[00:00:05]
>> It's my pleasure to read Dr Gerson. That you see in her in the physiology of this little heart there and he knows that. If you were watching the l.. And one in China at the nationals and says and a.j. I don't know if you can that's going to hurt either of the Quester before the link you can cause else I don't because he would let me off but that's what I'm for.

[00:00:41]
Right and I think that's what I mean here. But. There. Really I mean if you know they're going to. Change go ahead. Yeah Ok just speak up if you can hear me and just let me know Ok And thank you Simon and thank you the law for the invitation to speak at your seminar serious I haven't been to Atlanta since 1906 and it's like I've been plunged into an alternative and so beautiful I say should all these very proud of the earth your beautiful campus and city.

[00:01:35]
And so today I'd like to tell you about the work that we do in my lab which is really focused around the topic of precise motor control and we absolutely must reaching movements as a model for the us and we investigate questions of precise motor control when you're.

[00:01:55]
Studying Seldon's and out of service in the place. Quantitative characterization of the behavior so that we can alternately understand disease models I'm taking it all the way back to Sultan. And so it's I guess common these days to make disclosures I do not have any financial conflicts of interest or importance however I am going to be talking about Cyrano and the reason I require sun amount of disclosure is that apparently it really borders on people like this illustrator who has battle is the color of the cerebellum and Greg being that.

[00:02:37]
You know sells a little short the real cerebellum is only slightly less gray. But I get it really and in fact when we start to look at the micro structure and that we come to any patterns we can serve on that it's really spectacular beauty is revealed and so this is just a cross-section of the human services I mean you can see what are called the trees of light and in Lat Am I mean if you sue me.

[00:03:04]
And at a microscopic level you'll see where every Monico all really was activated with which is a gorgeous make research that's come close to a distinct apparatus of my fibers they contact the supernumerary Randall's cells which outnumber all the rest of the cells you're going to you. Will solve projectile the cerebellar of molecular layer with a crown fact that's been tackled and really trees uprooted in her hands which are really the work horses of Sarah Miller company.

[00:03:35]
And they'll send their absence down to the bottom of the survey done which is stably work out of the structure in the cerebellum you have. Been so much about only telling you today is really focused and circle in the eye and then the end we're going to circle back to trying to understand how Mr Miller and circuits actually compute information that helps movements become to sites now I've been alluding to Chris a smooth events and so I'd like to.

[00:04:04]
Take you back to another contemporary of Ramoni off or Holmes who is a position who operated around the time the 1st World War and I'm like he was interested in that the neurological deficits that emerged with what turned out to be girly precisely Asians at the cerebellum in this case or other regions of parts of the brain that were produced by gunshot wounds m during the war and what's remarkable about this image is that it's a essentially a card was asian of a movement that was made in the painting fighting if you just by catching a label to the finger of the man with the cerebellum or the agent and then using a long photographic exposures so that he can actually parked right it's the motor deficit that emerged in this industry people or.

[00:04:56]
So what he did is a classic neurological exam from disorders called finger to think the finger nose tip to bigger task So essentially you got problems moving his finger these targets and the vision was asked repeatedly point from his nose to these 3 different points in space. And what gives earth is that there is a distinct off lesion and occur around the end of these movements I mean he thought that these operations which turned out to really characterize terrible reviews and this work interesting from the point of view of how I feel that there are certain forces.

[00:05:35]
Though as a bunch of engineers you'll be more familiar than I am with this idea but it's true that when you want to be back errors to guide the outputs of control systems there's typically some level of oscillation That's the classic example of this is something like the thermostat when she uses an error to control the heater and the room temperature will oscillate around settling and this is true in reaching movements as well where if you were to say yes this will be back by the end you would find your little Karim's caps that start again because visual on information is rather slow to get even through the retina.

[00:06:19]
And so be ideal that is that. Engineers as well as the brainy must use some level prediction to implement the forward control so the cards can be achieved without these kinds of oscillatory errors. And because these are solutions emerged when serial or damage and has been hypothesized that some of the people controllers implemented by this are about.

