 Let's go back to where we were yesterday. So I think yesterday we were, sorry, at the point where I was talking to you about the role of I presented an experiment that was meant to prove the causal involvement of a visual area, which is called V5MT, which is a visual area that is sensitive to visual motion. And we prove basically that this area, we show that this area is quite important to perform a visual temporal discrimination task but not an auditory discrimination task. So again, there is an importance of the sensory specific courtesies in temporal perception. Now I want to just present to you this following experiment which goes a bit beyond just proving that these areas are important to visual perception, visual duration perception. But it's an experiment that wonders which circuits in these visual areas you basically engage in this duration perception. And I will explain to you with I think an animation that it's pretty eloquent, so it's pretty straightforward. Okay, so here you see this schematic brain. So this is the front and this is the back. The visual cortex is this, okay? So a visual cortex, it's a wide piece of the brain cortex, but what I'm talking is the primary visual cortex, which means it's really the first stage of cortical processing of information that comes from the retina and this is basically here. Okay, so what we know, it's the following. So we know that, for example, this is space. So if I present something at the upper left visual quadrant, so something like this diamond here, this piece of information is basically processed by neurons that lie in this lower lip of the chalcarine sulcus. So this is where these are the neurons engaged in this type of processing. And if I present something, if I flash something on the lower left, this should be in the upper right. Okay, so we know that this is the way the retinal space and the external space is mapped into the cortex, into the primary visual cortex. So, but we don't know. So if we add then a temporal dimension to it. So if we add time, so if I flash something that lasts 200 milliseconds or 800 milliseconds here, what does it happen? So where in the visual cortex these 300 or 800 milliseconds is processed? So is it processed? So in all this piece of the cortex, so basically the temporal dimension of the stimuli follows the spatial representation of the stimuli. So I basically engage the same neurons that are engaged for representing this spatial object, this object in space are exactly the same neurons that are engaged in extracting the temporal information of this object. Okay, so this is the question, this experiment I'm going to present to try to address. It's in the same line of yes, the experiment. So it's basically an experiment where they use transplants magnetic stimulation. So I stimulate primary visual cortex while basically I'm asking participants to discriminate two different temporal intervals. One, it's a standard duration, so it lasts 200 milliseconds and one, it's basically a comparison duration that could be either not, that could be it's always longer than the standard. But what I do, I just swap the order of presentation between those two stimuli in order of course for the subject to don't guess that the first, which one is the standard and which one is this duration that it's longer than the standard. So and the task is always the same. So which one of those two was longer, was presented for longer time. And what from the TMS point of view, I decided to focus on two pieces of the brain. So one is the left door cell V1 and one is the right door cell V1. And what I do, I just decide to target with the TMS a very specific piece of this primary visual cortex, which is the piece that I know is involved in the processing of a stimulus that appears at a certain distance from the fixation of the subject. So eight degrees of visual angle. So basically I know, so what I target is the area that I know process the information. When this is present, eight degrees from this point. So here, and how do I know that I'm targeting this brain regions? Because I have retinotopic mapping because I use another technique. I use fMRI to map to know which part of the brain responds to information that comes from different portion of the visual field. So I know, for example, that this pinky or bluish, so if I have a stimulus that appears here, this is the portion of the cortex that processes this type of information. If I represent something in green, I know that it's this. If I present something in the reddish part, this will be processed here. So basically because I use fMRI, I know which part of the brain is involved in the processing of information that I present here. Eight degrees of visual angle. So I have maps of the visual cortex of the special representation for each individual subject. And I know where I have to place my coil, okay? Then what I do experimentally, and my subject, what my subject does is just fixating this cross while the stimulus flashes eight degrees far from the fixation. And this is, so basically, and this is the flash and the subject has to decide the duration, has to discriminate the duration of the two flashes of the two intervals that are presented on the screen. And I do this in one condition, okay? In one block of trials, my subject does this and I flash the stimulus at eight degrees of visual angle. But in other