WEBVTT 00:00:03.537 --> 00:00:10.497 This is the Convergent Science Network podcast. Leading researchers in the domain 00:00:10.497 --> 00:00:16.777 of neuroscience, brain theory and technology are interviewed by Paul Verschure and Tony Prescott. 00:00:24.317 --> 00:00:29.137 This is Paul Verschure with the Convergent Science Network podcast and we're 00:00:29.137 --> 00:00:33.237 here at the 10th anniversary summer school, the Barcelona Cognition, 00:00:33.237 --> 00:00:34.477 Brain and Technology Summer School. 00:00:34.857 --> 00:00:41.697 And I'm here with Lynn Jurek and Bashir Yaraya, who are presenting this morning 00:00:41.697 --> 00:00:47.577 their collaborative work on anesthesia, the brain, the thalamic cortical system, 00:00:48.477 --> 00:00:52.057 and also how we can develop animal models of this. 00:00:53.437 --> 00:01:02.997 So now Lynn how do you think the kinds of systems you're looking at are helping 00:01:02.997 --> 00:01:07.277 us understand how the brain works so the particular angle that you take of anesthesia 00:01:07.277 --> 00:01:09.117 how is this really helping and how is it strategic, 00:01:10.297 --> 00:01:16.357 I think it's very important to know what we are doing in everyday clinical work 00:01:16.357 --> 00:01:21.337 because so I'm an anesthesiologist at the beginning and when I started my My 00:01:21.337 --> 00:01:23.577 residency in anesthesia, 00:01:23.717 --> 00:01:26.017 the first question I asked was at one moment, 00:01:26.297 --> 00:01:28.537 how anesthesia works on the brain? 00:01:29.017 --> 00:01:32.157 And the answer I got at the very beginning was, okay, it works, 00:01:32.277 --> 00:01:33.617 why do you ask this question? 00:01:34.217 --> 00:01:42.157 So that's why I came to the idea to study anesthesia, because at some moment 00:01:42.157 --> 00:01:45.657 the problem of anesthesia was to make it safe for patients. 00:01:45.657 --> 00:01:53.057 So there was a lot of improvement in the hemodynamics of anesthesia, 00:01:53.277 --> 00:01:56.617 in the respiratory field of anesthesia 00:01:56.617 --> 00:02:01.017 and so on, but not really in the neuroscience fields of anesthesia. 00:02:01.017 --> 00:02:06.297 So it was like at some moment you can get a shutdown of the brain. Okay, it's nice. 00:02:07.077 --> 00:02:12.937 People don't remember what happens, but then at the beginning when I started 00:02:12.937 --> 00:02:16.517 my residency, there was not much more that was known about anesthesia. 00:02:16.917 --> 00:02:22.377 And so at some moment it was like I want to know better what I'm doing to take 00:02:22.377 --> 00:02:25.717 at some moment perhaps better care of the patients I have in charge. 00:02:26.157 --> 00:02:30.217 So it's fair to say that actually we don't really know why anesthesia works. 00:02:30.217 --> 00:02:35.137 No, I think today we know more than 10 or 15 years ago. 00:02:35.197 --> 00:02:41.057 Okay, but still today we don't really know why people lose consciousness with 00:02:41.057 --> 00:02:43.417 different drugs we use in everyday life. 00:02:44.117 --> 00:02:47.577 And so what was really impressive in also your talk is the overview you gave 00:02:47.577 --> 00:02:53.437 of the different, let's say, interventions we have available today. Okay. 00:02:53.935 --> 00:02:58.455 To induce these states of consciousness that you might use in the surgery ward. 00:02:59.235 --> 00:03:03.495 So what are the most dominant interventions that you're using? 00:03:03.835 --> 00:03:10.515 I think between the most dominant drugs we use in everyday clinical practice, there's propofol. 00:03:10.555 --> 00:03:13.215 Propofol is used everywhere in the operating room. 00:03:13.475 --> 00:03:20.955 Also for the volatile agents is most sevoflurane, sometimes isoflurane, which is an older drug. 00:03:20.955 --> 00:03:25.795 And then if you have patients in the intensive care unit that are not very stable, 00:03:25.835 --> 00:03:29.715 you use ketamine because it doesn't act on the hemodynamics of the patient. 00:03:30.035 --> 00:03:36.055 So I think these three are mostly used and probably in the future what will 00:03:36.055 --> 00:03:40.155 be used at least in the intensive care unit probably dexmetadometin because 00:03:40.155 --> 00:03:42.895 it's thought to induce a sleep-like state. 00:03:43.095 --> 00:03:47.615 But would perhaps be more physiological condition for the patient. 00:03:48.275 --> 00:03:51.335 But that's right now more in experimental phase, so that's already being used? 00:03:51.655 --> 00:03:56.135 It's already being used in intensive care units, yes. When you wake up patients 00:03:56.135 --> 00:03:59.515 at some moment to induce like a sleep-like state. Right. 00:03:59.715 --> 00:04:05.875 But now the pathway of action of propofol and ketamine, and this also will be 00:04:05.875 --> 00:04:11.595 relevant later when we look at you and understand and analyze your data, is rather opposite. 00:04:11.875 --> 00:04:18.775 Yes, yeah. It's completely the opposite on the brain. So we can say propofol 00:04:18.775 --> 00:04:22.035 is essentially driving the inhibitory system. 00:04:22.215 --> 00:04:25.855 Yes. Right? This is fair to say in a rather nonspecific way. 00:04:26.035 --> 00:04:32.735 Yes. Okay. Well, ketamine is more specifically acting on the excitation on the inhibitory system. 00:04:32.955 --> 00:04:36.815 Yes. Right? Because it acts on the NMDA receptors on the GABAergic cells. 00:04:37.135 --> 00:04:40.695 But is there some sort of regional specificity to that in any way? 00:04:42.595 --> 00:04:47.535 Let's say for the ketamine drugs, it's mainly acting on the cortex, 00:04:47.595 --> 00:04:54.075 not on the subcortical regions and on regions like the amygdala of the hippocampus. 00:04:54.375 --> 00:04:59.435 And for propofolates, it's thought that at the beginning it acts on the cortex 00:04:59.435 --> 00:05:06.095 and in a second phase of action, it acts on the subcortical regions. 00:05:06.875 --> 00:05:12.695 There are human studies, So one human study by Leonel Bede who showed this like 10 years ago. 00:05:13.475 --> 00:05:18.375 Right. Patients. So this is really one mystery that we should try to clarify, 00:05:18.515 --> 00:05:20.215 right? We have these two drugs. 00:05:20.415 --> 00:05:25.095 They in some have an opposite effect on the inhibitory system of the cortex 00:05:25.095 --> 00:05:28.695 while still leading to the same outcome with loss of consciousness. 00:05:28.935 --> 00:05:32.935 So we have to solve that one before this podcast is over. 00:05:32.935 --> 00:05:37.835 And this is also what you have been trying to do, following very specifically 00:05:37.835 --> 00:05:44.315 a very unique paradigm that has been pioneered in the lab where you both are 00:05:44.315 --> 00:05:47.195 working, which is a local and global paradigm. 00:05:47.475 --> 00:05:50.155 Right. So I don't know. 00:05:51.577 --> 00:05:54.657 Which of you two would like to speak for the local global paradigm? 00:05:54.937 --> 00:05:57.457 But Bashir, you have been very active in that. 00:05:57.837 --> 00:06:01.797 So what's so interesting or unique about this local global paradigm? 00:06:02.557 --> 00:06:07.057 The local global paradigm is a series of sounds that you can listen to. 00:06:07.217 --> 00:06:08.617 This is what we call a paradigm. 00:06:10.317 --> 00:06:16.097 The idea came in from our friends and colleagues, Sénissas Dehaene and Yonel Nakache. 00:06:16.417 --> 00:06:19.437 What does it mean, local versus global? 00:06:19.437 --> 00:06:23.457 So if you listen to small sounds like beep, beep, beep, boop, 00:06:23.537 --> 00:06:27.837 you have the same sound, identical sound coming in, and then what you call a 00:06:27.837 --> 00:06:31.297 deviant, so you violate a rule at the local level, 00:06:31.417 --> 00:06:35.657 because it is a small, tiny scale of time. 00:06:35.657 --> 00:06:42.997 But this has been investigated for decades, and if you do an EEG called ERP, 00:06:43.357 --> 00:06:47.017 Nathaniel described what we call famously the mismatched negativity. 00:06:47.957 --> 00:06:56.437 People were excited about it, tried to perform this paradigm in anesthetized 00:06:56.437 --> 00:07:01.617 people, in people emerging from coma, but unfortunately was not discriminative 00:07:01.617 --> 00:07:02.697 of loss of consciousness. 00:07:02.697 --> 00:07:07.917 So even without consciousness, your brain, you can still process this mismatch 00:07:07.917 --> 00:07:14.097 negativity or the local part of this local global paradigm. 00:07:14.097 --> 00:07:18.717 So the trick there was to multiply the sequence, this local sequence, 00:07:18.757 --> 00:07:23.257 several times, and this time introduce a violating sequence. 00:07:23.757 --> 00:07:26.977 So the new sequence that comes by the end, like for example, 00:07:27.137 --> 00:07:33.657 this last sequence, is violating the other one. 00:07:33.777 --> 00:07:38.797 This is the global violation. And it turns out that it works fantastically because 00:07:38.797 --> 00:07:44.557 you really need to be conscious to process this global effect and to have your 00:07:44.557 --> 00:07:47.057 brain realizing this sequence violation. 00:07:47.457 --> 00:07:55.877 If you have, for example, a vegetative state with a patient that never recovered 00:07:55.877 --> 00:08:00.197 any consciousness, then you definitely cannot realize this global part of it. 00:08:00.857 --> 00:08:05.537 And do you think you can stack these sequences indefinitely? Is it like recursive? 00:08:07.417 --> 00:08:14.757 Currently, the structure of the sequence by itself is not aimed at recursive detection. 00:08:17.317 --> 00:08:23.577 This is definitely, there are variety, we are developing varieties of this paradigm 00:08:23.577 --> 00:08:28.817 to check for syntax, for numerosity. We did that even in animals. 00:08:29.157 --> 00:08:36.017 But so far, we consider that as a marker of conscious access. 00:08:38.117 --> 00:08:38.297 So... 00:08:40.523 --> 00:08:45.783 So now we have this paradigm where in some ways we have local and we have global deviations, right? 00:08:45.943 --> 00:08:53.423 So you see this as a specific probe of states of consciousness or aspects of consciousness. 00:08:53.503 --> 00:08:56.603 The global part of it as opposed to the local. 00:08:56.783 --> 00:09:02.183 This was demonstrated by Tristan Bekenstein, a paper with Dylan Akesh and Seyes Doan. 00:09:02.183 --> 00:09:05.583 And they showed that in humans, at least, 00:09:05.683 --> 00:09:12.923 healthy people distinguish very nicely the global effect from the local effect, 00:09:13.023 --> 00:09:20.303 meaning that the global effect will need a large-scale cortical activation, 00:09:20.603 --> 00:09:22.543 will induce large-scale activation. 