[00:06:48]
The basic idea is that the cerebellum takes in through its model matter of fact way. Sensory feedback information as well as corollary destruction and that is in commission enters the cerebellum through the use of modern drivers which then are transformed in the grand old so parents can provide an average sensory motor contacts that can be used by Indians out in the circular cortex to learn what to call in from all models are sort of representing each hand and predictions about there's a body.

[00:07:27]
In your and then Jack in the cerebellar nuclei including a structure called the interpose nucleus what you're then 5 the size the generate the forward control of the body through its outputs to line the brain stem and adults which ultimately influence I hold water in our so if this is the case how can we how can we look for evidence of people who are controlled from the cerebellum so I already showed you that reaching movements are known to be.

[00:08:00]
In here and certainly are needed and so those kinds of behaviors are going places well and I also go on. Low point by point out that a number of lines of research have indicated this particular structure the interposed and the glee as I'm in the control of the reach him.

[00:08:20]
So this is a study from Clive got the spot where they. Impacted you small into the end your interest was nucleus and the fact it was in the past became somewhat had a metric and a more recent study from our chance lab until the company. Out in theory was that we haven't found that region remains actually became somewhat hyper metric in these mice now what are the limitations in the in sort of understanding the reason that these kinds of the reaching errors occur in long activation say is that this doesn't really distinguish active control from that it's an initiation we're planning on nor does it identify any kind of control policy that's really been a domain of questions for busy ology and indeed people have recorded from the anterior because nucleus reaching animals so this is a study from Peter both hands grew where they identified Non's in the end Syria has not really has that are strongly model even during reaching However when the same group.

[00:09:38]
Had monkeys use of the device to push in different directions in the same way that they were reaching in your corrections insulter and very importantly much related and so this presented something of a conundrum because if you think of these experiments they were not able to attract and commit acts of the reach and when there was a great deal of my relation to the study when they contract it can access using devices there was very little modulation to say so they either have the right to be in a can imagine or they have a few Maddux or new neural activity.

[00:10:14]
And so this is sort of where the feel was even though there have that other observations relating kinematics. There is really no consensus about what was or dollar output structures in code that's valuable for reaching the parents and so we decided to take advantage of that revolution in motion tracking technology to try to get a better handle on what is the interior pose nucleus.

[00:10:43]
That contributes to reaching men's and started out with just recording neural activity during I'm just reaching up and smites And so our partners in years I can backtrack ing a lot of the grain from the answer because yes so basically we use a and I repeat a reflector that allows us to Amtrak motions very very precisely so we get real time you can imagine that in 3 dimensions and this allows us to do I'm characterize reaching movements across many many reaches so that we get a rich I can have a data set that we can then use to compare neural activity times when the 1st thing we do serve is that many neurons in the Ensure goes against our mightily very much so this is just a wrapper class showing all activity a lie from the start of the reach you can see that there is clearly modulation we can see this among the entire population of neurons that we record that there is a good deal of neural activity that is occurring there in the region.

[00:11:52]
Ok but you can also appreciate if there's this activity it's not particularly well structured and if we return to be observations of board it forms as well as the guest lab and the child lab get sees that need I'm going to reach start right now you know why this thing to look at because they're really preferential levels it's in the end points of movements rather than anything at the beginning and so when we.

[00:12:21]
Align our normal activity instead of into the restart if we were widely employ the 5 of the neural activity is much more clearly structured associated with the end point and this was true across the population where you find this population of neurons there are various strongly endpoint associated I mean you can see that this population is under clearly is as good as going to the celebration of the lamb and as shown here of the population level and where the system is and it really spans the endpoints so both of or and after in this entire nice alteration our He's all racial.

[00:13:06]
So this is sort of the place where a lot of neurosciences essentially run aground is to say well you see this is not to be throughly than to be kinematics in the following way. But does this have any sort of cause or relationship was a producing puts and point and so to address the question we've utilized and the fact that we have kinematics in the real thing I do under sensually prettiness and a closed loop my relation of the output of the cerebellar nuclei just during reach him.

