 I have to be really freely moving here. Thanks, Sean. Thanks to the organizer. It's a great pleasure to be here. I want to talk today about cellular determinants of spatial representation in the cortical hippocampal system. And there'll be three parts to this lecture. In the first part, I will talk sort of about structure function relationships in the medial and torino cortex. And this is work primarily pushed by Andrea Bogalossi, a postdoc in the lab. I will then in the second and third part, which are pre-for switch to the hippocampus, where I talk about hippocampal activity, in particular, play cells, activity and silent cell activity, and how it is affected by experience. And finally, discuss the possibility that there might be an internal trigger for global remapping. And these parts were done by Albert Lea and Schaubabstein, and they were still in the lab. And this is currently work underway done by Moitz von Heimdall, a postdoc who is also a poster here and who you should talk to for more detail. So the study of the cellular activity in the cortical hippocampal system, I think this has been one of the most fascinating stories in neuroscience in the last decades. It somehow didn't end up in the Nobel Prize yet. But who cares? What was so stunning about the work in cellular neuroscience of the hippocampal system is a series of discovery. And the first discovery was the discovery of play cells by O'Keeffe and Rostovsky, then head erection cells by Tobi and Muller, and finally the discovery of crit cell activity by the Morses and their colleagues. So why were these discoveries so important, and why is it such a success story? The reason is that the activity in the hippocampal system, the cellular activity, it was just unexpectedly abstract and explicit, sort of this idea of a cell firing in a certain spot and another cell firing in another spot. It really made it possible to think at a neurophysiological level about the idea that the animal was using a map. And before that, there was sort of behavioral evidence to suggest that rats, rodents, can do map-like navigation. Then when people discovered these head erection cells, Tobi and Muller, it sort of became actually difficult to think that the animal is not using a map. So the first there was this correlate, the play cells, but then when you find things that look like a compass and that's what these cells do, it would actually kind of be surprising if the animal would not be using map-like navigation. Then again, grid cells, I think they did another job, at least for me. And there was a debate started off by the discovery of O'Keeffe. He suggested that play cells are part of an internal construct, an internal spatial representation system that the animal applies to the world to sort of break it down and make explicit representations. And he had a lot of emphasis on the idea that this is something that is inside the animal. Now, people didn't want to believe that. In particular, psychologists, they have a very hard time with the idea that an animal is not empty. They think it learns everything and stuff. And I think that's sort of where grid cells came along and eliminated another large class of theories. Obviously, an animal has never seen a grid. These firing patterns, these hexagonal firing patterns sort of iterated through space, this is something a rat has never seen. So it's not a goddamn memory. It's not something the animal acquired. It's an internal system much like predicted by O'Keeffe. So OK, the cell activity in the cortical hippocampal system, it's fascinating, it's rich, it's explicit. Turns out it's very inspiring. A huge amount of work was done to sort of understand how the system works. Here, I have an atomy diagram. You don't need to understand that in detail. Blue structures are structures where people found head-direction cells. Here, yellow, blade cells, grid cells here in pink were found in enterinocortics. But it turns out that they're also in the post-subiculum and the parasybiculum, which is not depicted here. So there was a lot of anatomical work triggered, even more so. And many of you will know that or may have contributed. There was a huge set of models triggered to explain how the animal makes blade cells and how the animal makes these spatial representations. So we have literally hundreds of models for how these cells come about. Now, being an experimentalist, what I see as a challenge then is to eliminate these models, to reduce them ideally to one. This is not where we're at. We still have hundreds of models. A lot of models are compatible with the data. And this is something not so easy to understand. When you think about the fact that we have studied these cells like hell, there are thousands and thousands of studies. Why do we have so many models? Why do we have so many alternative explanations? I think a major reason is that we typically know relatively little about the individual cell under study. And this is both true for the synaptic events that contribute to blade cell, grid cell, head erection, cell activity, and for the microcircuits. And that's where our work is focused on. And today, I want to talk about hippocampus and enterinocortex. Enterinocortex is an interesting place. Head erection information, which is very heavily linked to the vestibular system, which scans from here, is integrated. And we see here in the enterinocortex for the first time allocentric coding, sort of spatial responses that are not egocentric, like the ones that we would find in the sensory courtesies. And then I will also talk about the hippocampus, which has this special ability to form memories very fast, sort of in single trial learning. OK, why is it so difficult to, why do we know so little about the individual cells we study? I think the reason is that most of our techniques use deliver only very limited information