 Is there a laser pointer or, or if there isn't I can, okay, if that doesn't work I can use the mouse. Is, is it already on? Hello? On the square thing? Yeah, yeah, yeah. Here you mean, right? Oh, I'm going to, okay. All right, good. All right. Okay. Very good. Thanks. Okay. Thank you very much. I'd first like to say I am going to be talking about topography, but in a very, very different sense from what Peter was discussing very little on the sort of a neuroanatomical topography and more about topography in the external world. I will have some speculation at the end that does sort of address some of the topics that were, that were in Peter's talk. I also just want to thank the organizers for putting this together in such a beautiful location and for inviting me to participate in it. Finally, and perhaps the most important thing I want to say is that I am essentially a messenger here. I'm not, not the, the most important contributor to this work by a long stretch. This is really from beginning to end the brainchild of a very talented ex postdoc in the lab, Cam Kang, Ming Dong Kang. And he, you know, he showed up in my lab with sort of the embryonic version of this project already in his head and proceeded to do just wonderful things with it. So I'm going to be telling you today, my presentation right here. Okay, good. A little bit about comparing olfaction to some of the more long, or systems that have been studied for a longer period of time. And that is the visual systems and auditory systems. And one of the things that we make great use of in our visual systems is the ability that with a single glance I can look out in this room and not only might I recognize familiar faces and know the identity of people in the room, but I also have immediate direct access to your position in space, right? I know where you are from a single glance. I don't have to search around the room to localize each one of you. And that also happens in other sensory systems too. And one of the most famous examples of that is in the auditory system where, for instance, we can certainly do a version of this ourselves, but specialists like this barn owl are really, really good at in complete darkness the ability to localize their prey just from the sounds that their prey is making. And so they would be able to again identify something about the characteristics of the sounds and maybe make some guesses about what it is, but also the sounds themselves convey positional information about where that object is in the outside world, okay? Now, how does that happen? And really there are two very fundamental ingredients in being able to localize cues from the external world. The first one is that there has to be some mechanism to preserve any spatial information that's present in the physical stimulus itself, right? So in the case of the visual system, the light coming from the sun scatters off elements of this tree here when it hits a point on the tree. Light goes in all directions as it bounces off of it unless it's a mirror or something like that. And so light goes all different directions and two rays that emerge from the same point. If your eye did nothing, they would strike different spots on your retina even though they were coming from the same physical external position and it would scramble the spatial information available in the light. But your eye has optics which essentially, if you will, performs a computation on the wave fronts coming from the tree and focuses the rays that are emitted at two different angles backed out none to a point on the back of your retina. And light rays coming off at different angles from a different point of course get focused down to a different point on your retina. And so the optics of the eye are the physical apparatus that preserves the spatial information present in the stimulus itself. Now that would all be for naught of course if the nervous system didn't go to some care to maintain that kind of information, right? And so you're all well familiar with one of the most famous sort of observations in neuroscience and that is the so-called retina topic projections, right? So that different spots on the retina project ultimately through the LGN to different positions in visual cortex, right? And so that spatial map, neural map preserves information that's present in the spatial information of the physical stimulus itself. And that's not of course a complete answer to how you localize stimuli but these are two necessary ingredients in order to be able to localize stimuli. The situation is a little more complicated or less obvious maybe in the case of audition. Audition seems like a difficult case to do full localization because you essentially have two sensors and you would imagine that your two sensors, your two ears would let you make say two measurements, two independent measurements of something related to the position of a sound source in space. And you might therefore wonder if you only are making two measurements and yet you want to localize let's say the direction to a sound source somewhere in the outside world you'd need to triangulate the position. You'd need at least three measurements in order to actually do that computation. And in fact you can do that with your two ears because your ears can actually extract multiple parameters from the sound stimuli that arrive. So for example in this horizontal plane the slight differences in the arrival time of the sounds to your two ears gives you an indication of the angle in the horizontal plane of the sound to the source the external shape of your ears and this is taken to an extreme in the barn owl where they're actually asymmetric one ear sort of tilted more upward the other one more downward gives you two different intensities and that difference in intensities is then used to estimate the angle in this direction. So in a sense you have two sensors, the two ear holes but you're making four measurements of one kind or another and that's enough to compute an estimate of the exact direction of the stimulus in the outside world, right? Okay, so that's the sort of physical side of the auditory system how might you be able to preserve the