[00:13:41]
And so the basic idea is that we have a representative marker on top we know where it is in real high now and then we can define a landmark at which point we deliver a very brief training of stimuli of light. To the empirical is going to be as bad as expressing.

[00:14:00]
The light activated absent genre so the brain is very brief it's just good if you know the seconds want to millisecond calls with and a 100 hertz and we can see whether modulation of the b.s. actually has any affects kinematics in real time so here's an example of this experiment so you'll see in the corner here a little light up on or a little dot in fact that's just indicating the duration of that stimulation and you can see that the mouse gauges in the movement in this case it's not successful.

[00:14:39]
Now there's a lot of reasons why I'm asking I may not be successful at this this task so it's helpful to look at many of these bridges and sort of compile so this will illustrate the introductory understandingly and regions and the factory of simulated ones these are reactively interspersed along the course of the instruments and so that mask he had disappeared.

[00:15:04]
Stimulation. And so what you'll see is that he still is but initially simulated enhanced in the late and regions are clearly intermixed But right after the stimulus you see that the simulated regions are lower and is not as far out as the unstimulated regions I mean that persists although there are some some late sort of corrective responses than stimulated reaches so we can summarize that show that the endpoints in our simulated conditions are consistently high go metric compared to the unstimulated regions.

[00:15:43]
So this provides I'm not these some evidence that activity in the interpose nucleus can actually influence point but these kind of positional data aren't as satisfying to look at when we're thinking about the temporal evolution of objects and so for the rest of the talk I'm going to be really focusing on velocity curves so for those if I were uninitiated Yes I think.

[00:16:11]
They will be reminded pretty quickly so this is a very bad hell on wheels for you to reach but in these ideas that when we reach out to something there's a very natural bell shaped velocity curve that occurs and then when we reach back there is a Belgian law to be her that there crosses 0 and indeed it might just when the liege I mean they still ship a lot of the profile and now we can look at the effect of stimulation on an outward colossally and observers are temporal of pollution of those changes and not surprisingly given the positional fact we see that there is a rapid decrease in the outward velocity with stimulation which is shown in Boyle and that was true across all of the animals and it's also Truman had it's mice we don't think this is an off target of fact.

[00:17:08]
We saw the events that were consistent both in the our in the upward direction such that the that there of change was consistent the reason the outward outward quasi this we also observed in the in mightily restricted absent expression 2 people are out there on that reject the rest and so this shows that interposed neuron act as a fusion alters reach can imagine is in real time in that.

[00:17:40]
Reproducible directional. This raises the question of whether this might reveal a mechanism for precise cerebellar control can act so to begin to address that question we took advantage of the fact that we can alter the power of our light which on alters the amount of bytes that we add to the system.

[00:18:03]
And we can also alter the position of our of our simulation so when we simulate either girlyman or late in the reach we always see the same direction of change and similarly when we when we have very kind of power of our light we find that we can vary the magnitude of the velocity change the suggests that when we look at neural activity in the environs stayed there perhaps not activity in the evening when and how fast the limb needs to decelerate and so the next thing that we wanted to investigate was instead of just introducing this kind of critter mation we wanted to take away spikes to say if the in darkness mechanism that we can record in vivo producing some cars will control of limiting.

[00:18:58]
And some to ask this question we use essentially the identical method except now instead of exciting neural activity we use an OS and they can turn down a neural activity. And that's called arch in this case that we used to argue are dogs and that is a proton comes they can paper gold rays now and we can show the spirit of practically hybrid polarizes neurons so what happens when we do this during reach so when we when we are stimulating during the reach we're actually able to accelerate the land wherever we put our labels in and we can see that clearly in this difference bar where we increase the amount we're going up or glassy and so you can see that bird change is in the opposite direction as the challenger drops and stimulate sort of accelerating the limb as if we're moving some kind of a great i'm a system and that ultimately previous is significantly higher metric and.