about the single cell. So we can extercellory court from awake behaving animals for decades, but we typically could not identify these cells and could not identify the microcircuits. And that's something that Andrea Bogalossi addressed. So he used a special micro drive that was friction based that we would implant on the animal because there's a lot of friction between the pipet holder and the pipet guide. The thing is very stable in place. He would approach a cell. He would stain it and then cement everything in place and let the animal run around. And this cementing in place strategy is also something that we use. It gives a lot of additional stability. And it's also a strategy that we use for the patching that I will talk about in about 15 minutes. Now, the way we do the experiments as yet and all the data I will talk about today come from formats where we anesthetize an animal, we then stain a neuron or we patch a neuron. We give the animal an antidote against anesthesia and we have it run around and then look at spatial correlates of firing. Now, this format has its problems because the animal first gets the anesthesia, then it gets the antidote. So when it's waking up, it's full of drugs. The behavior is quite variable. The reason why we do it is because it's easy. Before Andrea sort of started to do this, there were maybe in the whole history of neuroscience, I tried to count it. There was something then less than 50 cells identified in freely moving animals. And Andrea has now identified about 100 cells already and actually more people in particular, the Klausberg group is now also coming up with identified cells from freely moving animals. So we think this is something where we'll have a lot of movement. At the same time, we're trying to improve these experiments. We try to, in this format, we just get a single cell per animal or actually to be honest, it's more like half a cell per animal because not every experiment works out. We're trying now to improve this, to get several cells per animal, to do it in fully awake animals and to have animals that are trained to run around because this leads to better spatial discharge properties. This underway, it's promising. The cost is that it makes the experiments even slower. Good, so now we're gonna use this technique and ask how do microcircuits of cells in the entorhinal cortex look in particular? We are interested in head direction information and crit cell information. The idea was that these cells are mixed in entorhinal cortex and we're gonna evaluate this idea on the basis of our evidence. Before I really get into it, I wanna give you some background on the anatomy of these structures. So in particular in the human, here you look at the temporal lobe of the human, here would be the occipital lobe, here would be the frontal lobe. So this is the entorhinal cortex of the human. This brain structure is incredibly patchy. Sort of there are islands, half a millimeter islands or so, that contain about 3000 cells. In each hemisphere we find one to 300 of these little islands. In between there are basically zero cells. So as far as I know, this is the patchiest cortical structure that has ever been described. Now because it's so stunning in humans and monkeys, it was first actually in the original publication I was saying that there are no patches in red entorhinal cortex. Turns out when you look closely at it, you find that it's also patchy and what is very interesting, so you see certain spots of cytochrome oxidase activity in layer two. This would be a section through entorhinal cortex, dorsal being here, ventral being here. This is layer two, the dark region. This is a major output layer of the entorhinal cortex. In this area the outputs come from the superficial layers. And it projects to the layer two cells, in particular project to the dentate. So what was particularly interesting here is that we found differences in the patches when we went from dorsal where the patches are often very small to ventral where the patches are large. And that was interesting to us because there was evidence from multiple sources, first anatomical evidence from Menovitus Group and later physiological evidence from the Moser Group that there are functional differences here, namely the ventral entorhinal cortex being sort of concerned with large spatial frequencies, having big distances, several meter distances between firing spots of grid cells and the dorsal parts being concerned with sort of smaller spatial distances having sort of firing fields that are sort of 10 to 30 centimeters apart. Now what we also found is that there is not just one type of patch but that there are sort of multiple types of patches, small ones and larger ones, these larger ones we later understood are related to the parasybiculum and I talk about these in a bit more detail. So these patches were very intriguing to me because I worked with patchy structures before in like red barrel cortex. I have to say that we still don't understand the functional significance of these patches all that well. The reason is, as far as I'm concerned that we have too few techniques that allow us to relate sort of cellular activity to individual patches. So we made progress on describing these patches in the red. We found that actually multiple different stains reveal sort of this is a stain for cells, an islet stain. This is a chat stain for acetylcholine S-transferase, a sort of cholinergic enzyme that is involved in cholinergic action there and cholinergic action we know it's very closely tied to thetherism or cytochrome oxidase stain. So if you look at these different stains, they all reveal the same kind of patterns. So for example, these three patches we see in cell density, metabolic activity, cholinergic activity. So we have now a better take on what these patches are and from doing these