00:09:24.403 --> 00:09:29.003 Actually, our contribution came in with the animal aspect of it. 00:09:29.003 --> 00:09:33.143 The question was in that time, there were two questions. 00:09:33.223 --> 00:09:40.483 First, could we infer the same thing in non-human animals and our closest cousins, 00:09:40.843 --> 00:09:43.163 let's say non-human primates? 00:09:43.403 --> 00:09:50.463 And second, starting from an animal model in which we can induce anesthesia, 00:09:50.623 --> 00:09:58.343 could we manipulate this global detection with the manipulation of consciousness? 00:09:59.003 --> 00:10:06.003 Right, okay. So if we now look at this, the human case, what we look at is now 00:10:06.003 --> 00:10:11.783 a paradigm that allows you to access, if you want, informational aspects of conscious processing. 00:10:11.983 --> 00:10:16.363 So we're not saying much about the experiential, phenomenological aspects of 00:10:16.363 --> 00:10:18.603 consciousness, right? This is not part of the discussion now. 00:10:18.823 --> 00:10:21.143 Exactly. So we're really focused on access consciousness. 00:10:21.603 --> 00:10:26.423 And then what you see is a very distinct correlation between, 00:10:26.703 --> 00:10:32.103 let's say, these local sequences and, let's say, A1 processing in the temporal loop, 00:10:32.844 --> 00:10:40.664 or more broad activation in what in your center is called the global neural 00:10:40.664 --> 00:10:42.424 workspace. Absolutely. Right? 00:10:43.144 --> 00:10:49.124 So what are then the core nodes in that global workspace that you observe in this paradigm? 00:10:50.964 --> 00:10:57.244 So if the regions be activated with this local global paradigm, 00:10:57.424 --> 00:11:03.664 and especially for this global effect, There are parts of the prefrontal cortex, 00:11:04.044 --> 00:11:05.924 there are parts of the parietal cortex, 00:11:06.404 --> 00:11:10.964 there is the anterior cingulate cortex and the posterior cingulate cortex, 00:11:11.184 --> 00:11:14.504 at least for the cortical regions that we activate with this paradigm. 00:11:15.444 --> 00:11:19.904 If you go to deeper structures, there are structures of the thalamus, 00:11:19.904 --> 00:11:25.324 like the paraphysical and the nuclei of the thalamus, and part of the striatum 00:11:25.324 --> 00:11:29.144 that you can activate with this paradigm. that. 00:11:30.044 --> 00:11:32.984 Okay, so now we have a starting point, right? So then you say, 00:11:33.064 --> 00:11:36.904 look, there's a form of conscious processing we can now manipulate. 00:11:37.324 --> 00:11:40.924 We have a neural substrate and indeed the big step, as you also earlier said 00:11:40.924 --> 00:11:44.704 Bashir, the big step was now to bring that to an animal model because then you 00:11:44.704 --> 00:11:46.724 would have more experimental control, right? 00:11:46.764 --> 00:11:51.304 So how well did that generalization to the animal model really work? Lynn? 00:11:52.964 --> 00:11:58.264 I think it worked quite well because at the beginning we had no idea if the 00:11:58.264 --> 00:12:00.904 animal could detect this global effect. 00:12:01.244 --> 00:12:04.584 What was quite true that it could detect the local effect because there were 00:12:04.584 --> 00:12:08.624 older studies with electrophysiology that showed that if you, 00:12:09.464 --> 00:12:15.564 do invasive electrophysiological studies in the auditory cortex that you can 00:12:15.564 --> 00:12:20.404 get mismatch negativity that you can record in the auditory cortex. So we. 00:12:22.664 --> 00:12:25.664 That at least what we expected. Then was the question, could 00:12:25.664 --> 00:12:29.244 it detect something more complex because 00:12:29.244 --> 00:12:37.944 if you want yes to learn like a rule like if you have like five identical sounds 00:12:37.944 --> 00:12:44.124 then you get this as a rule and at some moment you get this global deviant and 00:12:44.124 --> 00:12:46.884 in all the animals that were tested, 00:12:46.964 --> 00:12:52.064 they were tested several times in the scanner there are three animals in total 00:12:52.064 --> 00:12:56.084 because for monkeys often you it's difficult to have more animals because you 00:12:56.084 --> 00:13:02.284 have to be trained at the beginning And the data we show where group analysis of this monkey, 00:13:02.384 --> 00:13:06.144 but if you look at the individual level, all the monkeys could get activations 00:13:06.144 --> 00:13:08.544 in the prefrontal cortex, the parietal cortex, 00:13:08.844 --> 00:13:14.464 anterior cingulate, posterior cingulate, and some also had activations in the hippocampus. 00:13:15.524 --> 00:13:21.164 But now, what's the magnitude of these differences between the local and the 00:13:21.164 --> 00:13:26.804 global variations of the task? and how does this magnitude compare to what you observe in humans? 00:13:28.118 --> 00:13:32.618 There's actually two remarks there. First, there is a strong hierarchy. 00:13:33.258 --> 00:13:40.918 So, the local global paradigm introduces the notion of hierarchical levels of 00:13:40.918 --> 00:13:43.138 violations, of sequence violations. 00:13:43.538 --> 00:13:48.858 And that is really important. This is in line with what we call the predictive coding theory, 00:13:49.158 --> 00:13:55.518 and what we saw already in humans, we see exactly the same homology in monkeys, 00:13:55.678 --> 00:14:02.758 is that because the global violation is hierarchically superior to the local violation, 00:14:03.178 --> 00:14:11.878 it also activates a high cortically organized the frontal parietal network, 00:14:12.338 --> 00:14:20.878 whereas the local effect was really activating the low-level auditory pathway, 00:14:21.018 --> 00:14:23.538 including within the brainstem until A1. 00:14:23.698 --> 00:14:27.698 And we saw that exactly in a homologous way in the macaque. 00:14:28.238 --> 00:14:34.898 Meaning that in the macaque, we saw the auditory pathway from the brainstem 00:14:34.898 --> 00:14:40.658 nuclei until A1 for this low-level deviant, which is a local effect, 00:14:40.878 --> 00:14:44.438 while when we go to the global violation, 00:14:44.858 --> 00:14:54.238 the second level of hierarchy, you see involvement of both prefrontal and parietal cortex. 00:14:54.518 --> 00:15:01.238 And we went even a little bit further to see how much this is really specific 00:15:01.238 --> 00:15:06.798 to the global effect by doing a technique in fMRI data called psychophysiology, interaction, 00:15:07.298 --> 00:15:16.338 PPI, and this put in a seed in A1, PPI could show that really global effect 00:15:16.338 --> 00:15:18.678 increased specifically, 00:15:19.198 --> 00:15:23.778 brain activity increased specifically in these frontal parietal singular areas 00:15:23.778 --> 00:15:29.878 when the global deviant comes to the macaque auditory system. 00:15:33.221 --> 00:15:39.421 Okay, so in the human case, we see some frontal parietal signature of the global task, right? 00:15:39.481 --> 00:15:43.201 And then in the macaque, you see something similar, also in fMRI, 00:15:43.341 --> 00:15:44.961 right? Exactly. We're doing fMRI in the macaque. 00:15:45.621 --> 00:15:48.701 But now we have, of course, this whole conundrum of homologues, 00:15:48.701 --> 00:15:52.301 right? What are homologues between macaque brain and human brain? 00:15:52.781 --> 00:15:55.721 And actually, if you just look at the pictures that you present, 00:15:55.841 --> 00:16:00.061 also in your work, it would appear that, for instance, these spots of activity 00:16:00.061 --> 00:16:03.621 are a bit more lateral in the macaw case as compared to the human case. 00:16:03.701 --> 00:16:07.421 Maybe I misinterpret, but how do you deal with the issue of homologue, 00:16:07.641 --> 00:16:10.201 of homology? It's really a tough issue. 00:16:10.861 --> 00:16:18.201 I like the question because we are interested in that, in brain-monkey comparisons, 00:16:18.301 --> 00:16:22.661 because this sheds light eventually on the uniqueness of the human brain, of course. 00:16:22.661 --> 00:16:30.141 And technically, it remains very, very challenging to do direct homology studies 00:16:30.141 --> 00:16:35.361 between the anatomy of the primate brain and the human brain. 00:16:35.361 --> 00:16:39.841 There are already some tools existing in the literature. 00:16:40.321 --> 00:16:46.321 There are still a lot of work to do there. And we are interested to do that in the future. 00:16:46.801 --> 00:16:56.941 Because it's tough to compare brain maps by direct visual, our visual system basically. 00:16:57.861 --> 00:17:03.781 And the notion of homology is not that trivial that we could think. 00:17:03.781 --> 00:17:13.461 What was important in our case is really to see that the two orders of violations, 00:17:14.081 --> 00:17:19.281 the two hierarchical orders of violations in monkeys, just like in humans, 00:17:19.581 --> 00:17:21.461 were paralleled by two... 00:17:22.168 --> 00:17:27.648 Hierarchical order of cortical activations. That was important in our case. 00:17:27.808 --> 00:17:32.968 Right. But now imagine we do something stupid, which I like because these are things I'm good at. 00:17:33.108 --> 00:17:38.508 And we just inflate or sort of in a very linear way map this macaque brain to a human brain. 00:17:38.788 --> 00:17:44.728 Would you predict that then the nodes in this global network in the macaque 00:17:44.728 --> 00:17:47.628 would align with those in the human brain or not? 00:17:48.188 --> 00:17:53.068 I'm not sure we can really fix the problem this way. There are tools in Caret, 00:17:53.068 --> 00:17:59.468 for example, Caret software offers tools to do just this direct alignment by simple inflation. 00:18:00.128 --> 00:18:07.868 However, you need to keep in mind that the human development of a human brain 00:18:07.868 --> 00:18:12.528 was not just about inflation of volume, as you know, but also of specification 00:18:12.528 --> 00:18:14.708 and local specification. 00:18:14.988 --> 00:18:18.868 And on top of it, all the education and training. meaning. 00:18:19.728 --> 00:18:25.968 This is clearly one of the challenges of the next years is how to develop really, 00:18:26.588 --> 00:18:29.328 proper tools to do these direct comparisons. 