 Okay, so good afternoon, and today I'll just basically, I will continue basically, I will keep going with this transition that we, yesterday we had, that then we observed at neurophysiological level, so at neural level, right? So we went from ramping to see in the striatum, the striatum, so the basal ganglia, how the activity, the whole activity of a population of neurons could be linked, can be linked to perception of duration, and then from considering the whole population, the behavior of the whole population, at once we saw that in the first place, there are neurons that seem to prefer certain intervals compared to others, and we move into considering not the activity of the population in a static way, but to try to see how the dynamic of a network of a neural population changes over time, and this most recent paper by Gouvea and Paton that has been published in Neelife, it seemed to show that really, so that what seemed to distinguish between the different durations is the speed at which the network reach a certain state. So it's not the amount of firing rate that matters, but how the neural activity, so how this firing rate basically unfolds over time, this seemed to be the key message of those papers. And now we go into this duration specificity, so this tuning that we saw yesterday. Tuning is interesting and we saw yesterday why, because tuning is something that exists in the brain for features that are very concrete like orientation, so speed, and so it seemed quite bizarre to find exactly the same type of tuning for a feature like time that is so abstract. Anyway, so this is another study in the show, basically the same thing, so the same duration selectivity, but in monkeys. So yesterday we saw it in rats and we saw it in the striatum, so subcortical area, okay? Whereas now we're looking at monkeys, so human primates or non-human primates, and sorry, and we look at this, I will show you briefly this study, that is interesting in investigating the existence of this tuning in premotor cortex, so cortical region, okay? Area that is active when you're planning to a movement. And here the monkey had a reproduction task, so the monkey basically here, the trial starts, here you see a sort of timeline of the trial. The monkey gets a cue, which is a visual cue, it's a color cue that could be blue, pink, or yellow. And the cue is, so each of these color is associated to a certain duration that it's in the second range, so it's either eight, four, or two seconds. And the monkey task is basically to hold the response according to the cue that receives. So if it's blue, in this case, he has to hold for two seconds. If it's pink for four seconds, if it's yellow for eight seconds, let's say. And the color, the association between color and time is changed basically in the experiments also. Anyway, so, and then they have to hold the response for the amount of time appropriate according to the cue that the monkey learned to associate to a certain duration, and then they get a reward. So this, I think now we are becoming expert in this. So here you see the behavior, the reproduction of the monkeys, there are two monkeys I think here. And you see that basically for longer duration, there is the classical shape that it's wide. So this is the distribution of the reproduction in the eight seconds task. So you see the monkey, the mean here is nine. So there is a slight overestimation. So it's 9.2 milliseconds, 9,000, so 9.2 seconds. So this is for the four second duration. This is for the two. So here it's Weber law, right? So the width of this Gaussian scales with the interval. This is the four second when the cue is reversed. So it's a different cue, it's a different color that is associated, this is eight reversed. And this are basically duration selective neurons. So here I see that this I plot. So there are raster plots, these are single cells. I'm showing just the single cells among the cells that they basically take into consideration, which are not many, I think it's 30 cells, not many. But anyway, this is the raster plot. It tells you in each trial the firing rate of this neuron. And this is the type of cue that is associated with eight seconds. So you see that this cell responds to eight seconds, but not much for four seconds, because you have to look at the rows, yeah. The columns, the first column. The first column is the same cell. So you see that there is a lot of activity for eight seconds, not much for four or two or four seconds, but there is again a lot of activity for eight seconds, even though the cue has changed. And this actually I forgot to mention, this is what you might have guessed from these pictures, is that this is the activity of the cell before actually holding the response when the monkey is just seeing the cue, okay? That tells the duration. And here is a cell that prefers four seconds, for example. You see four seconds here and four seconds here. So it doesn't depend. So the color changer, so this manipulation is meant to prove that the response of this cell, it's not linked to the color of the cue. It's not that this neuron prefers blue, because when the blue is associated with a different duration, the cell doesn't discharge anymore, okay? It doesn't fires anymore, okay? And this is the same here. So this is a cell that prefers four seconds and not the pink, because pink color, when it's associated