[00:19:59]
Now I just alluded to this idea that maybe there is a break in the system and that we have big breaks little breaks small breaks or the brain's turned off rate but this is a set of findings is essentially consistent with 2 different models so just to orient you to how we're thinking about this we know that motor cortex is essential for producing region with brogues cause debilitate debilitating our.

[00:20:28]
Effects on regions you can't reach any more and I have to genetic manipulations as well as music mall in accusations of order or acts of balls of course reaching to the of what we think is that motor cortex comediennes reaching movements but it does so through generation override your brain some strong force and so once we see that there are these effects by the cerebellum What is it doing to these motor cortical make.

[00:20:56]
So what I've shown you so far is that we know that when we simulate that we reduce our debt comparable lots of humor. So that could be a great. Suggesting that we're the sort of scaling this control motor signal by some amount less than one. And that. Would also be observations and also be consistent with other producing some kind of a directional ma'am so that we have a vector of change that's in this direction that we're adding on to our control.

[00:21:35]
Now the best RB do not distinguish these models however we would make a prediction that if we were to simulate upon return that these 2 possibilities would produce differing predictions the other great model would predict that it was simulated uncontrolled would slow the can would slow the return of the reach however with the directional command model we would exhilarate the return very much because we live on the stepper radio on the top of the control movement and so we performed experiments and they did our own consistent with this directional command model so centrally when we stimulate during return we exaggerate the in the activity.

[00:22:26]
And so we think that essentially the output of the cerebellum is added onto. The cortical command center that understandings of the use of the brain stem and into it in a directionally consistent way. So these are these ideas that we're revealing with after genetics. With us to ask kind of our original question again and interface which is what are the principles of the field forward control by having interposed Is there something deeper that we can say then well there is a directional command that's added on to the motor particles and.

[00:23:10]
And so we went back into our data heard the other way and recorded in the nuclei and we said Ok So what is this actually in coded. So as you can see there are trial by file variability in the Aliquippa of this of the sound. And we can in fact.

[00:23:34]
Perform simple correlational analyses and show that the firing rates are related to the amplitude of the d celebration of. And that was true in a large number of the endpoint believe in cells we get this linear relationship and then we have a school and why why is it variable from reach to reach and so we need to address that question the group reaches by firing grades which are indicated in the skull or math so these are the lowest firing regions and these are the highest very free.

[00:24:13]
And consistent with the linear correlation we find that a high scoring rate reaches out the fastest deceleration and lowest firing rate regions have slowest. Though But why so what was interesting is that it happened high from this period in which were again I think they're all firing reads We found that after deceleration happened to occur in regions that had the highest peak velocity.

[00:24:47]
As is shown here. So this suggests to us that the fastest reaches have to decelerate a fast as and he that was produced by this variable change in the output from the cerebellum and I think it may be clear on or obvious that if you have a great fast reach and you decelerate that's your the you're going to produce stable and points and that's exactly what we observe so on so this was interesting and we looked at it carefully along views a longer range of firing rate produced by the circle nuclei and we found that there was this very nice matching of the amplitude of the people Asadi.

[00:25:37]
And the endless unit of these celebrations that's the fastest regions and asked as the celebrations this is an emergency and so there was there's a relationship between people our city and the celebration that was produced by this variable outward in the in the server name. That just emphasized sort of the relationship between There's Abia and symbol and points and we produce a very simple model in which we are live in the average the celebration 2 regions that were the slowest and when we did that we find if you celebrate through basketball reach the end of when the hyper matter of record reach and similarly if you have an average the celebration on the Master's reaches you end up with a matter.

[00:26:29]
And so it's really this variable tuner that's happening in the interposing years that we think produces this untangled range of and from. And so this observation of adaptive scaling by your ear and your Goes nucleus is we think evidence for be controlled by a survivor that is is on one violation of reach and that extends based on the absolute value of each kinematics but before you answer that we get these as kind of adaptive.

