 So, yeah, the title of my talk is, that's the mouse, you can see that it's running, and when I thought of this title, I was mainly thinking of, you know, interesting books, as was said, unfortunately I want to be the answer to many things, but I hope I can respond to interest anyway. So, yeah, the title of my talk is, that's the mouse, you can see that it's running. And when I thought of this title, I was mainly thinking of how this visual processing change when the mouse or when we are running. And, fortunately, I don't have to convince you that this is an important question, because the whole visual input changes when you're running and you still, your brain is still to make sense of this. Also, we heard your goals may be changing when you're navigating. So, it would make sense to really adjust to processing this. And also, actually, perception is changing when you're looking also this morning and maybe I should not call it perception action vision. But yeah, so we know from human studies that optic flow is perceived slower when you're walking, but also your discrimination between optic flow speeds is improving when you're walking. And what I want to ask today is, where in the brain does this visual input and behavioral information specifically about the promotion integrate. And the point I want to make is that it's happening very early, maybe earlier than it's usually efficient. The focus I have in my lab is this area called the superior thickness or optic tectome and non-linear species. And this is just a little overview. This is a pure thickness. It's one of two areas that we see image forming information from the retina directly. In the mouse, it actually seems to be very important because it gets input from more than 90% of the retina gain themselves from the mouse. Whereas LGM and the salamos receives about 25 to 50%, which is very different from the climate. But so in the mouse, it seems to be really an important visual in the brain. So I think LGM is also interesting because it doesn't just be vision, which is more happening in the superficial areas. The deeper layers then transform this visual information to control behavior, like innate behaviors like the rest when there's a threat or escape or approach movement or suit movement, or just turning your body towards something interesting. I think it's a nice area to study because whatever is happening in the visual area may actually have direct impact on the behavior of the animal. So how do we study that? We use calcium imaging, two-photon imaging in superior filters. Unfortunately the compass is not great, but it's a bit tricky actually to get here to put, you know, window there because at least half of the area is covered by cortex, which we do not want to touch. And the other half is covered by big blood. There's also surgery is a bit tricky, but once you're there, you can have nice optical access to the neurons. And we are imaging those very superficial layers where the retinal input is coming in and most of the inferential processing is happening. So this is one of the typical average image you get when doing calcium-photon imaging. So here you see who's the green sensor, G-Camp, and you can see identify single neurons in this area. So we do our imaging, we head-fix the mice, we surround them with several monitors so they can see stuff, and when they're in the experiment, they're free to run whenever they want to, they're sitting on a treadmill or a floating ball. And we're tracking their running in pupil size to get some information about what the mouse is doing. So here that pupil size and running are highly correlated with each other, so I'm using those lectures interchangeably to talk about active or inactive state. Yes, what I wanted to say is we analyze both things, and for most of the things I will show, I did the analysis for both, and the results are qualitatively very, very similar. So here's one data set that we connected in the superior clip, so here you have hundreds of neurons, each row shows the activity. This is just during spontaneous activities of the mice, it's just being a gray screen, and we measure pupil size with a video camera and the running speed of the mouse, and the neurons here are sorted by their activity with running, and you can see quite clearly I think that a lot of neurons are positively correlated with other neurons negatively correlated. These are just two example neurons picked out, and you can see more positively the other one negatively correlated. This is just the first principle component of the whole data, just showing again that most of the variation of the neural data is quite well correlated with the activity at the running of the mouse. Now when we stimulate the mouse visually with the spreading stimuli, you still see the same pattern. So in addition now you, you may be able to see some of the visual drive so we will see how the render stimuli are coming on, and the neurons are reacting, but still you see this clear variation of activity with running. Now we looked at the processing of the rating, especially the direction training, where the pupil is large or small, and you can see here's three example neurons. There are still traces to different orientations, you can see this neuron partly responds when the pupil is small, but really increases its response to its preferred direction when the pupil is large. This neuron is quite differently, so they should risk large pupil sizes with an active arousal state, the activity goes down, and you can see this in the direction turning curve. This one is a primary neuron which is actually suppressed by the rating, and it's less suppressed with arousal. So what we see is basically a gain change of the tuning curve. What stays the same is the preferred orientation, so that's stable across the activity levels of the mouse. But this is how the response to the preferred direction changes, so we see here in the black bars that about half of the neurons change the activity significantly, and again of those half tend to increase their responses to the preferred direction half decrease their responses, which is just the same data and the comparative history. One interesting fact that I want to note is when we differentiate between excitatory and inhibitory neurons, we see a big difference. So the excitatory neurons here tend to increase their responses with arousal or running, whereas most of the inhibitory neurons tend to decrease their activity. Unfortunately, in the superior calculus, we don't have a nice circuit diagram yet, like people have in the cortex, so we don't know yet what the relationship is between these populations of neurons, but it may be an interesting fact to figure out what's actually going on. This was work I've done in my postdoc, so I've seen responses change, gains are changing. When coming to SARS-6 two years ago, one of the goals of the lab was to expand the range of visual stimuli to see what is happening when we show the mouse different things. And one thing we did with Maria, the constituent engineer at the postdoc in