condition, I present the stimulus very close to fixation, so three degrees. And in other blocks is flashed far away, so 13 degrees. So basically I try to present stimuli that are processed in different portions of the visual cortex, but I target with TMS, I zap with TMS, I want to suppress the activity only in one of these, only when the stimulus appears in one of these positions, which is this one, eight degrees. And so what are my predictions, okay? So my predictions are the following. So here I call our code for simplicity, the spatial position of the stimulus. And basically what I expect is when I stimulate, and okay, sorry, my stimuli, so the stimuli are always presented in the left visual field. So I note that the important bits of the brain, the process, this is the right hemisphere, okay? And then my prediction is that if I'm zapping the right position, that I should observe a greater Weber fraction, so I should see worse performance for this spatial position, not for this one, which is 13 degrees, but of course to some extent I should observe also some effects, even when the stimulus is very close to the fovea. The fovea is the part of the cortex that is dense of receptors, is the part of, sorry, of the retina that is dense of receptors, okay? And it's here, it's where basically you focus your attention, where your eye converge. When your eye converge, that's the fovea, the bits of the visual field that is processed in this highly densely, an area that is densely represented in the cortex, okay? Because there are a lot of receptors for that piece of the brain. And this, the fovea in the cortex, I can have a brain here, I don't know if everybody see myself, but basically this is a brain, okay? So that's the posterior part. The fovea is represented here, so at the pole, whereas this... So we lost your Dominican, unfortunately. Hello, Dominican? I think we lost Dominica, sorry, for this problem. And yeah, so what I was saying is that it's obvious that if I place, if I stimulate from the outside the brain, I'm very likely that I will affect the pole where the fovea is before going into these deep structures. Although that since I placed the coil, you know, the projection, so the way I placed the coil should nevertheless be precise, so I should reach, since I use this co-registration method, I should reach the eight degrees of visual angle, and also, yeah, the eight degrees of visual angle, because this is my target area. And so my prediction is that if I stimulate the right point, I should see a worsening of the performance for the eight degrees and for the three degrees. And in the left hemisphere, so if I stimulate the other hemisphere, where basically the other hemisphere doesn't process anything, okay? Because I'm physiologically is impossible that if I present, if the subject stare at the cross and doesn't move the eyes around the screen, so if I flash something on the right-hand side, so on the left-hand side, the left hemisphere should not play any role, okay? So what I'm expecting is no effect for the left stimulation apart from the fovea again, because again, I'm inside, I'm from the outside, and if I'm stimulated from the outside, nevertheless, it is possible that there is a spread of the activity also, of the electric field also in the other, in the Y lateral hemisphere. So these are the prediction of the performance and basically this more or less, these are the results that more or less follows these predictions. So here is norm is the Weber fraction again. So worst performance means, so worst performance are those higher parts. So here is the performance for the right-door cell B1 and for the left-door cell B1. So as you see, the greatest effect are for the positions that I imagine were the position involved. And if I normalize the right with the left hemisphere, which is something that we normally do, this is actually more data. So these are 13 subjects. This is what I see. So that really the greatest effect, these are data where it's the difference basically from left and right hemisphere, okay? So it's the performance on the right compared to the performance they see on the left. And so you see that basically the greatest effect seems to be in the it's for the portion of the cortex I stimulated, okay? And what is also interesting here, this you might like it or sorry, it's not presentation mode. It's this one. So here you see this is the simulation of the induced electric field, okay? So red means this is the greatest point of the cortex that it's the greatest intensity for the induced electric field. And this is how it spreads according to, in this subject, okay? This is based on the position of the subject's, the position, sorry, of the coil on the subject's head. And what you see here is basically the, how it looks like this induced electric field for the different special position that I choose to stimulate with to, I choose to, where I choose to present my stimuli because I'm stimulating with TMS only this, but my stimuli are presented at three degrees and a 13 degrees. And as you see, even the prediction of the simulation of the induced electric field is telling me that the induced electric field should be stronger for eight degrees, but should also be higher for three degrees. So close to the pole and be lower for 13 degrees. So the points that is the farthest from the point that I'm stimulating. And this