[00:27:09]
And so this raises the question of how is that computation actually made himself for the next several minutes I'm going to tell you a little bit about our work in. The end I don't buy her getting a grip on how to use the or computations or being and so we will start with the premise that it's the people who are control and competitions made by the interposing readers are in part when we derive it by the inputs to the nucleus and so I have been talking about this as sort of a monolithic structure for a little while but in fact the interpose nucleus is composed of multiple cells.

[00:27:57]
And a monkey is our on our plate neurons that projects to reemerge or a guess like a revenue leaders and eventually I'll Palace among other places as well again inhibitory on a popular Asian project to be inferior all of which is another structure that I don't have my well I mean what we want to ask is What are the inputs to preorder targets and if you know that these are sufficient to marginally reach him things.

[00:28:27]
So to address this question we used a matter that was developed a how the ways lab I go sulk institution where we're able to introduce our brain is virus to specific cell types driven by a creepy promoter that rabies has been modified so that it can actually jumps in after is in fact the whole brain because a glycoprotein that's required 1st and have to jump just has been left out but in order to get once and after John the idea is that he would introduce that like a protein just to create expressing cells and so in trance that the gene like a protein can be complemented with the rabies virus and then it can jump over the rabies virus and stop in the next cell and it won't continue to back the brain because it doesn't actually have expressed in code that the glycoprotein itself and so we've identified on a creature line that expresses.

[00:29:30]
Premotor output neurons of the circular nuclei called N.T.'s are one these neurons are not. The same as those garroted neurons that I can project other places and so are you actually both of these lines and Ts are one tree line and they gather one reply in combination with these viruses that alternately allow us to emulate the inputs to these different cell types and so we perform these experiments and we can map inputs to these different cell types we can analyze the location of all the preview new neurons that go to these cells but more importantly for this experiment we can analyze the location of all the extras are going to still go to these tournaments and they all kind of information is probably giving rise to this that they're still.

[00:30:27]
Obviously I don't expect to feel these levels but what I want you to appreciate is this the colors of these bars and the relative colors of the arms and what we consistently observe is that structures that are considered motor structures are by far the dominant inputs to the cells in the cerebellum you react more so than sensory sensory structures and so if we think about this adaptive scaling observing the server nuclei we might think that perhaps some of that is derived from the fact that the whole motor system is turned up and down with the amount of gold or more vigor and that the nuclei might inherit some of that information directly.

[00:31:14]
But that presumes that the pre-puberty neurons are also moderately in concert and so we were interested in this idea and we turn to a motor structure called the red nucleus to better understand this idea that motor information which is also called corollary is from your earlier discharge I enter the cerebellum through these muscle or.

[00:31:38]
We consistently observe I have retrograde label in our am from the Israelis experience and so we chose to just inject Ament standard and her brain tracer into the star into the red nucleus and what we observed was really straight Kinkel she was cerebellum jets and should be straight into you too and in a 2nd so we had 7 there was very intense innervation and the enter in her because from the red nucleus however and what we what we found striking is that there was almost no innervation of the cerebellar cortex.

[00:32:14]
And that's unusual because in typical mocking viral kept way up the cerebellar are packed with countless we're told Marty by rosettes these are the endings of the modern pipers and so those were not they are from the red nucleus and this is just something of an alteration of the certainties that we draw all the time which is the idea that marking drivers enter the cerebellum and enervate both deserve going for a cerebral cortex really is marked by Instead we observe the 3rd if there are at least some sources of modifiers from motor structures in particular that only innervate interpose nucleus and so that's further evidence that there are Candy I'm sort of an independent My relation a serial or output that might scale directly but motor bigger to produce as scale deceleration now.

[00:33:18]
At the start of the release of her kid you know run out and and of course there are cortex out of it and we don't think that they're probably doing nothing. Like one remaining minute I'm just going to talk about how we are thinking about what the circular cortex might feel they've given you this model now a bunch of times where we have sensory information and water or mation coming into the circle or we're facts and ultimately reducing or control the idea here needs to be a little bit spelled out in that marking fiber information is possibly a highly unlikely signal so.