with eight durations, doesn't elicit any response. And this is a cell that prefers two seconds. Okay, so even here you see, but okay, and what is nice of this paper, but I don't have a thinker slide about it, is that this is not, and this is also to connect with what I was saying yesterday. So this is not the only behavior that you can observe in this premotor cells. Other cells had a different behavior, had a sort of graded response. So the same cell responded to the different durations and they respond, so the response was different. So they had a graded. So they were like increasing a lot, they had the amplitude of their response was greater for the greater duration and was gradually reduced for the durations that were shorter. So something similar to the ramping, like I said that we said. So this just to tell you that your neuron behaves in a complex way, right? And it's not just one type of behavior that you can see. Okay, so they saw duration selective neurons. So neurons that were responding only to specific durations but also cells that were modulated by the duration. So the firing rate was changing according to the duration presented. And some cells that didn't respond to duration as well. Anyway, so this is another example in cats. This is visual cortex. So another region, okay, visual region. And here is basically you see these are, here you see exactly what I was saying. So this is a cell that prefers different durations. Okay, so this is for example, prefers, yeah. So this prefers a long interval. This is 800. This is 400, 200, 100, okay. But here for example, you see a graded response. So this is the same cell that just changed the response according to the duration. Exactly, and these are, yeah, it's exactly these are duration to cells because they, so what I mean is that in different rows you see different cells. Here it's the same cell that has an incremental firing rate depending on the duration. But let's go to the, to what I wanted to show you today. So let's go now to human brain, okay. So in animals, we have lots of quite a few evidences that duration can elicit some preferential response. So that there is some tuning, form of tuning in the brain. And this is observed in rats, in monkeys, in cats, in bats, okay. So in the auditory cortex of bats, and even in birds, like, okay, for birds time, so for birds time is pretty important, right, the capacity of detecting very subtle differences in the duration of the sounds is very important because of the way of communications like via songs, okay. But in also in birds, so there are, there have been described this type of cells that are tuned to duration. But in the human brain, we didn't have any evidence that this was a possibility, okay. Especially when the prevalent, the most important somehow, the most followed model was this internal clock, right, where you believe that there is a single mechanism, a single clock that should be responsible of processing multiple durations, no matter the sensory modality, the task and the length of these durations. So in humans, there was no evidence whatsoever of this type of property. But I think it was an interesting question to address with an experiment. And this, because topography, it's a very basic, is a very widely present way of representing different feature in the human brain as well. So for example, here I provide an example. In the brain, there are body maps as well. So in the primary somatosensory cortex, which is basically this region here that you see here. So there is a really a map of your body or better of the receptors that are on your skins. So this means that, for example, if I stimulate you with a brush, your feet, the feet are close. For example, the stimulation of the feet, the tactile stimulation of the feet reach a specific portion of the somatosensory cortex. And if I stimulate the ankle, the bits of the cortex that receives this information from the ankle, it's very close to the representation of the feet. So basically the same for the hands, the different fingers, the information that comes from touching different fingers, reaches bits of the brain of the somatosensory cortex that are neighbors, okay? This is, they are specially contiguous to form really a map, but we don't have just a somatosensory map. We have even a motor map in primary motor cortex. We have retinotopic map in the occipital cortex. So topographies, as I said, I think in previous lectures, it's one of the key principle of organization in the nervous system, a functional organization in the nervous system. And so we basically, we ask whether it was possible to see this type of organically, the same type of organization even for time. And there is, I think a missing bit here that I didn't, something that I didn't say because one thing is being tuned to a certain feature like time. I have neurons that are, or population of neurons that are selective or they behave in a specific fashion when I present a specific duration, okay? And this is what electrophysiology told us with all these studies. But what electrophysiology doesn't tell us because of the limitations of electrophysiology doesn't tell us whether the neurons that are selective to very similar durations are also contiguous in the brain. So for electrophysiology, it's very difficult to prove the existence of topography, okay? So here there are two concepts, two