spatial information, the raw stimulus itself then of course you also need the neural hardware to make use of that and again a really beautiful set of stories many decades old at this point have identified many of the contributing neural mechanisms and for instance in this case of computing the time difference in the horizontal plane there are these elaborate circuits that use delay lines and look for coincidences of arrival times of excitation in order to essentially map the spatial position in the outside world to a spatially receptive field in the auditory system. Okay, so both of those share this physical apparatus and then the neural hardware to preserve that information. Okay, so what about olfaction, right? So I think olfaction has long been viewed as being in a very different category from these two senses in terms of the ability to infer information in the outside world. We talk a lot about the identification of odors in the sense of is it lemon, is it strawberry, something else like that we don't talk as much of course about localizing stimuli. That's not to say it's not a crucially important phenomenon and there's been a lot of work that's been done on that I'll remind you of a little bit of that right now but one of the main reasons why people think that the localization of external world objects from smell is a complicated one is one that's been mentioned I think a couple times already now here most recently in Tristram's talk last night and that is the sort of plume structure of odorants, right? So these show some very old images of a pipe that's streaming out a fluid with a tracker in it and what you can see is immediately at the emission point of the pipe these come out in these sort of very nice sort of jets but then they tend to break up over time and they form these dramatic filaments and the exact shape and structure of these plumes depends greatly on the parameters that are involved in the system but one of the key things to keep in mind is that if you're a point sensor somewhere out here the direction from which one of these filaments sort of crosses you is not necessarily very informative about the direction it doesn't immediately give you a vector to the source of the odor, right? I might be standing here, I might have a filament sort of slap me across the side and might think it's therefore over that away but in fact the source of the odor is essentially in that direction, right? And so it's not an immediately reliable indicator of the direction to the source of an odor if you simply looked at if you had some way of sort of sensing how these filaments sort of swept across your face, right? And so that's given rise to some active interest in exploring some very different ways of localizing stimuli and that is that through the accumulation of evidence over time and Massimo Vergasola had a really lovely paper on this essentially providing an algorithm that maximizes the amount of information with each step that you take in an environment sometimes you take daring steps and directions that you think might not be productive but in order to sort of check your hunch about where the source of this odor is and over time you can fairly efficiently find your way to the source of that odor, okay? And I think it's very clear that this sort of general kind of phenomenon happens, right? And that it's especially maybe important for let's say thinking about a moth in a large cornfield where the gap to a potential mate might be 10,000 or more times the body size of the animal so you're really looking for these rare distributed sources in sparsely distributed sources in these very large landscapes. However, I would say so, well, I have not a shred of doubt that this phenomenon happens I would actually say that for mammals this is not the only way that olfaction is used and in part I think this comes from a lifestyle difference mammals tend to live at very high density or often live at high density and so simply sort of finding each other isn't always the hardest problem but the other thing is that animals tend to pay a lot of attention not just the sense that come lasting through the air but sense that are right in front of their noses on the ground and this is a video taken by a head-mounted camera on the dog, now one on the side and this dog is just being led on a leash but it's a trained tracking dog and it's following the trail of someone who had just walked through here before and what you can see is that the animal keeps its nose right on the ground it's not sensing this thing from 300 meters away with its nose up in the air it's really trying to follow the trail on the ground as the animal progresses forward and these trained animals and wild animals can follow these tracks at incredibly high speeds if a human handler were willing this dog would be following this track on the run so that's a fairly impressive accomplishment when you think about it the ability to sort of parse out the spatial information about the trail at high speeds as the animal is breathing and therefore only episodically sampling the information outside the world so what Cam was asking when he first came to the lab is might there be more information than we thought about the sort of spatial direction of an odor than just simply this sort of integration of information by directed exploration and so one of the most important differences between this air-centing and this ground-centing is the distance between the source and the nose and in particular if the speeds of flow or if there's very little flow or if the speeds are relatively low air flow exhibits a property of being so-called laminar which means basically if the streamlines exhibit sort of very smooth characteristics and you can illustrate that actually quite easily so this is a mouse that's been tracheonomized we have two pipettes here that are both going to puff out essentially a bit of smoke we've actually digitally colored them differently but it's really the same tracer in both cases which are actually taken as two separate movies and then superimposed on top of one another and we tracheonomize the mouse so we can provide an inhalation as we puff odors out