Listeners:

Top listeners:

00:00 00:00

Go to album

Paul Verschure on consciousness and distributed adaptive control CSN Podcasts
Edvard Moser on grid cells and entorhinal cortex CSN Podcasts
Giacomo Rizzolatti on mirror neurons and action understanding CSN Podcasts
Robert Axelrod on game theory and prisoner's dilemma CSN Podcasts
Adrian Owen on disorders of consciousness and vegetative state CSN Podcasts
Jonathan Whitlock on markerless motion capture and posterior parietal cortex CSN Podcasts
Luis Puelles on neuroanatomy and prosomeric model CSN Podcasts
Zoltan Molnar on subplate neurons and cortical development CSN Podcasts

Mark Blumberg on rem sleep and twitching

Episode 9 15.03.2015

PLAY EPISODE

Season 2015

0 Followers 13 Episodes

Description

What if the twitches you see in a sleeping infant are not remnants of dreams but a systematic self-calibration process for the developing motor system? Developmental neuroscientist Mark Blumberg explains how REM sleep twitching may serve as the brain’s sonar , pinging muscles one at a time and listening for the sensory feedback that bootstraps the body map. Subscribe for more from the Convergent Science Network podcast series. Mark Blumberg joins Paul Verschure and Tony Prescott at the BCBT summer school to present his research on the relationship between sleep, twitching, and sensorimotor development in infant rats. During active (REM) sleep, neonatal rats produce highly discrete myoclonic twitches , brief activations of individual joints occurring against a background of low muscle tone. Blumberg argues these are not random byproducts but structured motor events with a high signal-to-noise ratio, ideally suited for the nervous system to map its own body. The sensory consequences of each twitch cascade through the brain in ways that wake movements do not, because during wakefulness a corollary discharge mechanism gates reafferent signals. The discussion traces the developmental trajectory of twitching from spinal-cord-driven activity in the fetus through brainstem contributions in the neonate, showing how the system becomes progressively more complex. Blumberg presents evidence that multi-joint twitches develop spatio-temporal structure over the first postnatal week, with certain movement combinations becoming more frequent and more organized , suggesting a selectionist process that shapes which movement patterns persist. He challenges the concept of motor primitives from a developmental systems perspective, arguing that no aspect of the motor system comes for free and that what appears primitive at one level is always a developmental product at another. Key topics include the distinction between REM sleep twitching and wake movement, how corollary discharge develops in early life, why the cortex appears uninvolved in producing twitches at early ages, the relationship between twitching and joint maintenance, and what the developmental perspective offers to robotics and motor control theory. Part of the Convergent Science Network podcast series from the BCBT Summer School.

Tagged as:
Corollary Discharge Developmental Perspective Motor Movement rem sleep Sleep Twitching Twitches twitching

About the author

CSN Podcasts

Both the triumphs of humanity and its most evil deeds have resulted from collaboration. In a time where humanity is required to aspire to the former and minimize the latter, the question arises of how collaboration arises and why it fails. Surprisingly, this phenomenon, so central to who we are, is not well understood. Hence, a collaborative effort is required to understand collaboration in its full biological, psychological, sociological, cultural, and economic complexity and to translate this understanding into operational impact. This series of podcasts is one step toward achieving these complementary goals. The Collaboration Podcast presents interviews with people who are central orchestrators of collaboration in various domains including business, government, science, art, health, sustainability, and the military. The discussions were conducted by Prof. Dr. Paul F.M.J. Verschure and members of the Program Advisory Committee of the Ernst Strungmann Forum on Collaboration (https://www.esforum.de/forums/ESF32_Collaboration.html) during 2021 and had the goal to sketch a map of opportunities, challenges, and obstacles in human collaboration. The forum took place in May 2022, and now we would like to share this series of interviews with a broader audience. The full report of the Forum will be published in 2023 by MIT Press. The podcast was produced by the Convergent Science Network (https://www.convergentsciencenetwork.org/). Context: The stability of social systems depends critically on realizing sustainable methods of “collaboration,” yet how and by which means collaboration is achieved is not clearly understood; neither are the conditions or processes that lead to its breakdown or failure. Collaboration can be understood as cooperation between agents toward mutually constructed goals. Part of the reason for our lack of understanding is that the phenomenon of collaboration is, by nature, a highly multidisciplinary problem, and effective research into its complexities has been difficult to achieve across the broad range of scientific and technical disciplines involved. The need for a fundamental understanding of collaboration, however, has become increasingly important. Not only does humankind demand answers as it attempts to address critical challenges at multiple scales (e.g., climate change, migration, enhanced automation, social and economic inequality), but ever-increasing technological and economic means of interconnecting people and societies are disrupting long-established, familiar patterns of how we interact. Radical technological changes that are ongoing have the potential to reshape collaboration in ways that are currently hard to predict or influence (e.g., by altering configurations in interaction, information creation, and modes of communication). On one hand, such changes could disrupt hitherto stable forms of collaboration by affecting critical communication channels and traditional roles, as can be observed in the rapidly changing patterns in governance, commerce, and social interaction. Conversely, technology could lead to the emergence of novel, successful forms of collaboration that deviate from traditional “hierarchical” architectures. Evidence of this can be seen in areas as diverse as highly automated manufacturing plants, the open science movement, collaborative software repositories, user-centered services, and the sharing of economy-based modes of organization. Without a fundamental understanding of the mechanisms, processes, and boundary conditions of collaboration, it is not possible to evaluate or predict which of these possible scenarios are sustainable or even plausible. The Forum “How Collaboration Arises and Why it Fails” (May 8–13, 2022, Location: Frankfurt am Main, Germany) Chairs: Andreas Roepstorff and Paul Verschure Program Advisory Committee: Jenna Bednar, Julia R. Lupp, Bhavani R. Rao , Andreas Roepstorff, Ferdinand von Siemens, and Paul Verschure