then concept. One is the duration to prove in the human brain the existence of duration selectivity. And the second is to prove that this duration selectivity is topographic and organized. And this is what we did actually. We did, and we did it with an fMRI experiment that we conducted at Seven Tesla. So you remember fMRI is the technique that measures the hemodynamic response function. So it measures the level of blood oxygenation what tells you because blood oxygenation is linked to neural activity because more if an area an area that it's active needs has metabolic needs, right? It needs more oxygen compared to an area that it's not active. And what also I show you is that this hemodynamic response function is seen to correlate with the post-synaptic potential in monkeys. So they seem to be a link with neural activity. But nevertheless, it's an indirect measure of neural activity, this you have to bear in mind. Anyway, so the reason why to go at Seven Tesla. So on ultra high fields is that ultra high fields allows you so not again, normally you conduct an experiment at Three Tesla. This is what 1.5 Tesla. These are the regular scanners, but there are also these ultra high field scanners. Now there are more in Italy, I think there is only one 70 in Pisa. In Europe, there are quite a few. Here the data were acquired in Lausanne in Switzerland. And the reason to go at 70 is that Seven Tesla images. So Seven Tesla scanners allows you to have a greater signal to noise ratio. So the signal you get, you have more chances in less time to get a lot of signal out of your scanning time. And most importantly for our purpose, it allows you to go smaller in spatial resolution. So you have to imagine that if you use a 3D MRI scan, the spatial unit of your images is a voxel. So it's a 3D pixel and the voxel size is normally three millimeter. Here you can go up below a millimeter, but in this case, in this study, the voxel size was 1.5 millimeter. Again, bear in mind that so in a voxel, there are millions of neurons, okay? Still the spatial resolution, still it's very different from recording from single cells, okay? So here you are somehow detecting the activity of an entire population, okay? And it's very difficult to know. So to be able to distinguish between the precisely between the behavior of different populations, but still it's a pretty good resolution. And you will see that with imaging the way you analyze the data allows you to somehow, to distinguish the different behavior of a certain brain region, okay? And to try to see whether a single brain region like the promoter cortex can show a different behavior according to the duration that you present. Anyway, we start with the first experiment that you see was mainly conducted by those two, by two post docs. So Masamichi Ayashi is now a PI in Osaka, University of Osaka and Fudini Protopapa was a former lab member. And basically here the task was simple. So we present the first grading. So this grading here that had a certain duration. This was a fixed duration. It could be either 200, 400, 600, or a second. So it could be one of those four durations. And after this S1 duration, there was a little delay. And then the presentation of a second duration that could be either shorter or longer. Sorry. Then why? Okay. I thought, okay. I missed the S2 values here. Sorry. But it was either longer or shorter than those. It was 40% of 200. So the Weber fraction was 0.4. So 40% of 200, 40% of, 40% shorter or longer. Okay. And after also another delay, that was the response. So subject were asked to decide which of the two stimuli was longer, was presented for longer time. And basically you acquired images through all the trial. Okay. And you basically present this durations many, many times. You acquired a lot of images and a lot of observation. And basically our goal was to try to see what was the brain response. So at the offset of the S1 duration. So we really try to focus on the brain response in this stage, because this was the stage, the purely encoding stage where basically the subject was just extracting information, temporal information from the visual stimulus. And we decided to focus on the offset because the offset, it's clearly the time, the moment in which the durations becomes available to the subject, right? Because the subject only at the offset of S1 knows if it was 200 or 600. Okay. So, and in this study, so we were, so try to see what was the bold response, the hemodynamic response in this stage, and try to see whether there was in the brain any brain regions that show differential response according to the different S1 duration that we use. And basically this is what we found in the whole brain. Okay. Here you see this brain needs a little bit funny. It looks like it's an inflated brain. This is just for visualization purposes, we inflate the cortical surface. And basically what you see here in dark are the solchi. Okay. And in lighter gray are the gyri. Okay. Because the brain, the cortex is made, it's made of gyri and solchi, right? So, but in order to visual, to better visualize this activation blob, it's better to use a flat surface. Okay. It's easy. And basically what we see here, what you see color coded here are voxels. So again, the voxel is the special unit in fMRI. So these are voxels that are maximally responsive to each of the different S1 duration. And