the lab, is to still show ratings stimuli that now change different trigger for different attributes of those ratings in this case temporal sequencing. I should say this is early, early days, early data, but I thought I'd show you this data set because I think it's really interesting but taken with a sort of brain is going to push you control so that I think it's already going in the right direction and I'm quite excited about it. You can see wrong responses from four different neurons to these different temporal frequencies, either when the mouse is quiet and gray or an insect is red. So in this neuron you see nothing is changing much, whether the animal is active or not, whether this neuron activity seems to go up when it's active most of the time. This neuron is different, responses go down with the mouse, and in the fourth neuron it's different again. So a bit similar to what we've seen before but the exciting thing is if you look at the tuning of the temporal frequency, it seems that no matter whether the activity goes up or down, the preferred temporal frequency is increased with running or loss. So the neurons like to respond more to fast stuff when the mouse is running, which may make sense because it's seeing a lot of fast stuff. So here's just the population result where we began just before, characterized how much the response is changing for the preferred frequency in this case. And here you see how this temporal frequency, the preferred frequency is changing with running. And so far in our central analysis, picking the grading that it is most of the response we see that many neurons the preferred temporal frequency goes up. In the second project, again, early results. This is by Grazia. Also in the lab, we are looking at temporal dynamics of neurons in the superior vectors here using electrical physiology. And again, we show the simple ratings, the changing orientation, but we have a closer look at how the response changes over time. And here I'm just showing two examples, which maybe point into an interesting direction. So these, both of these neurons are then in the very superficial areas. And here you can see when the mouse is running that the initial response is quite large. So these different colors correspond to different directions of the rating that we are showing the very large and then become very quickly quite small. And when the mouse is inactive or quiet, the initial response already much smaller, but doesn't go that far down as in the activities. So what we call adaptation seems to be faster or stronger in the active case. So another example neuron with quite no activity in the active case, much stronger activity in the quiet case and actually you see another interesting phenomenon for sensitization. So it gets the response is getting stronger with time. And I have not really thought about why that is the case what the purpose of this but I think it's these things we need to look at to really understand what's going on to make sense of this modulation which is happening so early on. So here's just some population data runs out of the distinguished data and the superficial layers which are, which get the direct input from the retina of very little deep in this which are more multi sensory and motor related. So that's, that's all interesting we still need to make sense of this, as I said, one question though. The answer during my postdoc is, why does the motivation come from. And one of the ideas was maybe from the show cortex because that's projecting back to the superior Nicholas, and since many, many years we've seen that activity in visual cortex is influenced by the promotion. So what we did is, again, you see a bunch of physiology, and we recorded the data in the strictness, maybe nothing to the brain, or when we inactivate the one. This is the neuron in the secure weakness. Again, we see here in the control condition that the activity goes up when running. So when we activate the one was one that's going down a lot, because you take away this input to this new one at least one, but this is more influence by local motion is still very visible. So we don't think that this inference of by local motion is inherited from one is some information data. So again, we measured how much of this response is changing for the preferred direction. You can see again, diverse effects happening in the population of such a click close, but it's looking very similar, no matter whether we are in a control condition or whether we're inactivated one. So, we think we extremely want what else could it be. We thought maybe it's directly coming from the right in them. So to look at this. We express again, it's helping indicator in the regular day itself, but instead of imaging the neurons in the retina which could be tricky in the way mouse which moves its eyes. So that image, the excellent terminals in the superior clickers as you can see here. So we inject the virus to come into the eye of the mouse and then image into the clickers. So you can see some info for images showing that you get a nice later name of the axon from various practitioners of the secure clickers consuming. And this is now an average of the functional imaging of single rational terminals. This was a conclusion from Leon's lab. So instead of expressing the GCAM in the whole cell and all the axons you could localize it to the single endings for the synaptic terminals and localized activity there. And what we found. Looks very similar. So, here we show again ratings. And this is now a population of hundreds of people. And you can see, in addition to this very strong visual drive, you have this motivation by behavior. This was very surprising when we saw this, we actually print this as a control experiment to say we can't be from the retina but then we saw this and thought, oh, what, what is happening. And one of the first thoughts we have, well, maybe it's just an artifact right so it's a pupil, it's changing its size. Of course there's more like coming in, and it's large and less than small. So, what we did is now putting the mouse in complete darkness, and the pupil is so large and you cannot measure changes, so we just measure running speed. But there's another same population of controls sorted in the same way, and you still see a very strong motivation by one. And we do think there's something going. So no promotion is influencing the retina output. There's another control. We looked to convince ourselves that this is probably not true to any light input coming to the retina. What we did is, we looked at different receptive capabilities of on-puters or those that mostly respond to black stimuli or to bright stimuli, and we didn't see a big difference. So we thought, okay, if the pupil is small, so maybe then the ones preferring dark stimuli should respond more and the other round with the on-puters, but there was no significant difference between those. So it's another indication that it's not just simply driven by the input light. And again, the tuning properties for mutations, for example, you see that with the