of these tubes now one of the things that I wanted to sort of say is that I think one of the reasons maybe that this hasn't really been sort of noticed before is that as olfactory say physiologists what we tend to do is we tend to sort of for instance if you're delivering a stimulus in sort of let's say even in the old anesthetized days to the animal we would put a nose cone over the animal we would flood the entire sort of area surrounding the nose with the odor that way we knew the animal was getting it with good kinetics and all this and that sort of rapid filling of the entire space of course would cause you to lose spatial information and the key thing about the way that Cam's puffing out the tracer here is he's letting it out in dribs and drabs to be very specific the flow rate outside of these tubes is significantly smaller than the total bulk flow rate into the nasal cavity so it's really that the sniffing is drawing the odor in it's not that these are hosing out massive quantities of odor and that's really crucial for being able to see these kinds of phenomena so if you just hear a little video you can see the two tracers here entering the nostril and I'll play it one more time and what you can see is that the spatial information from these very close sources seems to be preserved at least up to the entrance of the nose right and so that at least opens up the possibility that perhaps there might actually be some spatial information available in this queue that's coming from the outside world Can we get a couple of the sideways? Yeah, you're basically sagittally, yeah Yes, exactly, that's right and I should say please do interrupt me during my talk so I will do this so I keep note of time Alright, so And that would be the sniff? I would call these wussy sniffs in the sense that we're not really trying in everything I show you today we're not really trying to mimic the sort of onset and drop off of sniffing we don't do any exhalation or anything like this in this case this was just a steady draw through the trachea and we just simply puffed out the odor and so essentially there were vows that just simply gated the tracer in that case right so the actual sort of dynamics of the acceleration and all this kind of stuff I think are certainly going to be important but there's little reason to suspect that that dynamics is going to fundamentally change the trajectory the actual sort of positions that are traced out by the individual odor and molecules so it'll change the speed with which they traverse it and I'll get to that a little bit later on in the talk but it doesn't change the spatial distribution very much Yeah, yeah, we're pulling it through the tracheotomy we're mimicking the sniff but it's not a very realistic mimic of the sniff It was only into one nostril Yeah, you're looking for the side Yeah, exactly, sorry, yes It's essentially a theatrical smoke machine So, yeah Yeah, exactly There you go, there you go, yeah, exactly, yeah So if you want to find out can the animals do some kind of spatial localization with odors the most direct way is to ask the animal itself of course, right So what Cam set up was a head fixed assay and he had two tubes that were delivering constantly delivering a flow of air there was a suction tube not really shown here to draw away anything that's produced sort of symmetrically oriented on the body axis of the animal and he would train the animals to lick in response to the presentation of an odor based isoamyl acetate He would deliver isoamyl acetate either from the left or the right and he would swap them on different trials and the animal was rewarded only if it was isoamyl acetate and only if it was coming from the left, right He also did two different concentrations, both 10 parts per million and one part per million in air dilution and to make sure that it wasn't essentially just a concentration threshold that was being detected and so after a certain amount of training the animal could certainly tell left from right quite easily so shown in red here is the frequency of times that the animals lick the main part of the training is to get them to stop licking whenever it's not going to be productive what you can see here is the animals lick very reliably when it's isoamyl acetate at 10 ppm or 1 ppm they don't lick, you know, they hardly lick at all when it's coming from the left they hardly lick at all when it's coming from the right so if you use phenylopyl acetate, a different odor then there's no licking whatsoever, right and so this shows that the animals can reliably distinguish odors coming from the left and coming from the right that doesn't prove that it's olfaction that they're using they could be using the wind, for example, right from these two cues and you have to sort of come up with why would the flow velocity be different when there's isoamyl acetate rather than air but of course all these things are theoretically possible Cam had to do a number of controls to determine whether they were really using olfactory cues for instance if you cut the whiskers which might be one mechanism of using the can of sensory cues it doesn't really affect their behavior he did a number of other I'll talk about this one more in a second but one of the most important controls here is to close the right nostril in that case they seem, they behave we interpret it as they behave that the odor is always coming from the left that's the rewarded side so they lick on every single trial and then on the very next day you reopen the right nostril and close the left nostril then essentially they never lick and that's consistent with interpreting it as they always think the odor is coming from the right hand side that's where they're drawing in the odor from you can also inject them with this magical chemical whose mechanism of action really isn't fully known but it seems to reliably and seemingly fairly specifically wipe out olfactory sensory neurons and that really sort of shuts down the behavior altogether so that shows you that you require olfactory cues they at least have to be permissive that by itself doesn't prove that they're instructive but taken together I think