so, for example, in red, you see the ones that are greatly responsive to the shortest and in green, those were greatly responsive to long, to the longest duration on the range. Okay. And basically we see this, these blobs, you obtain these blobs by doing a general linear model. So you basically model, so you regress, so you have your own data. So you have the images, which are basically matrices of numbers where different numbers, basically are the different intensity that the signal that you acquire have. And you basically try to see in this data, the variability that you see in the change of intensity in the signal over time that you see in the whole experiment, how you can partition this variability and what are the the source of this variability that you see in the signal. And you basically model all the events that happen in your trial. So let's see, if I go back here, you, for example, take into account the variability that is due to the, to the offset of this first stimulus, then you have, you model, you have a regressor. So JLM regressor for the offset of this, you have a regressor for the onset of the second sound, and you have a regressor for the response. So you try to model and to regress out the variance of interest and the variance of no interest, but you have to really sort of make sure that the variance explained by the regressor of interest is significantly statistically solid. Okay. And so what you do, you also do, you do some, some contrast, some, you run some t-test and you correct for multiple comparison. And this is basically, so these are what you see here, these blobs are the results of this multiple regression analysis. Okay. And this is our basically team-ups that are corrected for multiple comparison using a false discovery rate correction. Okay. So you don't, you divide basically your alpha level, it's divided by the number of interconnected bosses. So just to say that this is a very conservative somehow, approach, you look at blind that whole brain, you basically say, okay, just so you feed your model with all the foxes that are in the brain. Okay. You just don't focus on that. This is the results of the GLM of the multiple regression that survives your statistical analysis, your multiple comparison. So, okay, we see this. Okay. And what you see here is the results also. So you do, you have team-ups. And you also have on which you apply the team-ups tell you which boxes are selective for each of these different durations, but then you run a winner take call. So because you want, based on team-ups, you classify the voxel in this area has responsive to each of the of these different durations. And you obtain this map basically. And this is basically an area that we encounter also when we talk about the single, the, the, the internet clock model. And when we present the experiment where the subject were asked to judge color and duration. Okay. When we see that was activity also in this media part of the motor cortex is called premotor area supplementary motor area. But we also see another area where we, we could appreciate this transition in duration selectivity. And this was the intra parietal sulcus. Also this one is an area that we encountered before it's the area where we saw the ramping of the cells when the monkey was how to hold the response for different amounts of time. Then we saw these maps and we wanted to make sure we do some, okay, this is just so the blobs are the results of linear multiple linear regression analysis. But we want to now to quantify somehow this spatial progression. Okay. Then now we just appreciate by. So we decided then what you do you draw borders. And then you basically measure the, the, the distance of these clusters of boxes of these duration selective clusters of boxes from one of the border that you, you draw the border is basically are the, sorry, I go back just to show you better. So the borders are these ones. Okay. Are the edges of the map. One needs on a physical landmark. Okay. So this is the pre central is the most anterior part of the pre central gyros. So we drew this border and then the, the most anterior part was at the edge of the map. So it was arbitrary. This was arbitrarily set and this was based on anatomical landmark. And then you measure the distance of these clusters from one of these edges. Okay. Just to, to, to see whether this there was really this progress, this spatial progression. And you wait also this distance by considering the, the, the, how, how cluster eyes are also those boxes. So you wait. Yeah. Exactly. So you give more weight to this, let's say, rather than to those two little clusters. And this is what you see. This is the median and the interquartile range of the group level. No, we basically, because what you saw in the map, where the map in all the subjects that we tested. And we tested 11 subjects. And this is, but we do this computation in each individual brain. Okay. Because this is important to stress that brains are different in the morphology. Right. So you want really, since you want to appreciate this differential response in neighboring clusters, you really want to, to preserve the spatial. Yeah. You want to, to, to run this analysis on, on each individual subject brain in their own space. You don't want to normalize the brain to a template. You don't want to transform their brain into a template. Okay. Which