00:18:29.568 --> 00:18:35.548 Because with the ability to train monkeys and especially cac monkeys to sit 00:18:35.548 --> 00:18:39.688 in the scanner and to listen to the exact same paradigm in the same condition 00:18:39.688 --> 00:18:41.008 than the human counterparts, 00:18:41.408 --> 00:18:48.168 we start to, for a while already, with Vandafil, Nikos Drogathidis and Dois 00:18:48.168 --> 00:18:52.688 Tsao and other people we start to have this unique opportunity, 00:18:52.928 --> 00:18:57.988 we have two maps for exactly the same experimental setup, how can I deal to 00:18:57.988 --> 00:18:59.748 have a direct comparison, a computational, 00:19:00.608 --> 00:19:06.368 comparison, that's something we are also working on But what is interesting 00:19:06.368 --> 00:19:07.428 is that it might mean that, 00:19:08.197 --> 00:19:12.757 that the midline structures of the macaque brain might be organized somewhat 00:19:12.757 --> 00:19:17.157 differently as compared to those of the human brain, even though, 00:19:17.277 --> 00:19:20.777 let's say, from a phylogenetic perspective, they're relatively close. 00:19:21.077 --> 00:19:22.257 Would you agree with that? 00:19:25.057 --> 00:19:33.457 Again, I would say that not only comparative anatomy is needed there, 00:19:33.457 --> 00:19:40.977 But we need to automatize tools based on tracer studies, for example, 00:19:41.237 --> 00:19:45.017 to match the two brains. 00:19:45.857 --> 00:19:52.637 Will there be a clear transform between the two brains? 00:19:52.637 --> 00:19:59.157 If we think about all the millions and years of evolution that made this strong 00:19:59.157 --> 00:20:06.337 job, I'm not quite sure that we have an easy transform, direct easy transform. Right. Okay. 00:20:06.997 --> 00:20:13.657 So we have this paradigm. We see that we go from a local process to a more global process. 00:20:14.397 --> 00:20:20.697 But now what you're relying on is a paradigm that assumes that the animal is 00:20:20.697 --> 00:20:22.137 basically passive but aware. 00:20:22.637 --> 00:20:25.477 More as the human is passive and aware in this resting 00:20:25.477 --> 00:20:30.457 state paradigm and it's not that the resting state was invented because it's 00:20:30.457 --> 00:20:33.757 behaviorally so interesting it's just the simplest thing that people can do 00:20:33.757 --> 00:20:38.337 as they sit in a scanner or lay in a scanner uh right surrounded with this noisy 00:20:38.337 --> 00:20:43.777 equipment so now we're forced into the monkey as well so that means and also i brought this up in 00:20:43.917 --> 00:20:48.757 the talk right that the monkey is sitting there we assume the monkey is sort of at rest and aware. 00:20:49.757 --> 00:20:55.237 But we have actually no real idea whether this is in any way pertaining to the 00:20:55.237 --> 00:20:58.377 consciousness of this macaque monkey, right? 00:20:58.437 --> 00:21:03.497 So how do you, do you see this as a limitation of these studies that we have 00:21:03.497 --> 00:21:08.237 no way, that there's no reportability, that there's no overt behavior coming out of this monkey? 00:21:09.077 --> 00:21:12.377 Do you see this as a limitation or you don't think this is a problem? 00:21:12.657 --> 00:21:16.437 It is, of course, a limitation, very clearly. so 00:21:16.437 --> 00:21:19.577 how to do to fix 00:21:19.577 --> 00:21:22.537 the problem one way is to 00:21:22.537 --> 00:21:29.077 train massively the monkeys to report for the detection of the local violation 00:21:29.077 --> 00:21:34.517 versus the global violation of course we thought about that the problem with 00:21:34.517 --> 00:21:40.377 that is that it implies a lot of training with the paradigm And in our case, 00:21:40.617 --> 00:21:49.437 we try to, at least this was our strategy, we discussed a lot before starting 00:21:49.437 --> 00:21:51.357 this research program. Yes. 00:21:52.731 --> 00:21:56.691 We try to keep really the monkeys naive to the paradigm because overtraining 00:21:56.691 --> 00:22:02.191 with the paradigm is always a profounding source for the interpretation. 00:22:02.711 --> 00:22:06.111 Now, I agree, you pronounce the word reportability. 00:22:07.291 --> 00:22:13.931 But I would say also that we start to accept fMRI as a reportability now in 00:22:13.931 --> 00:22:16.831 humans with disorders of consciousness. 00:22:16.831 --> 00:22:21.111 When you see all this work by the Liège and Cambridge groups showing with this 00:22:21.111 --> 00:22:27.751 tennis imagery paradigm in highly disabled people with minimal conscious state, 00:22:28.051 --> 00:22:35.091 actually these people really do not report except through fMRI. 00:22:35.671 --> 00:22:40.391 So fMRI activation is also being acceptable reportability. 00:22:41.431 --> 00:22:45.351 Well, but that's maybe more a message of hope, right? Because it doesn't work in all patients. 00:22:45.971 --> 00:22:51.571 No, no, no, no, it doesn't. So, but okay, so here we have our paradigm. 00:22:51.971 --> 00:22:53.551 We have a local global effect. 00:22:54.411 --> 00:22:58.511 It's sort of, let's say, let's call it weakly analog to what you see in a human. 00:22:58.651 --> 00:23:02.851 I mean it weakly because we cannot really nail whether it's homologues or not, 00:23:03.611 --> 00:23:07.451 but you see a local global effect involving frontal and parietal systems. 00:23:07.991 --> 00:23:11.571 So fantastic, right? Great. So now we have calibrated, we mapped the paradigm 00:23:11.571 --> 00:23:16.011 to the macaque monkey, But now the real stuff starts because now you start to 00:23:16.011 --> 00:23:21.591 manipulate these states of consciousness with propofol and ketamine and other drugs. 00:23:21.931 --> 00:23:24.771 So, Lynn, what happened when you started to do that? 00:23:26.371 --> 00:23:32.591 So, at the very beginning, we had to make sure already that the anesthesia protocol 00:23:32.591 --> 00:23:38.851 would work because you cannot like put your monkey in the scanner, 00:23:38.991 --> 00:23:41.011 put one drug and then see what happens. 00:23:41.011 --> 00:23:45.611 Happened there was a lot of work that was done outside the cancer scanner before 00:23:45.611 --> 00:23:50.571 going to the scanner to make sure that already the anesthesia level was stable 00:23:50.571 --> 00:23:58.171 at every during the whole experiments so the anesthesia model was based like 00:23:58.171 --> 00:24:00.771 for propofol it was and also for kadamine after, 00:24:01.431 --> 00:24:05.991 it was based on behavioral scale of the monkey because he want to go to general 00:24:05.991 --> 00:24:10.311 anesthesia so So normally general anesthesia, you are not conscious. 00:24:10.631 --> 00:24:14.011 And then we had like behavioral tested with it in the monkey. 00:24:14.331 --> 00:24:24.111 And also you had to make sure that you have a stable EEG pattern for your anesthesia. 00:24:24.331 --> 00:24:31.131 Because you cannot say that you use like X milligram per kilo of a drug and 00:24:31.131 --> 00:24:32.411 then it will work for everybody. 00:24:32.771 --> 00:24:39.631 Of course, it didn't work for the monkeys. So we had to adapt the anesthesia 00:24:39.631 --> 00:24:44.991 concentrations on the behavioral scale and on the EEG before doing the whole 00:24:44.991 --> 00:24:47.631 paradigm after in the scanner. 00:24:48.860 --> 00:24:52.580 And then, but, so, okay, first you have to fine-tune the paradigm, 00:24:52.780 --> 00:24:56.540 you have to get to the right dosage, so you sort of nailed that problem. 00:24:56.640 --> 00:24:58.040 Of course, it's a very hard one, right? 00:24:59.820 --> 00:25:04.240 But then, in some way, you saw a differential effect of the ketamine and the 00:25:04.240 --> 00:25:07.860 propofol was not identical in its impact, right? No, no. 00:25:07.920 --> 00:25:14.560 It was like really closely, for the local effect was quite close. 00:25:14.740 --> 00:25:20.780 But let's say we weren't really interested in this local effect because there's 00:25:20.780 --> 00:25:23.620 a lot of literature showing that if you do electrophysiology, 00:25:23.800 --> 00:25:29.600 even in humans, that if you go to general anesthesia, often you lose some mismatch negativity. 00:25:29.760 --> 00:25:34.620 So we're not surprised that at some moment you don't get a local effect in these animals. 00:25:34.620 --> 00:25:40.080 What was quite surprising was for the global effect, because it could have been 00:25:40.080 --> 00:25:45.060 that, okay, you have no local effect, you will have no global effect at all, and so you have nothing. 00:25:45.180 --> 00:25:50.080 You will activate your auditory system when we present all the sounds or do 00:25:50.080 --> 00:25:51.900 the contrast for all the sounds for fMRI. 00:25:52.140 --> 00:25:56.840 That was at least what we were expecting, because it is reported in literature 00:25:56.840 --> 00:26:03.800 that the sensory cortexes are still active under anesthesia, but not beyond. 00:26:04.620 --> 00:26:09.720 And then at the beginning, we were quite surprised that we had no global effect 00:26:09.720 --> 00:26:15.900 at all for ketamine, and that still we had activations, especially in the prefrontal cortex of propofol. 00:26:16.600 --> 00:26:21.660 So that was one explanation could be that they act on different receptors, 00:26:21.760 --> 00:26:27.500 and what gives us explanations, but it's not completely sure. 00:26:27.580 --> 00:26:33.500 We cannot completely make sure that this… So the main thing that you observed, 00:26:33.640 --> 00:26:37.780 that you also described, is that with ketamine, you don't see the activation 00:26:37.780 --> 00:26:40.840 of this frontal parietal network, right? Well, with Propofol you do. 00:26:41.220 --> 00:26:45.700 But now in the Propofol case, is this frontal parietal network then still identical 00:26:45.700 --> 00:26:48.600 to what you might see in the control case? 00:26:49.100 --> 00:26:53.840 No, no. When you compare it to the wake state, what you see that you still get 00:26:53.840 --> 00:26:58.940 activations in the prefrontal cortex, but activations you never get is in the parietal cortex. 00:26:59.280 --> 00:27:04.600 So whatever condition you test, whatever analysis you do, all the time with 00:27:04.600 --> 00:27:08.780 popofol you block the activations of this global effect in the parietal cortex. 00:27:09.060 --> 00:27:12.860 You can get it in auditory system, you can get it in prefrontal regions, 00:27:13.240 --> 00:27:15.100 but never in the parietal cortex. 00:27:15.500 --> 00:27:22.000 Is that a significant observation in your mind, that it is in particular affecting the parietal cortex? 