Timestamp

00:00:00 - Oh, this is a warning. We're in trouble, Tony. This is the Convergent Science Network podcast.
00:00:09 - Leading researchers in the domain of neuroscience, brain theory,
00:00:13 - and technology are interviewed by Paul Verschure and Tony Prescott. You ready? All right.
00:00:20 - Paul Verschure with the Convergent Science Network podcast together with Tony
00:00:24 - Prescott. And we're talking now with Mark Bloomberg, who was a speaker at our
00:00:29 - BCBT 2015 summer school.
00:00:32 - And Mark, you were linking in your talk and also in the research that you've
00:00:37 - got some time, sleep and the development of the motor system.
00:00:44 - So, and in some ways also telling us that sleep is doing a lot more than just sleeping.
00:00:50 - So what's really the link here between the development of the motor system and sleep?
00:00:56 - Well, what you see in early development, one of the things that marks sleep
00:00:59 - in early development is the fact that you, first of all, you're sleeping a lot more.
00:01:02 - And even more, when you're young, you're having a high proportion of REM sleep,
00:01:07 - what we also call active sleep.
00:01:09 - And what makes active sleep active sleep is the fact that you are engaging,
00:01:13 - you have an activated nervous system, you have a loss of the ability to have high muscle tone.
00:01:19 - But even though you have low muscle tone during active sleep,
00:01:23 - infants are producing a lot of motor activity in the form of these myoclonic
00:01:28 - twitches, just twitches for short.
00:01:30 - And so what happens during these twitches is that you are activating muscles,
00:01:35 - you're moving limbs, you're moving eyes, you're moving whiskers.
00:01:37 - And there are sensory consequences to those movements, which,
00:01:41 - as we've shown over the last 10 years or so, cascades through the nervous system.
00:01:46 - And so the question we've had is what, you know, although these twitches have
00:01:49 - largely been thought of as remnants of dreams, you know, the old wives tale
00:01:54 - of dogs chasing rabbits in their dreams,
00:01:58 - what we think is happening is in fact that these twitches are not epiphenomenal,
00:02:01 - are not byproducts, but are products of the nervous system that play a role
00:02:05 - in self-organizing or bootstrapping the sensory motor system to create the fundamental
00:02:11 - structure that makes wake movements possible.
00:02:13 - So now what are the basic properties of these twitches?
00:02:17 - The basic properties are first that they are discrete.
00:02:20 - That when you look at twitching in slow motion as we have you see that individual
00:02:24 - joints are activated at the same time.
00:02:26 - Some animals the flurry of activity can be so great it's been they've been confused
00:02:30 - with seizure activity or people think that all these movements are happening simultaneously.
00:02:36 - But in fact they are highly highly discrete. We rarely see simultaneous twitches
00:02:42 - in multiple joints. It's a rarity.
00:02:44 - Another feature is that they are occurring against this background of high muscle
00:02:48 - tone and of low muscle tone, rather.
00:02:50 - And so that becomes important when you're thinking about producing a movement
00:02:53 - and getting feedback in return of that movement.
00:02:56 - You want it to happen in a situation where there's a high signal-to-noise ratio,
00:03:01 - where the signal you're getting back is clear.
00:03:04 - Um, the metaphor I like to use is a submarine where people are doing echolocation or sonograms.
00:03:11 - Um, um, sonogram? No. What is it called?
00:03:15 - Um, you know. Sonar. sonar thank you thank you word choice
00:03:18 - problem here um when you do sonar you know like every movie
00:03:21 - that's ever been done with a submarine the captain tells everybody
00:03:24 - to be quiet and then they send out you know a single ping and they
00:03:27 - get a they get a feedback from that i think that's
00:03:30 - what's happening during twitching you're shutting down the system and
00:03:32 - then you're sending you're pinging your your limbs you're pinging your
00:03:35 - whiskers you're pinging your eyes and you're getting feedback in
00:03:38 - response to that that then allows you to make sense of uh
00:03:41 - of how your your your body is constructed and then
00:03:44 - finally the twitches themselves have spatio-temporal structure
00:03:47 - that is that they there are certain movements that are
00:03:50 - joint movements that are more likely to happen than other joint movements
00:03:53 - and we've chronicled some of those changes across development so
00:03:56 - all these things together tell us that twitching is not some you
00:03:59 - know mere random activation it's a highly structured
00:04:02 - a highly organized system so there's there's a
00:04:05 - trajectory of twitching um through development
00:04:08 - is there sort of in that it does it become more complex
00:04:11 - or is it just yes we've seen that the that the movements become
00:04:14 - you you end up with more multi-joint movements you end up
00:04:17 - with more complex movements over the six day differences that we've looked at
00:04:21 - from two days of age in a rat to eight days of age there is a documentable change
00:04:25 - in the structure of the twitching but now can we so what was the lowest level
00:04:31 - of at which the twitch can be expressed breast.
00:04:34 - That's a single muscle fiber. It's a group of muscle fibers.
00:04:37 - It's a, it's a synergy between multiple muscles.
00:04:40 - Yeah. Great question. What's the lowest level? Um, we don't know.
00:04:43 - So to move a limb, it probably takes at least many muscle fibers.
00:04:47 - We've, we don't think it's probably, we do not think.
00:04:51 - We suspect it's not at the level
00:04:53 - of single muscle fibers. It's probably at the level of muscle groups.
00:04:56 - But just due to the problem of measuring individual muscle fibers,
00:05:01 - we haven't been able to technically work that one out.
00:05:03 - Okay, but then how far up would that go?
00:05:08 - Would it go to the level of the movement of a whole limb?
00:05:12 - Like I move my arm and my hand and my fingers?
00:05:14 - Or is there an upper bound to what you would call a twitch?
00:05:18 - Well, a twitch is a movement of any given, in any given direction for any given joint.
00:05:24 - But you can have a bout of twitching in which you can have multiple joints,
00:05:28 - even across multiple limbs, that occur in a very rapid sequence.
00:05:31 - And we've used about 50 milliseconds as our cutoff for defining one twitch to
00:05:36 - another twitch to another twitch as defining a bout.
00:05:39 - And when you look at that very fine time scale, I mean 50 milliseconds is pretty
00:05:42 - short, what you see is there's certain sorts of movements that are very likely.
00:05:46 - So one thing that you you might have is that your shoulder might bring your
00:05:49 - fist towards your body of your right limb, and very quickly after that,
00:05:53 - you may have the same sort of movement in your left limb.
00:05:56 - And those sorts of movements, what we call homologous movements,
00:05:59 - a movement towards the body of your right, movement towards the body of your
00:06:02 - left, that happen with a great frequency.
00:06:04 - And that can be attributed to certain aspects of the spinal circuitry connecting
00:06:08 - the right to the left limb.
00:06:10 - So we've been trying to dissect out those sorts of highly probable and low probable.
00:06:15 - Events in these more complex twitches.
00:06:20 - What do we know about the development of the motor pattern circuits that are
00:06:27 - underlying some of this?
00:06:28 - Are they developing during this period, or are they more or less in place?
00:06:32 - They're greatly developing. So initially, even across the early ages that I
00:06:35 - was talking about today, the spinal cord plays an enormous role initially in
00:06:39 - the production of a lot of these movements, especially in the fetal situation.
00:06:42 - By the time these animals are born, you're already getting brainstem contributions to that as well.
00:06:49 - So you're now adding a brainstem motor component to your spinal motor component.
00:06:53 - For the purposes of twitching, that's all you have.
00:06:56 - But even for the purposes of wake movement, at those ages, the cortex is not
00:07:00 - playing any role at all, as far as we can tell, in the production of movement.
00:07:04 - The cortex is mostly receiving information and laying down those early structures.
00:07:08 - They do not have a lot of motor outflow. So for the most part,
00:07:11 - this period of infancy in rats, which is roughly the last trimester in humans,
00:07:15 - this is largely a period of brainstem control of movement.
00:07:20 - Very little cerebellar control directly cerebellum doesn't really do a lot of
00:07:24 - direct motor control um so the the brain stem is doing most of the heavy lifting
00:07:28 - are we talking about the first 10 days of rat roughly maybe the first week or
00:07:33 - so yeah so they're mostly in in the the nest,
00:07:36 - not moving around very much presumably they do move i mean they the movements
00:07:40 - that they have during wake their eyes are closed uh the visual system is limited
00:07:43 - they're getting some light but not a lot because their eyes are sealed until
00:07:46 - they're 15 days of age their their interactions are mostly huddling,
00:07:50 - so they're interacting with their litter mates,
00:07:52 - they're diving into the huddle, they're coming out of the huddle,
00:07:54 - and then they are seeking the mothers for milk and protection.
00:08:00 - And if you put them outside the nest, they can engage in some locomotor activity.
00:08:05 - But it's fairly limited at those ages. They're dragging themselves along,
00:08:08 - aren't they? Yeah, exactly.
00:08:10 - And it's not until they're about, you know, weaning is a rough,
00:08:12 - you know, when they actually get independent.
00:08:14 - But around 21 days of age, they're very ambulatory and they're moving as you
00:08:18 - would expect of an adult, pretty much.
00:08:20 - But to qualify as a twitch, we are talking about, let's say,
00:08:26 - a time window of about 50 milliseconds duration.
00:08:29 - But also, I would assume there's a lack of coordination because otherwise I
00:08:35 - can make a goal-oriented movement and say, oh, this is a sequence of twitches.
00:08:39 - So there must be a minimal level of, let's say, coordination across multiple twitches.
00:08:46 - So where where do we draw the boundary there in terms of the coordination across
00:08:50 - twitches okay well what defines a twitch that when i said 50 milliseconds i
00:08:53 - meant the difference between two twitches two independent ones the twitches
00:08:57 - themselves are very fast on the order of five,
00:08:59 - milliseconds but it's hard to know when you start when you stop those sorts
00:09:02 - of things some of the movement is active and some of it is passive in return um the the.
00:09:09 - So it's easy to distinguish a wake movement from a sleep movement.
00:09:13 - I think I showed you some examples today.
00:09:16 - When you see a baby or a rat when they're waking, what you see are very high amplitude movements.
00:09:21 - The movements of the limbs are typically, they have some relationship to one
00:09:25 - another. It can be an alternating movement.
00:09:27 - It can be a retraction of both limbs. It could be a yawn.
00:09:30 - It could be kicking. and those sorts of movements are happening when the muscle
00:09:35 - tone is high and the nervous system is in a different state.
00:09:38 - What distinguishes a twitch has to be the nature of the movement itself,
00:09:42 - how quick it occurs, and also the background activity, the low muscle tone,
00:09:48 - that general sense of relaxation.
00:09:49 - Anybody who has a dog or a cat or a baby or a person and you're watching them sleep,
00:09:55 - what you'll notice is, or if you've
00:09:56 - ever been on a bus and you've tried to sleep in an upright position,
00:09:58 - you feel terrible when you wake up and your neck
00:10:01 - has your head has fallen over and you're because
00:10:04 - your neck is completely relaxed and that is what happens in
00:10:07 - REM sleep it's that relaxation of the muscles that defines
00:10:10 - REM sleep in addition to this this sort of
00:10:13 - this phasic this fast activity that occurs with twitching all
00:10:17 - of those things under normal conditions have to occur for
00:10:20 - you to say that it's a twitch but for just to say