is what you need to do when you, when you do group level analysis, because what you do, you just basically move all the activations in individual brain into a common space. But this, to do this transformation implies a lack of spatial precision. Right. And because implies some sort of error in the, because brains are slightly different. And these are basically those slopes that are calculated for each individual subject. So the gray lines are those slopes. These are weighted relative distance from the posterior border that you see in different duration selective cluster. So this just tells you that the red clusters are the far from the posterior border compared to the green fox cells that are more, that are closer to the posterior border because this transition goes from anterior to posterior for shorter to longer duration. So you see, there are some subjects that seem to have a totally different slope, but in eight out of 11, the slope is negative. So as we see in the group level, which we present this red line. And if we, when we did the same, the same computation in the, in the intraparietal sulcus, we saw a greater degree of variability. As you can see from this, the interquartite range. And as you see also in the slope, that there are, there is a lot of variability across subjects. So only four out of 11 show a positive slope. So this means that basically only in SMA, we actually can say that there are maps, because these maps are consistently present, across subjects. Whereas in IPS, some subjects show them up, others don't. Okay. So then there is another way of loop. So now there is another way. There is another. Yeah. Yeah. And the paper that we discussed that had as its topic, the modulation of attention that then in turn produces different estimations, different precisions and the estimation of time versus color. Right. Yeah. The anatomical, the functional anatomy of what's its name. There were others. There were at least, I remember three or four cortical areas. Precisely cortical striatal areas that were involved. There was the PFC. There was the PMC. Not only the pre supplementary motor area. There were also, they mentioned the basal ganglia and these were not seen. At least I didn't recognize them in these fMRI images. So that was one thing. And then also you mentioned that there is a topological mapping of the body onto the brain. I suspect that's also functional because it kind of makes sense that the tip of one finger would be mapped near to the tip of another. Because if I'm getting useful information from the tip of one finger, I might get also useful information from the tip of another. And maybe I as a brain don't want to search a million drawers to find them. And I'm interested whether this sort of progression from point two to point four to point six to one in the pre SMA in the supplementary motor area might be useful. And might the usefulness of it be the reason why you can see that in the SMA, but not in the intra parietal cortex, maybe in the intra parietal cortex, that's sort of a organization would not be useful. And I'm interested whether you would be so bold to, to give us sort of an interpretation. Yeah. These are very good questions actually. First to answer your first questions about other areas. What are the other areas that are normally part of a circuit of time? Well, the truth is that we saw and actually we brought in the paper and this is a paper that has been published in plus, but once that I'm presenting in plus biology and it's one in the reference that I gave it to use Proto Papa and it's one of my colleagues. If you go through the paper, you see that we show that indeed, when we run this GLM, this multiple regression analysis, those were not the only areas active significantly active at the offset of the first test one durations. There were other areas. And those areas where the cerebellum. It was the primary visual cortex. There was a intra parietal cortex and there was the same fear from the gyros. No basic idea. Okay. At least with the disconservative threshold that we use that there was no, no such activity in this subcortical regions. I run a few experiments in my so far in my career fMRI and I think not always this space. I think it really depends on what is the task. This is a very perceptual task. And here I'm focusing on a very perceptual stage. It's really the first level. So it's just encoding the duration. There is a, you know, there is no decision to make yet. Okay. So we saw yesterday that basically are often linked to reward prediction. And so to somehow predicting a reward, it implies, no, it's, there is no prediction here to make. No, there is no proximity with the response level. We are far from making a response here. So this could be an explanation why we don't see. Anyway, we saw the other areas. But we didn't see those areas are clear topography. Okay. So we didn't see. So what we did since we were, I mean, this was really the first study on that. We wanted to be in the first place blind. So we want just to rely on this on the stats. And we want. And since we didn't know where to expect. So we had some, some ideas, but we really want to be blind. So we decided to just focus on areas. Then for further analysis on areas that clearly show a group level. Some topography. Of course, doing things that group level can be misleading. As you saw now, right? Because