields in, in neuroscience, uh, the science of consciousness and well beyond science, what I was going to say is that, um, consciousness, another difficult part of it is that you actually do need to know if you want to study it at all. Uh, you do kind of need to know consciousness, uh, admits much more philosophy than other, uh, empirical neuroscience fields, I suppose. Um, so are you, do you feel like you’re constantly kind of going back and forth and having to think about the philosophical implications and, and positions and where you stand on that? Uh, and also the empirical side of things?
Hakwan 00:20:07 Yeah, definitely. Um, I think there are two sides of it. I think I, because of my background, I, uh, my undergraduate training has a bit of philosophy. And then I lived in New York city for, for a good few years where I interacted with a lot of the, the major philosophers, um, in active in the field. So I, I tend to actually sometimes just go in and try to steal ideas and borrow ideas and then try to translate them into those language of science and even sometimes go about testing them. So I, I benefit from that, but I think even at a more basic level, just thinking about these very, very basic issues about necessity and sufficiency, and how do you define what is a theory and how, how to avoid conference? These are things that we scientists supposedly know. These are terms that, uh, we don’t, we don’t think of them as philosophical terms, but often we don’t really think deep enough about them.
Hakwan 00:21:03 Uh, in fact, I would say maybe Mo maybe he’s my, maybe is a problem in my view, but I think in neuroscience, in general, I think most people make mistakes about these, these terms quite regularly. And, and because most of the time, it doesn’t matter if you, if you, if you do, if you, if you’re in a more mature feel, you know what experiments are good and what experiments are bad, even sometimes you can’t articulate it. So, so clearly, you know, that certain things are bad. You know, that this won’t get you published, and then you stop. Whereas in consciousness as a more of a kind of wild west situation, and there’s so many things you can try and people are trying new things and, and some of the existing ideas are turn out to be completely flawed. But then the field is so small that, that the recognition of the problem hasn’t happened, hasn’t been, become so widespread yet. So, so you constantly have to apply these what I call almost like basic critical thinking skills. And once you put it this way, people would say like, no, come on. We have basic critical thinking. Uh, we don’t need this. And I think part of the book, at least early chapters to try to convince people, no, we really do. And we in print make a lot of really embarrassing mistakes. And if we don’t go clean them up, we will be in quite a lot of trouble.
Paul 00:22:21 It turns out this stuff is hard.
Hakwan 00:22:23 It’s very hard and, and, and convincing people that they are in trouble. It’s a thankless job. And, and that’s, maybe that’s why I want to write it as an academic monograph, because I don’t think that would make any money. Anyway, if I tried to make it popular and more, more aim for general audience, basically I’m a downer is, is not, is not going to sell very well.
Paul 00:22:47 Yeah. Well, that’s what, that’s what I alluded to my cold sweats that you induced in me. All right. So I’m going to, uh, here, I’m going to read another little passage, uh, and this is from the last chapter chapter nine. Uh, and so you write when I was young, I thought I would one day write a book to solve the hard problem. I still have not given up on the problem itself, but by now I recognize that it was a diluted idea at the moment. The best I can offer is a summarized in this final chapter and extended version of perceptual reality monitoring, which includes some elements of quality space theory. So we’ll talk about two of, uh, both of those things, um, during the course of the conversation. Uh, but, uh, you know, I’m keying on the, when I was young, uh, I thought I would one day write a book to solve the hard problem. Really. Do you have thinking about your own mistakes and your own trajectory and thinking about consciousness and working on it, all the, all the empirical work that you’ve done that you’ve done, do you have, um, advice to someone who is thinking that they will one day write a book that solves the hard problem, et cetera, someone’s, who’s like starting off in their career thinking of, of your own.
Hakwan 00:24:01 Yeah, I think, I think at least I can, maybe not sure that advice will be much useful, but at least I can share one element that I totally misjudged and I didn’t think it would be so hard. Um, so when I started out my graduate advisor, Dick passing him was a great teacher and really trained, taught me everything I needed to know. And in, in neuroscience, or maybe even in being a good person, he’s really a very, very influential, um, figure a mentor to me. And, but one of the constant friction was he didn’t really want me to go into this field. He thought that basically, I think most of the first couple of years of working with him, a lot of our conversation goes back to this issue. He would basically try to convince me why don’t you work on something that is more promising because this really feels like it will be wasting your time.
Hakwan 00:24:53 And yeah, he will be trying to talk me into a number of different arguments, like sometime about how tractable something would be. And in the end, I sort of won that debate in terms of convincing myself. I never convinced him that he should be doing that, but I think I, I, I soldiered on still looking back, I’m still in it. Yeah. So I, maybe in a sense I want, but, but I think one thing I wish I, I wish I listened more carefully or I, I wish I realized was that the hardest part is not just a science. So science is hard enough certainly. And the philosophy is hard enough, but another part is just social. Um, you’re in science, you, you have to, you’re basically constrained by, by the peer culture, by their peer group culture. I mean, ultimately getting funding and getting published, or even just like getting feedback from your peer group depends on what the peer group thinks. And I think that, yeah, basically it’s a bit of a, um, uh, complaint or, or, or, or one of my down a moments. I think that the culture of the field needs to change a little bit if, if, if we need to really move towards a solution to the problem. So
Paul 00:26:07 Do you think that this is, um, particular to the study of consciousness? Or do you, because everyone faces constant a constant barrage, constant friction, right. Do you think it’s just accentuated?
Hakwan 00:26:21 Yeah, I think, I think it’s a kind of, it’s not a categorically new problem. I think the problem is access in some other fields too, but I think having, having moved around different views a little bit, even though I never left consciousness, but in order to attack a problem, I would go to different disciplines now. And then I, at some point I would, I would just doing vision psychophysics for a few years. And then I, my wife is a sociologist. So I also interact with her colleagues and, and look into how they think about consciousness. And I, at some point I do human neuroscience. I sometime collaborate with people where, of animal models. And I tried to dabble a bit in computation and modeling, having looked into how different fields, where I think our future has some really uniquely problematic aspects and, and, and, and explanation, or those stories clear about how historically we got to this place.
Hakwan 00:27:12 Um, so basically there’s a, there’s a historical story of how we got to this place where the emphasis of the few often plays so much emphasis on stardom and the end of popularizing work, which otherwise is really important. I think that the, the, the power and influence we, we allow those, um, um, populous ideas to together is really out of proportion. And, and, and the other problem is because of the problems of the few being so bad that we have basically gathered together a group of people with a culture that we don’t want to talk about those problems. We have a bunch of people who are so good at soldiering on when other people told us that what we do is hopeless. And so we have a culture of people who are kind of resistant to criticisms, and some of them basically just ignore criticism and let it roll off their shoulders and say, okay, you guys been saying this for a long time. I I’ll do whatever I do anyway. And I think cultures like that are not really helpful to us
Paul 00:28:15 When, when Dick, uh, passing him was a pro I guess, trying to help you out by a warning you against it was, was that sociological, or was it the science part that he was warning against?