00:27:22.844 --> 00:27:27.304 Part of this frontal parietal network? I think it's quite significant because 00:27:27.304 --> 00:27:34.584 there is a lot of literature showing that you have a prefrontal parietal deconnection with anesthesia. 00:27:34.664 --> 00:27:39.364 There's a lot of work done by George Meshour in the US showing that you have 00:27:39.364 --> 00:27:44.364 really a deconnection with all the anesthetics, whatever anesthetics you use, like Propofol. 00:27:44.404 --> 00:27:47.984 He showed it with ketamine, with sebofluron on EEG data. 00:27:48.124 --> 00:27:51.304 And he shows that all the time you get a deconnection between the prefrontal 00:27:51.304 --> 00:27:53.784 cortex and the parietal. But you show a bit more, right? It's not disconnection, 00:27:53.784 --> 00:27:56.044 it's disappearance. Yeah, it's disappearance of activation. 00:27:56.884 --> 00:28:01.404 It's actually more extreme than that, right? Yeah, it's disappearance of activation. 00:28:02.704 --> 00:28:06.204 But that would suggest that it's often maybe something we can get back to later, 00:28:06.344 --> 00:28:09.764 that this parietal part of the frontal parietal is maybe much more of a hub 00:28:09.764 --> 00:28:11.864 in that system than the frontal part. 00:28:12.024 --> 00:28:14.144 Would you agree with that? Yeah. That's a working hypothesis? 00:28:14.524 --> 00:28:19.784 Yeah, yeah. Okay. So, okay. okay, so we have this differential effect of ketamine 00:28:19.784 --> 00:28:24.764 and propofol, and then you start to analyze in much more detail, 00:28:24.844 --> 00:28:28.944 which is really very interesting, the specific organization that responds. 00:28:29.284 --> 00:28:32.144 So I just started to look at the effect of connectivity, how is the effect of 00:28:32.144 --> 00:28:37.304 connectivity then changed due to propofol in the context of this task. 00:28:40.204 --> 00:28:46.724 So because then you could use that then again as also a marker for reinstating 00:28:46.724 --> 00:28:50.404 this excess consciousness, right? This is where we're going with that. 00:28:50.524 --> 00:28:53.724 So how useful was that analysis? 00:28:53.884 --> 00:28:57.504 We look at this effective connectivity and how it is then changed by these different 00:28:57.504 --> 00:28:58.904 drugs. What did you observe there? 00:29:04.810 --> 00:29:09.630 Interestingly, if you want to study anesthesia or even consciousness, 00:29:10.270 --> 00:29:14.610 there are basically two different ways. 00:29:14.770 --> 00:29:21.430 There's a way where you challenge the brain with a task, auditory stimuli. 00:29:21.470 --> 00:29:26.490 It's even easier in people with eyes shut down. 00:29:27.730 --> 00:29:31.250 As Lina mentioned people should 00:29:31.250 --> 00:29:35.110 really be aware of that under anesthesia obviously 00:29:35.110 --> 00:29:37.730 there is no shutdown of the brain there is 00:29:37.730 --> 00:29:42.910 still a strong brain processing of a lot of information a lot of information 00:29:42.910 --> 00:29:49.550 does it mean that there is a conscious access does it mean that there is a memorization 00:29:49.550 --> 00:29:57.630 of this data processing but there is at least for example with the sound we display and with fMRI, 00:29:57.770 --> 00:30:03.070 we see very strong activation of all the auditory pathway. 00:30:04.330 --> 00:30:08.970 Very strong. So I said a word to my colleagues, surgeons, be careful of what 00:30:08.970 --> 00:30:12.550 you are talking during operations because patient is listening. 00:30:14.770 --> 00:30:23.810 Well, now another way is to study the spontaneous fluctuations of brain activity, the so-called, 00:30:24.610 --> 00:30:31.530 resting state, a very, very extended field of neuroscience now, 00:30:31.850 --> 00:30:34.830 since the pioneer work of Marcus Reiko and St. Louis. 00:30:35.010 --> 00:30:41.550 And if you look at these spontaneous fluctuations, there have been work in re-anesthesia, 00:30:41.710 --> 00:30:48.690 but here we introduced a new way of analysis of these resting state networks 00:30:48.690 --> 00:30:51.270 called dynamic resting states. 00:30:51.430 --> 00:30:56.070 What does it mean? It means that people who are familiar with resting states 00:30:56.070 --> 00:31:02.330 know that you scan your subject or your animal for 10 or 20 minutes without a specific task, 00:31:02.550 --> 00:31:08.710 and then you analyze what we call functional correlations. 00:31:08.710 --> 00:31:14.190 We would prefer the world of functional correlations over functional connectivity 00:31:14.190 --> 00:31:16.310 because it's more objective, but still. 00:31:16.530 --> 00:31:22.070 And you will describe networks that we call, for example, default mode network, 00:31:22.370 --> 00:31:29.370 attention network that will pop out very easily through some code. And here. 00:31:30.371 --> 00:31:35.251 Here, you're dealing with the whole picture of your 10 or 20 minutes of acquisition 00:31:35.251 --> 00:31:41.691 at once, while we know that even the functional correlation between two areas 00:31:41.691 --> 00:31:44.331 fluctuates during all the session of acquisition, 00:31:44.611 --> 00:31:46.551 sometimes could be stable. 00:31:46.871 --> 00:31:54.391 So, here comes what we call dynamic resting state, because through a sliding 00:31:54.391 --> 00:31:58.471 window phenomena, it can also be done without the sliding window, 00:31:58.471 --> 00:32:01.651 you can cluster brain states. 00:32:02.471 --> 00:32:07.451 Here we used an unsupervised method called K-means. 00:32:08.431 --> 00:32:12.651 Cluster all this resting state data into several brain states. 00:32:13.591 --> 00:32:18.511 And these brain states prove to be extremely useful. What happens there? 00:32:18.751 --> 00:32:26.991 So we could see that if we, let's say there's seven brain states that explain 00:32:26.991 --> 00:32:28.871 your resting state configurations. 00:32:29.671 --> 00:32:36.451 And with the clustering of your resting state, you will do the same clustering 00:32:36.451 --> 00:32:40.031 in the awake state, then androanesthesia, then propofol, ketamine, 00:32:40.211 --> 00:32:44.051 whatever the anesthetic, and with different dosages also of these tracts. 00:32:45.411 --> 00:32:54.231 Suddenly, we saw that there are brain states that are very close to the anatomical 00:32:54.231 --> 00:32:55.691 connectivity of the brain. 00:32:56.971 --> 00:33:03.571 And those states, we will call them rigid states because at that moment when 00:33:03.571 --> 00:33:09.651 your brain is in this state for say a few seconds, your resting state is 100% 00:33:09.651 --> 00:33:11.571 explained by your connectomics. 00:33:12.391 --> 00:33:18.311 You have no freedom. You are jailed by your structure, your brain anatomy. 00:33:18.591 --> 00:33:22.751 So you have a high similarity between structure and function. 00:33:23.853 --> 00:33:31.053 And when it comes to the other part of the spectrum, we found a state that is 00:33:31.053 --> 00:33:31.973 completely the opposite. 00:33:32.173 --> 00:33:39.273 This is a configuration of resting state that is completely uncorrelated to the structure. 00:33:40.333 --> 00:33:45.553 And we called a flexible, it's a very high, highly flexible brain state. 00:33:45.813 --> 00:33:50.273 It means that at some point, your brain jumped from a state to another. other. 00:33:50.413 --> 00:33:54.953 And these states have completely different properties. 00:33:55.533 --> 00:34:03.033 Now, the nice finding we made in the group is that when you are awake, 00:34:03.353 --> 00:34:10.093 most of the configuration of your brain can be explained by many, 00:34:10.133 --> 00:34:16.293 many brain states and not only one with a heavy shift to these flexible states. 00:34:16.293 --> 00:34:20.693 Whereas when you go to endo-anesthesia propofol, 00:34:20.833 --> 00:34:23.913 ketamine even others you see 00:34:23.913 --> 00:34:26.893 the exact opposite it means that your resting state 00:34:26.893 --> 00:34:33.633 is explained mainly by the rigid state and it means that you have a strong influence 00:34:33.633 --> 00:34:40.873 of your connectomics to explain your spontaneous brain fluctuations it means 00:34:40.873 --> 00:34:44.173 that these spontaneous brain fluctuations are not so spontaneous, 00:34:44.493 --> 00:34:46.993 they are shaped heavily by the structure. 00:34:47.393 --> 00:34:55.333 So to summarize this description, let's say that we could say somehow that being 00:34:55.333 --> 00:34:59.313 conscious, meaning being completely free out of our brain structure. 00:34:59.833 --> 00:35:03.873 Right. But the thing is, if you go to the brain structure itself, 00:35:04.493 --> 00:35:07.633 that's an estimate of a DTI. Yes. Right? 00:35:08.313 --> 00:35:12.853 And DTI is relatively incomplete in sort of assessing the structure of the brain. 00:35:13.213 --> 00:35:24.293 Absolutely. That's why actually in our study, we used a database that is not a DTI-based database. 00:35:24.393 --> 00:35:31.413 This is COCOMAC that summarizes all the literature of tracing in macaques. 00:35:31.833 --> 00:35:35.053 And there's a strong literature of neuroanatomy. 00:35:35.233 --> 00:35:40.653 And so in our case, this structure matrix, we're coming from these studies. 00:35:41.759 --> 00:35:45.839 Okay, so then we have these seven dynamical states in which you could cluster 00:35:45.839 --> 00:35:49.799 the resting state activity that you measured in the macaque brain. 00:35:50.879 --> 00:35:55.419 Then we see that in the low entropy case, and you also showed that, 00:35:55.579 --> 00:36:01.199 let's say we're not conscious or we're close to sleep or we're on severe sedation. 00:36:02.519 --> 00:36:06.919 It's perfectly matched to the structure. and in the case of wakefulness, 00:36:07.179 --> 00:36:12.039 we have a much higher variability that cannot be related to the structure directly. 00:36:12.359 --> 00:36:16.919 But do you see then a very discrete transition in those seven states? 00:36:17.899 --> 00:36:23.239 Like, is there a sharp threshold? Like, if I'm mildly sedated, 00:36:23.279 --> 00:36:28.199 I'm immediately below that threshold somewhere or it's a more gradual transition? 00:36:29.719 --> 00:36:33.519 It's a very good question. I think for the moment with the data we have, 00:36:33.619 --> 00:36:38.819 we cannot answer the question because when we did these experiments, 00:36:39.119 --> 00:36:43.719 the animals were either in one state or another state. We didn't do transition studies. 