[00:33:59]
Say any kind of information about a lot continuously encoded by Mark Ayers and there would be this huge amount of information coming in but this kind of information is hard to learn on its side and so it's hypothesized that this context layer which is the grandson layer decomposes the signal into tiny little bases and then be used a little signals coming out of the grain all cells are more open to search a learning by the particular Hans then otherwise wouldn't would occur and so I'm not going to go into servo or learning theory except to say that we think that there is probably some kind of learning that happens at this level it probably contributes to the forward control the to address that we began a recording from for kanji runs in head fits my needs that are reaching for a fluid call.

[00:34:54]
And so when we record comprehension or ons the words of the u.p.a. we notice that there are many neurons that are beyond reach as us and what we ultimately want to know was whether this kind of activity can help produce stable end points. And so what we have begun to do all is stimulate the market fibers that ensure the system but we're doing this I'm going to a slightly.

[00:35:21]
Different way than we're doing that stimulation up to genetically in the in the interposed So here we are using oxygen ethics and what we're doing is of blocks of trial rather than randomly interspersed so that mouse is able to reach for the target $25.00 times and then we begin a block of stimulation where every reach we stimulate at the same.

[00:35:47]
Landmark for 50 reaches and then we take the stimulation. And the idea is that the stimulation because it's occurring at the same landmark every time should become a predictable signal of the particular on's could use to help guide. To guide the motion. And it so what we observe with respect to end points is that we have a spread of endpoints before we start the stimulus block then we start stimulating on every trial and work were able to actually push the end point around however we noticed that the it indeed as we predicted that the endpoint actually begins to shift to me Can you match Rick to the original endpoint.

[00:36:41]
I mean again that when we turn the stimulation off that there is a predictable rebound such that the reaching the end points are again our hyper metric river and so together these data suggests that indeed there is probably some kind of an avocation that's going on in particular on the ultimately guy endpoints in the produced by the interference nucleus but work analyzing the neural of facts of these perturbations is ongoing.

[00:37:16]
And so so that's where I'd like to end that all hate of questions afterwards but the basic idea that I want to emphasize is that we identify a thief or an old policy from serial killer outwardly as the intervals. And that policy is essentially true that if we scale the desire a change of the limbs through variable firing rates and we've made some progress in I don't know motor signals from the brain maybe introduce there a learned behavior actively on it and that Africans coming into the service are cortex really can be skill as such that they control points in.

[00:37:59]
Either actual lengths. An insult I don't have to stay tuned into we can actually tell you more about how those pretty nice in the halls are produced. And the intimate knowledge the very hard work of the students that I have had the privilege of working with never get all the work in the anterior in your hands I don't call it is working on the grid kinda styles ole John is my technician this done a lot of the rabies tracing and analysis and conceive a field in the work of the red nucleus and so I think we have I'm playing time for questions so if you would like to.

[00:38:39]
Let me know if you have any questions thank you for your attention. Peter you have very clear. Of 34 of them all but I think there are about. 3 blocks of the right at your front after the 1st approved cards at the present proxy for the same kind of course the.

[00:39:11]
Royal Guard circuit is the same as the one you are riding and complex but I would. Never Ok So Cindy either is describing just laugh a cold cereal or adaptation experiment in which. You can throw darts at a target and go you know you're on her getting hundreds of goggles which destroys your visual field by some number of degrees and then you know you start you start throwing off target you think you're Breitbart you're really awkward and then quickly you're human adapt and you're now I'm forgetting you take the goggles up now you over cracker box.

[00:39:58]
And it's counted there are other patients that they are not going to use that at each. And so so 1st of all yes I think this is exactly in line with with that whole model and what's interesting about this is that this particular version of this is that we are producing that is a version with Martin by overstimulation itself and the cerebellum seems to be able to adapt so that isolates it again to the cerebellum which is good news for us and studying the continued fires but I'm I think that once you know I'm clear in all of these experiments What is the information being used and the kid you know runs to meet their computation and so all in all of these cases we have cortical him but through the pines he knew we had them undoubtedly.