Hakwan 00:28:28 I think mostly the science part. I wish he wanted me more about the sociological probe, but, but at that age I wouldn’t have understood anyway, I cared. Yeah. Yeah. I thought like, why do I care about my peers? I would just do my signs if they are wrong, I would just be right. But no, you, I think I come to learn. You’re, you’re really ever only as good as your peers and because he’s a, is a group activity. So, so the last few years I’ve been trying to maybe try to try to change that culture a bit without, without much success, frankly, but I think it needs to be done.
Paul 00:28:59 Yeah. Well, that’s how you actually finish the book, talking about the sociology of all this as well. And, uh, yeah. All right. Well, uh, enough of the culture here, and maybe we’ll get back to it, uh, in a little bit. Um, so, uh, like I said, the first half of the book you, you devote to, um, explaining all of the intricacies and difficulties in, um, empirically studying consciousness, one of the things that you do is you, you kind of lump, um, a lot of, well, in some sense, you lump everything into local versus global views of consciousness. Um, and I know that’s, you know, not necessarily a fair distinction in your own perceptual reality monitoring, you’re, you’re trying to kind of go between the two as well, but, uh, can you explain the difference between local and global views and then we’ll get into a perceptual reality monitoring?
Hakwan 00:29:51 Yeah. So these are somewhat characatures or, you know, goalposts, uh, there are, there are not a fair, detailed description of the entire lay of the land, but I made an analogy that sometimes you to get the lay of the land, you, what you just want to learn a few landmarks and some, some, you know, very core screen orientation. So global theories, um, as the name suggests was suggested, uh, consciousness informs involve some kind of informational broadcast for out the many areas of the brain. So if you’re conscious in the sense of having a very simple, subjective experience of just seeing a color patch that is red to seeing that redness involves way more than your, your visual neurons that are selectively tuned to, to, to redness in, in the, in the early visual areas, it involves pretty much the whole brain, many areas, uh, frontal parietal, those higher, uh, association cortex areas. And then local theories basically says the opposite. They said, if you see, right, those neurons and to read is, is why you see red and those higher cortical areas in prefrontal brighter areas. Those are just for the further thinking about that experience or the reporting of that experience, or maybe even attention to those, to that experience, but it’s not the experience itself. Um, so that’s the major distinction I think that has been very useful to, to focus on because people do fall basically at least along the spectrum somewhere.
Paul 00:31:21 And so we talked about perceptual reality monitoring last time that you were on when you were on with, uh, Steve Fleming, but, um, to, to spare people, to have to go, um, you know, search for that and listen to it. Could you just kind of recap what perceptual reality monitoring is and how it fits into the local and global views?
Hakwan 00:31:37 Yeah, so basically, um, perceptual relative monitoring theory is a variant of another class of theory sometimes go higher order theories. And the idea basically is that the, the local guys are not wrong. Um, there is a sense that the, when you see red to early visual activity that directly selectively tuned to, to these content, they, they represent the, what is called the first audit content. So you need readiness, uh, to see red, but this is not sufficient. And on top of that, presumably you need some higher order activity and PRM perceptual reality monitoring. It’s just kind of like a, an empirical story or a neuroscience story, trying to explain why that philosophical notion that you need something higher order. Um, why, why is it plausible and the idea as well when you’re early visual areas, uh, when the neurons coding, uh, tuned to rad when they fire, there are a number of possibilities that, that they fire, right?
Hakwan 00:32:35 It could be because you’re actually seeing red. It could be because just imagining, right? So we know the top top-down imagery also activate some very similar activity and it could be that those neurons just fire spontaneously because noise is everywhere in the brain. And so because of these differences, uh, then it would, it is very plausible that there’s some downstream areas that try to read out or monitor to early sensory activity, representing rat, and try to make the inference and say what it really reflects. Right? So if you say like these neurons are, these fret tuned neurons are firing at 50 Hertz to higher order monitoring mechanism might decide, well, this is just baseline noise because baseline noise is about this level. So let’s just ignore it. Then you would have no conscious experience. You shouldn’t have any conscious experience anyway, and when it makes a mistake.
Hakwan 00:33:30 So if it actually knows a bit of a thing that, oh, this is actually higher than a baseline noise level, maybe it did the statistics wrong, then, then you will be hallucinating. So that basically is a way to say, well, you, you, you need this kind of a late stage self monitoring engine. And when that engine makes a mistake, it accounts for why you would have hallucination, why you would dream, uh, why you sometimes see things normally. And Y Y you sometimes don’t see things when you shoot and et cetera. So, so that helps to argue why, why D to pre-fund to, and parietal mechanism are important. Um, but the first older is also important.
Paul 00:34:08 And you’ve, um, in your earlier manuscript, you have tied this together with the notion of generative adversarial networks and how a similar mechanism, uh, works with, with Gans as well.
Hakwan 00:34:21 Yeah. So, so part of the, um, at least one, one criticism, the, the philosophical theory of high-order theory has been, uh, um, have to phase and, and some, actually some authors because of some philosophers, because of the argument actually left a high or the cam was that they think that it’s not plausible, that the brain would grow this higher order organ. Right? So you have a first older signal that, that sets red. What are you just tack it there, or use some dynamics, which is distinguished the cases. Why would you grow another thing on top to try to monitor it? It sounds so clumsy. And so, so much like a hormone colors. And so some people think that it’s just empirically not plausible. And so using the Gans argument is to say, well, actually, if you look at just an artificial neural network where people just try to solve engineering problems, they end up doing something very similar.
Hakwan 00:35:12 So, so it may not be so implausible that you’ve illusion equipped us with this kind of higher order monitor. So in the context of, um, of Gans, uh, the idea is that where you to, to train a top-down generative model, going from the concept of a cat to the actual, uh, first sort of representation of a cat, an image like reputation of a cat to do this kind of top-down self gent self generated imagery would be very useful. We know that you can allow you to do predictive coding and all sorts of nice stuff. And the problem about this is training. Those networks takes a lot of time and turns out there’s a trick called Gans generative adversarial network, I think, has been called, um, one of the coolest trick and machine learning in, uh, in the past decade, uh, by, by ear, like who, and I think, and, and, and, and the trick is basically to set up one network that would do this.