00:36:44.019 --> 00:36:48.979 I think it's something very important. We had a lot of discussions to do future 00:36:48.979 --> 00:36:53.839 experiments with transition states, so to going from awake state, 00:36:54.179 --> 00:37:01.799 but very slowly to an anesthesia state, even during SED, which can last for several minutes. 00:37:01.939 --> 00:37:05.619 So to go from one state to the other, and also the other way around, 00:37:05.719 --> 00:37:10.959 to see what happens when you are in anesthesia and you wake up, 00:37:11.019 --> 00:37:17.519 but I think it's studies we have to do to really to know if it's something you 00:37:17.519 --> 00:37:22.999 go from one conscious state to the unconscious state until you lose some of 00:37:22.999 --> 00:37:27.839 these brain states we described or if there is a transition that could be slowly. 00:37:28.959 --> 00:37:34.419 Okay. But then they also highlighted the fact that you see very specific kinds 00:37:34.419 --> 00:37:41.499 of correlations disappear under sedation and then it seems to be dependent on 00:37:41.499 --> 00:37:43.079 the kind of drug you are using. 00:37:43.919 --> 00:37:47.859 Right, so we talk about the positive correlations or the negative correlations. 00:37:48.119 --> 00:37:52.179 So here we have this dysfunctional connectivity diagram. That means I look at 00:37:52.179 --> 00:37:55.139 the correlations across all the voxels that I'm looking at. 00:37:55.339 --> 00:38:00.439 So how is this then specifically affected by either ketamine or propofol? 00:38:01.539 --> 00:38:06.439 What we saw for this correlation study is with propofol and with ketamine, 00:38:07.079 --> 00:38:10.259 especially when you look at stationary connectivity, connectivity, 00:38:10.579 --> 00:38:17.299 but also when you look at the brain states that are still present in anesthesia, that you mostly use, 00:38:17.639 --> 00:38:22.559 that you have a loss of mostly the negative correlations in the brain. 00:38:22.779 --> 00:38:27.419 There are still a lot of positive correlations that are still present in the brain. 00:38:27.859 --> 00:38:32.439 They are less compared to awake state, but there are still many that are present 00:38:32.439 --> 00:38:35.839 and what you lose completely, at least for these two drugs, are the negative 00:38:35.839 --> 00:38:38.059 correlations. Do you find it surprising? 00:38:40.866 --> 00:38:46.966 Not completely, because there is a thing like one hypothesis that negative correlations 00:38:46.966 --> 00:38:49.846 could be important for information processing on the brain. 00:38:50.026 --> 00:38:54.606 And then when you're not conscious, you have no information processing, and then we lose them. 00:38:54.946 --> 00:38:59.986 So that's one idea. That's why I was not completely surprised to find this. 00:38:59.986 --> 00:39:04.686 But you could also argue that since you are manipulating the inhibitory system, 00:39:05.066 --> 00:39:09.326 right, that that might have a specific effect on your negative correlation because 00:39:09.326 --> 00:39:13.006 these are the guys who are sort of managing that part of the process. 00:39:13.326 --> 00:39:16.346 Yeah. Would that be reasonable as well or you think that doesn't make sense? 00:39:16.806 --> 00:39:20.206 No, that could be another explanation. I never thought I'd think of this one, 00:39:20.286 --> 00:39:22.566 but it could be a good explanation too. Okay. 00:39:23.186 --> 00:39:27.206 Let's try that. You tell me next time. But now the other thing is that then, 00:39:27.266 --> 00:39:33.026 which was really exciting because you are coming from, both of you are fully 00:39:33.026 --> 00:39:35.606 committed to the global neuronal workspace idea. 00:39:36.446 --> 00:39:41.886 And so you then started to look at the nodes, the frontal parietal nodes that 00:39:41.886 --> 00:39:44.206 would make up this global neuronal workspace. 00:39:44.466 --> 00:39:50.886 And you started to look at how they would be specifically affected by either propofol or ketamine. 00:39:50.886 --> 00:39:56.386 I mean, and then in some sense, surprisingly, the effect was sort of like indistinguishable, 00:39:56.466 --> 00:39:59.026 like coupling across the global 00:39:59.026 --> 00:40:02.766 workspace nodes disappeared in both cases sort of in a comparable way. 00:40:03.566 --> 00:40:06.246 This is a fair summary, I think, though. Yeah. 00:40:07.245 --> 00:40:14.505 So, if you want to link both findings, for example, with the task, 00:40:14.725 --> 00:40:17.545 with the local global task, and in the resting state, 00:40:17.945 --> 00:40:25.085 it's true that findings were more homogeneous with the resting state approach. 00:40:25.085 --> 00:40:32.125 So all anesthetics share the same signature, which is the loss of independence 00:40:32.125 --> 00:40:35.065 of resting state from brain structure. 00:40:35.405 --> 00:40:42.025 Whereas with the task, it was more subtle and more, there's a myriad of possibilities. 00:40:42.025 --> 00:40:46.785 But actually, there is an apparent paradox. 00:40:46.945 --> 00:40:55.705 And a way to match both is that the idea that the integrity of this global neural 00:40:55.705 --> 00:40:57.325 workspace is really key. 00:40:58.165 --> 00:41:02.145 If you take a piece of it, then the whole system falls down. 00:41:02.145 --> 00:41:07.785 And it looks like, I mean, there was always a mystery. 00:41:07.925 --> 00:41:11.605 How can anesthetics that have completely different molecular targets, 00:41:11.765 --> 00:41:18.085 different neuronal population targets, different pharmacology and post-receptor 00:41:18.085 --> 00:41:22.085 molecular events converge to the same result? 00:41:22.305 --> 00:41:26.065 I mean, it looks impossible in terms of pharmacology. And actually, 00:41:26.165 --> 00:41:33.405 one of the potential explanations could be that because they deprive the global 00:41:33.405 --> 00:41:36.665 neural workspace in a different manner, partly, 00:41:36.885 --> 00:41:41.905 completely from its information processing, 00:41:42.225 --> 00:41:48.485 but it's enough to disorganize part of it to lose completely the conscious access. 00:41:50.276 --> 00:41:56.296 Okay. But what I found surprising is that the effect is so non-specific, right? 00:41:56.356 --> 00:42:00.656 So here we have two pharmacological agents that have very different impact on neural circuits. 00:42:01.316 --> 00:42:05.176 And at this very macroscopic level in which we analyze the system, 00:42:05.356 --> 00:42:09.956 the impact is actually indistinguishable, which worries me a bit. 00:42:09.956 --> 00:42:14.956 Because from a global workspace perspective, it's an information processing 00:42:14.956 --> 00:42:19.016 view on the brain where you say, well, we have all these processors that are 00:42:19.016 --> 00:42:23.416 competing for access into the conscious buffer, if you want. 00:42:23.956 --> 00:42:32.096 And they start to broadcast into this buffer when they are sufficiently driven, right? 00:42:32.096 --> 00:42:38.216 But now, in some sense, here there's no signature really of such a buffer being 00:42:38.216 --> 00:42:41.796 maintained in any systematic way. It's sort of it's there, it's not there. 00:42:42.356 --> 00:42:50.016 It might be also, Paul, that the measure of resting state and its relation to 00:42:50.016 --> 00:42:53.876 structure is not really specific to the agent. 00:42:54.116 --> 00:42:59.896 It's just because you end up having a complete loss of consciousness. 00:42:59.896 --> 00:43:03.136 Consciousness is like a phenotypic biomarker, let's say. 00:43:03.956 --> 00:43:11.296 This is also a way to explain why when you do tasks, you have more subtle differences, 00:43:11.436 --> 00:43:16.476 and while when we do the resting state approach, it looks all the same. 00:43:16.976 --> 00:43:21.076 This is actually an important point, right? Because for instance, 00:43:21.136 --> 00:43:23.236 in other studies of consciousness in humans, 00:43:23.576 --> 00:43:29.336 you might see very distinct kinds of subnetworks for consciousness that's more 00:43:29.336 --> 00:43:33.796 metacognitive oriented towards the self and conscious states that are more oriented 00:43:33.796 --> 00:43:36.076 towards the external world, right? 00:43:36.216 --> 00:43:40.896 So if you have actually the kind of paradigm you're using here, 00:43:41.076 --> 00:43:42.316 none of that is being controlled. 00:43:43.533 --> 00:43:47.833 So maybe you're not driving the system efficiently enough to actually see anything 00:43:47.833 --> 00:43:53.493 that is really specifically impacting that excess consciousness that you're after. 00:43:53.873 --> 00:44:01.473 Because again, something I like to say is that consciousness is an ambiguous word also. 00:44:01.693 --> 00:44:09.753 So you mentioned metacognition and self-monitoring, which is not necessarily 00:44:09.753 --> 00:44:11.473 what Local Global is probing. 00:44:11.473 --> 00:44:20.733 This is again another view of consciousness of self consciousness even though 00:44:20.733 --> 00:44:23.893 maybe it didn't appear like that but I was trying to support your point, 00:44:24.693 --> 00:44:31.713 by basically I would not be surprised if this sort of resting state paradigm is not helping you, 00:44:32.593 --> 00:44:36.793 it's more pulling you away from where you want to be because there's no specific 00:44:36.793 --> 00:44:42.593 the brain is not configured towards a certain goal so do you also see that as 00:44:42.593 --> 00:44:47.933 a challenge I agree and actually what goes also in this direction is that this 00:44:47.933 --> 00:44:50.573 signature with resting state similarity, 00:44:51.493 --> 00:44:57.413 is now explored by other groups for other circumstances with loss of consciousness 00:44:57.413 --> 00:45:03.873 the group of Oxford who sleep for example and at ICM with. 00:45:05.333 --> 00:45:11.073 People with disorders of consciousness consciousness, and they are very exciting 00:45:11.073 --> 00:45:16.053 data, finding again this same signature. 00:45:16.393 --> 00:45:20.113 And here we are beyond different pharmacological agents. 00:45:20.253 --> 00:45:25.773 We have different conditions, different even brain anatomy, sometimes with lesions. Right. Okay. 00:45:27.671 --> 00:45:31.591 So to close this part up, and then we go towards the thalamic cortical system, 00:45:31.891 --> 00:45:35.771 the other thing that I find really interesting about the data you present, 00:45:35.971 --> 00:45:42.651 because it's so specific, is that since you're also in the global workspace tribe, 00:45:43.011 --> 00:45:49.651 in some sense, your data shows that we should take global a bit with a grain of salt, right? 