multiple stains our estimate of how many patches they are came down. We first started with several hundreds. Now we think sort of in the upper third of the MEcBF maybe 30 of these patches, we think there's about 200 cells per patch. Yeah, it's a reasonably small number and quite different number from the 3000 cells per patch that we would find in the humans. What is also intriguing to us that we find patches not only in the interrhinal cortex but also in the structures that project to them like the post-subiculum and there are a lot of structural similarities. The patches much like in the interrhinal cortex are also restricted to the superficial layers. They also have size gradients indoors or ventral and we think this is something that iterates in cortex and is a feature that is very typical for cortical processing. As far as I know we don't have any evidence of patchy structure in hippocampal system or very strong patches. Good, now what's happening in these patches? What do we see when we identify cells there? So here is a recording from a layer two cell that we identified. We have the animal runner around in an omaze and what we find, we find sort of multiple firing spots where the cell is particularly active in the corners here, the cell actually fires basically no spikes at all the five turns that the animal made and at these spots we have then other spots where the animal fires and most of the runs. Okay, this is similar to a grid cell in a linear environment. We haven't studied these cells in an open field where you see the grid pattern better and the reason is that we typically do not get good enough spatial coverage with our animals who just woke up from anesthesia. They run for 10 minutes or 15 minutes but for a grid cell, for sort of robustly seeing this grid pattern, you really need the animals to have run like hell and we don't have that. So we restricted ourselves to these linear masses where we can assess the spatial modulation better. A very typical thing is that these, their two cells show very little head direction modulation. Yeah, they are sort of not directionish and they're very strongly cedar modulated. Okay, now how do other cells look like? I think I'm actually running late on time. We skip this sort of the, layer three cells look similar to the layer two cells. In the deep layers, we often find silent cells, yeah. Or we were very puzzled by the large fraction of silent cells, about half of the cells are silent in our experimental conditions and we initially thought that it's related to experience. We see in some cells an experience effect so that when the animal knows the place better, they fire more, but it's not as robust as we initially thought. Actually many cells also would fire more in a unknown environment, so we're not totally clear about that. What is striking is the architecture is very different from the layer two cells. The layer two cells actually made only very moderate local connections, yeah. And quite restricted connectivity in the superficial layers, in the deep layers which actually receive the hippocampal feedback, there is very rich local connectivity. So we think that the microcircuits of these deep layers that get hippocampal feedback and the superficial layers are very different. Another thing that is very interesting is that these cells, basically the layer five cells, we never saw them connect to these big patches here. Now what's happening in these big patches, in these big patches we found a lot of head direction cells, yeah. And these cells are strange cells, often very small cells. They very much respect the borders of these big patches. So we think these big patches, the parasympicular big patches, they have very, very sharp borders. So the cells never cross, that's why it's basically never crossed these border and we think it's one of the hardest borders in the red cortex. Now these cells have a very interesting axonal architecture. First of all, they make a single axon that goes, sort of, we find these cells at the border. This is, let me start here again. This is the side view. This would be dorsal. This would be ventral. These are the different layers of the entorhinal cortex and at the dorsal end of the entorhinal cortex we find these bigger patches and they send an axon to layer two and actually it turns out that they often send select with their axon a single or two in this case of these smaller patches. Now this is one of the few things that we know that might help define these patches. So they get very private directional input. So what that made me think originally is that maybe these patches can sort of, as an ensemble, they get specific direction information and we think that the direction information, the head direction information is involved in updating the map so that when I turn my head, yeah, not everything jumps but it stays in place. So we think that the head direction information might help rotate these patches and we wondered if individual patches might rotate. Now from talking to Edward Moser, my friend Moser and people who record from ensembles of great cells, I'm not sure this is a viable hypothesis because they would typically say that sort of all the typically most of the grid cells would rotate together. It would not be, they'd have seen little evidence for restricted rotations. I went to first here, I didn't go through the spike plots so this is a spatial modulation of the cell. It fires everywhere but it's strongly modulated as a function of where the animal looks so when the animal looks south, it fires and that makes it a head direction cell. Okay, these cells, they do not just have this one exon that goes to the layer two but they also have a very long exon that we call circumcurrent exon because it runs around the entorhinal cortex. So it would connect actually many of these big patches together and