Hakwan 00:36:05 They would try to do this generative, predictive coding and another network that was basically to try to compete with the generator and they call it a discriminator. It basically look at a generated output and try to tell whether it actually is just generated or whether there’s actually a real image of the world. So you basically trying to do forgery detection, they call it. So you take, you take an image and then say, well, this is generated by the generator, or is it actually a sample from the world? And then the two content can then compete, uh, because if the discriminator catches the generator, uh, making a forgery, uh, if it doesn’t look enough, like a look enough, like a, like a real world sample, then they will try to punish the, uh, the generator. If the generator managed to slip into forgery, uh, passing the detection of the discriminator that is generated something that look enough, like, uh, look, look like a, an external real stimulate enough. Then, then the discriminator will lose a point when you set them up to compete, then they both would grow very quickly. So is that just for solving an engineering problem, but you see that the discriminator is a bit like the high or the monitor that I talk about, right. He essentially look at your own sensory activities and say, wow, this is generated by myself, or it is generated, or is triggered by the external world,
Paul 00:37:25 Please, correct me if I’m wrong. Cause I’m still trying to wrap my head around kind of where you situate subjective experience. Right? Because I think most theories or most approaches to thinking about consciousness kind of hit it at this highest level of being right. That it’s like at the end of the processing stream, wherever that, whether that’s a local processing stream or a global, uh, thing is kind of like the end. All right. Oh, and then there’s consciousness, but yours seems, uh, that the PRM version, the discriminator a reality monitoring type version, uh, seems kind of in the middle of cognition, right? Because then you have, you have these lower order, uh, cognitive, sensory modalities, things like that, processing that then the reality monitoring system monitors that, and then am I right? Passes it along to do some other higher order cognition that is not part of subjective awareness.
Hakwan 00:38:21 Yeah. I think that’s exactly right. It’s a great question. I think that’s the problem or the fundamental tension between the two camps, the Localist, the first older guys said, well, we’re just talking about seeing red. So of course we should look at the visual cortex and look, look for those neurons are clearly tuned to red or these later on cognition is just, it’s just, it’s just cognition. What does it have to do with seeing red? And then the higher the guys does sometimes come across as if we’re over intellectualizing consciousness. We sometimes come close to because of the jargons in philosophy, we come close to almost saying why seeing red is a fought. Uh, and, and it’s about really high order cognition. You have to have sense of self and you have to be self aware to do all of this. And then the local guy just said, well, you just talk about something else.
Hakwan 00:39:09 I’m not talking about that kind of consciousness. Like, um, like what the social scientists would talk about, like, like, you know, false consciousness, collective conscious. And I’m not talking about that kind of highly intellectual self narrative about, about things I’m not talking about your rationality. I’m just talking about the simple feeling of seeing red. So obviously you guys are just talking about something else, so we ignore you. And so my thing, my, my my game plan has been a, tried to tell them, try to convince the local guys that well, even when we lesson, we just really talk about simple experiences. Uh, sometimes not even things that you are attending to, not even things that you conceptualize and think about. I think there’s already a case to be made that there is some higher or the impact, uh, late, late against like, self-monitoring, you need to you just to sort out whether you’re actually hallucinating red or seeing red or, or your retinue rides are just firing randomly because of noise.
Hakwan 00:40:03 You need some sort of implicit self-monitoring. And I emphasize that it’s really implicit. It happens automatically. It’s not something you have to think about. And presumably very young children and smart animals would have the same mechanism, even though they may not even know explicitly what a self is. It’s just that it’s like a Ganz. It works automatically. It’s not something you have to activate and, and spend cognitive effort to do. And then there would say, well, yeah, okay. So if that’s true, then how does it relate to all the other really high order stuff that, that the sociologists and people talk about then I think, well then there’s probably because you have these Gantz like, um, um, decisions to, to decide whether your first orders input, um, decide what they reflect, because then that, that outputs to the higher, really high order stuff to self narrative and et cetera. So your self narrative and the higher order cognition, text and input from the, from the implicit higher order monitoring. Um, so exactly, so, so consciousness is exactly the interface between the, between the earliest early century perceptual stuff to the really higher order, um, cognitive stuff.
Paul 00:41:14 But, you know, so the brain is highly, highly, highly, highly recurrent, right? So, uh, this was higher order, um, areas where we do sense of narrative and sense of self, uh, can those, uh, reverse and become inputs into the reality monitoring, uh, interface as well, so that they are like first order, uh, systems in that respect.
Hakwan 00:41:38 Yeah, that’s a great question. I think day two, but the petunia extended that reverse input happens, it is probably modulatory. So it provides some context in my impacted a little bit, but the impact is not so strong. And I think it’s not really constitutive to having experience. And so when, when you change your culture, when you’ve just moved to a different country and speak a different language and immerse it for, for decades, presumably changes your whole brain to some extent, um, um, I think just like, or drinking wine people talk about if you become a really wide Buffalo condos here, uh, which I’m not
Paul 00:42:19 That’s
Hakwan 00:42:19 Right, right. Scotch actually. Yeah. Scotch is a good idea. I used to not like scotch and now I like scotch, so it probably changes. So in my early, um, kind of olfactory and taste processing to some extent. Um, but, but that’s kind of like an indirect route. It’s not like that as soon as you think of scotch and everything changes at that level.
Paul 00:42:43 Yeah. One of the reasons why I’m asking about this and you write about the, the functions, the potential functions of consciousness in the book. And so positing it like situating, the, our subjective experience sort of in between, uh, gives the worry that, um, maybe we, our consciousness doesn’t have as much functional relevance as we would like it to have. Right. It’s a little scary. Right. So, so first of all, just to understand, um, consciousness, is it important to understand what function it might may subserve? Or can we, can we learn about consciousness without understanding its functions without an account of its functions?
Hakwan 00:43:25 Yeah, I think, I think, yes, but it’s also not so straightforward. I think the, yeah, in general, like all the constructs in psychology and cognitive science, maybe even neuroscience, most of the, um, the, the target phenomena are defined functionally, right? So we say like, we want to study memory. Nobody was steady a kind of memory that doesn’t remember things, right. So memories are the way they are because they store information for future retrieval and use, et cetera. So, so attention people will sometimes say it’s vague, but as a functional concept, we also know it. Try to filter us out some useless stuff or irrelevant stuff for the current purpose and try to boost the signals and gate them for, for downstream processing. We can, we can talk about a function like this is probably good to have, and then we can go and map out how to bring my implemented function when it comes to consciousness is trickier because there’s a lot of, kind of, hand-wavy saying about all consciousness must have evolved to do X and live it.