00:45:49.691 --> 00:45:54.791 Because like in the macaque brain, we talk about 12 identified nodes, more or less, right? 00:45:54.791 --> 00:46:00.171 So how global, in your opinion, should we really think about this global neuronal 00:46:00.171 --> 00:46:05.751 workspace in the context of the task and the kind of manipulations that you have been using? 00:46:07.011 --> 00:46:16.231 The term global actually refers also to the global broadcasting of information. 00:46:16.651 --> 00:46:22.231 And actually there is no single paradigm that activates all the global neuronal workspace. 00:46:22.231 --> 00:46:29.311 So, obviously, if you are assessing, for example, an auditory paradigm, 00:46:29.771 --> 00:46:33.971 you don't see visual activations and vice versa. 00:46:33.971 --> 00:46:45.111 So, the GNW theory should not be seen as just a clear anatomical network, 00:46:45.511 --> 00:46:56.431 as DMN, for example, or so forth, but as a general idea of cortical broadcasting of an information. 00:46:56.431 --> 00:47:03.351 And the fact that here the auditory information went much beyond when there was a global violation. 00:47:04.455 --> 00:47:08.955 And went to prefrontal areas close to frontal eye fields, for example, 00:47:08.955 --> 00:47:17.955 to intraparietal sulcus with area VIP, shows this second level broadcasting of information. 00:47:18.295 --> 00:47:26.635 But it doesn't necessarily tell you that we have X nodes or Y nodes in specific species. 00:47:26.915 --> 00:47:31.195 This is the way I see it. Well, but I understand that you could argue that the 00:47:31.195 --> 00:47:35.195 global workspace in the end is talking about broadcasting it to some sort of buffer. 00:47:35.275 --> 00:47:38.315 You broadcast it to some sort of memory system, right? 00:47:38.415 --> 00:47:42.195 And then the question is, is that memory system what you now actually reveal 00:47:42.195 --> 00:47:47.695 in these paradigms, frontal parietal, as a memory buffer in which these broadcasts land? 00:47:48.635 --> 00:47:52.355 Or is this frontal parietal system you visualize, let's say, 00:47:52.355 --> 00:47:57.595 an auditory specialization of a much larger kind of buffer in which you can 00:47:57.595 --> 00:47:58.575 potentially broadcast? 00:47:59.275 --> 00:48:04.675 It's really impossible to answer definitely this question. But if you would 00:48:04.675 --> 00:48:05.795 speculate, what's your hypothesis? 00:48:07.235 --> 00:48:17.135 If I speculate, I would say that the apparent homology we see between the macaque response, 00:48:17.615 --> 00:48:21.075 the macaque cortical response, and the human cortical response, 00:48:21.075 --> 00:48:27.755 Let me know that we are really having a description of, between brackets, 00:48:28.035 --> 00:48:29.855 a macaque global general workspace. 00:48:30.715 --> 00:48:38.615 Okay. So we have a starting point now to start to look at the details of the 00:48:38.615 --> 00:48:40.475 mechanisms of anesthesia, right? 00:48:40.675 --> 00:48:44.235 So this local global paradigm has helped you in doing that. You revealed it 00:48:44.235 --> 00:48:45.995 now in the macaque. So now we have an animal model. 00:48:46.195 --> 00:48:50.335 So now we can become more specific in manipulating it. That's also what you 00:48:50.335 --> 00:48:53.575 do with your deep brain stimulation, right? 00:48:53.635 --> 00:48:57.655 So why do you believe deep brain stimulation might help you to then manipulate 00:48:57.655 --> 00:49:00.195 these states of excess consciousness? 00:49:01.980 --> 00:49:09.440 So, just let me say a little bit about the story and why we got into there. 00:49:09.940 --> 00:49:16.960 From a personal perspective, I'm a neurosurgeon and when I started my residency 00:49:16.960 --> 00:49:19.300 neurosurgery end of the 90s, 00:49:19.300 --> 00:49:27.120 early 2000, I was surprised by a category of patients that have a strong, 00:49:27.340 --> 00:49:32.160 big trauma or catastrophic stroke. 00:49:33.340 --> 00:49:38.380 Many of them die, unfortunately, but some of them, and more and more, 00:49:38.440 --> 00:49:45.120 thanks to modern medicine, we are making these patients living. 00:49:45.120 --> 00:49:48.560 But unfortunately, many of them are in very bad conditions. 00:49:48.960 --> 00:49:54.000 And here, it's not about motor recovery, but actually it's about consciousness recovery. 00:49:55.580 --> 00:50:00.720 It's an increasing issue in our modern world. Think about it. 00:50:01.080 --> 00:50:07.360 Fifty years ago, there were no such patients. It's like almost an invention of modern medicine. 00:50:07.860 --> 00:50:13.940 Because these patients, until the 50s, with the appearance of the modern intensive 00:50:13.940 --> 00:50:19.400 care, and surgical techniques, died in 100% of the cases. 00:50:19.740 --> 00:50:29.300 I mean, the stroke was so strong, the trauma was so ugly and dangerous, all of them died. 00:50:29.500 --> 00:50:35.760 So modern medicine is producing a new medical condition, which is a vegetative 00:50:35.760 --> 00:50:39.440 state and or minimal conscious state. 00:50:39.440 --> 00:50:50.400 And it's really intriguing, because once the acute phase is done and they survive, 00:50:50.780 --> 00:50:52.780 they start opening eyes. 00:50:53.800 --> 00:50:58.280 Family are happy, but unfortunately it doesn't mean they are conscious, they are aware. 00:50:58.960 --> 00:51:04.420 They are awake, they have wakefulness, but not necessarily aware. 00:51:05.311 --> 00:51:10.411 And I wasn't tried by these patients because they could stay weeks, 00:51:10.531 --> 00:51:11.551 months in the department. 00:51:11.751 --> 00:51:18.071 Of course, rehab centers don't take them. And now they start to have specialized 00:51:18.071 --> 00:51:19.711 centers for these patients. 00:51:20.031 --> 00:51:23.591 So there is an issue. There is an issue. And at the same time, 00:51:23.631 --> 00:51:25.231 there are famous personalities. 00:51:25.311 --> 00:51:31.871 We think about Aaron Sharon, who passed away a few years ago. 00:51:32.211 --> 00:51:34.011 Currently, Michael Schumacher. 00:51:35.311 --> 00:51:42.011 So in the headlines, we see more and more these conditions of what we call disorders of consciousness. 00:51:43.071 --> 00:51:51.071 In the same time, deep brain stimulation, which is a surgical technique that 00:51:51.071 --> 00:51:53.811 consists of implanting electrodes inside the brain, 00:51:54.571 --> 00:51:59.791 became more and more fashionable and more and more effective in another disease 00:51:59.791 --> 00:52:01.131 called Parkinson's disease. 00:52:01.131 --> 00:52:03.911 And thanks to the work of Prof. 00:52:04.091 --> 00:52:09.931 Annemarie Benhabide in Grenoble, this makes it very popular and now there's 00:52:09.931 --> 00:52:17.431 something like 150,000 people in the world who are walking around you with these 00:52:17.431 --> 00:52:19.711 electrodes with Parkinson's disease or tremor. 00:52:19.711 --> 00:52:28.091 So, because this revealed to be a powerful mean to recover from Parkinson's, one of the ideas was, 00:52:28.211 --> 00:52:38.471 could we actually identify brain targets to do brain simulation to recover from consciousness loss? 00:52:38.471 --> 00:52:44.451 Of course, it's much more challenging, but because consciousness now becomes 00:52:44.451 --> 00:52:49.531 a neuroscience issue and not only a philosophical question, 00:52:49.931 --> 00:52:56.871 and now it's in the lab, in many, many labs in the world, it starts to be considered 00:52:56.871 --> 00:52:58.871 as a function, like motor function, 00:52:59.251 --> 00:53:04.851 like language, and because the patient lost this function, it might be that 00:53:04.851 --> 00:53:06.851 there are techniques to recover from that. 00:53:08.471 --> 00:53:14.191 This is general context. And actually, there are already some few reports in 00:53:14.191 --> 00:53:19.611 literature back to some decades ago with some groups in the world who tried 00:53:19.611 --> 00:53:24.331 to do thalamic stimulation to make vegetative patients recover. 00:53:24.691 --> 00:53:26.651 Why thalamic? Actually... 00:53:27.933 --> 00:53:35.393 Thalamus have very strong connections, very widespread, to large-scale cortical networks. 00:53:36.073 --> 00:53:42.973 Frontal, prefrontal, parietal, so, singular cortex, precuneus, 00:53:43.193 --> 00:53:48.973 all these key areas for conscious treatment. 00:53:50.493 --> 00:53:55.673 So, the idea was there. And we know also that thalamus is highly involved because 00:53:55.673 --> 00:54:00.713 some strokes of the thalamus can, even small strokes inside the thalamus, 00:54:00.813 --> 00:54:05.353 can make consciousness disappear without any other neurological problems. 00:54:06.573 --> 00:54:14.433 That's how there's a convergence toward these DBS of thalamus to try to restore consciousness. 00:54:14.933 --> 00:54:21.093 So then, how successful have you been so far in restoring consciousness with 00:54:21.093 --> 00:54:21.913 deep brain stimulation? 00:54:23.153 --> 00:54:27.433 In our lab, we are doing preclinical work. 00:54:27.733 --> 00:54:34.453 We are not doing a clinical trial, but we are using exactly the model that has 00:54:34.453 --> 00:54:37.733 been developed by Lynn using the anesthesia. 00:54:37.913 --> 00:54:45.153 Of course, anesthesia is not the perfect model of a vegetative state or a minimal 00:54:45.153 --> 00:54:46.913 conscious state after stroke, 00:54:47.253 --> 00:54:57.953 but this is still a model where now we control very well, including with behavior, with fMRI, with EEG, 00:54:58.253 --> 00:55:05.373 the level of conscious access and conscious state with the local global paradigm on top of it. 00:55:05.453 --> 00:55:14.153 So we started from these models in the mechanics and we inserted 00:55:14.433 --> 00:55:20.033 deep electrodes within the thalamus with several areas actually of the thalamus 00:55:20.033 --> 00:55:22.513 to test the specificity, which part of the thalamus. 00:55:23.388 --> 00:55:28.688 And to our big surprise, actually, we found that the manipulation, 00:55:29.008 --> 00:55:35.488 the electrical stimulation of antralaminar thalamic nuclei could suddenly induce 00:55:35.488 --> 00:55:40.068 very strong phenomena, a very strong reaction in the monkey. 00:55:40.288 --> 00:55:46.868 It means that the monkey who was deeply sedated, really with deep general anesthesia, 00:55:46.868 --> 00:55:53.388 starts having twitches with sometimes strong movement that were not seizures 00:55:53.388 --> 00:55:55.048 because we controlled with EEG. 00:55:55.788 --> 00:56:01.128 They start to have a spontaneous breathing reaction, increased heart rate, 00:56:01.208 --> 00:56:08.008 as if they were awakened from the anesthesia while they still receive in their veins a lot of propofol. 