00:10:23 - it's REM sleep there also has to be some oscillatory property in the
00:10:26 - brain i mean it's not necessarily okay i mean
00:10:28 - there comes a point when you can measure eg activity and
00:10:32 - that eg activity as bears more of a resemblance to
00:10:35 - waking than it does to to uh um to slow wave sleep or quiet sleep um so waking
00:10:41 - and REM sleep have very similar eg patterns right but it's not rhythmic it's
00:10:45 - highly desynchronized um so there is you know there are rhythms that occur during
00:10:50 - during active sleep but they're not the ones we so there is a little,
00:10:54 - risk of circularity here if you define REM sleep in terms of exhibiting twitching.
00:11:00 - And then you want to say twitching happens during REM sleep?
00:11:02 - No, it's also with the muscle atonia.
00:11:04 - So the muscle atonia is the key for that. That's right.
00:11:08 - I mean, you need more than one component to define a state, and those are the two.
00:11:13 - But now we have plenty of other examples of brain activity patterns that we
00:11:17 - see during REM sleep that we don't see in other states. Even in these pups? Yes.
00:11:21 - And they also do the sort of passive sleep as well as the active sleep?
00:11:26 - What's called quiet sleep? Yeah. It's very brief in these animals,
00:11:29 - and it's very difficult to distinguish from quiet waking because they're so immature.
00:11:35 - And so there comes a point around 11 days after age when quiet sleep can be
00:11:40 - defined on the basis of EEG activity.
00:11:42 - But prior to that age, it's very hard to distinguish. A quiet animal is just
00:11:46 - a quiet animal, and their eyes don't open.
00:11:49 - So you don't have a lot of the components that we have in an adult.
00:11:54 - But in some sense if we follow your your sonar example now if we if we remove
00:12:01 - muscle tone and i see the twitches in some sense this is if you want the central
00:12:05 - nervous system talking to the skeletal muscle system and then you could say
00:12:09 - well the muscle twitch or the twitches i observe.
00:12:13 - Are if you want the lowest level of resolution the brain has available to really
00:12:18 - talk to the skeletal muscle system and as such this would be let's say a movement primitive okay.
00:12:26 - Would you be happy with with that interpretation or you
00:12:28 - look at it differently is it more like a random driving of the skeletal muscle
00:12:32 - system in a way that is not related to actually the control of the skeletal
00:12:36 - muscle system well so now so so we've we had a little bit of discussion in the
00:12:39 - talk about you know primitives and what they mean and so let me just it's important
00:12:43 - for me to state where I come from philosophically on that issue.
00:12:47 - So I come from a school of thought that, you know, my training and my background
00:12:51 - is in an area that's, you know, it's a fundamental developmental perspective.
00:12:58 - And the framework that's adopted by people like me is called the developmental
00:13:02 - systems theory perspective framework, however you want to call it.
00:13:06 - And the point of it is, is really to sort of to get away from using certain
00:13:09 - types of words that we think can be traps to understanding the origins of any given system.
00:13:16 - And for people like me, I mean, I'm very interested in where things come from.
00:13:20 - And the problem with saying that something is a primitive is that it immediately
00:13:24 - evokes the notion that you get it for free, that you get it without developing it.
00:13:28 - But there is no aspect of the nervous system, there is no aspect of a biological
00:13:32 - system that does not first develop.
00:13:34 - So when you say primitive, So I would have to throw it back at you and say,
00:13:38 - well, what do you mean by a primitive?
00:13:40 - Because even the fundamental wiring up of that system has to occur through a
00:13:45 - variety of genetic, epigenetic, and activity-dependent processes.
00:13:50 - So there are no primitives in the nervous system. But I gave you a definition, right?
00:13:54 - Because I told you it's the lowest level of resolution in which the central
00:13:58 - nervous system can talk to the skeletal muscle system.
00:14:02 - If you want to define that as a primitive, then we've now defined primitive
00:14:05 - in a way that it's typically not defined.
00:14:07 - The way it's typically defined is to say that the wake movements,
00:14:11 - the goal-directed movements, certain things that animals do,
00:14:13 - alternating leg movements, for example, central pattern generators,
00:14:17 - that these are primitives.
00:14:18 - But as people are now showing more and more, even central pattern generators develop.
00:14:23 - So the thing that used to be called a primitive has now become a developmental product.
00:14:27 - And I think that that's, I think it's healthier to stay away from those sorts
00:14:30 - of things. but, you know, to the extent that you want.
00:14:32 - What's that? We need a word. We have to say, this is the lowest level of resolution
00:14:36 - that cannot be further divided.
00:14:38 - It's like an atom, right? But I think I agree with Mark here that it's relative.
00:14:42 - So you could say that from the point of view of the cortex, you might have a
00:14:45 - most primitive instantiated in the brainstem.
00:14:48 - But from the point of view of the brainstem, you have to assemble that out of
00:14:52 - components, which may be at the level of muscles.
00:14:56 - And then at the muscle level, you could break it down further.
00:14:59 - So yeah so i think you can use primitive but
00:15:02 - you have to say what you mean primitive with respect to what a little
00:15:05 - bit that's what i was saying where this is the lowest level at which the
00:15:07 - central nervous system can drive the skeletal muscle
00:15:10 - system i'm not sure that's true so let's let's let's explore that for a second
00:15:15 - so we're very very early in development in rats in the fetal period that would
00:15:18 - be true for the maybe that's true for humans as well but earlier in the fetal
00:15:22 - period um muscles can skin it's called myogenic activity you can have muscle activity
00:15:28 - that occurs without any neurogenic, any neural, any brain, any central nervous system input.
00:15:33 - And as you, as you get, as you move, move through development,
00:15:36 - the spinal cord is doing a lot of this activity on its own.
00:15:39 - And then eventually, as I said earlier, then the brainstem starts to do more.
00:15:42 - So the mode, there could be a motor primitive very early in fetal development
00:15:46 - that's just at the level of muscle.
00:15:47 - Then you can have another motor primitive that's at the level of spinal cord,
00:15:49 - another one of the brainstem.
00:15:50 - And then at that point, I think the concept of a primitive loses its force.
00:15:54 - You know, at that point, I would rather just say what we are
00:15:57 - describing rather than using a word that has such a
00:15:59 - history you know so that's that's my problem and i think
00:16:02 - we do need a word and then we have to invent it right here
00:16:05 - right now because no that's not good enough i think because what we have to
00:16:11 - deal with is the question are we driving these muscle groups or muscle fiber
00:16:17 - groups in some random way as our sonar example would be look Look,
00:16:22 - I'm exploring my environment.
00:16:23 - I'm exploring my muscle, muscular system.
00:16:25 - I'm picking it. I get stuff back randomly.
00:16:29 - Or am I driving it in a way that already anticipates controlling that same system?
00:16:34 - And now I have to drive it in a way that is specific to the control I will perform
00:16:40 - later during wakefulness.
00:16:42 - And then it must be a unit, a basic unit, that I can assemble into behaviors.
00:16:49 - Well, so I'm going to argue for a more iterative process in this whole thing,
00:16:54 - where you start off with these, and they can't even be random movements,
00:16:57 - and there's some evidence in fetuses that they start off when they're spinally
00:17:01 - generated, that they look random.
00:17:03 - And they change over time. And then the question becomes, why do they change?
00:17:07 - And when you start getting into the issue of any kind of a loop,
00:17:10 - you know, anytime you have an output and an input, you can't say what causes what at that point.
00:17:16 - So, you know, you have an activity pattern, you have a motor outflow,
00:17:20 - but then you get information about that motor outflow and that modifies the
00:17:24 - next iteration of that system. And so the system keeps getting more and more complex over time.
00:17:30 - So what initially starts random starts to look structured and eventually becomes goal-directed.
00:17:35 - And then the process of, well, what does that mean to be goal-directed?
00:17:38 - And now we're talking about how you produce models of the outside world inside
00:17:42 - of your head that allow you to control your limbs in very, very productive ways.
00:17:46 - And now we're getting into the issues of all the computational processes that are involved.
00:17:50 - We're only beginning to understand that system from the perspective anyway that I'm taking with this.
00:17:56 - And so I'm trying to avoid certain assumptions that come out of non-developmental
00:18:02 - work because I want to try to understand how the thing develops without being
00:18:05 - bound by concepts that are outside of that perspective.
00:18:09 - I guess that one of the differences may be that we are interested in assembling
00:18:15 - control systems for things like robots, and there the notion of primitive is very useful.
00:18:23 - So you can say, I have some device that I know can control the arm and move it to a certain place.
00:18:30 - Therefore, I can just, at a higher level, just command the position I want to
00:18:34 - move my arm to, rather than worry about how I coordinate the different joints to make that movement.
00:18:40 - And that definition of primitive is really useful then in robot control.
00:18:45 - And I guess it's an interesting question as to whether that's useful then to
00:18:51 - think about in terms of hierarchy of brain systems.
00:18:54 - For instance, people often think about the superior colliculus as being a device
00:19:00 - that can generate head and eye movements to a particular location in space,
00:19:04 - so that a system talking into the colliculus just has to give it a coordinate.
00:19:08 - Right. I think that's absolutely right. Right. So the point I think my challenge
00:19:13 - back to the robotics perspective is that.
00:19:18 - If you have the notion of a primitive, if that becomes a useful concept in robotics,
00:19:22 - but it turns out not to be a useful concept for actual biological systems,
00:19:27 - then might there be some value in thinking about how you get to a primitive
00:19:32 - and how you imagine the developmental process, and might that play a role in
00:19:36 - creating robots that are more flexible and adaptable?
00:19:40 - So I can understand that the concept of primitive is useful,
00:19:43 - perhaps, and safe from my perspective. As long as you say that it's a primitive as of time T.
00:19:49 - As of this moment, it's a primitive for that.
00:19:52 - But then it doesn't tell you how you got there. To agree on the fact that these
00:19:56 - primitives are acquired is not a problem.
00:19:59 - But it is a stable representational structure that helps you to perform a certain operation.
00:20:05 - And it's also a commitment to saying brains are controllers in the end.
00:20:08 - They control the skeletal muscle
00:20:09 - system. Well, I guess I agree with you on that point. I mean, I'm fine.
00:20:12 - I wish that all people use the word primitive the way you're using it.
00:20:15 - Unfortunately we found each other mark unfortunately unfortunately
00:20:20 - people have used the word primitive to do a lot of mischief sure and
00:20:23 - and it's the mischief i'm trying to avoid by by being
00:20:26 - cagey about the word okay but but at least we resolve this obstacle now i think
00:20:30 - we have we're ready to move on exactly so this is very good but now through
00:20:35 - development do you see that twitches the twitching behavior becomes more stereotyped
00:20:40 - in an acquired fashion i emphasize So it loses variability over development?
00:20:45 - We don't know. We did do an entropy measure for the multi-joint activity.
00:20:51 - And we did find interesting relationships between how entropy and the frequency