Hakwan 00:44:27 And, and, and it’s really not a lot of empirical evidence that it does. And if you think about how to even get empirical evidence for that, then it kind of goes into this kind of chicken and egg problem that if you define consciousness as self-reflection, well, then yeah, then you will find this self-reflection stuff to be the correlate, but then the, your opponents are going to say, well, but why do we define consciousness as self-reflection, it may not, you may not have those functions. And so it’s a pretty tricky problem. Um, and I, so I, I tend to side with the local theorist there in that they, they basically assume that consciousness has very little function, right? At least in terms of higher cognition. So they would say basically if you’re in the more extreme version that, that some people actually hold, uh, but maybe not love, but they were sitting, if you cut out your visual cortex and put it on a Petri dish and keep it alive, it would have, you know, red qualitative experience, even though there’s no downstream cognitive circuits to appreciate it.
Hakwan 00:45:30 Some people would do things so, right. So basically that would have no function. And, and some people think so as long as it had the right network properties, even though is, is completely detached to input, output is not even functioning. Uh, it would have, it would have read experience, even if it doesn’t do anything. I think that’s sounds extreme, but I think it’s a good starting point as in, we should define a minimal notion of, of what, what, what raw read experiences, maybe not as extreme as the Patriot dish guys, the pen psychosis. Um, but if we can at least say, well, read experiences, just to being able to potentially report about read experience, if you, if you’re able to report, right. So assuming that you can report about stuff, then, then having read experienced is to have to propensity to be able to report it.
Hakwan 00:46:22 And that’s it, maybe, maybe that’s a kind of more minimal definition, uh, or maybe you could add a bit more, right. You can see, well, being able to see read is to be able to see it as not blue. Right? So that, that also seemed to make sense and be able to see red maybe is also like, you can see it as more like orange then blue. That also seem almost like a given if it, if it doesn’t have these functional properties, I don’t think it’s the same phenomena anymore, but what should we assume that being able to see red is to be able to associate it with danger and, and to be able to slam on your brake when you see it on the, on the, uh, traffic light. I think those, I would try to not assume that it’s a case. And then, and then having this minimal definition, then map out a circuit, then we can actually then empirically investigate. What are the, what are the normal functional advantages associated with that more minimal circuit?
Paul 00:47:18 This gets into the I, the idea of, um, mental quality space, right? So, um, I guess my opening question would be if seeing red was the only, this is an impossible thought experiment, but it was, it was the only experience one could have then I guess you would pause it. That one is not having a conscious, the first time you see red and you’ve had no other, no other experience. You actually don’t consciously, uh, perceive red that you have to see red and then a little something different from red.
Hakwan 00:47:52 Yeah. I think for humans, I, I’m not sure, but, but I think for, for, let’s say for a creature that is, has evolved without any history of seeing any other color, uh, then I think it would not see, right. It’s ready. We’ll just see a flash of light. Yeah. It doesn’t have the dimension. Right. So I think that I mentioned comes about, because we have maybe also in our evolutionary history, we have equipped our, our sensory cortex, uh, to, to be able to, to, to distinguish these different things. Then seeing red is about seeing it as not something else seeing red is seeing it as it, as it is rather than everything else. So in relative terms, it, it has a distinct quality. And I do think that that is part of the, um, almost like the two basic function that a minimal functional definition that he should have otherwise, otherwise the theory runs the risk of basically being a theory of reportability it doesn’t explain to qualitative aspects.
Paul 00:48:56 Well, so, um, can you explain a little bit, or describe a little bit what mental quality space is, and, and then talk about how you’ve incorporated it within the perceptual reality monitoring account, because this is something you kind of thought about more recently than the PRM idea itself, right?
Hakwan 00:49:12 Yeah. Yeah. That’s an example of where I totally went into philosophy and, and, and borrow something from, from, from philosophers. So the, yeah. And then try to mesh it with the science. So the PRM story, despite his controversy and all that, I think, I think we have a pretty good, um, it has gained some scientific traction, so it has made some impact in a few. And I think I can, I can give talks and lists out the empirical evidence in favor of it. It’s not, it’s not a done deal as in science, many of these things won’t ever be so, so soon, but at least I think there’s enough scientific support. I can say that it’s a scientifically tested and partially supportive theory, but the problem is then it doesn’t address this issue of why red looks the way it does. It tells you, okay, if you have a brick, if you have some rep to activity in the brain, then you know, when it will be conscious and when you will be hallucinating it, you know, when you can not report it, that’s great, but it doesn’t seem to get it to the deeper issue about how, why have this qualitative aspect, why, why it looks different from the other colors.
Hakwan 00:50:21 So it turns out, uh, mental quality space, uh, is an eye, is a concept that can be traced back to least, um, at least Wilfred sellers. And then in more recent times, um, Austin Clark and David Rosenthal also wrote a lot about it. So the idea, and sometimes I’m some other people call it a structuralist, uh, view of perception, which is to say the perceptual content has to be understood in relation to the, all the other perceptual content that you can have. So read looks the way it does because well, because it is more similar to orange and purple and pink and brown, uh, more similar to these than to black and white and silver and, and blue and, and stuff. So because of the, um, if you, if you want to express what it is like to see red, you can basically find out all the pairwise comparisons to all the other, maybe not just color or the other possible sensations you can ever have and write down or exhaustively how, how each pairwise similarity is.
Hakwan 00:51:22 Then, then you would basically describe what red is. Like. There’s nothing more to be known. Uh, and I find it quite plausible and strangely my, my strongest critics in the Localist, the first, all that Kames also seems to think that there’s some traction there. So I, so I liked the idea, I think conceptually it works, but then the problem is, but it’s so philosophical, right. So how do I combine that with PRM? That that was the, the, what I did in the, in the book and the final chapter. So, so the idea is, um, so PRM, as I mentioned earlier is about your sort of higher orders. Implicit self-monitor maybe like the discriminator in, against like architecture. It basically tried to look at, or try to monitor the, um, all of this is mechanistically done. It just basically tried to look at the early sensory activity and, and make a decision as to say, well, this activity has these dynamic properties that given my statistical understanding it is reflecting the state of the world right now.
Hakwan 00:52:24 So you better be conscious of that. So that’s the decision. And if you think about that, then the, the high or the monitor has to have some sort of address book. It has to know what these neurons and those neurons are in the earliest sensory cortex, right. And interestingly do early, early sensory contracts. We look at it so often we forgot how, how unique it is is, is a very, it has some, yeah, some nice properties there are, that are not necessarily, um, they’re in, in artificial perceptual mechanics, right? So the neurons are very sparse. So at the same time, if let’s say you’re seeing red, only few neurons are a small number of neurons that are involved in that, not your end, your entire sensorium has other neurons landed in a touch and tactile and auditory qualities. They are not involved, or even within the color sensitive neurons, not every neurons involved.