00:56:08.808 --> 00:56:13.468 But how can we be sure that you are now not in the same position as the family 00:56:13.468 --> 00:56:18.028 members of the coma patient that opened the eyes? Excellent question. 00:56:18.368 --> 00:56:24.288 So, this is the advantage of the lab also, is to control the experiment and 00:56:24.288 --> 00:56:27.168 to add experimental tools. 00:56:27.648 --> 00:56:32.508 There are several ways. So first, we have an EEG monitoring, 00:56:32.808 --> 00:56:42.188 and we see clearly that we enhance cortical activity with dynamic DBS to a level 00:56:42.188 --> 00:56:52.068 that starts to resemble to wakefulness of the same macaque when he's completely awake and conscious. 00:56:52.348 --> 00:56:57.148 This is not enough. And here comes again our local global paradigm. 00:56:58.288 --> 00:57:03.428 That is, as you could guess, our favorite paradigm, 00:57:03.608 --> 00:57:09.468 because the global paradigm is a very good way to interrogate the conscious 00:57:09.468 --> 00:57:15.028 access of the animal during this awakening by DBS. 00:57:16.108 --> 00:57:22.728 So we performed local global paradigm in this period of time when the animal starts to be, 00:57:24.162 --> 00:57:30.402 I cannot tell all the results because this is ongoing work, but I can say that's 00:57:30.402 --> 00:57:32.122 very encouraging results there. 00:57:32.342 --> 00:57:38.102 Right. Of course, in that sense, it's fantastic how systematic you have been in building this up. 00:57:38.142 --> 00:57:40.462 This is really taking a lot of discipline and effort. 00:57:40.542 --> 00:57:44.142 This is really not easy, and I respect that a lot. But one thing that you also 00:57:44.142 --> 00:57:49.562 shared with us this morning was that you also found an explanation of why the 00:57:49.562 --> 00:57:51.682 stimulation intensity has to be so high. 00:57:51.762 --> 00:57:58.762 Because interestingly enough, you could sort of only induce this sort of reawakening, 00:57:58.922 --> 00:58:03.822 these reawakening signatures, but high levels of stimulation of 100 hertz and 00:58:03.822 --> 00:58:06.002 above and not with lower frequencies. 00:58:07.062 --> 00:58:11.302 So, in the interlaminar nucleus or close by in the thalamus, 00:58:11.362 --> 00:58:15.782 you must be stimulating at a huge intensity to trigger the system to start to 00:58:15.782 --> 00:58:17.562 reawaken. Why is that? Yes. 00:58:18.102 --> 00:58:22.902 So, in term of frequency, even if movement disorders, with Parkinson's, 00:58:22.902 --> 00:58:24.902 you need to be higher than 100 Hz. 00:58:25.782 --> 00:58:34.922 Then comes the intensity issue. The intensity actually will play with the activated tissue volume. 00:58:34.922 --> 00:58:41.422 So, we can now model that and have an idea about which nuclei of the thalamus 00:58:41.422 --> 00:58:43.102 are involved with this stimulation. 00:58:43.642 --> 00:58:46.802 Very importantly, we control this condition. 00:58:46.942 --> 00:58:54.382 It means that with the same setting and programming parameters, 00:58:54.762 --> 00:58:57.362 with the same high intensity and high frequency, 00:58:57.802 --> 00:59:02.462 we manipulated areas of the brain that are really close on the same electrode, 00:59:02.462 --> 00:59:12.042 actually, the same condition with absolutely no behavioral manifestation and no EEG changes. 00:59:12.402 --> 00:59:15.582 That is really key in these kind of experiments. 00:59:15.942 --> 00:59:21.542 But how do you then explain that we need these huge intensities? 00:59:21.782 --> 00:59:27.162 So why don't you get this kind of reawakening at lower intensities of stimulation? 00:59:28.562 --> 00:59:32.902 So in terms of frequency, for example, this is actually a general question. 00:59:33.002 --> 00:59:37.702 It's still not solved, even for Parkinson's patients who receive the same therapy, 00:59:37.842 --> 00:59:41.522 DBS, every day in many places in the world. 00:59:41.622 --> 00:59:46.662 We did some pretty many work in our models, not the anesthesia models, 00:59:46.902 --> 00:59:55.102 but in another program, and we could see that if we compare high-frequency versus low-frequency... 00:59:56.201 --> 01:00:03.841 Using fMRI again, which is also a favorite tool, only higher frequency could 01:00:03.841 --> 01:00:05.441 activate cortical areas. 01:00:06.661 --> 01:00:11.821 And now in the field of DBS, there are a lot of arguments and a lot of evidence 01:00:11.821 --> 01:00:18.041 that explain that most of DBS effect is not deep, it's superficial. 01:00:18.381 --> 01:00:23.241 It's large cortical networks activations. And it looks like you need higher 01:00:23.241 --> 01:00:31.361 frequency so that this retrograde solicitation of cortical areas happens with DBS. 01:00:31.721 --> 01:00:36.821 And that might explain why interlaminar would be particularly well-placed to do that. Absolutely. 01:00:37.221 --> 01:00:43.541 Okay. But on the other hand, you could also argue that what you see are just, 01:00:43.661 --> 01:00:47.541 let's say, more the brainstem systems reacting to the stimulation, 01:00:48.301 --> 01:00:52.181 and the signal travels downward towards midbrain and brainstem, 01:00:52.181 --> 01:00:55.961 And indeed, what you're looking at is a monkey that is comparable to that coma 01:00:55.961 --> 01:00:57.941 patient. So how can you exclude that interpretation? 01:00:59.161 --> 01:01:02.721 Actually, it might very be that it also happens. 01:01:03.021 --> 01:01:11.001 However, we have evidence of cortical strong changes in cortex through EEG and through fMRI. 01:01:11.001 --> 01:01:20.601 So it would even be the best scenario that you have both reticular activating 01:01:20.601 --> 01:01:25.481 system modulation and cortical modulation. Okay. 01:01:25.721 --> 01:01:32.401 But still the variability that you see under propofol and deep brain stimulation in a macaque, 01:01:32.541 --> 01:01:38.781 the kind of effective connectivity in cortex looks a lot more regular than in 01:01:38.781 --> 01:01:41.101 the awake state. It's not really identical. 01:01:42.261 --> 01:01:47.161 Yes, because actually I showed very preliminary data. Okay. 01:01:47.461 --> 01:01:54.921 The work is ongoing and we have also another animal. 01:01:55.061 --> 01:01:58.881 We have much more acquired data now that is being analyzed. 01:01:59.601 --> 01:02:04.161 It's extremely interesting because maybe by driving this interlibrary system, 01:02:04.461 --> 01:02:09.781 you are reawakening cortex, but still in a more, let's say, structure-dependent way. 01:02:09.781 --> 01:02:13.521 And that's why I have these much more restricted islands of activation that 01:02:13.521 --> 01:02:17.381 you showed in your functional connectivity than the more broad connectivity 01:02:17.381 --> 01:02:22.041 you might observe under normal conditions. You mean in non-specific way? 01:02:22.481 --> 01:02:26.361 Yeah, exactly. Actually. In the resting state. Absolutely. And we do see that. 01:02:26.461 --> 01:02:30.861 If we do just resting state without any behavioral consequence, 01:02:31.341 --> 01:02:36.921 resting state on minus resting state off, we see cortical activations. 01:02:36.921 --> 01:02:40.161 However, here, what we could see is. 01:02:41.006 --> 01:02:48.866 The controlled cortical activation for the global effect versus the global deviant. 01:02:49.006 --> 01:02:53.346 So the global deviant versus the global standard. 01:02:53.666 --> 01:02:58.726 So it's double controlled because you have within the paradigm global deviant 01:02:58.726 --> 01:03:00.566 versus global standards. 01:03:00.866 --> 01:03:08.786 And we control also, we have two conditions of stimulation at different intensities, 01:03:08.786 --> 01:03:12.506 both of them inducing cortical activations. 01:03:12.606 --> 01:03:20.606 And it shows a real specific difference for the DBS condition as compared to the other one. 01:03:20.826 --> 01:03:25.346 Okay. Fantastic. Lynn, do you think that we can ever get away from using drugs 01:03:25.346 --> 01:03:27.146 to get people anesthetized? 01:03:28.006 --> 01:03:33.366 I hope so that one day we will not use drugs to anesthetize people because it 01:03:33.366 --> 01:03:38.346 would be much easier because then you don't need to intubate your patients to 01:03:38.346 --> 01:03:42.926 have mechanical ventilation or to have side effects on the hemodynamic part. 01:03:43.286 --> 01:03:46.286 So I'd hope that it could work one day, 01:03:46.406 --> 01:03:53.486 perhaps not with DBS because we will not implant DBS systems to the patients 01:03:53.486 --> 01:03:59.506 who go to surgery for another reason, 01:03:59.606 --> 01:04:03.406 but perhaps like systems with, I don't know, 01:04:03.566 --> 01:04:07.446 perhaps TDCS or so on, perhaps with specific parameters. But what do you think 01:04:07.446 --> 01:04:08.946 is a realistic target here? 01:04:08.986 --> 01:04:12.406 Let's say 50 years from now, let's take a reasonable time window. 01:04:14.346 --> 01:04:19.766 I don't know. For the time window, I have no idea. Okay, but what do you see as the next big step? 01:04:20.666 --> 01:04:28.166 The next big step, I think it's already understanding the consciousness better 01:04:28.166 --> 01:04:31.606 as we do now because there's still a lot of things we don't know about, 01:04:31.646 --> 01:04:33.386 about consciousness and consciousness processing. 01:04:33.386 --> 01:04:40.646 For the anesthesia part, what could be reasonable is to do all the analyzes 01:04:40.646 --> 01:04:45.806 we did also on EEG because we will not put fMRI in the operating room. 01:04:46.166 --> 01:04:51.226 And then to get analyzes to get like a closed-loop anesthesia with anesthetic 01:04:51.226 --> 01:04:52.426 drugs that are completely, 01:04:52.826 --> 01:05:01.366 that the EEG pattern or the ERP patterns control the anesthetics drug delivery 01:05:01.366 --> 01:05:02.826 and so in a closed-loop manner. 01:05:03.386 --> 01:05:07.246 I think that it could be reasonable in a few years. There are already devices 01:05:07.246 --> 01:05:14.246 on the market who can do it based on the spectral index, but perhaps like a 01:05:14.246 --> 01:05:16.126 better organization of this. 01:05:16.386 --> 01:05:21.966 Okay. So, Bixi, one problem I'm still having, and then we're going to the finish line. 01:05:23.326 --> 01:05:27.466 So, okay, we start with proper fault ketamine. We know what it does. 01:05:28.446 --> 01:05:32.886 We're manipulating the GABAergic system, and the GABAergic system is actually 01:05:32.886 --> 01:05:35.006 pretty fast. The time constant is there in milliseconds. 