a group level. It looks like that IPS has a map. But then when we look at the individual subject, this map disappears. So at least some subject show other subject. Didn't show. But that was our starting point. A group level. There was no maps. Apart. There were no. Visible maps. Apart from IPS and SMA. That's why we just decided to focus on those regions. Then later on for a, when we then. Publish this work, we, we went back to this data. And we actually realize that in those regions. That were active. That were significantly active. At the offset. That were active. That were significantly active. At the offset of the duration. We actually observe duration selectivity. And we have a bio archive. We have a paper that it's not published yet, but it's a different story. So if you're interested, I can send you the link to that paper where we saw. Selectivity in those regions. But not a very clear topography. Okay. So topography. And then comes. To your question. Why topography in SMA. And we will see. Why topography in SMA, I think. Because I think it's SMA. It's a key structure. For. Read out time. Compare. So it's probably compared to what. So probably. The regions that are at the back of the brain. Do something slightly different. I don't know. Visual cortex. Maybe. Just extract this information. It doesn't read it. Parietal cortex. Maybe it's an intermediate stage. Of this process of recognizing this duration. And SMA. It's the level of the decision in the sense that. Okay. This is it's the highest stage. Of the duration recognition process. And in this bio archive. Manuscript. We run also a connectivity analysis. And we actually see this flow of information. From back to front. Okay. So it's. But you will see how I have other. Specification to make about SMA. That makes this a likely structure. To take a decision about duration. And one of these. Now. Okay. What time is it? It's four already. I can just go quickly. So one thing that I want to show you. This thing. Which is. You can. You see duration selectivity by. What is interesting is not just to see. What's the response. Of this cluster to their preferred duration. But it's also interesting to see what, what is the behavior of this duration selective cluster. When I presented the non-preferred duration. And this is what we see. We see that there is a sort of. Gaussian like. Response. Which is something that. Again, it's not new. So this is also something that is present widely presented in the brain. So this, what, what it means is this, this is the, what I see in the Y axis. It's the. Normalize signal change. This is bold response. Okay. So really the ammo dynamic response. This is. The row signal. Average across the conditions and across the subjects, but this is the signal. In those different duration selective clusters. That it's a plotted against the different. Presented duration. Where there are diamonds. It's clearly what we expect. So this is the. The cluster that prefers 200. Second duration. And this is a, of course, the peak. The, the, the, the, the, the peak of the bold. Is where we expected. Okay. But you see that the, the bold. The signal decreases. For durations that are closer. You know, it deeps here. Here it increases slightly, but look at also at the green cluster. So this is basically the, the signal is greatest. Where it should be. But it slowly decay with distance from the preferred duration. So this is something. About the behavior. Of those duration selective cluster that seem to be similar to what, what you would, you would see in a, in a duration, in a orientation selective cells, no. When cells that are tuned to speed. That normally. Cells that are tuned to similar feature. Are contiguous. Okay. And they have overlapped. They have overlapped. They have overlapped. They have overlapped. They have overlapped. They have overlapped. They have overlapped. They have overlapped. They have overlapped. They have overlapped. They have overlapped. They have overlapped. And they have overlapped. Contiguous. Okay. And they have overlapping tuning functions. Whereas they are far from cells that are tuned to. Very distant future. And this is the duration of the tuning. In. So as you see, even if we don't see an intraparental cortex topography at tuning level, we see exactly the same behavior that we see in IPS, so in SMA. So this therefore doesn't mean that IPS doesn't do any job. It does a job that it's slightly different from the job that SMA does, okay? Because it needs less topography than SMA. And then we wonder, okay, this is a second experiment that is in the paper of cross biology, and where we use a wider range of durations. We run a totally different type of analysis that it's... And this is again, you see the map. So you see the progression from red to blue, from anterior to posterior. There is no time to go into the detail, but I want to say something about the question and about, so these are the different slopes again. Okay, so here you see the variability. Here I'm showing maps in individual subjects. Okay, so you see this is the progression, this is anterior, so you have to see from red to green. So from red to green, from red to green. These are different hemispheres, okay? So from red to blue, because here we have a wider range of durations. Here we go from 200 up to three seconds, okay? So it's a, this is another subject, but more or