another feature of these cells is that when we rotate the animal in forced rotation we see strong responses very much suggesting that these cells see a lot of vestibular input. So a few concluding words on sort of the micro circuits of these head direction cells. So what we wonder if these circumcurrent exons are involved in a global computation of head direction. So when we go back here, they have this exon that connects many, many of these big patches and we wonder if this circumcurrent exon is involved in aligning all the head direction vectors, right? A key thing about the compass is that the needle shows in one direction. So we think that this might be something that enforces sort of a common head direction vector across the system. Certainly when you deposit a little bit of dye in one of these bigger patches, a stunning observation you make is the whole parasybiculum, a huge structure is filled with exons. So there is very strong global connectivity. When you do the same experiment here, you find there's a bit of local connectivity, but it's never that the entire entirinal cortex lights up. So there seems to be a big micro circuit difference in sort of the exons that these head direction cells form compared to the more crit-like cells that we discovered in other layers. Another feature of these cells is that we really have seen no deep layers associated with these parasybicular patches. So when you go back here, these cells, these patches are fat. They actually almost touch the surface. There's very little layer one above and they get very thin and basically we see very little layer five connected to it and layer five is what gets a hippocampal feedback. The deep layers in particular feedback from CA1, these cells do not seem to get. So what we wonder is given the lack of feedback, maybe the animal doesn't memorize this. Maybe the animal doesn't consolidate this. A key idea sort of in hippocampal research is that the hippocampal feedback allows the cortex to sort of consolidate memories about what the animal did. Now what we wonder is if there is no hippocampal feedback or at least no CA1 feedback, maybe the animal doesn't memorize where it looked. It just memorizes its path. Actually, this is a very good spatial, a very reduced spatial representation but when we draw maps of where we were, we don't specify where we looked. We just specify where we have to go. So this is an idea and it's maybe compatible with some other evidence that actually also shows a lack of replay in head erection cells, et cetera. Good. I spent quite some time now with structural function stuff in entorhinal cortex. I also wanna talk a little bit about hippocampal circuits. So I'm gonna shift gears a little bit and this is work done by Albert Lee and show him Epstein and sort of a key thing about hippocampal representations is when you have the animal run around, you would have a group of cells set fire in certain places, so-called place cells, certain place cells and other cells that are silent. So now we investigated these cells with patch recordings and I show you a movie of such a recording. In this case, a place cell recording. What you're gonna see right-hand side is the animal running around and left-hand side, the membrane potential. So we start with the animal sitting in the place field and in black, you're gonna see quite a bit of spiking. The animal sits here in the place field. There's a lot of very massive activity. The animal runs out of the place field which is here and actually the cell shuts down. This is the zero line, this is the membrane potential. Now it comes back into the place field and you're gonna see substantial activity. Yeah, the cell firing quite a bit. Now we're gonna show a run where the animal runs, enters the place field from the other side and you're gonna see some mighty spiking events, complex spikes, we think, that is a gigantic depolarization of the cell. Very, very impressive events and I show you this now in a more formal format. This is the cell that I just showed you, the firing map of the cell. So the animal runs around here. The cell would fire only in this corner and here we have the animal covering some space. This is when it's in the gray area in the place field and you see this massive depolarization. In particular, we see these gigantic huge bursts that look like they have a big calcium contribution. They can be so strong that it actually brings the cell into the depolarization block. So we do not see any sodium spikes anymore and then actually it relaxes again as the animal moves out of the place field. These big events, these big depolarization events, they were a surprise for me. I had recorded for years in freely moving and also in head fixed cortical cells in these gigantic events that occur when there are such that the depolarization is such that the cell goes in depolarization block. I never saw this and I actually, I only saw that in vitro protocols when people did certain plasticity paradigms and I always made fun of it because I thought these people have no clue but I changed my mind. So in hippocampus, you really often see these gigantic events and we wonder if this is a difference of cortex in hippocampus. In particular, might be cellular mechanisms, these very massive depolarizations might be a mechanism for instantaneous plasticity. Sort of the single trial learning that hippocampus is good at and that you don't see like this in cortex. Good, so this is a one class of cell that we see, the other class of cell that we see are silent cells. So if the animal runs around and there are very few spikes fired and the membrane potential is just rock steady. So we recorded these first cells in Holland and they still remind me of it. The membrane potential is just flat, just flat. And this was kind of surprising for us. Most of all, we have very few cells, I