is, you see how it looks like, how they look like the different retinal position for this simulated induced electric field. So it should be very low. And this is basically the sum of the induced electric field values that is plotted over the performance of the subjects in the two basically for the two, conditions. So when I stimulate the right and the left. So you see that whereas if I stimulate the right, I mean, are three points, right? So fitting a line with three points is not ideal, but nevertheless you see that there is, if you just look at the points, you see that there is a linear increase somehow. So the greatest, the electric field simulated, the higher is the Weber fraction. So the worst the performance in the subject. And whereas for example, if you see these three points for the left stimulation, you basically don't see such a linear increase. And this is basically, this is a single subject. No, the performance I show in the slide before was the group level. Here is just an example, how you can use the simulation of the induced electric field to look at and try to correlate, to make a correlation with the performance. Okay, so this again, this results basically show that not, show not just not only that the visual cortex, the primary visual cortex is important to perform a duration discrimination task, but also that this involvement of primary visual cortex seem basically specific, seem to be specific for the, so that space and time goes together in this task. So that basically is the portion of the cortex that is processing the visual information that also process the duration information. These are unpublished data, the experiment is still going on. So let's see what's going on with the, we have to just finish data collection just to make sure that the data are robust enough. Okay, and then we will try to publish, but this is the take home message. So now I will switch with, so to the state dependent network model and to some other type of data, not humans, but are gonna topic slices. So slices of cortex, not brains. Let me share something different, which is this one. Okay, so you should see this one and I'll go in presentation mode. Okay, state dependent networks model. This is basically, it's one of the reference that will come in the next slide by Dean Bonomano. He basically claims that this is the most extreme, it's the opposite of the internet clock model because it assumes no modularity at all. So it assumes that basically time is encoded by the same circuits encoding other stimulus properties such as color, motion, for example, individual modality. So basically according to Bonomano, there is no modularity of time perception, no areas that is really specialized to tell time, but time is also, it's basically distributed in the brain. So any brain region is in principle capable of telling time because time uses to tell time, the brain uses the temporal features of neural network communications, the short-term synaptic plasticity, long-term synaptic potentiations. So really the time lags of neural activity of the dynamics in the networks is used to tell time. And so basically there is a prediction that this model makes. So time is encoded as a unique event in the network, in the network dynamic. So there is no linear metric of time, there is no accumulation, there is no clock that accumulates time, but time is in the dynamics of a network. And in the network state at a certain time. So basically, if I present, it's like he always made this, I think, fun, nice metaphor of throwing a pebble in a pond. A pebble in a pond. The pond makes ripples, okay? So the ripples represents basically the state of a network at the given time point. And by time is basically, it's told by counting the ripples that this pebble produces when you throw it in a pond, okay? And so this means that every single time duration is a unique event and it's encoded in the context in which the network is at a given state, okay? So that's why there is no linear metric and there is sort of, yeah, there is no linear metric, no need of accumulation or clock. And, okay, this comes from one of the references that I give it to you, that is a paper published in Neuron quite some time ago, 2007, it's Carmarca in Buono Mano. And he uses basically a neural network. He uses a simulation, it trains a network, but here it's not the reason why I'm showing this. So this is the network that he trains. So there are some excitatory, inhibitory neurons. This is basically this neural network has been trained to discriminate two different durations, like something that lasts 100 milliseconds from something that lasts 200, they tried many different things. But what I want to show is basically what it means for, so the way he often represents this network dynamics, okay? So here is basically is this principal component. So it basically reduces the complexes, it reduces the side dimensional space that is the network, the dynamics of a network by using this principal component. And what it shows is basically represents this. So this is basically is the network dynamic when a certain event is presented at time zero, okay? So it creates a certain trajectory. And this fast peak are basically due to this fast depolarization that are due to the fact that the system is excited by the incoming event. And then the slow waves, this slow activity, this slow trajectory reflects