these controls argue that the animal is using olfactory cues for its behavior to solve this horizontal task I want to say that this is not really a surprising result there was a previous publication in 2006 again suggesting that rats can smell in stereo it's always simultaneous and in fact there's always flow coming from the two directions it's just a question of whether it's been doped with odor or not yes right and they meet in the middle and it seems likely that there is a little bit of crosstalk we don't have a direct measurement of that but yeah exactly that's right so it's exactly right and that's what we think is happening here so this is just essentially a replication of that previous work yeah so it's entirely possible that they are using additional cues to do this right so I mean the narrous occlusion simply shows that something entering a nostril is an important component that the dichlobenil injection argues maybe a little more directly that it's probably olfactory but because we don't know the mechanism of dichlobenil and because it could be permissive and that's only instructive you can't rule out but what it does show is there's certainly some sort of chemisensory interaction there that is driving this behavior right so sorry what I'm still having a hard time hearing you so you're saying the concentration oh you're saying what if it's slightly different oh that's the reason for the control of 10 ppm versus 1 ppm right oh yeah yeah right so you're absolutely right that you could essentially come up with a lookup table for doing this if you imagine that there's a slight bias in the concentration between the two sides is basically what you're saying right and our ability to calibrate it there isn't such a difference right but you could argue that there might be yeah exactly right exactly and I guess I'd say that one of the other things that so this tubing swap I guess I'll bring it up right here right now those kinds of things are actually turned out to be really important controls because animals can also cheat all over the place and the chem had to work quite hard to make sure that they weren't cheating on this task under certain conditions I can get too many questions so the tubing swap essentially controls for the possibility that just the different sides are emitting different contaminating odors for example okay alright so I want to address one I often get questions about this kind of thing is one possibility is that this isn't really being solved by sterile faction from the two nostrils maybe it's actually being solved by sequential sampling very much in the same way of the vergesola mechanism which is the general concept of accumulating information over time about where the odor sources and so here you can see that even though it's a head fixed preparation there's a little bit of nose movement at the end and one might wonder is that actually what they're using to determine the side that it's coming from and there are a couple of things so we can't rule out the possibility that they're getting some information from that but one of the things to point out is that the sequential sampling works essentially equally well whether you have one nostril or two I think nostril manipulations that the animal's behavior acts as if it's really a stereo effect that's happening argues against the interpretation that nose movement and accumulating evidence over time is a major contribution to this behavior yes so I don't have any slides to show you but Cam did some interesting experiments actually where he kept it closed for about three weeks or so and over the course of a little more than a week or so the animals start getting to about an 80% criterion success rate on the stereo task which we thought was pretty interesting and to be honest that was actually what Cam was his initial hypothesis was more fine scale discrimination in the horizontal plane and the project almost died when he did a control and that was to do the tubing swap and it turns out that if you have one so if you have two nostrils and tubing swap does nothing animals behavior essentially stays perfect right if the animals have learned the task to perform an 80% or so over three weeks you say wow they can do it even with one nostril but then you just simply just swap the tubing they actually flip in their preference so it's contaminants in that case they're cheating in the task when there's only one nostril and they're not they don't seem to be cheating when those two exactly yep that's a systemic injection it's not we're not like injecting it in the nose or anything like that is there what yeah you're right I guess we could do zinc sulfate we haven't thought about that so you would say that that would be that's essentially just to argue that not only is nostril are nostrils important but the actual factory that's true that's a great idea that's what this is yep no problem it's an important thing okay so where I think the so this is cool but it's not new there have been publications previously on sort of stereo faction the horizontal plane what was really new was that Cam discovered that like the barn owl not only can they do the horizontal plane they can also do the vertical plane and in fact they only need one nostril to do this right so here's an animal where one nostril has been permanently closed now the odors are being delivered dorsally eventually rather than left and right and you again take the animal through the same training procedure and again the animal can perform really really well and again we so now you can't close the one remaining nostril because the animal can't breathe but you can do a number of a number of controls to again convince yourself that the animal is really using all factory cues here a lot of some of these are the exact same same controls he did a few more in this case so for instance the pressure reduction is essentially just again yet a new way of looking for mechanosensory controls or timing information and then you know the one that sort of makes the behavior go away is when he dopes the clean airline with a small percentage of isoamilacitate and that I think really argues very convincingly that