should say, because these experiments, in particular the patching and the freely moving animals is exceedingly difficult but it looked very much like they come in two flavors, the cells, sort of the flat ones, the silent ones and the ones with the peak. And this is a finding that intrigued us a lot and that we worked on quite a bit and what was actually quite special about this difference between place cells and silent cells is that we then also discovered intrinsic differences between these cells. Namely, what we found is when we fire the cell and actually we would fire the cell when the animal is still anesthetized. It hasn't, it has never been in the maze that it's gonna explore in 10 minutes when we wake it up here. So we would patch a cell, we would inject current and what we would find is that some cells give regular spiking patterns and other cells burst when we inject current. And once more, the animal has never been in the place. What we then found is that the cells that would burst would come out as place cells and the regular spiking cells would come out as silent cells. Sort of suggesting that there were intrinsic differences. To me it was a big surprise, for years I suggested not to analyze this data because it's anyway from the goddamn anesthetized animal we're doing here the awake stuff, so why bother? But then in the end we did it and we found this difference and very surprising to us. Another thing that was also very stunning and was you could anticipate this finding from extracellular data but I think it's more clear in our data is that the cells very much come as they are. Sort of here we show the first time the animal runs and we show the data from the entire session and we show the membrane potential traces and the spiking and what you see here, this is sort of and we average across all runs. This is the first run. Clearly this here is a place cell, yeah. It has a firing peak. So clearly this peak is already there, the first go, another even more dramatic case. The first time the animal runs through it, the entire session, they're just the same. So this made us conclude that the place cells, they come as they are, yeah. So probably this whole place field formation and it's not a learning business, yeah, as we would have or as at least some theories might have liked to think and this is in line with what people had concluded from extracellular recordings earlier. Good, Albert and Jerome, they then analyzed more properties of the cells, more intrinsic properties and found quite a bit of further evidence for intrinsic differences between silent cells and place cells and these were entirely unexpected as I already said. Even better, what Albert recently did and what was just published in Science is he fired cells at a certain spot and could show that he could switch on such a peak, that he sort of, by just a manipulation, intrinsic manipulation of the excitability of the cell, he can get place cells and I think this will tell us much about how these firing fields come about, yeah. These are un-predicted findings and we think that people need to pay more attention to cell intrinsic factors in place field formation and memory formation and I wanna leave it here with this. The last thing I will talk about is very preliminary. I sort of talk about it because I was asked for an abstract and I somehow put it in and I didn't wanna leave it out these early days and it are findings on hippocampal remapping, yeah. And what is global hippocampal remapping? This is something that was realized early on but then became more clear in the 80s and it's simply the fact that when you have an animal in an environment A, you have a certain set of place fields one in the corner here, one in one here in the middle, one in the other corner. When you put it then, the animal in another environment, you get a totally different set of firing properties. Some cells would totally lose place fields. Some cells would sort of have the place field in a very different sort of global location. Some cells would move the place field from the border to the middle. Now, when you then put it back into the old environment, you see the exact same place fields as before, showing you this is not just random. And now this ability to totally rearrange, totally rearrange place fields from one environment to the next, we think that this is part of the hippocampus ability to form highly context specific memories, yeah. It's just the way the cells fire and actually as a spatial relationship of the cells is very different, is very much context dependent. Now, this is very different from cortex. At least so far, what people would say, what the Moses would say about enterinocortex, they would say, you put the animal into do different environments, sort of the crits as a first order approximation, they transfer the spatial relationship of cells as they say. And so this is a unique hippocampal property. I also say here, it's very magic. What I mean by that is that it's deeply non-understandable. So basically in cortex what we are used to, the way we try to understand cortical circuits in a microcircuit fashion is that we make little wiring diagrams for, say, receptive fields. They're like thousands of models for how you wire up a V1 complex cell or a V1 simple cell or, say, direction selectivity. And this is what we typically do in sensory physiology. We make little wiring diagrams for how response field response property comes about. Obviously in hippocampus you don't need to draw little wiring diagrams for these things because they just do what they wanna do. So this is a very magic, very dark ability of the hippocampus. Initially when I understood this, I decided never to work on hippocampus, but I changed my mind. Okay, so we looked into this a little bit and what we did is we, let me give you a bit more of background. What we