this slow co-synaptic inhibitory potentials and short-tap synaptic plasticity that occurs in the networks, okay? So in the network one that once basically the network has been, the network has been changed by this input stimulus. And basically this is when the network is when the first stimulus is delivered and this is when a second stimulus is delivered, right? So the second stimulus finds the network in a totally in a different state, right? When the second stimulus secured the network is in a different state. And this is described. So the second stimulus is described by this second trajectory. But this is just for visualization just to visualize his idea of different network states. And the temporal dynamics within a network, some are fast like depolarization events when cells are excited. So they discharge action potentials when they, there are some action potentials and then these slow consequences of the action potentials, okay? Now, one of the prediction of this model is that basically that, as I said before, a duration is a single event that is represented in a specific context, right? And if you change this context, you change the event. So there is no reset. You can run reset because the network, so because, sorry, I go back. So the network takes time for the network to go back to the initial state. So if you then, so with an event you create a certain dynamic, right? But if something occurs before the event, before the network has went back to the initial state, the network would be in a different state, okay? So the prediction is then that if I run this type of experiment that I will tell you about. So the prediction is that I cannot reset the clock, okay? And so if I introduce, if I introduce a disturbing element, if I introduce a stimulus that disturbs my temporal discrimination and before basically the network go back to the initial state, then my performance should be impaired by the occurrence of these task irrelevant stimulus. And I'll show you maybe this will be clearer when I explain the task. So basically he's asking participants to decide whether this interval is longer or shorter than a reference interval that the subject has basically learned before. So of course, this basically this interval can be is t plus or minus a delta t, okay? So in one condition, this interval is basically preceded by a first stimulus. But this stimulus is always presented at the first, as a, sorry, at a fixed interval. So you have those two, so the subject has to focus on those two sounds. So this beep, beep, and in the interval between them. But sometimes those two beeps that are basically the marker of the interval is preceded by a first task irrelevant beep. Now, in one condition, this task irrelevant beep is occurs at a fixed duration. Whereas in this case, it happens at random times. It is presented at random times before those two task relevant beeps. And so the prediction is if there is no reset of the clock, basically I should see a difference between a fixed and a variable condition. So if the clock reset, I shouldn't see any difference because the subject knows that has to reset the clock at the second sound. So he has to ignore the first, reset with the second and then estimate the time that is between those last two. So if on the other hand, the network, so sorry, the time is given by the network state and I don't have any resetting mechanism of any clock because there is no clock because the clock is in the network state. Then I should be, my performance should be disturbed by this task irrelevant stimulus because the task irrelevant stimulus arrives and creates a certain dynamic within the network. And so when the second stimulus arrives, the second stimulus, the network will be at a certain state that will be different according to the temporal distance that is between the first and the second sound. And so another sorry prediction, so because this state-dependent network model so it can be good to explain how time is represented in the brain included in the brain but time in a short range, right? Because this dynamics, the dynamics in a network cannot be very long, right? There is a constraint in the time range they can hold basically. They happens in a certain temporal range, right? Which is hundreds of milliseconds, not more. So basically the model should work for durations that are below a second but not for durations that are above a second, okay? So, and that's why basically the authors use this type of task with two durations with 100 milliseconds and a second. And these are the results. So here you see the discrimination threshold which is an indication of how good is the performance of the subject sense of the greater these values, the worse is the performance, okay? So the higher the bar, the worse the performance. And so you see that, okay, for two interval fine, you have just an interval to discriminate. So for two T means two sons, there is a good, so the discriminant initial thresholds are more or less similar in the, but if you go at the three T condition where you have three sounds, there is a really big difference between the random and the fixed conditions. So basically there is a lot of, so there is a, the threshold is much higher for the variable condition. So subjects performance is impaired in the three T variable compared to the three T fixed. I hope you understand that the three T