01:05:36.919 --> 01:05:40.019 And now we try to make inferences about what this does to consciousness, 01:05:40.219 --> 01:05:46.679 also at the mechanistic level, using a measurement technique that has time constants of seconds. 01:05:46.939 --> 01:05:50.039 So we're order of magnitudes away from it. 01:05:50.699 --> 01:05:55.519 So isn't this possibly a problem? 01:05:55.599 --> 01:06:00.039 And also with the animal model you have, you could use other techniques that 01:06:00.039 --> 01:06:03.159 bring you closer to the time constants of the systems being manipulated. 01:06:03.159 --> 01:06:08.979 So first, do you see this as an obstacle that you use these techniques that 01:06:08.979 --> 01:06:12.079 are so slow that maybe you have 01:06:12.079 --> 01:06:16.139 no idea what's going on really at the mechanistic level in this brain? 01:06:16.339 --> 01:06:19.599 So how do you square that circle? 01:06:19.959 --> 01:06:26.379 It's true that fMRI scans the brain every two seconds, and that's obviously a different timescale. 01:06:27.979 --> 01:06:31.339 I firmly believe like many 01:06:31.339 --> 01:06:34.539 neuroscientists that there is no technique in neuroscience that 01:06:34.539 --> 01:06:37.719 can solve all the problems and only multiple approaches 01:06:37.719 --> 01:06:41.199 are there to move forward 01:06:41.199 --> 01:06:44.559 so in our case we also have 01:06:44.559 --> 01:06:47.939 EEG monitoring of these 01:06:47.939 --> 01:06:57.199 animals which have a quicker life time scale but definitely our objective is 01:06:57.199 --> 01:07:04.659 to go further and to have access to the intracranial recording with the, 01:07:05.359 --> 01:07:08.559 challenge to have multi-site recording simultaneously. 01:07:09.299 --> 01:07:15.579 And that's one of our next moves. Okay, right. 01:07:16.479 --> 01:07:23.279 So you leave the option open that maybe everything we discussed so far is irrelevant 01:07:23.279 --> 01:07:26.519 because of dynamics that we have to understand is faster. 01:07:27.299 --> 01:07:34.479 I don't think so. Because take, for example, the intracranial electrophysiology. 01:07:36.459 --> 01:07:41.979 If you combine fMRI and electrophysiology, if you know, for example, 01:07:42.059 --> 01:07:47.199 through fMRI, the most relevant science to record, then this is... 01:07:48.001 --> 01:07:53.741 Clearly the way to go to O4. Otherwise, you would not put thousands of electrodes 01:07:53.741 --> 01:07:56.801 on the mechanic brain to have access to the whole brain. 01:07:56.961 --> 01:08:05.281 And second, the dynamic resting state is now being modeled at the multi-second level. 01:08:05.721 --> 01:08:09.781 People, groups like Gustavo Deco are doing that. 01:08:09.981 --> 01:08:15.661 Of course, it's a model that is put on the fMRI signal, No, but it's there, 01:08:15.781 --> 01:08:18.581 and there are some interesting results. 01:08:19.241 --> 01:08:24.201 Okay. But, of course, it's also a matter of how good the predictions are of 01:08:24.201 --> 01:08:27.641 these models and how well they can be validated and so on. Absolutely. 01:08:28.081 --> 01:08:31.861 But it's an incremental step. This is where we… Absolutely. But I believe firmly 01:08:31.861 --> 01:08:40.441 in these models, providing that there is a mutual exchange with biological data. 01:08:41.721 --> 01:08:48.701 That's one of, let's say, my dreams, is that we are able to model, 01:08:48.841 --> 01:08:54.581 for example, the consequence of DBS almost whatever the site in the brain. 01:08:55.301 --> 01:08:59.361 Look, DBS is now expanding for psychiatry disorders, 01:08:59.781 --> 01:09:04.521 for consciousness disorders, on top of Parkinson's and other diseases, 01:09:04.901 --> 01:09:11.201 eating disorders, and the current paradigm in medicine in neurosurgery is mainly 01:09:11.201 --> 01:09:13.621 to rely on single observations, 01:09:13.821 --> 01:09:20.001 because one day there was a stroke here, there is a stimulation there, in unwanted nuclei. 01:09:20.721 --> 01:09:25.781 So we move almost by chance. And the idea is to change completely the paradigm, 01:09:25.941 --> 01:09:31.641 is to say, well, if I have a kind of flight simulator of the brain for DBS, 01:09:31.941 --> 01:09:39.061 then this model, if it is well established, and I believe we can achieve that, 01:09:39.361 --> 01:09:47.241 Then, I start defining and rationalize and suggest targets for diseases in a 01:09:47.241 --> 01:09:48.581 completely rationalized manner. 01:09:48.801 --> 01:09:52.821 And here, artificial intelligence, of course, will play a big role because it 01:09:52.821 --> 01:09:56.821 will really can say and suggest, 01:09:57.061 --> 01:10:01.301 well, for this disease, with all what we know about brain imaging in this disease, 01:10:01.541 --> 01:10:04.341 very clearly these two targets are key. 01:10:04.341 --> 01:10:08.641 Or this one is probably the most efficient or the most convenient. 01:10:09.081 --> 01:10:16.781 So I see it as a main way and a shift in paradigm in terms of normal duration 01:10:16.781 --> 01:10:18.221 and neurosurgery in the future. 01:10:18.521 --> 01:10:27.221 Right. Very good. So, Lynn, you are now in this field also as a physician, right? 01:10:27.381 --> 01:10:29.301 On the one hand, you have these concerns about your patients, 01:10:29.461 --> 01:10:32.661 then you understand these animal models, you look at the mechanisms of anesthesia. 01:10:33.521 --> 01:10:37.821 What do you see as Lynn's law that we should follow to understand the brain? 01:10:40.530 --> 01:10:43.650 That's a very good question. Don't look at Bashir. Yeah, that's a very good question. 01:10:44.590 --> 01:10:51.250 I'm not sure that I have a law to follow. At least not for the moment. Perhaps one day. 01:10:53.290 --> 01:10:57.690 I don't know. It's a very tough question. So I think I have to… Okay, 01:10:57.730 --> 01:10:59.550 yet a bit more time. Bashir, what's Bashir's law? 01:11:01.330 --> 01:11:06.770 So again, for anesthesia, I think this field… For the study of the brain in general, right? 01:11:07.630 --> 01:11:10.050 Yes. We're going macroscopic now. Yes. 01:11:11.890 --> 01:11:20.490 Something I find interesting from a philosophical point of view is the finding 01:11:20.490 --> 01:11:27.830 that consciousness emerges when resting state is freed from the structure. 01:11:28.910 --> 01:11:34.510 Think about we are building neural networks, machines, robots. 01:11:34.510 --> 01:11:42.110 Robots, we may start inducing some resting state activity in these circuits. 01:11:42.750 --> 01:11:50.450 What would happen if we start inducing ship configurations that are completely 01:11:50.450 --> 01:11:56.230 independent from the structure of the microprocessors and the transistors? 01:11:56.410 --> 01:12:03.270 Would we, at least at some time, assist at the emergence of an artificial consciousness? 01:12:04.910 --> 01:12:06.910 That's a law? You see this as a law? 01:12:08.630 --> 01:12:16.610 It sounds like a prediction. I don't know, but this would be a huge change, 01:12:16.770 --> 01:12:19.550 because if machines start to acquire consciousness, 01:12:20.730 --> 01:12:24.190 then there is a big shift, I think, in that case. Absolutely. 01:12:24.450 --> 01:12:26.730 Does it worry you when machines… No. Okay. 01:12:27.510 --> 01:12:30.930 But that means what you're talking about is some sort of virtualization, right? Absolutely. 01:12:32.030 --> 01:12:36.290 It's a virtualized system. Absolutely. virtual machine absolutely okay so then 01:12:36.290 --> 01:12:42.270 lynn um now that we don't have lynn's law which is a real failure so we should 01:12:42.270 --> 01:12:47.130 work on that um now five years from now i'm going to come to paris, 01:12:47.870 --> 01:12:54.550 i'm going to visit you at neurospin and um i want to see whether a prediction 01:12:54.550 --> 01:12:59.870 you make today was falsified or verified five years from now so what's the one prediction 01:13:00.270 --> 01:13:05.050 that you feel is most important to see tested in that time frame. 01:13:06.420 --> 01:13:11.620 I think what we want at least to test in the five years or the few years that 01:13:11.620 --> 01:13:18.100 will come is to see how the feed-forward and the feedback works under anesthesia. 01:13:18.160 --> 01:13:24.940 Because if you use all the fMRI studies, you can make prediction about feedback 01:13:24.940 --> 01:13:28.420 feed-forward, but it doesn't give you a real answer. 01:13:28.580 --> 01:13:33.000 So the only thing that you can do is during invasive electrophysiological studies 01:13:33.000 --> 01:13:35.820 to verify is that you have really, 01:13:36.480 --> 01:13:41.280 that feed forward is present on anesthesia as to what the assumption is today 01:13:41.280 --> 01:13:45.700 that feedback is completely blocked so I hope that in the next five years we 01:13:45.700 --> 01:13:49.500 can get answers on this Great, very good. And Bashir, what's your prediction? 01:13:50.240 --> 01:13:53.800 Because I'm going to find you in the same room five years from now Pleasure 01:13:53.800 --> 01:13:58.020 But then with another prediction What is it? 01:13:58.180 --> 01:14:05.040 Prediction is the following I'm not sure we'll keep Thalamic DPS as the ultimate 01:14:05.040 --> 01:14:06.620 goal of consciousness restoration. 01:14:07.120 --> 01:14:11.900 I believe the computational model would predict where should we stimulate. 01:14:12.140 --> 01:14:17.480 It would be deep, it would be superficial, to be combined with non-invasive 01:14:17.480 --> 01:14:24.200 and invasive, and being able to build on this technology for consciousness restoration 01:14:24.200 --> 01:14:26.620 based on the computational model. 01:14:26.860 --> 01:14:33.020 Okay, very good. Well, then Ulrich, and Rochelle Yarriott, thank you very much for this conversation. 01:14:33.240 --> 01:14:39.040 Thank you. Thank you very much. The CSN podcast was produced by the Convergent 01:14:39.040 --> 01:14:45.160 Science Network of Biometrics and Biohybrid Systems, a project funded by the 01:14:45.160 --> 01:14:47.820 European 7th Research Framework Program. 01:14:49.480 --> 01:14:54.660 For more interviews, recorded lectures, or upcoming conferences in the field 01:14:54.660 --> 01:15:00.600 of biometrics and biohybrid systems, go to csnnetwork.com. 01:15:01.040 --> 01:15:09.360 Music.