Hakwan 00:53:19 So, so, so has this part sparse coding property that is not, not, not really so trivial. And it also, um, um, it’s, it’s white, it’s smooth, right? The representation as we call it, a neural networks is it falls into a continuum. So that means that for, um, two sensory experiences, you can almost like trace down to address. You can say, well, this experience is basically mostly driven by these neurons. And then the other experience is mostly driven by those neurons. And you can talk about these two addresses. They are like a population profile. And then you can just talk about how these two addresses are, um, how, how much they overlap, because it’s a smooth continuum. So for any to experience, there’s some meaningful degree to which they are similar and some degree they’re not similar. Basically they overlap a lot. If there’s a lot of cross talk, then, then the new might actually confuse them. So that means that the higher order monitor to the extent that it can do is hire all the job by, you know, pointing at these early century coalitions and say, well, this is reflecting the world. This is not, this is hallucination, et cetera. Then it must know this address layout, right. In some way, no, in a implicit sense. I mean, the, the, the, the computations at the higher order level has to somehow be respecting and tracking the whole address, the sensory layout of the sensory cortex, or
Paul 00:54:43 Is it the, is the whole address or is it the differences?
Hakwan 00:54:46 Yeah, I would say it actually holds the address. Uh, almost. So if you, again, I, I use these kinds of natural language analogy. Sometimes it, it get read to write. I don’t really mean there’s a little man sitting in the prefrontal cortex doing this, but it’s as if, as if you, if you tell the prefrontal cortex that, you know, this experience is coded by these neurons, even number them like these Dakota by neurons 57 to 78, uh, with, with some, you know, fussy edges and et cetera, and you would tell them then that then the prefrontal cortex should know, okay, those are color neurons. I know that, and those color neurons are very far from another experience, which is coded by neurons 120 to 238, because these two populations are physically wired in a way that they are they’re quite far apart. So in a sense that your prefrontal cortex kind of knows what an experience is like, it almost like you get it for free.
Hakwan 00:55:45 Um, you is the analogy I use in the book is like, like a dictionary. If I tell you that a word is on page five of the OED, you basically can guess a bit what that word is. And if I tell you, well, how, how similar is that word to another word, which is on page 505 in a dictionary, you will know they will be very different. The first few letters will be probably different. Um, and so likewise, if you know the layout address, um, which the prefrontal cortex I argue, it has to because otherwise, how can it send all these top-down signals to do attention and cognitive control? It has to have some internal, maybe implicit grasp of the whole address system. Then, then in a sense you, that that’s maybe how you get the mental quality space kind of for free. Um, and I say for free, it’s not really free. I mean, evolution probably equipped us to do this because there are advantages to sparsity and smoothness and, and this kind of architecture.
Paul 00:56:44 I have a few more questions that are just kind of popping into my mind because, um, I think I thought about this reading your book, and I don’t remember if you, uh, address its relation to, um, the cognitive map. So, right. So, and, you know, navigation has been really hot in the past few years in neuroscience and that then has extended. And, and the fact that we, the idea is that we have this cognitive map and how does it get encoded, but, um, then multiple studies have extended the idea of the cognitive map to abstract spaces as well. And, uh, um, I don’t know if you’ve thought about how, um, this address kind of approach relates to cognitive maps.
Hakwan 00:57:23 Yeah, great question. Uh, it’s not in the book, but it’s in a paper coming out. Um, and the, and the story is nice. I, so he’s one of those online meetings during the, during the pandemic where I presented this, uh, mental quality space ideas. So it actually, I got into mental quality space, right at the beginning after pandemic, when, when there was nothing else to do. And I just sat on my balcony,
Paul 00:57:46 That’s like a primacy effect. Right? You remember the beginning of the pandemic and what you were doing.
Hakwan 00:57:52 Yeah. Basically this whole is a pandemic child, this whole MQs idea. And part of the problem is now, okay. Having gone through her, a semi empirical, uh, uh, story, having constructed this semi empirical story and has to inspire us and, and, and, and smooth coding and, and Ganz, then how do we test this idea further? I mean, it sounds like it’s still, it’s kind of almost like a philosophical story. It has some empirical assumptions, but how do we test it? We’re not going to be able to change the cortex to make it less virus, I suppose it will be tricky to do. Um, but then my friend and Steve Fleming, uh, pointed out, well, maybe we can actually hypothesize that the higher or the S the higher the side, when I say the addresses has to be incorporated. Somehow I was being very vague about this and thinking about the implementation is not easy.
Hakwan 00:58:45 And I was hoping some other clever computational modeling people would, would sort it out. But then Steve said, well, but there is an idea which is exactly like this, um, explicit or sort of explicit, uh, grit, grit, Colt, uh, that is used for, um, um, spatial navigation in the brain. And we know that how, when we actually physically especially navigate physical space, doing locomotion, we know that in rodents and in humans, there are, there are grid cells that, that, that, that code that encode the space. And we allow us to do this navigation, and we know that they have been used for abstract stuff. Like you pointed out. In fact, there was one study, uh, that, uh, I think is the first one was a bow, the AAO. And I think they actually found it in auditory scent, or sorry, in olfactory sensory space, if you make people navigate over an old factory space, that is, you know, a space defined by the intensity of different odors.
Hakwan 00:59:40 Uh, so it’s not an actual physical space, but it’s just an abstract space defined by, by different, uh, auditory olfactory stimulation. But navigating through that, you seems to activate some grit, light representations into prefrontal cortex too. So then Steve thought, well, why don’t we just go and test it? So we wrote a paper and maybe we’ll be doing that. Um, I didn’t put it, I didn’t try to update a book because the book at that point, it was already written. And I also think it is probably a very long shot. And so that’s one possible way. The prefrontal cortex handle these address, which is to put these different, uh, sensory addresses onto an, a space and abstract space, and actually then try to use other other machinery or, or mechanisms for, uh, spatial navigation to actually try to do, to navigate around that space. It will be wonderful to bring actually does that. And if it does that when whenever a conscious experience occurs, right, it’s not just that in some train tasks, the brain can be trained to do something like that. I think, yeah, I can be trained to think about color in terms of a 2d space, but, but when I just see color, I’m not sure my brain already instantiate, uh, this kind of space or this kind of spatial navigation activities, but it’s a, it’s a nice hypothesis that we can test. So when we’re doing that now.