00:20:56 - of movements changed over development.
00:20:58 - So it appeared as if the multiple joint movements that were composed of multiple twitches,
00:21:07 - though the entropy that is the organization of that system
00:21:10 - um it became more ordered over time over
00:21:14 - a very short period of time so that it appeared and and the ones that were more
00:21:18 - ordered became more frequent over time so there seems to be an interesting almost
00:21:22 - like a selectionist like a darwinian process and choosing which types of movements
00:21:27 - are going to persist to the next day or the next week of the animal's life well we don't yet know.
00:21:33 - Is when you start getting into systems that are much more highly developed,
00:21:38 - much more complex, capable of really complex goal-directed reaching movements,
00:21:42 - locomotion, you know, much more complex things.
00:21:45 - We don't have a good sense yet of how well twitching looks in those situations.
00:21:49 - I have a colleague sent me a video of a racehorse who appeared to be in REM sleep.
00:21:55 - I mean, the horse was on its side and it was moving its legs,
00:22:00 - you know, in this beautiful synchronous racehorse-y kind of a way.
00:22:03 - And racehorses rely more heavily
00:22:04 - on their hind limbs than they do on their forelimbs for propulsion.
00:22:07 - And so then the question is, well, you know, somebody might say that the racehorse
00:22:11 - was dreaming about its race, and I would say that it was engaging in spontaneous
00:22:15 - activity during REM sleep that is sort of a way of training up its or practicing its system.
00:22:21 - But we don't, that has not been documented.
00:22:24 - But those two things could be the same. They could be, yeah.
00:22:26 - Yeah. So there was some cortical activity that went alongside of that.
00:22:30 - But here we get into this really interesting debate that I'm having with some
00:22:33 - colleagues about whether or not the motor cortex plays any role at any age in
00:22:37 - the production of twitching. And I would argue there is no evidence that the
00:22:41 - cortex plays a role in twitching. But it might not even be motor cortex.
00:22:43 - It might be visual cortex, the animals visualizing, you know, the track.
00:22:48 - Yeah, but how do you go from the visual cortex to motor control without engaging the motor cortex?
00:22:52 - Oh, well, you go via the brainstem, sort of via the basal ganglia.
00:22:56 - Yeah, no, there is no evidence that any part of the forebrain is playing a role
00:23:00 - yet, but it remains to be proven.
00:23:03 - But my argument to my clinical and other sleep research friends is to say that we need evidence.
00:23:10 - There is no evidence right now. And there are recordings from the visual cortex,
00:23:14 - which also support that claim.
00:23:16 - Right. But now in development, do I lose muscle fiber that I'm not twitching enough?
00:23:20 - We don't know. I mean, we do know that twitching is happening close to the age
00:23:23 - that you start to lose polyinnervation of muscles, but there's never been a
00:23:29 - direct link between those.
00:23:31 - Okay, but that's a potential functional role as well, right?
00:23:33 - It is, I mean, and there's another one as well, which has been established for
00:23:37 - movement, which is that if you don't move your limbs as a fetus, your joints will fuse.
00:23:42 - Right. And so it's important for maintaining healthy bones and healthy joints
00:23:46 - when you're very early in development.
00:23:48 - And you need to move them or they fuse and then you're out of
00:23:51 - luck yeah but now you you you contrasted your your view on on on the role of
00:23:56 - twitching in in sleep with the idea of reference copy yes right or corral discharge
00:24:01 - and you you don't at least in in the way you presented it you you seem to see
00:24:07 - a difference between the two.
00:24:10 - So what would be that difference between twitching patterns,
00:24:14 - the role of twitching in the developing nervous system, and the role of an afferent copy?
00:24:18 - Well, so the results we have so far suggest that what distinguishes movements
00:24:23 - that we make when we're awake, this is just the data are only for young animals.
00:24:28 - So it's important to be clear about that. But the movements that an infant makes
00:24:33 - when it is awake actually involves,
00:24:37 - when you're moving your limb in the air or whatever, the sensory information
00:24:42 - that could be coming back from that limb, the proprioceptors and so on,
00:24:47 - the information from that is being shut down,
00:24:50 - prevented from making its way into the brain.
00:24:54 - Whereas when you twitch during your sleep, that information flows freely through.
00:24:58 - And we take that as evidence that that's why twitching is distinct and why it could be important.
00:25:05 - Now, in order to gate that information when you're awake, we posited that this
00:25:10 - corollary discharge process is engaged, and we now have direct evidence that that is the case.
00:25:15 - If we inhibit that corollary discharge phenomenon from occurring,
00:25:19 - we now allow information from during wakefulness to move into the brain from the wake movements.
00:25:24 - So that's one aspect of how the corollary discharge system can modulate sensory
00:25:30 - inflow into the brain during sleep versus wake.
00:25:35 - But that's just one feature of all this.
00:25:39 - I think of this as, you know, this is very much what you would expect on what
00:25:43 - people do with what people have found in adults, where the amount of reafference,
00:25:48 - the amount of information coming from waking limbs is very, very low when we're awake.
00:25:52 - And I was at first a little bit surprised by this.
00:25:54 - You know, I'm sitting there, people listening to the podcast can't tell this,
00:26:00 - but if I'm moving my arms in the air, and anybody can move their arms in the
00:26:04 - air, they can even close their eyes while they move their arms in the air.
00:26:08 - They have a very strong sense that they are moving their arms and they have
00:26:11 - a sense that that's sensory, that something sensory is happening.
00:26:14 - But motor control people who I've talked to about this say that not only are you not,
00:26:20 - the reason why you have that, what they would call an illusion of that,
00:26:23 - is because you have a computational model in your brain which is telling you
00:26:27 - that this is what you should be doing. This is what you should be feeling.
00:26:31 - So it becomes an illusion of a sensation that you are moving your arms in the
00:26:35 - air as opposed to what's happening, we think, in twitching when you actually
00:26:39 - are getting feedback that is changing brain activity.
00:26:43 - So I think that's what was surprising
00:26:46 - to us is that we saw this system functioning as early as we found it.
00:26:50 - So in your opinion, then, the twitching feedback is really a signal that is
00:26:56 - closely driven by the muscles themselves,
00:26:59 - while the corollary discharge is more based on, let's say, a forward model that
00:27:04 - is making predictions about what you were supposed to do or what you are doing, etc.
00:27:08 - This is sort of the transition that you see there. Yeah, I mean,
00:27:11 - we think it's going to be, you know, we haven't identified the source of the
00:27:17 - signal that is shutting off, gating that activity during wakefulness,
00:27:21 - but we suspect it's going to be tightly related to exactly what you said about
00:27:25 - something related to a forward model.
00:27:29 - But we need to, you know, we suspect a few areas, but we haven't established
00:27:32 - yet. But it might mean that it also can be a cascade of incrementally more prediction-oriented
00:27:40 - feedback signals as opposed to just two different systems.
00:27:44 - Yes. So on what side are you there in the interpretation?
00:27:47 - Well, it's all one system with different states. So, you know,
00:27:50 - the whole system is simply being engaged differently when you're awake versus when you're asleep.
00:27:54 - So when you're awake, certain, I mean, we don't know the full range of parts
00:27:58 - of the brain that are active.
00:28:00 - We know a lot about sleep circuitry, but not nearly enough.
00:28:03 - But, you know, we're getting to a point now where we can say that,
00:28:05 - you know, with respect to certain aspects of the motor system,
00:28:08 - it is being engaged differently during wakefulness than it is during sleep.
00:28:12 - And understanding what we're trying, we're trying furiously to understand in
00:28:17 - the lab right now is what are the critical parts of that circuit that are responsible
00:28:21 - for these differences between the states.
00:28:24 - So the claim, just to be clear in my mind, is that when you generate an emotive
00:28:29 - command when you're awake, a copy of that command.
00:28:32 - Goes to a part of the brain which then
00:28:35 - uses that to compute the sensory experiences i
00:28:38 - might expect as a consequence of that movement right and
00:28:41 - then the ascending sensory signals are matched with that prediction and anything
00:28:47 - that i have been able to predict then gets you know cancelled out and doesn't
00:28:52 - go up to my brain that's now when i'm asleep and i twitch wherever that motor command is coming from,
00:28:59 - I'm not going to send a copy to this comparator system so that any sensory signal
00:29:06 - that's coming in is going to carry on right up to, say, cortex.
00:29:10 - So although I initiated the movement, I still get the feedback as if I hadn't,
00:29:16 - you know, as if it moved separately.
00:29:18 - That's right. That's the claim. Yeah. And I think the further claim that you're
00:29:22 - making is that that's useful because it helps then parts of the brain to self-organize, perhaps,
00:29:30 - because the sort of activity-dependent aspects of development,
00:29:35 - you need to know that you have muscles out there that are working and operating
00:29:40 - and that have feedback signals.
00:29:42 - That's a wonderful synopsis of exactly what I'm saying. And I want to try to
00:29:49 - drive this home to a listener.
00:29:51 - Remember, I started with certain types of anomalous individuals,
00:29:54 - whether it was that at Johnny Eck, who was missing his two hind legs.
00:29:58 - The dogs who are missing their two forelimbs or hind legs, that sort of adaptability
00:30:02 - is not something that you can program in.
00:30:04 - You can't say to the brain, you know, before you know that you're going to have,
00:30:07 - you're not going to have two limbs and now you're going to remap the brain.
00:30:11 - The point I try to make is that whether you're born atypically as a dog with
00:30:16 - just two forelegs or two hind legs, or you're, you're a typically born dog with
00:30:21 - four, um, with four legs,
00:30:23 - you need to map the body you have, not the body you're supposed to have.
00:30:27 - And really what happens in development is no matter whether you have a typical
00:30:32 - body or an atypical body, the same exact developmental processes are going to
00:30:36 - be involved in making sense of your body and in mapping your body and making it a functional system.
00:30:42 - You know, we've lost sight of the, you know, in the 19th century,
00:30:45 - one of the biggest things were these, you know, freak shows and side shows,
00:30:48 - as they were called at the time, the Bartholomew Fair in England,
00:30:51 - where all these different anomalous people and animals would be displayed.
00:30:56 - And people could see for their own eyes how amazing people could be,
00:31:00 - the so-called armless wonders, people who could use their feet to sew and to
00:31:04 - play the piano and to play the violin and shuffle a deck of cards,
00:31:08 - things that most people imagine you could never do with your feet.
00:31:10 - And yet, if you're born without arms, you learn how to use your feet in very unique ways.
00:31:15 - And I think this aspect of plasticity in a
00:31:18 - developing system is something that we that i just you
00:31:21 - know like to emphasize in this context but now in this
00:31:24 - so now we have to the concept sort of clear
00:31:27 - and also how these different control systems would then work together with
00:31:30 - respect to feedback from twitches and from an inference copy um now we can start
00:31:36 - to look also more at the structures that underlie this right and and one of
00:31:39 - the first structures you pointed out there was the red nucleus which is a brainstem
00:31:43 - motor nucleus and you see then the red nucleus as both the trigger of Twitch's.