rousal, the responses go down, it's even nice. Tunica for direction, and this is the population. And you can see here that actually, surprisingly, most of the retinal ganglion styles actually decrease their response during the rousal. Which is very different from cortex and also different from what we saw in the superior clinical neurons. What that means, again, I don't know why you would want less input or less visual drive in the retina when you're running. I don't know, but that's what we see. And we're actually not the only people who saw this, which is nice. So other people in the animal lab, they did a very similar experiment that instead of looking at retinal terminals in the superior clitoris, they looked at retinal axon in the LGM, so the other area, getting image forming input from the retina. And they saw a very similar thing, namely that during high rousal activity mostly going down. They also saw another interesting thing, namely that there seems to be a difference in tuning for different spatial frequencies, namely that the suppression is much higher for no spatial frequency than high spatial frequencies. And that there's some kind of gating mechanism, especially for lower spatial frequencies that can't get through during high rousal states. Okay, so again, we were very surprised. So we tried to get a step further and think of, you know, how is it possible where could this information come from. And we came up with two major hypotheses or pathways. One is that new motivators act directly on the retinal terminals and they indeed have three sectors, for example, for Sarah Shonan that modulate how much calcium actually get into those retinal terminals, which we then pick up with the calcium imaging. And the other possibility is that these new motivators go directly to the eye, which is I think the more exciting hypothesis. And, and there are indeed a few brain areas that do project back to the eye in memos. So what we did to, to try and tackle the question is use again electric zoology and put our world into the optic tract where you could measure from retinal axons and measure the activity there. It's going to be tricky to experiment and doesn't yield a lot of data so across lots of animals, I got like five axons, single axons. And we went through a lot of trouble to make sure that this is actually like the axons, but we convinced ourselves. And here again, put that mouse into darkness, we recorded it's running and here the final rate. So you see it's pretty much all over the place. So it's not that clear, but when we look at the cross correlation between running and the final rate we see that in these two places that is a positive correlation. Okay, so somehow running into the final in this axon. Here's the populations and see that quite a few of those acts of positive correlation and two of the negative correlations. So, there is the possibility that locomotion behavior already influences the activity in the retina itself. That doesn't of course exclude the possibility that it's also acting on the retinal terminals is is just some data from a recent study showing that history, which is projecting back to the retina. What is not locally produced in the eye can change the activity the firing rate of different retinal gang themselves when you show them different issues. So here, for example, increasing activity that's increasing activity. Right so in summary. I'll show you that the rational modulates retinal output and neurons in superior papers. It affects the motion changes in response game, but also changes in three preferences, possibly. And changes in temporal dynamics. And these effects differ across retinal terminals so we saw up mutilation and down mutilation across the whole population. One of the questions that would be, yeah, what I'm particularly interested in is trying to figure out what the purposes of what the computational advantages of this behavioral modulation happening so early in the visual processing stream. And some ideas we have is in the direction of the decision coding, so that you want to hone into those features of the visual input that are actually important in the task or in the behavior you're currently involved in. So maybe the past patients, you want to see them better when you're running. So what we are doing now is to look at different visual features like speed tuning or frequency tuning and how they are affected across the population in the retinal exons and the neurons in superior clip loss. And we're also looking for trying to look at different functional cell types to see whether you know cells that have a certain task to analyze the, the scene, whether they are up or down mutilated in the different areas. And, but last I would like to thank all the people involved in these different studies. So, this land of people that was involved in the Rousin story, first paper, and these are members of my lab who collected the new exciting data. Thank you. So, I assume it mouse the security. Yes. So, can you see any regional information that's coming into this approval, for example, drive your cells with a central vision or things you might expect like that. That is a good question. And so far we only look at it for television because of this treaty that I mentioned at the beginning that most of the superior clip loss is covered by cortex and we so far we didn't want to just get rid of this. So, we are restricted to the area that is concerned with very poor. Yeah, the only question was, did you go back to slow. Yeah. Yeah, but there's a five second offset in those two time points. I'm not sure which one's leading to that. Well, I'm not set on that time. So how do you look at that. So I think this time cost is mostly driven by the slow time force of the running, when you do simple. Yeah. I don't remember which way, but I wouldn't. I wouldn't interpret too much into it because of this very slow time cost of the money, I think if you. I think it would be better if we had more data to look at like one of the sides and just look at those. Yes. So, we show that the game of the nation, the change of game of the nation, and then in the near future, we will have this shift in our industry, in our temple of people. Do you think maybe that shift can come from the audience for this game of the nation, or from us? My guess is as good as you. There is a paper showing that the similar things are happening in primary mission projects, where there seems to be a shift in the preference for temple of frequency, but again, whether it's inherited or not, yeah, I don't know. So, it might seem like to be strictly not true. I don't know if people still think that's true. Two circadian nipples have a big effect on the business system. Maybe the next. I don't know. I don't know. My day I heard about what you were serving there. I think it has an effect. I'm calling that manual on sensitivity. That's as far as I know, beyond that, I think what we definitely can say, there's no shortage of things that will probably have an effect on the early business. I think we're just starting out.