Paul 01:01:01 Cool. And so, um, I don’t know if this is even a, well, we’ll see if it’s even a question worth asking, but, um, just, um, you know, uh, I’m continuing to try to flesh out, uh, kind of the situation, uh, where, you know, subjective experience is situated. So for PRM, it really does not depend on any model of the world, right? So, uh, it doesn’t need to connect to something like Basian brain, where we are constantly having to try to model the world. It’s Mo is really more like a gating, uh, discriminator mechanism. So does it depend on a model or relate to, uh, models?
Hakwan 01:01:42 I would say it is, it has a model of your sensory activity, right? It has the statistics of your sensory activity, but it’s not, it’s not explicitly a model of the world. I think the, the, the model of the world is probably downstream.
Paul 01:01:55 So not in the conscious realm,
Hakwan 01:01:57 Um, or consciousness will fit into that model of the world and the model of the world, my feedback to this mechanism to, to some lesser extent, but, uh, the, the, the constitutive mechanism for having the simplest, uh, conscious experience requires having a model of your own sensory activity rather than it doesn’t require a model of the world
Paul 01:02:21 Of the world. Right? So your science career, my science career also my academic career was devoted to, uh, visual neuroscience, essentially. Although I tried to, I tried to study metacognition, but in a visual paradigm, right. And the vast majority of, uh, consciousness talk is about visual, you know, seeing red is the, uh, paradigmatic, uh, example. Um, are there, is it useful to think about, uh, other sensory modalities? I know you just mentioned olfaction and the kind of cognitive map that, that can take place in all our olfactory senses, but is there an in testing theories of consciousness? Is it you, is it a useful approach to try to triangulate among the different sense modalities because, you know, vision is laid out topographically, right. And you were just talking about how, uh, olfactory, uh, has this kind of topic, topographical smoothness as well, but it seems if I’m a robust theory of consciousness should be able to account for, uh, all the senses, right?
Hakwan 01:03:25 Yeah, definitely. And, and that’s, uh, one of the limitation of current research that, that we are, we should be, uh, we should be very mindful of this limitation. Much of it is done in vision. There’s a smaller degree that is done in auditory and tactile, and olfaction is really tiny. Um, I see it as a, it’s a problem and should be addressed, and hopefully the, the theory were generalized to the other cases. And I sometimes think that that hope is, you know, is, is hopefully, is hopeful, wishful thinking, but at the same time, he’s also the norm in science, wherever we start with a paradigmatic case and hope that a generalized Intuit doesn’t, uh, that does how science is done, usually, basically. So, and the, and the excuse for doing that elbow or the reason for doing that is mostly just experimental convenience, right? So we know that the, the visual system is well understood and it’s easy to control the stimulus and you can stimulate the entire visual field pretty easily.
Hakwan 01:04:24 So doing tactile is a lot harder. You are stimulating a much smaller portion of the, of the whole, of the horse, uh, tactile cortex because of the way that it’s spatially organized. So you’re just stimulating a small part of it. So your signal, if you do whole brain imaging, it’s going to be small. Um, and, and, and when it comes to a fraction, the sensory fatigue is very, it kicks in very easily getting people to smell like 50 different stimuli. They usually want it, they want to, they want to get out, whereas making them to click on a few thousand Gabor patches, surprisingly people are pretty good at doing that. So, so mostly these kind of convenient reasons we, we started a vision, but eventually, yeah, we do have to test whether it generalize to all of them.
Paul 01:05:12 You’ve worked on a lot of different experimental modalities, right? So you’ve used largely MRI, but you’ve also stimulated with transcranial magnetic stimulation. Whereas our limit, I know we have multiple limitations, but, uh, it, you know, can you imagine it’s technology holding us back if we had a better recording devices or some technological breakthrough, would that actually help, or do we have the necessary tools? Now we just need to ask better questions and have better theories to test.
Hakwan 01:05:40 I think that technology would certainly help a lot. It will be game-changing, but I also think that the people who are very keen on pushing those boundaries, they’re good for them, but I think they often also neglect, uh, the, the simpler problems. They can’t found some meat with a better scanner. You still have the same compound. So if your, if your concept’s on, right, uh, it would still be problematic. So I, so in a way I’m, maybe I’m in a few, I’m in a good place. I th I think the field is moving. There are new technologies. I can get my hands on them, but I’m also not like frustrated for more frustrated by not getting to those new tools sooner, because even at the current technology, there are still a lot of conceptual cleaning up that needs to be done.
Paul 01:06:30 You know, I kind of gleaned over the what, what is a massive part of the book, uh, in talking about all of the different, uh, important issues related to the, and thorny issues related to the, uh, studying consciousness, um, empirically. So I’d like to ask you just about a couple of those, and then can leave readers to discover the rest of them in the book. Um, one of the things that you argue for is that task performance capacity, uh, is a, a crucial issue. So, um, could you just describe what you mean by that and why it’s important?
Hakwan 01:07:02 Yeah. So is one of those compounds that are, um, maybe not so obvious and some people denied that even the confound. So the idea is that if you want to study, um, uh, consciousness in the sense of subjective experience, then you really want to find experimental manipulations so that you just change subjective experience, but nothing else, right? You want everything else as much as possible to be held constant. So people quite often would focus on a stimulus confound. So when you change the subjective experience, if you change it by changing the stimulus itself, well, then, then you find some differences in the brain. You don’t know whether it’s because the difference occurs because of the difference in subjective experience, or maybe just a physical stimulation itself changes to input. So you need to control for that. And I think it’s not a trivial problem, but I think the whole, the entire field is aware of this problem.
Hakwan 01:07:53 And a lot of effort has been spent to address that. And you also need to address things like reporting, right? You don’t want, you don’t want to be, um, let’s say you turn an experience on and off. You don’t want it to be the case that when their experiences on they have to report, and when there’s no experience, they don’t report, then you have to same reporting confound. And again, I think it’s not entirely trivial, but people have dealt with, and I think it’s pretty, pretty good. And I think the problem is people it’s a subtle point, but th th th confines that are subtle, I actually the worst conference because the compounds that are relatively obvious, that means that other people would have taken care of it. And if you look at the literature, then, then the answers would be mostly immune to those conference.
Hakwan 01:08:38 If you look carefully and pick those studies or control for the contract and task performance capacity is to compound, it is so subtle that I think very few people have taken the trouble to control for that. So what I mean is when you’re, you’re aware of a stimulus, uh, you’re, you have a conscious experience. That’s good, but you’re also more able to do tasks, uh, in relation to that stimulus. And you’re you’re, you can identify it. You can remember it better. You can talk about your, your, your capacity for doing things related to that stimulus is also stronger. And that applies to all the paradigms, uh, in consciousness, even, even though in some cases, when you don’t have to do a task, right. And if I knock the rivalry, I think when I, when I first raised this question back in almost over a decade and a half ago, people would say, well, but by not cause the rivalry doesn’t have this problem because you are not asked to do a task.