less you see that in most of them, there is this progression. But there is a difference, I think, between those maps and the sensory maps that you could see, for example, in the special maps, for example, in primary visual cortex, no? In primary visual cortex, you have a representation of the whole retina. So for each portion of the retina, you have a corresponding group of neurons in primary visual cortex. You have really the representation of the receptor space into the cortex, but there is, compared to this maps of time, compared to those maps of space, those maps are much more variable at the individual level. So there must be something special. They're not really, and you see, okay, the special map, no matter the variability, you see it in every different, in every subject, you more or less see a special map. Whereas here, you don't see the map in every single subject. And we wonder what is the source of this variability, right? Why is, what could explain this variability? And then we wonder then whether this variability could be linked to behavior, could be linked to perception. So we wonder whether maybe in subjects that have a better map, they are also better timers, okay? They are more accurate in judging the duration and more precise in judging the duration. So for this, we basically decided to correlate. This is what you see. So we correlate the slope. So the slope of the describes the spatial progression in each individual subject with the accuracy of the perception of the different durations. And we've actually did see this correlation, okay? So we see that the slopes is steeper, the steeper the slope, the more accurate is the subject. And here it's reverse because the steeper is the slope, the lower the variance. CV stands for variance, for variability, okay? So the more variability is a measure of how precise you are. So the more precise the less the variance. And so basically it means that these maps are in SMA, are linked to behavior. And for example, we don't see in a different experiment that we run, we don't see the same correlation in between the maps in parietal cortex and the maps in SMA. So SMA for this reason can be considered an area where that recognizes duration and it recognizes in order to perceive, to make a decision and because it's, so the mapping SMA, it's closely linked to perception compared to the other three regions. And I think, okay, it's eight. And I want to just conclude by showing something and why I'm saying that there are maps in parietal cortex because then we run another experiment and today unfortunately I don't have the chance to tell you about and it's an experiment where I can show you, I can conclude by showing this because this is related to, so okay. We run a second experiment, a third experiment using auditory stimuli this time because the question was how time is mapped in these areas. It's in a relative fashion or it's a relative or absolute the time that is represented in those maps. So what I mean is if, so here you see two maps that have different ranges but there is overlapping, the two range overlap. And one interesting question is if I use non-overlapping duration ranges and I make just a single duration to overlap between the ranges, what's the destiny of this overlapping duration? So what matters is the position that you have in the distribution of the duration or it's just the absolute duration that matters. And basically the new results are telling us that what matters is the position that you have in a distribution. So the selectivity that you see here, it's not absolute but it changes according to the fact that this is representing the longest duration in the range and this is for the shortest duration in the range. And in this new experiment, by changing the way we identified the maps, we were able also to identify maps in IPS. That's why I'm telling you that and whereas the map in SMA again show close correlation to behavior, the map in IPS don't. But I want to say something to, these are all the position in space. This is mean distance from the posterior border. So basically we calculate the distance of each of these clusters from this physical border. And color coded are the different clusters and let's focus only on the orange. So the yellow, sorry. And these different shapes are clusters that respond to the same physical duration, which is I think 600 millisecond in the two experiment. Okay, so this is the same physical duration, but in one case, so in this yellow, this 600 is in a wider range. In this case, it is in a smaller range. So this, what I'm saying is here that different clusters of voxels responds to the same physical duration when this duration is in a different range. This is already an indication of what I just said that time is relatively represented in these maps. It's not. So the square points are when the range is up to 600 milliseconds. It's from two to one second. Okay. It's the smaller range. Whereas the circums is when the range is wider. Yes, yes. Okay, so you see that, so this means in the first place that these are different voxels. So different neural populations, let's say, that responds to the same physical duration when it is in different context. So these maps are formed in the brain depending on... I think so. The hypothesis is this, is that they are in place when you need it, especially in this very frontal area, like