fixed, the network is always at the same state when the second stimulus arrives. That's why it's not affected by the, by this task irrelevant stimulus. But this, let's say pattern of results, it's not present when you use much longer stimulus range. Okay, so somehow the prediction of the model that there is no reset has been proved by this behavioral experiment. And there is another prediction that the model makes. And it's the fact that the model assumes that time is represented in the dynamics, the local dynamics of networks, okay? So if, so if I'm, let's say, if I'm using, if I'm asking participants to discriminate time of different frequency of sounds of different frequency. And we know that frequencies, different sounds of different frequencies engage different populations of neurons, okay? So we know that there are certain neurons that responds to low frequencies and other that responds to intermediate or higher frequency of the frequency of the audible spectrum of frequencies, okay? So we know that there is a specific network for each of these frequencies. So according to the network model, time should be represented and coded in the network, in each of these networks separately, okay? So if I use like in the visual experiment that just presented, so if I know the different spatial position are recruit different neurons, I expect that the time, if I present duration in this, this in spatial position, these thing neurons are involved to tell time for this different spatial position, okay? So it's exactly the same logic. And how can you test that? You tested by presenting subjects with sounds that have sounds of a given frequency. So here you present two sounds, okay? Like beep, beep. So subject has to decide which one of the two was longer. And here what you manipulate is the distance between the first and the second sound. So sometimes it's beep, beep, sometimes it's beep, beep. So the lag between the two stimuli is bigger. Now, if the lag between the two stimuli you have to discriminate is too short, basically you are not able to tell time because the stimuli, this is a limitation also of attention resources. So you are not able to discriminate very well the two stimuli. So this is what happens here. So short ICI means short interstimulus interval, long ICI is long stimuli. So if the lag is big enough, you are able to discriminate the stimuli. If it's short, you are not. But because why is this? Because the network, so there is not enough time for the network. So this entangle between those two stimuli. So the activity that is elicited but each of these stimulus cannot be properly read out. But if I engage a totally different pathway by presenting a stimulus of a different frequency, then things should change, right? This is the prediction of the model. And this is what happens in reality because this is what it shows here. So if this is short ICI, and this is long ICI, when I use this exactly the same sounds of the same frequency. But if I change the frequency, so if I use a beep that is high frequency as a first stimulus, and I use a beep of a lower frequency as a second beep, then you still are able, even if the lag between the two stimuli is pretty short, you are able to discriminate them. And this is basically what is showing this absence of difference between long ICI and short ICI. Comparison and standard same frequency, comparison and standard different frequency. This is what the label is telling us. And all these seem to be specific for duration discrimination because if I use a control task and I ask to discriminate the frequency and on the duration, this difference doesn't happen. Okay, another prediction of this model, which is I think it's that basically you don't need. So the time is the brain tells time through the dynamics of networks, to the temporal dynamics of a network. So you don't need different modules, but time is represented locally in network activity. So you don't need an extensive, you don't need a brain to tell time somehow. And basically you test that, you test that by using basically by testing, slices, so this is an in vitro study. So you use the slices of cortex and what you use cortical slices, this is another paper, very interesting one, use cortical slices that are transfected with the virus. So basically, because he uses these slices, implants the slices with the electrodes and basically he uses electrical stimulation via the electrodes, but he also by injecting this virus, he makes the neurons to be sensitive to light and basically he also uses optical stimulation. This is optogenetic, I don't know if you heard about and it's nice because you can basically by combining the two techniques, so by stimulating electrically and also optogenetically, you're able to understand which are the cells you to target because the cells are sensitive to light. You can use basically, you know, which are the neurons that are actually, that are actually affected by your electrical stimulation as well. So here what he does is basically train this neurons in the cortical slices with an electrical pulse. So it just stimulates electrically and after a lag, he stimulates optogenetically, so with lights. And this pair of electrical and light stimulation has different lag. It's 100, 250 or 500 millisecond, okay? And it does this for four hours and it records from the more super from pyramidal neurons. It's a