Hakwan 01:09:30 And that’s why I realized, okay, I must emphasize is performance capacity. Compound is not performance confidence, even when you don’t ask them to do a task. Obviously when let’s say you’re seeing a dark in the cat rivalry in, in a, in a binocular riflery, uh, paradigm, when you see the cat, obviously you’re much, much more able to detect subtle changes in the cat image than the dark image. And in fact, it has to be done, has been done. So we can, we can make people do tasks in the context of binocular rivalry. And of course, unsurprisingly, the, the image that is conscious is much, much better process. So that kind of found is a pretty deadly, because I think it means that if you’re not careful your theory of consciousness, my end up just being a theory of how the brain works. And I suspect that most current theories are actually just how the brain works because that compound is not being taken care of.
Hakwan 01:10:19 So when you say I compare to conscious condition in a non-conscious non-conscious condition, these activities in the brain, uh, show up when you’re conscious, well, maybe those activities are just driving information processing in some ways. And yeah, of course, when you’re conscious, then you can do things you can remember, you can, you can do cognition. So, so you need to control for that. And, and some of my colleagues would say, no, no, no, you don’t. You absolutely don’t need to control for that. Because being able to process information is consciousness. So there’s no way you can control for that. It’s like saying that I tried to compare tall people versus short people and let’s control for the length of the bones. Right? Well, you can like, because 12 people are tall because still their bones a longer, like you can control for the very thing itself. And that’s, I think the, when, when the issue of, uh, the phenomenon of blind sight comes in because blind sight to me, even if it happens very rarely and some of the cases are controversial to the extent that it has ever happened in the entire world, to some extent is a, is an empirical counter example to the conceptual identity between consciousness and performance capacity.
Paul 01:11:32 Let me, let me stop you there. But I just want to mention, because I don’t know if I’ve told you this before. Um, and then we’ll, we’ll, we’ll talk more about blind sight, but, uh, some of your early work on, on blind sight was inspirational to me when I was trying to form my own graduate, uh, studies. So, uh, thanks. I’m not sure if I’ve ever said this, but thank you. I didn’t know those early studies.
Hakwan 01:11:53 I didn’t realize that. I didn’t think that, I didn’t know. Now I realize I’m much older than you are. I always thought I’ve read your paper, the single cell recording paper on, on metacognition. That actually was very useful to me too, but I didn’t realize that my work somehow has an impact on YouTube.
Paul 01:12:11 Oh, I was trying to, it actually helped me devise the task, um, thinking about the, uh, you know, patient UI and those early blind site studies. And, um, yeah. So anyway, I, I just wanted to make sure that I didn’t forget to express my gratitude to you, so
Hakwan 01:12:25 Very glad, very glad to it makes my day.
Paul 01:12:28 Yeah.
Hakwan 01:12:29 Yeah. So in plain sight is, is a case where lesion to the primary visual cortex cause some patients to, um, um, deny that they are there. They have any sense of subjective experience. Sometimes the scenario is not so complete, right? Some people will say, well, they can see motion, but not static stimuli. And sometimes it applies only to some stimulus, uh, parameters, but to the extent that you find those stimulus, and then they say they really don’t consciously see them. The interesting part is they can guess what the identity of the stimulus. So, and when I, when to meet guests, they just, you just give them two buttons and they press buttons. So data they’re basically doing psychophysics tasks like, like you and I do. Why? Because when I do psychophysics tasks, usually people don’t ask me if I’m guessing or seeing, they just asked me to press the damn button and those patients can do exactly the same.
Hakwan 01:13:22 And then they can achieve a fairly high level of accuracy, like 80%. Correct. Uh, so D prime of about one. So to have a solid signal to noise ratio or more than one. Uh, and, and so that means that, that, that, that is an case that demonstrate that the subjective experience is not always required for them to process information. So the two things are not the same, so they are not the same equivalence relations as being tired of having taught, having long bones. Because now I find a case of a person who’s very tall, but doesn’t have very long bones. So to the extent that that one case has been shown, that means that while the two are different than you need to control for them, uh, but very few people do. And I think that that is maybe one of the biggest problem that the field is facing now.
Paul 01:14:08 And how does a perceptual reality monitoring account for blind sight?
Hakwan 01:14:13 Yeah. So, so the nice thing about, um, these kind of higher order, uh, theories, instead, we assume that most of the functional work is done by the first order representations. So, so just having the, um, a strong first order signal would allow you to be able to guess the identity of the after of the stimulus. So we can do the, the, the blind side guessing tasks or any psychophysics tasks that way the higher order monitoring or the perceptual reality monitor is always like a common tree on the first order signal, right? So you have a, you have a strong, ongoing, um, first order signal in the sensory cortex that might go through your parietal action channels and allows you to do those tasks. And, but on D on top of that, you almost have a, you have a higher order commentary and say, wow, this signal now reflects a state right now. So you have to conscious experience, but if that is missing, that’s where you have blind side. So maybe your higher order monitor makes a statistical error. And even that at, in the presence of a very strong first the signal, we said, well, no, that’s just baseline noise. Uh, let’s ignore it. So you don’t have the, um, uh, conscious experience. But if the, if the first order signal is in fact strong, then you can drive the, uh, the performance. So you have a performance with our awareness in that case.
Paul 01:15:30 So presumably there are, um, quote unquote normal, uh, subjects in the population who, you know, may experience what would amount to blind sight on every 15 trials of some task, right? Um, because you have this ongoing monitoring situation. And sometimes it fails. I mean, is that plausible that a it’s just sometimes failing to discriminate correctly and therefore would lead to something like a blind site trial or something
Hakwan 01:15:57 I would think so. I would think theoretically and intuitively it is quite plausible, but the empirical demonstration turned out to be really difficult because we don’t know what, when people just say that they’re not conscious in that trial, maybe they mean that there is, it’s just faint. And yeah, so, so the empirical demonstration has been hard and a lot of people are very skeptical of blind sight. Some people were even skeptical whether Blindsight existed ever. Um, so that, but, but I think, I think if you go for the more solid evidence to show the blinds have existed, at least in some conditions, under some rare patients. I think that that is a defensible position.
Paul 01:16:37 So a Hogwan, I want to, um, we’ll wrap up here with just a couple of, sort of broader questions again. Um, you know, like I, I’ve repeated multiple times, you really analyze a lot of these studies and the potential confound, the issues to empirically studying consciousness and, and throughout your career, you have spent, you’ve devoted part of your career to testing other people’s theories, right. Uh, and, and testing your own, but, and also generating your own, which you’re in the midst of with the perceptually perceptual reality monitoring and the mental quality space, and that’s going to continue, what’s the right balance between one could spend a lifetime, an academic lifetime, just testing other people’s theories. What’s the right balance between testing other people’s theories and searching for, you know, the right controls, et cetera, versus, uh, generating your own and working on your own pushing forward.