premotor cortex. And according to the range that you use, they change the selectivity somehow. And I have this new data that unfortunately we didn't have. So I was too ambitious for today to present all the second bunch. But yeah, and we saw really this shift. We have, yeah. So this is like a region of spatial, of temporal representations. Exactly. This is a regions that cares a lot about town because we saw these regions across many studies. We, as Carlo pointed out before, we saw it also in the experiment where we compare color versus time, you know? It's an area that very often it's correlated to temporal tasks. It's a premotor region, but yeah, but most of the areas that are in parietal cortex is somehow linked to motor, because we saw, you know, do you remember when we talk about LIP neurons, those are active when you make my movements. So parietal cortex is an area that process visual information with the purpose of making an action, okay? And all the circuits, even the Cerebellum, the Bessel-Gandia are all motor regions. And I think it's not by chance that this area that tells time are areas that are important for motion. But because I think if there is no motion, if there is no change, there is no time for us. How can we get time if we don't see a change? This is, of course, it's a speculation, but I think it's, yeah. So I wanted to conclude that I have also to conclude everything by just making sense of what we basically, again, the purpose of my course was just to guide you to neuroscience and the way neuroscience research is conducted and the logic we use and the methodology we use. And for me, time was a sort of an excuse for that. I told you about time because I'm working mainly on this topic. But basically, if I have to make sense of what I say, if the question is, how is time represented in the brain? How is processing the brain? Well, the question is, I don't know, I don't know yet. And I think this is something that honestly, most of my colleague would tell you about. So what we know, I think is that there are multiple brain regions and you see all these regions that are colored here that they seem to be engaged in temporal computation. But what I also tried to show you by presenting all these papers is that it seemed that we really move from a very old-fashioned idea of having a single clock in the brain, like the internal clock model, to an idea where complexity plays a greater role. So there is no single clock, a universal clock in the brain, but time is more likely to be distributed and an intrinsic property of the brain, where which part of the brain, so it's responsible for time processing, it really depends on what you have to do and why you need time for. So it depends on the sensory modality, it depends on the task at hand, it depends on the time scale. And what I also like about our field is that we really moved into considering, again, not just single-neurons behavior, but we also moving to a more microscopic scale by considering also the behavior of a neural population. And this is also what I do with my research because I don't do electrophysiology, so I look at stuff at microscopic level. And thank you for your attention and I hope you enjoyed. Okay, thank you very much. So time for questions. So I don't see questions in the chat. Yes. Okay, so congratulations. So, yes. Yeah, time is relative. Time is relative. Well, I don't know if time is relative or not. What I'm saying is that probably the way the way time is expressed in this populations seem to be relative. So the map we see seem to be relative. So time is not represented in an absolute fashion there, okay? This is what I said. I'm not saying so and what I think also I don't know is how similar those maps are from sensory map. These are also, it's an open question. How different are those maps? So because somehow the mapping in lower level region, the map, so the orientation map, the retinotopic maps, those are more absolute, okay? So how similar or different those map are from those sensory map is something that needs to be explored with more experiments. Okay, so one thing which I think we learned in this course is also a sense of how sophisticated you should be in the question you ask to such a complex system and what is the level of noise when you do experiments in such a system? So, which is very... Yeah. Yeah. Okay, thank you very much. So... And if I may add something, the system is so complex that that's why we need expertise. So somehow neuroscience is a very interdiscipline work where there is room for lots of expertise from physicists. So there is more computational side, there are biologists, there are psychologists. And even the tools that we use, we often need, I don't know, super expert, super computational collaborators that help us out to run certain analysis. So it's very difficult also for me that I run experiments to know everything about what I do, what I'm handling, right? So you try to really know as much as possible, but tools also are becoming so sophisticated the way we analyze the data that often it's very difficult to keep track and you need lots of... So it's really a collaborative effort to do this type of experiments and to conduct this type of research where each of us give a contribution according to the background, according to the expertise. Yeah. So thank you very much. Ciao. Grazie, Matteo. Yes.