obviously were interested in is how were these novel maps generated and sort of one classic idea is that there might be a pattern separation involved. Yeah, a pattern separation operation involved in the Lloyd gaps and Moses have, and also other peoples have very nice evidence for that in cortex when you put the animal into similar but distinct environments you get in the entorhinal cortex, you get quite similar receptive fields. When you then in the hippocampus in CA1, you get very different representations just like I showed you. And the idea is that maybe a pattern separation operation in the dentate contributes to that. Now, we followed another idea. We followed the idea that maybe internal factors could induce remapping. We already saw that our place fields, there seems to be quite some stuff in the cell. It's excitability seems to be a major determinant if it will form a place field or not. So what we thought is, hey, let's not look for this environmental stuff, but let's look for factors inside the animal. And what we looked at first was dopamine, which is thought to be a novelty in teaching signal. And more interestingly, manipulations with the dopaminergic system are greatly disruptive to map-like learning. So interestingly, a task that doesn't really require a map like a teammates go left, right, the animals can still learn when you put them full with dopaminergic agonists or antagonists, they still can learn these tasks. But when they need to use a map like a Morse-Watermase task, they get very bad under interference with the dopaminergic system. So what we did is we used what now turns out to be a very good or a very bad decision. We used a very dirty agonist, apomorphin. It acts strongly on the dopaminergic system, but it actually also acts on other systems. And maybe I come to talk about this and looked how this would affect the hippocampal representations. And what do we see here? Sort of we would do the following experiment. We have an animal run around. We record from multiple cells in the hippocampus and we see here place fields in this case. Then we would inject a little bit of saline. And what we would typically find is that sort of the place fields are kind of similar. They're still a spot here, still a spot here. They're not exactly identical because the runs of the animal are also not exactly identical and because these cells are quite variable in their discharge. Here again, another field that kind of stays relatively similar. You still have this firing spot in the corner. You still have this firing spot. This firing spot was not there, but there is a lot of similarity when we do this. When we then inject apomorphin at a high concentration, sort of one place field would be totally lost. Another one would remap, yeah? And it turns out that this is very prominent. If you look at the population statistics, so when we do saline injections, basically three quarters of the cell either fully stay or look sort of like they stay. And I actually think if we would push our behavior, give two more runs, assesses more carefully, I still think this is an underestimate of the stability, but 75 to actually more cells would do the same and very few place fields sort of would appear or would sort of remap to other locations, et cetera. Now, when we inject apomorphin, whole health breaks lose, yeah? Very few place fields stays the same, yeah? About 20%, many remap, many appear, many assert or so, 25% are lost. So this is very striking. Now, as I said, these were very preliminary findings and there is much about this that we still don't understand. We are still not really clear what dopamine receptor subtypes are involved, actually, when we give more specific drugs than apomorphin, we typically get less good effects. So this is very unclear to us. When we give antagonists to antagonize the dopamine allergic system, we also get quite variable effects. So this is stuff that is underway. What I think it tells you, however, what is interesting here is that you don't necessarily need two environments to make two maps. I think this is clear from our data. So we have here, this is presumably not a pattern separation operation because the animal is just running in the same environment and we get two maps. Let me summarize here. I talked in the first part about structural function relationships in the medial enterinocortex. We find different microcircuits for head direction cells versus cells that look more like grid cells. This is encouraging, I think, also our methods to sort of derive structure function relationships and freely moving animals, they are improving. It's still that we, from this, we do not understand, obviously, how it works or even the most obvious questions, what is happening in a certain grid patch, in one grid patch as opposed to the other, we still can't answer. I told you about the idea that hippocampal blade cells and silent cells, that these cells are, to some extent, internally preset, that they arise in just in no time. As soon as we measure, we see them. And I told you about the idea that maybe there is an internal trigger for remapping, or at least our data suggests this possibility as opposed to getting remapping exclusively from different environments. Thank you very much for your attention. We have time for a few questions. Yes, there's one here, and then Crystal. Do you think that every silent cell has a chance to become a place cell, or is there something like, is naturally different between place cells and silent cells? Okay, this is a very interesting question, and I oscillate on the answer. So from the X-cellar, obviously, there are tons of remapping experiments done in the X-cellar literature. And from these, you very much have the feeling it's random. So it looks, you put the animal in another place, and then either it loses or it gets, so you get this feeling