00:31:50 - And the recipient of this first level of feedback from the twitch,
00:31:54 - is this really the first station in this whole cascade?
00:31:57 - The first station is probably a sensory motor loop within the spinal cord itself.
00:32:01 - But the first one within the brain is going to be the red nucleus and all of
00:32:05 - its associated structures.
00:32:06 - So the red nucleus is one of the most important, most studied of these structures.
00:32:11 - But what's really interesting is that, you know, people have looked comparatively
00:32:13 - at different animals with weird morphologies, you know, like an elephant.
00:32:17 - And what happens is that the trunk is actually controlled by a nucleus that's near the red nucleus.
00:32:23 - So different animals with different structures have different emphases on different
00:32:28 - parts of those brain areas.
00:32:30 - But the red nucleus is certainly part of it. It's important for twitching.
00:32:34 - If we lesion that area, twitching goes down precipitously.
00:32:37 - And it's also involved, of course, in waking motor activity.
00:32:41 - So, this is a nucleus that's going to do one sort of firing pattern when the
00:32:46 - animal's awake, and it's going to do a very different kind of firing pattern
00:32:49 - when the animal's asleep.
00:32:50 - So, the same basic system is being manipulated for different purposes at different times of the day.
00:32:56 - But on the brainstem, we also have other motor nuclei, like you might find motor
00:33:00 - in some particular formation, or the peduncle pontine nucleus.
00:33:04 - So, do you see the red nucleus as then a specialized nucleus for twitch, let's say, control?
00:33:11 - Or is it just one among several at that
00:33:14 - level of brain organization so if you lesion the red nucleus
00:33:17 - not only do you not not only do you
00:33:19 - not wipe out all twitching but your effect
00:33:23 - is only temporary so that there are other parts of the
00:33:25 - brain stem that can make up for it there are multiple sites in the
00:33:28 - brain that are responsible for producing twitching because the
00:33:31 - red nucleus is you know plays a bigger role with the forelimbs than it does
00:33:34 - with the hind limbs and there are going to be other areas they're going to play
00:33:37 - a role for different types of movements and so So what we're really talking
00:33:40 - about is a network of premotor nuclei that are interacting to produce and interacting
00:33:44 - with the cerebellum as well to produce the full range of movements that we're capable of.
00:33:49 - So then what's the next level? So now we close the loop over the red nucleus.
00:33:53 - And indeed, now you mentioned cerebellum.
00:33:56 - Do you see now the cerebellum again talking to the red nucleus to trigger twitches
00:34:00 - at the different spatial temporal scale?
00:34:02 - Or is structure now processing the feedback signals that come from the twitches
00:34:07 - generated by the red nucleus?
00:34:09 - So we think that the twitches are going to play a role in shaping the cerebellum,
00:34:13 - but we haven't seen any evidence yet that the cerebellum is affecting twitching.
00:34:17 - And we've done this by lesioning the output nucleus of the cerebellum.
00:34:23 - But that work is still ongoing. going. So we haven't published that yet.
00:34:26 - That's still happening.
00:34:27 - So the cerebellum, though, is certainly part of this whole process.
00:34:30 - How it actually shapes twitching, we have not yet established that. But you did show that.
00:34:35 - The twitch correlates very strongly with a complex spike-like response in the
00:34:41 - cerebellum, at least the multi-unit level in the cerebellar cortex.
00:34:45 - So is that the link that you... Yes. Well, the cerebellum, the Purkinje cells
00:34:49 - in the cerebellum are definitely paying attention to twitches.
00:34:53 - Twitches are a very strong driver of cerebellar activity.
00:34:56 - And we're also recording from the deep cerebellar nuclei, which are the output nuclei,
00:35:00 - and they also are strongly um i'm receiving um
00:35:03 - information about twitches so the whole system is very
00:35:06 - much attuned to what's happening with these uh twitches during
00:35:09 - REM sleep but then you move up to other that's that's
00:35:12 - another sensory motor loop that gets closed with it
00:35:15 - that runs over the inferior olive essentially and it goes back to the
00:35:18 - red nucleus so the output the deep cerebellar nuclei project back to the red
00:35:22 - nucleus and presumably is modulating its output and all of these systems have
00:35:26 - to be you know to mapped in a topographically organized way so that a signal
00:35:34 - coming out of the red nucleus is going to activate the forelimb,
00:35:36 - and then the forelimb is going to feed back and affect certain parts of the cerebellum,
00:35:39 - which is going to go back and affect a certain part of the red nucleus that
00:35:42 - produced the signal in the first place.
00:35:44 - If you don't have that kind of convergence of output and input,
00:35:47 - you're going to have a disorganized motor system.
00:35:49 - But is that the example of the comparator also that Tony talked about earlier in your mind?
00:35:54 - The comparator is really more about what happens when you're awake.
00:35:58 - Yeah, but still you have to train it up. You do have to train it up.
00:36:01 - And you have to keep it in sync with your skeletal muscle system.
00:36:03 - And I think it's still a mystery. You know, the relationship between, I can't, I mean,
00:36:07 - we have to be careful because we don't have any information about how that comparator
00:36:11 - works, how it's being trained up, and whether or not wake movements are part
00:36:16 - of the training process or whether twitching is doing the work for that.
00:36:19 - We still don't know the answer to that question.
00:36:21 - And we're pretty far from answering it probably. But now you did mention that's
00:36:25 - obviously complex spikes.
00:36:26 - There'll be a signature of inferior olive involvement and also changes in the
00:36:31 - simple spikes of Purkinje cells, which would suggest that we have mossy fiber
00:36:35 - inputs to the Purkinje cells also conveying information on the twitch.
00:36:39 - So how would, should we, now the inferior olive is usually seen as conveying
00:36:42 - error signals and driving learning in the cerebellar cortex.
00:36:45 - So that would mean that we actually have two inputs coming from the red nucleus-induced
00:36:49 - twitches into the cerebellum, and one is in, let's say, a sensory state,
00:36:53 - and the other one is more an error.
00:36:55 - So could we interpret this as the first one reflecting the initiation of the
00:37:00 - twitch, and the second one reflecting, let's say, the recurrent feedback we're
00:37:05 - getting from the execution of the twitch, driving, for example, the error signal?
00:37:09 - Or you would, at this point, not try to differentiate so specifically?
00:37:15 - Things get really tricky when you start. So the concept that the inferior olive
00:37:18 - is computing error signals comes from work in adults.
00:37:22 - And it's possible that that concept can be imported.
00:37:27 - You know, with precision into development. But it's also possible that the role
00:37:31 - of the inferior olive changes developmentally.
00:37:33 - So it's conceivable, I think, in our mind, that really what's happening with
00:37:36 - the inferior olive is you're still setting up the topography of the cerebellum.
00:37:41 - So that what you're getting is a signal from the inferior olive,
00:37:44 - whose only purpose at that age is to line up with the sensory input that's coming
00:37:49 - from the twitching limb or whatever.
00:37:52 - And then you have a convergent input that allows a cell in the cerebellum to
00:37:56 - say, oh, I just got a signal that comes from this part of the red nucleus and
00:38:00 - this part of the body, and they're all linked together.
00:38:02 - If that linkage process is what development is about, that could be it.
00:38:08 - And remember, the cerebellum at these ages is extremely depleted in terms of
00:38:11 - its complexity. There are no granule cells or parallel fibers that have yet emerged.
00:38:16 - And you have this opportunity to simply converge inputs.
00:38:20 - And that may be the functional role of the inferior olivate at this age.
00:38:23 - But it's interesting because it might actually also help the cerebellum,
00:38:26 - which is in that stage also developing very rapidly, to obey a certain somatotopy.
00:38:31 - Because this is actually one of the characteristic features of the cerebellum
00:38:35 - and the climbing fiber inputs in it, that you have multiple somatotopic maps
00:38:38 - of the body that, of course, you have to lay down in some way.
00:38:41 - And your twitching signals might be essential to actually get that done.
00:38:45 - That's exactly what we think is the key about that.
00:38:47 - Whether there are other functions about the inferior olive or any part of the
00:38:51 - system that we're missing by virtue of our methods or anything else,
00:38:55 - you know, that's something we have to take care and make sure about.
00:38:58 - But as of, you know, given our methods and given what we've seen so far,
00:39:02 - we think it's mostly for somatotopic organization at this stage.
00:39:08 - Okay. But that would imply that then for adults that have a somatotopic organized cerebellum.
00:39:13 - The twitching signals would not be really having a strong impact on cerebellum anymore.
00:39:17 - Is that the case? Well, the short answer is nobody knows.
00:39:22 - And the longer answer is that you not only have to develop and refine systems,
00:39:29 - but you have to maintain them and you have to repair them.
00:39:32 - And so it's conceivable that what happens in adulthood is maintenance and repair.
00:39:36 - And it also could be that when we do certain types of skill learning,
00:39:39 - learning to play a piano, for example, that you could actually engage some of
00:39:44 - these mechanisms again in distinct ways. Has this ever been looked at?
00:39:48 - Absolutely not. So we do not know the answer to that.
00:39:51 - It's conceivable that twitches are important for early development.
00:39:54 - They go away and they're not heard from again from a functional perspective.
00:39:57 - But it's also possible that we've been missing out on a very important aspect of adult function.
00:40:03 - So twitches could have multiple functions, I think you're saying that.
00:40:07 - I mean, if we If we go back to our robotics analogy, then I think you've already
00:40:11 - mentioned the fact that a twitch could help me keep my joints free so that I
00:40:16 - don't use my arm for a while.
00:40:17 - It could seize up, so a twitch could give me some flexibility.
00:40:21 - A twitch could let me know that my control system is working.
00:40:26 - You know, sort of, I need to know that when I command an instruction,
00:40:29 - the arm will actually move.
00:40:30 - And, you know, we do this in the lab all the time. You know,
00:40:33 - can I still move his fingertip? We send a small movement and it works.
00:40:38 - Something like a twitch could allow you to configure the
00:40:41 - pattern generation system and you know
00:40:44 - in terms of where the primitives come from they have to self-assemble in
00:40:48 - some way and they could assemble by making you
00:40:51 - know just fire a few neurons you get some movement and then you you're you figure
00:40:56 - out that if you fire these groups of neurons together you get a more complex
00:40:59 - movement or two limbs move together that sort of thing so uh and that uh in
00:41:05 - robotics people have used this sort of motor babbling,
00:41:09 - to try and work out how to control your body.
00:41:12 - So you can learn to move a robot arm to a point in space by sending different
00:41:18 - random commands to the joints of the arm and seeing what happens and building up a map that way.
00:41:24 - And then I think you were emphasizing in your talk about twitches generating
00:41:29 - sensory feedback to the brain.
00:41:32 - And then that's another, I mean, that's similar in a way to this learning how to control the arm.
00:41:37 - But would you say it's the same thing or another thing i
00:41:40 - think it's in the same ballpark i think that we're talking about a similar class
00:41:43 - of of ideas what you know the motor babbling
00:41:46 - idea i think is is a it's a form of exploration yeah
00:41:49 - and i'm talking here about using twitching as a form of exploration too the