[00:40:49]
And we have all of this from the back on the great stuff that's coming in from us the papers pm the spinal cord and so isolating you know which happens doing this which happens doing that is almost intractable I think under our sort of in the guy Venus conditions and so our hope using this model is that we can just basically hijack a mosque in 5 years and we don't know which one it is we don't necessarily care and we can produce the can actually nation that's either going to be helpful or hurtful to the system and then and then test whether it actually is usable though we think it would be exactly the same mechanism and it's probably the same mechanism that's going on in other parts of the cerebellum.

[00:41:33]
So that I can think Yeah I've. Got 5 question so I'm. Very happy for my view of the. It worked exactly as per. Your young and I thought and I started thinking Ok Ok so many times the vision is you know pushing off on certain subjects who were born into the past great as opposed to what you're feeling was her vision or the warmth of your previous story and so are we are all pretty fine trajectory is there is a good article we should offer to the 3.

[00:42:21]
Ok so we I think your question is like the root of a whole set of experiments that were trying to address because we've been wondering why is it a long history. Of I'll get to get to that in a 2nd so 1st of all. We really wanted to do some mix up exact in the script or Asians in a variety of directions I have and have spent a certain amount of time developing these crude life closed loop pumps and.

[00:42:51]
The magnets they call the land so that we can do these kind of acts get an experiment I'll spare you all the gory details that they haven't worked that well but. But the question remains a good one is why why is this and why are the direction of changes so certain natural and so on so 1st of all we don't know whether it has cerebellum as incoming something concerning the bodies and or brain which with the time possibly on what we're calling actor a change in this direction really and general are active after this and her and so we're doing experiments now to alter the targets and we would expect that pending on that movement that some of these directional changes that we found to be a variant with one target are actually variable So for instance if the animal reaching down we might predict that.

[00:43:48]
Actually And rather than that so those experiments are ongoing. Series. That says. Depending on what the overall. We're going to control. In the word God in all her patients in the same way that it will rain and direction but we have to have a response with that 1st before we can start to be clear how we should respond.

[00:44:28]
That. I'm sure was the that it was deliberate I think a lot of the time when you're about likely or feels. It could cinematics matters a lot like water beach for that but is the is there a representation. But their belief. My question is how are they partly explained or should live like that these that take into account that season building a few for suddenly make the land or let that.

[00:45:08]
Go around us that we can reduce their own yes that is exactly what was predicted to hear. So all that the. Environment in the physics of the body are presumably learned of a service so the to. Predict. And really you know when you watch babies develop and things like that I recognize there's other things going on but a lot of that is probably training the internal model to be able to handle these you know complex physical interactions and so you see a lot of flavors like Hinds are.

[00:45:46]
Winning and cons. You would be good. All over yes or that's regular deal with. It with. Her I think the idea is that a transition perturbations happens and there is rapid correction and there's good evidence to suggest that a lot of that is cortically Even so I do know though is that the servant cares about that deviation from the action and then over trials where we train and so then you get application.

[00:46:22]
Over trials and then if that provision has now gone on there is they're asking reactivation but if I credible evidence. That is a learning you know that it's actually Bernie. Bernie. If it's yeah I mean we have to talk more people about the differences there I mean what more we can say is that in the end experiments in which.

[00:46:56]
Our people have been able to isolate the specific sensory community and our sensory stimuli perky runs and does in big you wait like this we'll hear or see that are equivalent to gravitational pulls and it's something that's really true combinations of includes that are playing hard to replicate that like for example if there is a this simular allusion of am.

[00:47:27]
That's the same as some kind of if you know when you're exhilaration different direction but the system can disseminate you a bad dream using comment real things and so on so I think maybe the answer is that it's not solving a physical problem in the same way that it is as well but it is evil to use these kinds of signals in ways that are ultimately helpful for the system to operate but I think it's maybe too complicated.

[00:48:01]
So let's say How do you Paula.