that it's random. Now, clearly, in our hands, it was very non-random, so there were cell internal factors. So what I think is that maybe, this is part of why we started to do these dopaminergic experiments. Maybe there are switches that allow the cell to change its internal properties. Now, given the fact that Robert just showed that you can, by just intrinsic, cell intrinsic manipulation, you can sort of switch on a place field, or perhaps even switch it off. So that makes me think that these cell intrinsic factors must be very important. Yeah, we don't know if some cells are sort of forever silent. It makes no sense to me, but yeah, who cares? Are there key morphological differences between the classes, or are they all the same, like a cellotype? Yeah, we don't know, we don't know. So we should do this sort of the place cell, silent cell data that I showed. These came from patching experiments in the awake animal. It's exceedingly few cells. It's sort of 14, 15 cells or so, and only about half or so that we recovered, so there's no point making statements about that. As I told you, we are now quite a bit more efficient in recovering cells, this is Chuck's cell approaches, where we just stain the cell, and there we have high recovery rates, so it's clearly traceable with these techniques, and one should be able to find that. Other questions? Christoph? No, no, no, no, no. They occur very much preferentially in the place field. Can you use the microphone please? Yeah, I repeat the question. The question was if the complex spike or these gigantic events are direction selective, if they would come as in the movie, only when the animal enters in one direction, the place field or not, this is not the case. So they occur more in the place field, the tuning. We have too few cells to really strictly compare the tuning, it seemed to me tighter, but I'm not sure. So we think that actually these are very common events. I think that if I understand the extracellular folks correctly, this is really what is a complex spike. That the spike amplitude dives down so much, we think that the spikes in the classically described complex spikes which are the feature of the hippocampal pyramidal cell, extracellularly is really these big events. So we think that they are very common. But if you look at the extracellular literature, they actually have also done a very poor job. They should be able to pull this out where the complex spikes are, where the spike dives down most and it's not been done very well. Okay, one last question, Neno? Michael, thank you very much for a very nice talk in which you summarize some of the major debates in the field. It's also obvious from your talk that the reason for you to leave Rotterdam because things were too flat, too boring, which is still a pity I think, but because it's a nice country. But I had a question about your grid cells and the relationship with the head erection cells. So you initially started to say head erection cells are typically related to the post-subiculum and then you discovered this very intrinsic, intriguing wiring of the parasybiculum. It seems to have these global effects that the directional system might be tuned into one direction. So do you know what the relationship is between the parasybicum and the presybicum? Have you seen any specifics about that? Or are they too independent? Yeah, yeah. Okay, so we looked at this a lot. So I dance around a bit about the question how head erection cells and grid cells are related because we are not really in a full position to compare our data to the ones that you and Edward published. From our reading of the papers from your labs, it seemed like a mixture. We think that there are more head erection cells in the parasybiculum. We, what is not so clear to us, what is clear to us is in layer two of entorhinal cortex there are very few head erection tuned cells and I think this is in line with the findings. What is less clear to us is how many of the really in the entorhinal cortex head erection cells we have. We also seen these, but few, yeah. Now the interaction between the parasybiculum and the entorhinal cortex. So this circumcurrent axon, we think makes a global connection across the parasybiculum. The interaction we say entorhinal cortex is not global. It's a specific stripe. So we think that the, let me go back to maybe I have one here. Just a second, it refuses to work. So we think that there are these circumcurrent axons. This being the parasybiculum there, we find a lot of head erection cells. This top few of the entorhinal cortex, yeah. And we think that they make very specific connections to one stripe of the entorhinal cortex. And we think this is probably where one sort of spatial frequency is represented. Sort of the spatial frequency as you demonstrated changes from ventral to dorsal, yeah. So we think that they make global interactions along here and they make rather specific interactions with layer two along here. What we also saw is that we get some feedback from, that seems to return this, in particular from layer three cells. What we don't quite understand is basically when we stain, when we do sort of large scale stains here, we get exactly this connectivity. The whole parasybiculum will fail when we do a small injection here. Also when we, also we would get from a specific site cells back labeled here. What we don't see so very much, what we don't understand is this back projection from layer three. We don't see these so very much. But we thought it were specific cells on both sides. And layer three probably projecting back and parasybiculum very specifically, one piece of, one slice of entorhinal cortex. Probably one spatial frequency. All right, so now we have time for a coffee break. And the parallel sessions will start again at 10.20. And let's thank again our keynote speaker. Thank you.