00:41:53 - details may or may not be critical you know so the the but you know we're in
00:41:57 - a robot system you have a much more you it's a much more low noise system right
00:42:03 - and typically compared to a biological system is Is that a fair thing to say?
00:42:07 - If you were sensory input? Yeah, I would say so. Okay. You know,
00:42:10 - so the challenge for a biological system is to reduce the noise.
00:42:14 - You have multiple modalities, you know, multiple, so many peripheral receptors,
00:42:18 - so many different things happening at the same time.
00:42:20 - Everything, all these different things you need to organize in order to figure out what's going on.
00:42:25 - And of course, you can program the motor babbling to be just move this arm in this way.
00:42:29 - And it could be that the biological system has to solve that problem differently,
00:42:33 - perhaps by going into a state like sleep.
00:42:36 - But I don't see any reason why you can't model that very idea in a robotic situation.
00:42:41 - Well, I know every robot has a massive problem always with calibration.
00:42:44 - Before you do anything, always first, okay, we have to calibrate that the robot
00:42:49 - really knows where its limbs are because otherwise things will go dramatically wrong, right?
00:42:54 - So you spend a lot of time on these calibration issues, which are often taken
00:42:57 - for granted and not really seen as part of the solution.
00:43:00 - But what you're describing here is also the brain's calibration system to stay
00:43:04 - synchronized with its body.
00:43:07 - I mean, my idea of a perfect world is where roboticists take these ideas seriously,
00:43:11 - sort of along the lines of what I was... My perfect world was very different, Mark.
00:43:16 - Well, I was a morbid idealist when I was a child, but now I care about this.
00:43:21 - I'm one thing only, you know.
00:43:22 - Life has changed a lot. That's right. I've become very narrow in my perfection ideals.
00:43:28 - He just wants obedient roboticists, you know.
00:43:31 - Yes, my idea of a perfect world is where roboticists take these ideas seriously.
00:43:34 - But seriously, I mean, it would become pretty close.
00:43:41 - So we have some clear ideas. Also, we have the beginnings of a circuit diagram,
00:43:47 - if you want, how these signals play out.
00:43:49 - But now what you also showed, which is really very curious, is that you really
00:43:54 - can see a very strong difference of how these twitch signals might actually
00:43:58 - impact thalamus, cortex, and hippocampus in wakefulness and sleep.
00:44:03 - And what you showed there, which I found very surprising, is indeed also these
00:44:07 - twitch signals, you can show like very strong responses in thalamus and cortex.
00:44:13 - So how dominant are those signals?
00:44:18 - Very dominant. it. The signals that we see in twitching in a cortex and in the
00:44:23 - parts of thalamus we've looked at, which are the parts of thalamus that process sensory input,
00:44:28 - they are arguably more activated during REM sleep than they are during any other
00:44:34 - stage of these animals' lives.
00:44:35 - Now, again, we are not looking at animals freely moving in a nest.
00:44:39 - And I think it's very important to say that because obviously there's a lot
00:44:43 - of things that go on in an animal that's engaging in its environment in in a
00:44:45 - more natural way than we are able to pick up at this stage.
00:44:49 - But nonetheless, in an animal under our conditions,
00:44:53 - the activity that we see is so precisely associated with twitching and so shut
00:44:58 - off during wakefulness that that gave us the clue to start thinking about this
00:45:06 - coronary discharge situation.
00:45:08 - So if we had gone immediately to a more naturalistic setting,
00:45:12 - we would have never seen the pattern of activity
00:45:15 - that we saw and we would have never been able to dissect out
00:45:18 - this aspect of the circuit but the thalamus and
00:45:21 - cortex are very different i'm talking about sensory motor cortex we've recorded
00:45:25 - from visual cortex and it has a lot of the same activity patterns but in that
00:45:29 - situation it's driven by retinal waves and the retinal waves are not occurring
00:45:33 - during sleep or wake they're just they're happening on their own cycle and so there
00:45:38 - are parts of development that have no tie.
00:45:41 - To the sleep-wake cycle, as far as we can tell.
00:45:43 - We do see activity in visual cortex that is independent of the retinal input.
00:45:50 - That is associated with sleep that we haven't, that we're still working on.
00:45:54 - But what you essentially see is in sleep, if I have a twitch,
00:45:57 - you will see a spindle-like associated response with a very short latency after
00:46:02 - the twitch in thalamus and cortex.
00:46:04 - Right. About 100, about 100 milliseconds, give or take. Okay.
00:46:07 - Yeah. So, but that, how do you interpret that spindle response?
00:46:13 - This is low frequency, high amplitude.
00:46:15 - Yeah. So what's it doing? Well, you know, in the adult literature,
00:46:19 - people talk about spindles playing an important role in memory consolidation.
00:46:22 - But here we're talking about movement-related activity.
00:46:25 - It appears that what makes cortex homogeneous from a functional perspective
00:46:30 - is it doesn't matter what kind of input you have to the cortex,
00:46:33 - it's going to show a spindle.
00:46:35 - The first thing it shows is this oscillation, this little reverberation in the circuit.
00:46:42 - But it is accompanied, certainly we've seen it in older ages,
00:46:45 - by unit activity, by actual neuronal firing.
00:46:48 - So it's not just the local field potential.
00:46:50 - You also have the action potentials accompanying it.
00:46:54 - But how the spindle plays a role in actually altering circuitry,
00:46:58 - we don't know. Would you see this as a reset signal?
00:47:02 - Because basically what's happening, if you pump so much energy into a small
00:47:07 - volume of cortex, you will entrain all the neurons in that volume.
00:47:11 - And whatever these guys are integrating, you really reset that. It's gone.
00:47:16 - Right? You don't mean resetting in the oscillation sense of resetting.
00:47:20 - You mean… Well, you actually, you're just resetting all these neurons to the
00:47:24 - same state, essentially, right?
00:47:25 - It's possible. We've never, I mean, we don't have any evidence for that,
00:47:27 - something like that yet.
00:47:28 - At it and but one thing that you see early in development that's so
00:47:31 - striking is that the the cortical activity is so
00:47:34 - quiet i mean it's like it's like the stillest of
00:47:37 - water and then a twitch happens and it's just a little ripple in the surface
00:47:41 - and then it goes away back to you know and if you record we what we need to
00:47:45 - do is to sort of get what you're asking about is really record from a much larger
00:47:49 - population of neurons to get a better sense of what the population activity of those neurons are.
00:47:55 - And there's been some work by Yuri Busaki's group on that going back 10 years,
00:48:00 - but there needs to be more of that sort of work to really see what's going on more dynamically.
00:48:05 - But now the, if, because I think for your model to work, I think it's.
00:48:12 - I would expect that you want to show is that I trigger my spindle, my twitch,
00:48:17 - I get a recurrent response via spindle, and now learning something at the cortical
00:48:24 - level because I'm actually wiring in or rewiring my cortical control over that
00:48:29 - little set of muscle fibers.
00:48:32 - Yes. Well, that's one dimension of plasticity that may be important.
00:48:35 - But another dimension, and that's what I tried to highlight at the end,
00:48:37 - was just that cascade of activity from area to area to area to area.
00:48:42 - And so what you may actually be doing, and this could be one of many possible functions,
00:48:47 - is that you're really connecting up these individual pairs of areas so that
00:48:51 - an area fires and then a very high probability of a neuron in the next area,
00:48:56 - that is what allows for the sort of Hebbian process of saying,
00:48:59 - you know, now I'm talking to you and we're part of the same system.
00:49:02 - We belong to the same club and the, and you, you don't belong to my club.
00:49:06 - You're, you're not part of the right forelimb. You're part of the right hind
00:49:09 - limb or, or you're just another part of the body and I don't care about you.
00:49:12 - So if you're, you know, the process of developing these systems,
00:49:15 - these topographic systems involves not only saying what, what,
00:49:18 - who, who I belong to, but who doesn't belong to me.
00:49:20 - And that process across areas may be just as important as what's happening within
00:49:24 - a given area. Okay. Does that make sense?
00:49:28 - Sure. Sure, because what you're saying is, look, what I do now,
00:49:31 - I'm twitching, I generate a strong response in my thalamus.
00:49:36 - This is now driving cortical volume. And now I get some ripple going through
00:49:40 - my cortical network, which allows neurons associated with that twitch to be linked together.
00:49:47 - Exactly. I think that's a perfect extension. Yeah, absolutely.
00:49:50 - And however, the alternative could be to say, well, when you twitch, just ignore it.
00:49:56 - Reset the cortex, whatever's going to happen, we're going to ignore it because
00:50:00 - we're just twitching. It's completely meaningless.
00:50:03 - So how can you refute that interpretation? I can't refute it.
00:50:06 - I can only use, I mean, you know, like I said, I mean, you know,
00:50:12 - I've been trying to be skeptical about the possible, you know,
00:50:15 - there's always the possibility that a system does things and it has no functional
00:50:18 - importance whatsoever. I mean, you have to keep that.
00:50:21 - Perhaps that has to be your null hypothesis all the time. And that had to be
00:50:24 - mine, you know, until very recently.
00:50:26 - And what convinced me that that wasn't the case was, first, the fact that Twitch-related
00:50:30 - information is getting into the brain and Wake-related information is not.
00:50:34 - But also, secondly, the sheer enormity of the activity we're talking about here.
00:50:39 - So, you know, there's a big emphasis in psychology and neuroscience about the
00:50:43 - quantity of input being an important player in developing systems.
00:50:48 - And I don't see how you can avoid the fact that hundreds of thousands of twitches
00:50:52 - are happening every day and activity is happening throughout the entire brain that is so profound.
00:50:59 - It would be extraordinary to me if this was not functionally important.
00:51:02 - If the nervous system could actually ignore all of that activity,
00:51:06 - that would seem to be unprecedented.
00:51:09 - And so it's not just activity, it's highly structured activity with real information content.
00:51:15 - It seems to me the brain is in the process of making sense of information, not ignoring it.
00:51:19 - So I just would be stunned if at this point it would be… Perhaps some analysis
00:51:25 - of the spatiotemporal patterns would help that.
00:51:28 - I don't know how far you've gone down this route, but, you know, for instance,
00:51:31 - you might hope to see that all the different potential ways in which you could
00:51:38 - move a particular muscle would be explored over a period of time rather than just one.
00:51:45 - Because you perhaps want to imagine that your motor maps and cortex are seeing
00:51:51 - the full range of potential ways of controlling that system.
00:51:54 - And maybe also to say to it, look, you've been doing this action a lot today.
00:51:58 - Remember there are these other actions that you could do and make sure you retain space for those.
00:52:03 - I think that's a great point, and we have made some inroads on that issue.
00:52:07 - We've been doing some filming of rat pups when they are in a harness,
00:52:12 - so we're filming them from below.
00:52:13 - And we're looking at if you imagine looking from
00:52:16 - below of a of a the tip of a limb that limb
00:52:20 - is going to occupy different parts in a two-dimensional space that you
00:52:22 - can see from below and what is very striking that you see is that when the animals
00:52:27 - wake up and they're kicking their their movements as you follow them you know
00:52:31 - represents a very small part of space compared to the twitches themselves which