Hakwan 01:17:38 Yeah. I might give a recursive answer to that. I, in the form that whatever balance did, one thing it is, I think we should just do more empirical stuff. So I think so if you assume, if you feel okay, I’m going to spend what X amount, like X percent of my time thinking of theory data down, you would, you would never regret it. I, I actually, when I wrote a book, I sent it to one of my, um, uh, good friend and colleague, uh, Richard Brown. And the first comment he gave was, but the book seems structural SIM seems to be structured funny. Like why would you introduce a theory only in chapter seven? Uh, and then the, maybe the most novel idea of mental quality spaces in the final chapter at night? Like, why would you put it in the, in the front? And I think it reflects that that sentiment, or, or, or advisor just gave, I think we should just place much more emphasis on doing the experiments.
Hakwan 01:18:34 Right. And I kind of reluctantly get into theory because at some point I needed to address some, some issues. Uh, and I, and I just say, I have a inclination. I hang out with philosophers maybe too much. And then I, and I like thinking about this, but every time I do it, I, I do try to remind myself is, is an indulgence. And it’s an indulgence that has not served a few too, too well. Um, so I constantly try to force myself to do less theory and more and more, more careful experiments. Even the theory is fine. I mean, everybody loves doing them, but
Paul 01:19:10 In your day to day thinking, uh, and research are, do you find more stubborn, the philosophical problems or the empirical problems or challenges?
Hakwan 01:19:22 I think ultimately the philosophical problems are harder, but in this, but I think the empirical challenges are more stubborn in some ways, it kind of relates to what I talked about earlier. I think when philosophy, at least you can be not so subtle. You can just, you can throw the kitchen, kitchen sink as you add your best friend, and you can still go for a beer afterwards. And they might like you take may continue to light news strangely. Whereas in science, I think you have to worry about stepping on people’s toes a little bit more. And given the state, it is, I think doing empirical work can be quite challenging for, for not, you know, not, not very necessary reasons.
Paul 01:20:05 Thinking back to your earlier career and you don’t have to go all the way back to the beginning, but do you feel like the problem of consciousness feels bigger to you now or smaller to you now than then?
Hakwan 01:20:23 Yeah, that’s a great point. I think it’s kind of both. Um, I think the, we have made some progress. I wouldn’t say we solved the problem. I say not acknowledged. I don’t think so, but I think we have made some progress in the sense of problem is easier, but in respect in my own life has certainly got a lot bigger. It become much more all consuming and, and, and, and the challenges, you know, grew into a different dimension. So when I started out, I didn’t think of funding. I didn’t think of the importance of closing my mentees, continue to grow the fuel and keep a healthy culture and go, you know, try to fight back on some of these, um, hold for emphasis on stardom and, and public engagement. And so that has become a much bigger problem, much bigger than I even even bigger than I could contemplate back then. Yeah.
Paul 01:21:16 So, uh, again, you know, you detail a lot of studies in the book. Um, what percentage of empirical studies, uh, about consciousness? Do we need to just throw away because of all the, uh, confluence?
Hakwan 01:21:31 I would think the majority of it, yeah,
Paul 01:21:35 It
Hakwan 01:21:35 Hurts. Yeah, it does. Uh, it does. I mean, maybe not throw away, but we have to really think about them in a much more skeptical way. Um, because just to the task performance capacity confound, it’s really everywhere. It’s really everywhere. So all the theories of consciousness has a similar flavor. That consciousness is something bigger, better, stronger, more stable, more global, more deeper, more predictive, more complex, more integrated. And yeah, that sounds like how the brain should work. And maybe that’s exactly because of that confound, giving you the flavor that you think you’re studying consciousness, but you’re just studying how, how information processing works most efficiently. So, so maybe we don’t throw them away, but the recognition of that means that we have to redo a lot of these stuff and it’s going to be a lot of work.
Paul 01:22:34 Let’s, let’s end on a high note here, uh, speculating on your own future trajectory. Where are you in terms of the maturity of, uh, the perceptual reality monitoring and including the quality space and, uh, as you project forward, where do you see yourself on the curve of your, of your theories and, and your empirical research to test the theories? Because that’s what you’re primarily concerned with, or at least a lot of what your concerns are. Um, where are you on that curve? Are you at the beginning? Are you, is it a halfway mature? What do you think?
Hakwan 01:23:07 I think, um, when it comes to perceptual reality monitoring, uh, I am feeling pretty good about it. I think that the details are probably my version would be wrong, but I already see other people picking up on my slack and, and innovating and doing better versions of it. So Sam Gershman wrote a piece that a similar Mrs. Zilkha white or another really majorly influential scholars in, in the machine learning world has also gotten into this problem. And I think they would, they would do well even, I, my version would probably fail, but I think the feel, uh, would do okay. But when it comes to the mental quality space aspect, I think we are really still at the very infancy. Um, and, and that’s good. That means that there’s more, more work to be done. I don’t, I don’t have to retire myself so soon.
Paul 01:23:56 You can, you can, uh, uh, yeah, let, uh, passing him now at the, the fight is still going on strong and for many, many more moons to come. Um, well, thanks for the book Hogwan. And, uh, like I said, I think among other things, it’s just going to be a super valuable resource for anyone getting, you know, thinking about getting into empirically studying consciousness, because damn it, there’s a lot of, uh, confounds and a lot to consider.
Hakwan 01:24:23 Thank you. Thank you for asking me.
Paul 01:24:30 Brain inspired is a production of me and you had, don’t do advertisements. You can support the show through Patrion for a trifling amount and get access to the full versions of all the episodes. Plus bonus episodes that focus more on the cultural side, but still have science go to brain inspired.co and find the red Patrion button there to get in touch with me, emailPaul@braininspired.co. The music you hear is by the new year. Find firstname.lastname@example.org. Thank you for your support. See you next time.
0:00 – Intro
4:37 – In Consciousness We Trust
12:19 – Too many consciousness theories?
19:26 – Philosophy and neuroscience of consciousness
29:00 – Local vs. global theories
31:20 – Perceptual reality monitoring and GANs
42:43 – Functions of consciousness
47:17 – Mental quality space
56:44 – Cognitive maps
1:06:28 – Performance capacity confounds
1:12:28 – Blindsight
1:19:11 – Philosophy vs. empirical work