type of neurons that has a certain, it's called pyramidal because the body shape looks like the body of the cell look like has a pyramidal shape. And those neurons are very long axons in the most superficial layers of the cortex. The cortex has six layers. Here you record from layer two and three. And this is what he records, okay? So he basically, these are of course different slices, no? That are trained with different lugs. And normally, if once you stimulate, you have to observe a post-synaptic potential, a post-synaptic, an excitatory post-synaptic potential, which is normally then followed by a lot of poly-synaptic activity. This poly-synaptic activity is basically normally reflects the activation of the network as a consequence of this first basically synaptic event. And it's also a measure of how the network is intrinsically active after this first input stimulus. So here he are basically, this is the traces, okay? Of a single cells are four different traces of a single cell. And these are the poly-synaptic traces. So after then this first depolarization event. And here you see that basically in this four different traces that the lag, so that there is a response to each of this stimulation, no? And the lag is different. It depends on the temporal interval between the two stimuli, right? So here after the training, you have two peaks that are more or less a hundred milliseconds apart. Here you see this activity that it's more or less 200 milliseconds. And this is, you see that the poly-synaptic activity tend to be delayed because of the learning. Sorry, this is what you trained, okay? This is the training. The pairing of the two stimulation is only on the training. This is what you see only in the testing phase when you basically don't have the second stimulus, okay? The second stimulation. But you see that the brain has acquired the capacity of reacting to the good point in time, the time that has been learned basically. And these are basically, this is a single cell. This is a single cell. Here you see multiple cells and each row is a cell. Actually each row, so you have the five traces for different cells and the cells are sorted according to the latency of response and they are normalized to their own peak. And so here you see how this activity evolves over time. According, and so you see that basically the cells in the tissue have learned the interval, okay? Because this poly-synaptic activity is basically delayed depending on the train interval of the tissue. This is exactly the same thing just represented. These are the mean trace, okay, over time. And the color code is for the different, basically learned interval, 100, 250 and 500. This is just the same thing as a cumulative distribution basically. So you see that there is really a difference. Here is a difference of the peaks. So the peak is delayed according to the interval. This is the same thing. This is the mean event times. So the mean poly-synaptic activity over time. Okay, now an interesting question is, is this a cellular process or it's a network mechanism? So in order to answer, so it's something that happens at cellular level that produces this behavior in the slices or it's what happens really in the network that matters. And in order to address this question, they in the first place try to see whether this behavior is in a certain pathway and not in other pathways. So what they do is the following. So they have this cortical slice. They basically implant two electrodes and in the first with the first electrode they basically deliver the stimulation and they pair the stimulation with the light. So they have both electrical and octogenetic stimulation at the 100 milliseconds delay. In with the other electrode, they don't do this pairing but they just electrically stimulate the tissue after I think they choose two to five seconds after this first pairing. And then they see what happens to the traces of this pyramidal neurons, poly-synaptic activity in the pyramidal neurons. And what they see, they see that indeed there are two distinct pathway, right? So if you do the pairing, you see the behavior as before. So you see those two poly-synaptic activity. The shaded basically area is where the light is supposed to be during the training phase. And these are the traces during the testing phase, okay? So you see that here you see what you observed before. So that there is a quite good temporization between first and second pair in the trace of the poly-synaptic activity of this cell. But if you look at the non-pairing, what happens when you don't pair the cortical tissue, then you don't see this behavior. So basically this specific pattern of response is happens only when you do this pairing and not when you do something different, when you don't have the pairing of electrical stimulation and light. And also if this was a cellular mechanism, you should see a change in the cell, in only in the cell that are sensitive to light, okay? Because of the, as I said before, they inject this virus and the virus makes cells sensitive to light. But not all the cells in the tissue become sensitive to light because of this injection. There are also cells that are not sensitive to light. But if what produces this behavior is a cellular mechanism, you should see basically this behavior only in cells that actually receive the light, the stimulation of the light and are