00:52:36 - are filling up almost like a spherical and elliptic elliptical region around that mid-rest point.
00:52:42 - And that tells me, along the lines of what you're saying, that these animals
00:52:46 - are really exploring the full range of biomechanical possibilities of their limbs.
00:52:50 - Well, there's another aspect to this, actually, which you often ignore.
00:52:54 - Often we think about our bodies as if it's a robot, right?
00:52:58 - But that's actually not really the case, because once you drive the same muscle
00:53:01 - fiber frequently, you deplete it of glycogen, it will start to get fatigued,
00:53:06 - and you must start to recruit other muscle fibers to maintain that same movement.
00:53:10 - I don't know if you ever ran a marathon, but you'll discover how many different
00:53:13 - gates you can generate if you run a marathon, at least at a decent speed.
00:53:17 - So the point here is, and this might be also an exploration of alternative,
00:53:23 - let's say, muscle fiber groups that can be recruited into a functional unit.
00:53:30 - So the point is, if you would only depend on what you really use during the day, you,
00:53:35 - You might also, in some sense, set yourself up for a fall because you become
00:53:40 - highly specialized and you cannot go for, let's say, alternative movement patterns
00:53:44 - anymore, recruiting associated or alternative muscle fiber groups.
00:53:48 - I think that's a great point. And there are similar phenomena where you could,
00:53:51 - in addition to what you're talking about of depletion, but also if you put a
00:53:54 - restriction on somebody's movements and how quickly people can then go to an
00:53:59 - alternative way of doing things.
00:54:00 - You know, you stick a cork in your mouth and you can figure out another way
00:54:03 - to voice, you know, what you want to talk about. Shall we try that?
00:54:08 - So I think the online motor plasticity that we have is extraordinary.
00:54:15 - And where does that come from?
00:54:17 - Even when you have never experienced something before, you very quickly can
00:54:20 - figure out how to get around that barrier.
00:54:23 - And I think that could be, you know. Right, because imagine you would have to
00:54:26 - figure out at that point in time how to innervate your muscle in order to drive
00:54:31 - it in some alternative pattern. Yes.
00:54:36 - I think when we were talking earlier, you mentioned that you thought it would
00:54:39 - be difficult to block twitches.
00:54:41 - So some of the obvious experiments where you prevent animals having twitches
00:54:45 - and then you look for deficits are not going to be easy to do.
00:54:49 - But are there other experiments that you're trying to do which will more directly
00:54:55 - show a causal link between the twitch and some of these benefits that you think are going to have.
00:55:01 - This has been an incredibly frustrating part of doing this work.
00:55:04 - I mean, you know, for the people who are looking at the development of the visual
00:55:07 - system and you have retinal wave activity.
00:55:11 - The retina wave activity, this is at a time of an animal's life when they have no wake.
00:55:16 - They're not able to receive light. So it's just retina, it's spontaneous activity,
00:55:21 - it has nothing to do with light.
00:55:22 - And I envy them the ability to stick a needle in the eye, sorry that sounds
00:55:28 - a little gross, and inject a drug that simply blocks all that activity and look at the consequences.
00:55:33 - We can't do that. I mean, first of all, what we're looking at is a sensory motor
00:55:37 - system that has these loops. loops and we're looking at a system that is specific
00:55:41 - to sleep, whereas retinal wave activity is not.
00:55:44 - So we have a double problem that, uh, that people working on pure sensory systems just don't have.
00:55:51 - And that has proven to be a, uh, an incredible barrier to nailing down something.
00:55:56 - So instead what we've done is we've relied on converging evidence and we've
00:55:59 - tried to look at, you know, sort of circling the drain of all,
00:56:02 - of all the various things that we might predict based upon, you know, our, our hypotheses.
00:56:07 - And we are trying to get at this issue and we have tried to use things like
00:56:11 - optogenetics, but there are limits, uh, for our system and using it.
00:56:15 - I won't go into the details, but there are, um, it's not as easy as it is,
00:56:19 - for example, if you're doing adult work. Right.
00:56:21 - So, um, I never thought I would say, you know, optogenetics is easy,
00:56:24 - but, you know, but it's relatively harder to do in an animal that,
00:56:27 - and that's, uh, very young.
00:56:29 - So So there have been these impediments to solving, you know,
00:56:34 - this problem of function once and for all.
00:56:38 - But I think we're going to get there. We have ideas. We have certain sorts of
00:56:42 - conditioning or learning paradigms that we've thought about being able to employ.
00:56:46 - And I think I'm hopeful in the next two or three years we'll nail it down. And is there some...
00:56:52 - Follow through or crossover into sort of treatments are
00:56:56 - there people for whom you know maybe kicking more
00:56:59 - in their sleep would well you know some benefit
00:57:02 - so we just got funded by the gates foundation to to branch
00:57:05 - out and to start looking at uh human infant
00:57:08 - development and we want to know first of all you know how much do human infants
00:57:12 - twitch and also are there individual differences and ultimately the reason why
00:57:18 - uh the the whole point of this particular project um was to conceive of using
00:57:23 - twitching as an early and sensitive indicator of developmental disorders.
00:57:28 - A lot of these psychiatric disorders, some might argue every psychiatric disorder
00:57:33 - or developmental disorder, has a motor component to it.
00:57:36 - Sometimes the motor components are enormous and have been largely ignored.
00:57:39 - Autism has a huge motor component. Schizophrenia has a huge motor component.
00:57:43 - And the question is, where do these problems come from? Well,
00:57:46 - if they start as developmental problems,
00:57:48 - then we should know what's going on developmentally, And it's conceivable that
00:57:52 - how the motor system develops, which is so important to the problems that autistic
00:57:56 - kids have or schizophrenics.
00:57:59 - I mean, if you understood whether or not there are problems in the early development
00:58:03 - of the sensory motor system, the early organization of corollary discharge systems,
00:58:07 - that we might gain a foothold either into predicting and then ultimately treating,
00:58:13 - better treating these sorts of problems before they become full-blown.
00:58:17 - That would be the, you know, the pie in the sky kind of dream for this sort of thing.
00:58:23 - And that's exactly what we're trying to do.
00:58:26 - So individual differences could provide a really important clue to that sort of thing.
00:58:30 - And I talked a little bit today about autism and the role that the cerebellar system may have in that.
00:58:37 - And so, you know, we don't pay a lot of attention to how healthily our infants
00:58:43 - are sleeping, you know, but the environment for sleep could be a critical determining
00:58:49 - factor for a lot of these things.
00:58:50 - Now, we have quite an interest in stroke because we think that if we're going
00:58:55 - to have any impact in the neuropathology, stroke should be the first one because
00:58:59 - it's the simplest one, right? Having a hole in your brain.
00:59:01 - But now in stroke patients, would you see more twitching?
00:59:06 - So we have a study that we're starting, that we've been starting.
00:59:09 - We're waiting for our first patient in humans.
00:59:12 - We're aiming to look at their sleep patterns immediately post-stroke.
00:59:16 - There are two ways to think about this. One would be that, you know,
00:59:19 - one reason why stroke is so problematic,
00:59:22 - you know, massive cortical stroke and very problematic paresis,
00:59:26 - as you know, the one way to think about it is that we're out of luck because we're older.
00:59:31 - I mean, I'm not talking here about perinatal stroke, just adult stroke.
00:59:34 - And we're out of luck because we don't have access to the same mechanisms as
00:59:39 - adults as we did when we were infants.
00:59:41 - You know, everybody knows that infants are able to recover from these things
00:59:44 - much more than adults can.
00:59:47 - Well, why can infants recover so much more quickly? I would argue that there
00:59:50 - are a variety of reasons.
00:59:51 - One of them could be because of the nature of how they're still in the process
00:59:54 - of developing these systems. They're still more plastic.
00:59:58 - The other possibility though though, that I think needs to be looked at is the
01:00:02 - one where you're actually able to recruit these older mechanisms under conditions of disrupted maps.
01:00:10 - So now stroke may be too high up in the hierarchy to recruit that.
01:00:15 - So it's possible. What happens when you have amputation and you have to learn
01:00:19 - how to use your remaining arm in a functionally new way?
01:00:22 - Or what happens after peripheral nerve damage where you've lost some control
01:00:25 - and those nerves are growing back?
01:00:27 - Under those conditions, it's conceivable that the nervous system is actually
01:00:30 - able to detect a very dramatic change in the relationship between the peripheral
01:00:35 - part of your body, your limbs, and your brain, and could re-engage the process
01:00:40 - of spontaneous activity.
01:00:42 - And what if it's possible that some of us are better able to engage that process than others of us?
01:00:46 - Then you could have individual differences in recovery from these sorts of things.
01:00:49 - Those are the kinds of questions I want to address because if,
01:00:53 - you know, even if it's partially right, there could be a way eventually to use,
01:00:58 - just like people are recording from,
01:01:01 - neurosurgeons are recording in many centers now across the country,
01:01:04 - including at the University of Iowa, in epileptic patients with chronic electrodes in the cortex.
01:01:10 - Cortex, it would be possible ultimately to implant electrodes into the brainstem
01:01:15 - and to stimulate what's happening in early development and to basically re-bootstrap
01:01:19 - that system from the inside out.
01:01:22 - So if we could recruit twitching, if twitching turns out to be important,
01:01:26 - it's not crazy science fiction to think that you could put in electrodes into
01:01:30 - the brainstem and stimulate different parts of it and reinvigorate the development of the system.
01:01:36 - But how about starting with just direct muscle stimulation?
01:01:39 - Well, that might work to some extent. I mean, we saw from Brian Kolb's talk
01:01:42 - that peripheral stimulation through stroking can have an important effect on plasticity.
01:01:47 - But I think that if we're right about the role that twitching plays as a self-generated
01:01:52 - movement, it may be important for the nervous system to both produce the movement
01:01:55 - and to get the feedback itself.
01:01:58 - Still a very interesting open question.
01:02:01 - So another issue that I was curious about, so if we're now thinking about using
01:02:06 - twitching to set up our control system for motor control, another important
01:02:12 - aspect of motor control is its rhythmicity.
01:02:14 - There are these ideas around different people that also motor control is very
01:02:20 - much segmented in time at a certain frequency, let's say 10 hertz or what have you.
01:02:25 - Would you see similar kinds of patterns or patterning in how twitches are generated?
01:02:31 - And how does that vary over development, if at all?
01:02:36 - So we're looking at that issue right now, and we're still analyzing data.
01:02:42 - We're very interested now in a variety of different rhythmic patterns that we see.
01:02:47 - And what we're primarily seeing as of right now at these particular ages and
01:02:52 - in the parts of the brain we're looking at is that the most rhythmic activity
01:02:55 - occurs when the animal is not moving at all.
01:02:57 - And then what happens when the animal starts to twitch is that those rhythmic
01:03:00 - patterns disappear and then the animal stops twitching or stops moving and the
01:03:05 - rhythmic patterns commence again.
01:03:09 - But and we also see evidence that you know at one age for example we're recording from the red nucleus,