it's again chemosensory driven ah no no sorry I'm not sure I'm not really showing that this is just showing reproducibility the training period for these two for these two tasks it's about the same it's about three weeks or so to get the animals up to I mean depends on where you set the criteria in threshold but roughly speaking that's the case oh yeah no this is really well well let me let me back up it it's as if it were proximal in the sense of the tube you're actually about a centimeter away but because of their large diameter and because where you were very careful these laminar flows in a sense it's it's it doesn't matter what the distance is as long as you don't break up into turbulence right yeah what is the flow well what is it what is the flow rate well I don't remember the number right now I can I can get that number too though yeah the airflow from that particular yeah yeah so it could be right but but if you just simply don't so it every so for instance like in this case here right everything is exactly the same it's the same pressures that are being the exact same sequences of valve you know valve openings and closing all the tubing is exactly identical there that there and the only thing that's happening is he squirted a little bit of isoamyl acetate into what had been the clean air jar ahead of time right and so and this completely wipes out the animal's behavior basically right and so so I I think that pretty convincing argues that's not mechanical anyway okay let me let me let me move on here because there is there is more to tell okay so there's argue that together like the barn owl even though there are only two nostrils the mouse seems to be able to infer the the direction of the of the odor source in both the horizontal and the vertical and the vertical plane one more haha no we don't control this is the animals freely behaving yeah I'm not fully sure can we talk about this during the coffee break yeah okay so so okay so cam wanted to so the natural question is of course how on earth does the animal solve this task you have only the two nostrils how do you get enough information to sort of triangulate the direction to an odor and so a cam reason is that you had to have the two components that I mentioned the beginning it has to be at the physical apparatus inside the mouse's head to preserve the information that's present in the in the actual external queue and then you know there must be the neural hardware that also preserves this information right for use by the nervous system correct it's only not into again there's there's no surprise about the horizontal the vertical is the really surprising one exactly and so try to understand how this happened cam decided that he had to take a closer look at the actual anatomy of the nasal cavity itself right and so what he did was was he took a high resolution CT scans about 10 micron resolution of three different animals doing their entire nasal cavity all the way back to the curved form plate and a little bit beyond and I was impressed I shouldn't have been if I had normal literature well enough but I didn't I was impressed at the extreme reproducibility of the turbinate structure of the of the three different animals it's an insane level of reproducibility and so then so he got as you know something like a thousand to fifteen hundred different slices from the CT scans and you know sort of towards the tip of the nose the structure of the nasal cavity is relatively simple and as you progress farther and farther backward and the turbinate structure becomes more and more elaborate you can see that they're just these astonishing astonishing labyrinthine structures here and he ended up deciding that because there's actually tissue and he wanted to include the actual space for the tissue and you can see that in the CT scans but a lot of the automated algorithms that we tried didn't work he actually laboriously traced by hand all of his all of his sections and and then if you essentially throw away the original images and just put together the tracings then at that point he was able to build a three-dimensional digital model of the morphology of the air passage ways in the nasal cavity that's what's being shown here so a medial view of the air passage ways here's the trachea so right so here's the tip of the nose here's the trachea down here and so you're looking at here from the medial face from essentially the centerline of the animal here you're looking from from the outside of the animal inward and here are the turboform plate would be right here and you can see here from the back and from the top and then what he did was he used this 3D digital model of the mouse nasal cavity and he had it 3D printed in a transparent material and he did that because he wanted to actually be able to trace the flow lines through the nasal cavity and actually see what the pattern was and so he had this apparatus for delivering odors from different angles both in elevation and azimuth here's the model itself it's not quite as transparent as we would have liked so we focused a lot of our analysis on the so-called dorsal miatus which is the part of it that had the best visibility both from above and from the sides so that we could actually map three-dimensional positions it's a scaled model it's scaled about seven and a half times compared to the actual wild type one and because we're going to do the tracing in fluid because there just aren't that good mechanisms for tracing in air you have to then of course match certain parameters and the important parameter match is a number called the Reynolds number and briefly without going into detail what you essentially end up having to do is choose the solution that this is immersed in in terms of its kinematic viscosity and set the flow rate of the bulk flow rate through the model so that this ratio essentially matches the size the scale magnification of the system and so here are the numbers that we used here I should say that to get the exact physical behavior there's a second number that you should match which is called the Schmidt number that takes into account the diffusion of molecules right this just