sensitive to the light, which are called CHR plus cells. But indeed what they see, they see that if you plot the activity of non-light sensitive cells, you still see this behavior over time, okay? So this means that basically they claim that this is the interpret this results as suggesting that what happens, it's really a property of the network. It's not a property of the cells, it's a change in the cell. But it's a change in the network. Also what is interesting, and then I'm ending with this is the last slide for today, what they also, what is interesting is they show that this dynamic change in the network activity over time is a consequence of the learning of a learning process. Because if you see this is basically each row is a cell, okay? And this is how the cell reacts to basically to the first stimulation, to an electrical stimulation before the training, okay? So basically before the training, there is no, so there is pretty diffuse activity, but there is no real, there is really few activity. The population, the cells in the slice don't really react to the first, to this first stimulus. But this, you start to see this activity in the slice only with the training. This is two hours training, I think, and this is four hours training. And this is for the slice that is trained with 100 milliseconds or with the 500 milliseconds. Even the activities seem to be, so here it's a two hours, it's more spread, right? Here it becomes more tuned to the 100 milliseconds because this is where this activity should be. Because if the network has learned this delay between first and second stimulus, you see activity here, here it's more spread. So it's something that you, so the emergence of interval-specific network dynamics during the course of training. And I stop here for your questions, yeah. Thank you, thank you, Dominica. Are there questions? I think there were a lot of new concepts in this lecture, so probably one is these imbibor cultures that you use, these cortical slices. Maybe can you explain what are these systems? So what do you mean? So basically this is piece of tissues. Is this what you mean? So what do you mean? Yeah. They just cut a piece of tissue that they take from the brain, for example, of rat, that is still alive, right? And then they basically put, so the network is still, so the neurons are still active. And they measure, of course, there is a limited amount of time they can do this experiment in vitro, no? And they just then use this, they implant the tissues with these electrons and they do this electrical and optogenetic stimulation. So I never did a vitro study, but I think this is fascinating how you can keep neurons to be, to train a network, to train a population of neurons in a limited amount of tissue, right? Because you lose all the connectivity with other parts of the brain. So these are studies that I use really to either investigate stuff of the cell, normally in vitro studies are used by neuroscientists they're more interested in understanding the molecular aspects of cognition, okay? Those they are looking for cellular changes in the synaptic activity. So here the level of investigation is very small, right? Because you really look at cells and you also measure the chemical properties sometimes if they, you do neurobiology, right? With this, but here it's, you can still use it, even to use a dynamic of, to study dynamic, not at a bigger scale but at still the scale of the tissue you are in your hands, right? Still there are, okay, lots of neurons in these tissues, right? It's not like one or two or three, it's still, so you cut like the six layers of the cortex in terms of millimeters, I don't know how much it is. But I guess if we go into, if I reread the paper and surgical details we know how much of, it's, I don't have it here. Yeah, this is the, so in terms of size, how big it is this tissue? I know I need to go into supplementary, you speak very... Yeah, I think the bigger part, I'm idea. Yeah. Any other questions, or questions? I have a question. Sorry. So, yes, please. Go ahead. So basically you say that there are two views on the problem, one says that there is a big network that extends time in the brain. There is another view that says that different parts collect data and they understand time. True? Yeah, so well, I think there are different types of, type of, so different levels of complexity, right? So you, in order, so one thing is performing a task, right? So, which entails a series of cognitive functions, right? So you need to pay attention to something, to the stimulus. You have to extract, you have to memorize. We say it many times, you have to sustain your attention over time. And all these tasks, right, are complex, are cognitively demanding. And for those tasks, you need the whole brain. So a slice cannot behave like a brain, right? So in that sense, you have a connectivity at a higher level. So the system, so different maybe cognitive functions are observed, but different maybe parts of the brain and these parts of the brain are connected, no, are connected. So you have, for example, the visual cortex, that we mentioned today, that receives the information from the retina. The visual cortex extracts the very simplest feature of the stimulus, like the luminance, the orientation. Then the more you go into the hierarchy, the more complex become the extraction of the, of the