01:03:16 - and we see rhythmic activity or actually I should say very little rhythmic activity
01:03:22 - but just a four day difference from eight days of age to twelve days of age
01:03:26 - all of a sudden, all of this rhythmic activity is coming in.
01:03:29 - And if we anesthetize or inactivate the cerebellum,
01:03:35 - We get rid of all of that. So it appears as if the cerebellum,
01:03:38 - just that four-day period, is now providing rhythmic activity into the red nucleus.
01:03:43 - And so we're trying to follow that entire circuit to understand how that occurs.
01:03:47 - But it's a very dramatic effect. The difference between recording from the red
01:03:51 - nucleus of an eight-day-old and a 12-day-old is night and day in so many different dimensions.
01:03:55 - So that turns out to be a really interesting, potentially sensitive period for
01:03:59 - the development of the system.
01:04:00 - But now, if we talk about seeing or not seeing this kind of rhythmic activity.
01:04:05 - I mean, if you look at the awake brain, of course, it's configured in a very
01:04:10 - different kind of dynamical state.
01:04:11 - The way the thalamus will operate is very different.
01:04:15 - Right now, it will be much more in, let's say, a mode where it is passing more
01:04:19 - high-frequency responses. It's not hyperpolarized so much.
01:04:23 - So it might be the case that you still see these kind of low-frequency events,
01:04:27 - but they're masked, if you want, or much more attenuated because of the ongoing
01:04:31 - activity of the thalamocortical system.
01:04:34 - So is that the way you think about it? Yeah, I mean, I think we're still very
01:04:38 - early in the process of trying to understand this.
01:04:41 - I mean, the motor system, the actual overt behaviors of these infant rats,
01:04:47 - I mean, if you think about, you know, like a third trimester human infant,
01:04:50 - the behaviors are fairly limited in their complexity.
01:04:53 - And, and, um, and so, and the thalamocortical contributions are minimal.
01:04:58 - So we're not seeing a lot in terms of the complex rhythms that you see later
01:05:02 - in development when behaviors become more complex.
01:05:05 - And so in the way you described the system, even though we're talking about the learning,
01:05:11 - the development of, of, of a sensory motor system and motor control in the,
01:05:15 - in the motor control literature, people are very much carried away by this more,
01:05:19 - let's say, pseudo formalistic perspectives, like it's all Bayesian and it's
01:05:23 - inference and it's forward models and so on.
01:05:25 - And so many things have given us, let's say, a handle on certain aspects of motor control.
01:05:30 - But what's interesting is that in your analysis of the system,
01:05:33 - you don't really seem to build on those constructs. They don't seem to give
01:05:37 - you leverage in trying to understand the development of this motor system.
01:05:41 - So how do you see what you do and your data and also your interpretation relate
01:05:46 - then to the more standard field of motor control?
01:05:51 - Well, that's a great and deep question. So, yes, you know, I've read the literature.
01:05:56 - I, um, and one thing that always strikes me is there's always a little box that
01:06:01 - says, you know, prediction, but there's no box that says where that prediction comes from.
01:06:07 - And I think that is the missing developmental piece. I mean,
01:06:09 - developmentalists always,
01:06:11 - I mean, it's like, it's, it's, it's what we talk about over beer is,
01:06:16 - is how it gets, it doesn't get integrated into these formal models because unfortunately
01:06:21 - the formal modeling and then from a developmental perspective,
01:06:24 - there's a very low, well, you can tell me, right?
01:06:26 - How much do you know about a developing model?
01:06:28 - And I've talked to computational modelers about it.
01:06:31 - Most people I've talked to have not been interested in modeling it that way.
01:06:35 - But here's my challenge then to that community.
01:06:38 - Wouldn't it be better if we understood how a system develops computationally
01:06:41 - and in the process add it on?
01:06:44 - So know what is the simplest computational aspect of a system?
01:06:47 - So take the system we're working on.
01:06:49 - Wouldn't you want to know whether or not the inferior olive is doing an error
01:06:53 - signal early in development or not?
01:06:56 - And when it starts to have an error signal, then when you add that on,
01:06:59 - what kind of change does that have to the behavior of the animals?
01:07:01 - And by understanding the development of the system, wouldn't you learn something
01:07:04 - more about what are the essential computational ingredients to the behavior at each stage?
01:07:09 - And what things are critical, necessary, and what things are not?
01:07:13 - I think that is a missing perspective from the computational community.
01:07:18 - I think we would gain so much from being able to take development seriously
01:07:23 - and try to build these models in a way that mimics the developmental process.
01:07:29 - I think something we might touch on before we finish that listeners might be
01:07:34 - interested in is the effect of sleep deprivation.
01:07:38 - So if I take smart drugs and I don't go to sleep for a month,
01:07:43 - am I going to see some... You can take stupid drugs for the same thing.
01:07:46 - Am I going to see any motor consequences? What would your theory predict?
01:07:51 - Well, I don't know how much... I mean, anybody who's engaged in self-deprivation
01:07:58 - for whatever reason, probably has noticed that motor coordination is one of the things that goes.
01:08:04 - You know, you'll stumble around a little bit more, but you also have a lot of perceptual problems.
01:08:08 - So it seems as if from that standpoint, for whatever reason,
01:08:12 - because it's not really known, you know, what's happening in that situation.
01:08:16 - But several days of sleep deprivation and you're going to look a bit drunk,
01:08:20 - which is interesting since drunkenness is very much mediated by the the cerebellum,
01:08:25 - but, um, at least the behavioral component.
01:08:28 - Um, but you know, the, the, the experiments that have been done on deprivation
01:08:32 - have largely focused on the physiological consequences.
01:08:35 - Um, I, I'm sure there's, and some aspects of a sequence learning and things
01:08:40 - like that have also been studied, but, um, uh,
01:08:43 - yeah, I mean, there, there was a drug that came out about 10 years ago now called modafinil,
01:08:48 - so known as provigil that people thought was going to be,
01:08:51 - you know the great the great keep you awake drug um but
01:08:54 - i don't think it's it's advisable to keep yourself
01:08:57 - awake on drugs like that you you know do
01:09:00 - so at your own risk so so mark
01:09:03 - and i mean you are really
01:09:06 - digging very deep in in in a phenomenon that that from the outset might look
01:09:11 - like almost trivial if you look at the brain as a whole right because we think
01:09:14 - about memory and consciousness or whatever but now you found that actually by
01:09:18 - really understanding the twitch Which you actually can really gain a deep insight
01:09:23 - how brains operate and configure themselves.
01:09:25 - So if we would like to follow, now in your tradition, what is Mark's law that
01:09:31 - we should write on the wall and follow every day? Yeah.
01:09:35 - Um, just one, I only get one law. Yeah, we're going to have to, no problem.
01:09:39 - Well, I think, you know, the, the, the, I'll say it from a, from a scientist's
01:09:44 - perspective, you know, it, it can be a bit scary studying something that everybody,
01:09:49 - everybody, uh, considered to be of no consequence.
01:09:53 - And one of the joys I get out of doing science is to try to do things that people
01:09:59 - find counterintuitive and surprising.
01:10:01 - There are risks with that. I mean, it could be that everything we've ever done,
01:10:05 - you know, turns out to be of no consequence whatsoever, in which case the egg
01:10:09 - on my face will never come off.
01:10:12 - But, you know, after you follow the data and you try to generate good hypotheses
01:10:16 - and you try to keep an open mind as best as you can, and you let things go as
01:10:21 - they have. And, and in this particular instance, I've, I'm, I'm personally relieved at this point.
01:10:26 - I think it's going to turn out to be, uh, of, of great value.
01:10:31 - Um, but there's always that possibility you're wrong and that's just the nature of science.
01:10:35 - So I guess my law would be, um, we need to encourage risk-taking,
01:10:40 - you know, I mean, it's very easy to, to, to, for, for all of us to study the same exact thing.
01:10:45 - But it's, I think it's also nice to, to enjoy being the first person to study
01:10:48 - something that no, you know, not only that people weren't interested in,
01:10:51 - but actually people thought were of no consequence whatsoever.
01:10:54 - And I get, I actually, I do get a lot of joy in, in, in sort of digging things
01:10:59 - up and seeing where they lead and being surprised. Right. So take risk.
01:11:03 - But then, so five years now, Tony likes traveling. Yeah.
01:11:07 - Tony will come visit you, and he will come and check whether you actually found
01:11:11 - evidence for a hypothesis you're going to generate for us today.
01:11:15 - So what's the most ambitious hypothesis you want to see realized and confirmed
01:11:20 - five years from now in your program?
01:11:23 - Well, there are several. So one would be that we want to show that Twitches
01:11:27 - are playing an active role in plasticity and the organization of the system.
01:11:30 - We want to understand what parts of the system are providing the critical inputs
01:11:35 - that lead to these different aspects of modulation.
01:11:39 - I want personally to understand how twitching changes is differentially expressed
01:11:44 - in different species and across the lifespan in different animals.
01:11:48 - And I want to know, I have, I want to understand how body morphology,
01:11:52 - like whether or not you have a trunk or whether or not you have a fin or whether,
01:11:54 - you know, whatever, uh, how your body, whether or not the most important parts
01:11:58 - of your, of your, of your bodies are the ones that twitch.
01:12:02 - And then I want to understand, especially in humans, whether or not we can see
01:12:08 - that twitching actually can play a role in the maintenance and recovery of function
01:12:12 - that happens after skill learning, stroke,
01:12:16 - peripheral damage, whatever, so that it actually might be playing a role in
01:12:20 - those sorts of processes as well.
01:12:21 - So really just sort of proving the foundation and expanding it into other domains,
01:12:26 - I think that's where we need to go. All right, great.
01:12:28 - Mark Bloomberg, thank you very much for this conversation. Thank you. Thank you.
01:12:35 - But I don't share your utopic dream of a world running around with submissive
01:12:41 - roboticists. Sorry, Mark.
01:12:44 - That was going too far. A little weak on the dreaming, I guess.
01:12:52 - I guess that's a price you paid, you know. I dream of a world of roboticists.
01:12:58 - So beautiful. My ideal organization.
01:13:01 - I have been known to engage in hyperbole, so, you know. Well, you succeeded.
01:13:08 - That was great. Very good. Thank you. The CSN podcast was produced by the Convergent
01:13:17 - Science Network of Biometrics and Biohybrid Systems,
01:13:21 - a project funded by the European Sevens Research Framework Program.
01:13:27 - For more interviews, recorded lectures, or upcoming conferences in the field
01:13:33 - of biometrics and biohybrid systems, go to csnnetwork.eu.
01:13:39 - Music.

CSN Podcasts

Mark Blumberg on rem sleep and twitching

Season 2015

Description

CSN Podcasts

Episodes

Matthew Diamond on whisker system and decision-making

Stephen Noctor on cortical development and precursor cells

Ranulfo Romo on decision-making and somatosensory cortex

Narender Ramnani on cerebellum and cortico-cerebellar loops

Kate Jeffery on spatial cognition and grid cells

Timestamp

Be the first to leave a comment

Leave a commentCancel

Legal

Follow us on

Mark Blumberg on rem sleep and twitching

Description

Episodes

Timestamp

Be the first to leave a comment

Leave a commentCancel

Tags

Legal

Follow us on

Login to enjoy full advantages

Please login or subscribe to continue.

Go Premium!

Enjoy the full advantage of the premium access.

Stop following

Cancel subscription