feature that the brain is able to, and then you, so there is an area that it's important, so the one network of areas that is important to attend this, to attend this stimulus, or that are not active when you just ignore the stimulus. So, so, and so there is, there is a connectivity at a macroscopic level of the different brain areas. But then there are all sorts of, you know, there is a connectivity at a macroscopic level of the different brain areas. But then there are also these very local dynamics that you studied with these slices, right? And you can assume that the slice or locally in a circuit, in a small circuit, you have all the tools to be able to extract temporal information because all this network has some temporal feature because temporal activity unfolds over time. And then you can use these local changes in different brain areas. Then to tell time at, with this more complex task, okay? Because I don't know if you understand, we really don't know how much what we see in these slices, for example, it's related to the perception. What these slices are telling us is that there are temporal dynamics in the, there are temporal dynamics in, that can be learned in a piece of a tissue, of a cortical tissue. But how this temporal dynamics then can be used for perception, it's something more complex and it's different. So somehow this model, this state-dependent network models believe that we use these tools. So these local dynamics in network states to tell time, but don't believe that time, in order to perceive time, you need other brain areas. You just don't need these local computations. You need something else to perceive it because perception is something that has a high degree of complexity, no? Yes. Thank you. You're welcome. So there's a question in the chat. How do you measure the action level of each neuron? Well, I think... It's a voltage, so you measure that exactly. Yes. It's millivolts, you see. It's here. There's a stimuli and also you measure the amplitude also of the response you can measure. There are several things you can measure. You can measure the latency. So when the neurons fires, you can measure how frequently they fires, how many neurons fires. Mm-hmm. Yeah. Thanks. So I have a general question, which is probably very naive. There are all these type of oscillation, the arm oscillation, delta, and et cetera, et cetera. So are these the outcome of this type of network dynamics? And how are they related with this? Or are they related to this internal clock? So... Yeah. This is a very good question, actually, Matteo. And in the first place... Okay. There are hypotheses about... Okay. The two things might be related, no? It's obvious. Even in this very sketchy idea of the internal clock, an internal clock is an oscillator. So we have multiple... We have this oscillator reactivity in the brain, right? So this might play a function. People are trying to relate any of these brain rhythms to that perception. I did it myself with an experiment. There are certain brain... There are oscillation in certain brain frequencies that seem more than others linked to perception of duration. For example, in the beta frequency, the beta band frequency, oscillation in beta band frequency are very wide. They can go... So the beta band is wide. It goes from 13 hertz up to 20... Yeah, 20 hertz. And in this range of oscillation, this seems to, for example, if you... If you ask participants to extract temperature to pay attention to a stimulus that lasts for several months of time, normally you see this activity in the beta band frequency during this... But this is, of course, it's just a correlation, right? So you know... But what is really the mechanism no one knows. No one knows why we have this oscillatory. So there are just... There are different hypotheses of why the brain oscillates, resonates at this different frequency. Why do we have it? Some people believe that we have it because... So they help the brain to keep a certain state. Right? It's just a way of keeping that state steady, right? So you are reaching a certain state and then these oscillatory mechanisms help you to keep this state. Others, for example, believe that oscillatory mechanisms are a way the brain could facilitate the communications between these parts of the brain as well. Like if these parts of the brain are starting resonating at the same frequency. But really, these are just hypotheses. They are not really a very clear answer to that. So this is just talking about generally of frequencies. And the link between frequencies and time is just a matter of investigation in these days. And as I said, it's just beta frequency, alpha sometimes. But the problem is that most of these frequencies, nobody knows why do we have it. And also they're linked to also several cognitive functions. Like let's say you normally, if you attend to a stimulus, you have a suppression of alpha frequency in sensory regions. So alpha is something that normally you see when you are relaxed, no? And it's relaxed but awakened. So it's... And then you have this suppression if you attend in something. So it's also difficult to say, okay, this is just for time and not for other functions. So the good news is that there are a lot of interesting questions. So this is the end of these lectures. We have more to come in the next couple of weeks. Yeah, next week is Tuesday, I think.