accounts for the flow so this should perfectly match the flow also diffusion we're not matching the Schmidt number right so this physical model does not take into account the diffusion of the molecules okay so using this we could get at least reasonable quality visualizations the flow patterns the nasal cavity I'll show you what the real data look like here so here we're looking at the nose model again trachea this way tip of the nose is over here and cam's going to release a little bit of just it's just a calcium carbonate tracer in this sucrose solution and when you when he releases that you can see the actual travel through the nasal cavity during again a virtual sniff this is just under constant draw and what you can see is that it doesn't fill the entire nasal cavity right not all portions of this sort of mock olfactory epithelium receive the odor and essentially just one small portion of the olfactory epithelium was exposed to that and again you would get a completely different answer if you hosed the odorant at high speed here right because you'd fill the entire passage away and then it would cover the entire epithelium we've we've we've done that this only happens when the draw the source of the draw is coming from sniffing right over the time so calcium carbonate is a little bit more dense but over the these timescale it won't play a role yeah I would actually guess we've actually a little bit of modeling I don't think I have any slides to show on that but actually I I think for certain odorants actually the absorption is probably an interesting relevant effect actually and I can comment on that maybe yeah yeah so they so in general basically if the odorant comes out slowly then in general you know and is not you know over the entire area but is a focal sort of point source of odor then it tends to map to a particular a small subset of the overall epithelium can actually try doing this with the gold particles but they all they tend to they're very ballistic and they just hit the walls of things very early on so we've we've we spent about three months trying to actually do it with real world objects and it just doesn't it's interesting we didn't try anything radioactive possible that would work yeah yeah yeah so so the there are two it's a great it's a great idea I'd love to do that experiment there are a couple of issues with that I mean so one of them is is that anything that you insert into the nasal cavity tends to disrupt the flow pattern unfortunately and and and these are all fairly tiny and so actually putting in enough apparatus to do imaging that I mean there isn't an endoscope in the world yet that exists that that actually could pull that off we would love to do that experiment so yeah okay so this is that one flow speed just to find out what kind of parameters this is to depend on one of the first things that we that he tried just during the flow rate and within limits so this is overlaying three different flow rates on in three different colors I see this key got a little bit covered up here but the but the highest sniffing rate if you will the highest speed sniffing is is is shown in red sorry the next the next slide shown in red and then in green and blue are slower rates and only the blue at the very slowest rate you actually see anything different and that's again because if the sniffing gets sufficiently slow that the odorant has time to spread out before it enters the nose then it will cover a little bit more of the overall overall nasal cavity right but by and large it's essentially over a physiological range the actual spatial pattern is invariant so now the really interesting experiment if you have essentially point sources in the outside world here he can select three of them here in primarily the horizontal plane here again color code them differently and then if you deliver the odorants then voila you can see that these different spatial positions in the external world map to different positions on the epithelium right and so this should gives you a physical mechanism whereby external space is the topography of external space is preserved it's not only true in this horizontal axis remember that this horizontal axis is sort of the boring easy example because you already have the two nostrils to solve the problem more interesting cases vertical distribution of odors and what can sound here is that the important thing to do is to watch the nose model from above in this case and it turns out that in that case that the different positions in the external world again map to different positions in the nasal cavity but it's in the medial lateral axis not the dorsal ventral axis and so overall can took a you know constructed a map I'm in the interest of time I should try to wrap up here he was able to do the highest resolution in just this dorsal meadis here there's a too much scattering to really do good 3d mapping elsewhere in the nose model and even in this one small passage let alone the availability of other passageways through the nasal epithelium he's actually able to essentially infer that there is this map even within this dorsal meadis and that the most sort of ventral locations tend to map near on the medial face of this dorsal meadis and the most dorsal locations map to the more lateral portion of this dorsal meadis okay so and so the thing it was really surprising is that there's as odorants enter the nasal cavity there's this 90 degree almost perfectly 90 degree clockwise rotation of the odorants so that this axis gets mapped to this axis here and that'll be important for some of the more speculative stuff that I'll mention if I have time to get to that I might skip over this but if there are questions I can address this. Suffice it to say that diffusion is potentially an important phenomenon we think it significantly reduces the strength of this effect and we've gone to some effort to try to actually measure the actual flow rates during the actual behavior itself by doing some separate experiments in which we we didn't want to disturb the actual flow pattern outside the animal so we would measure the actual flow rate using a thermistor and then we put a chest band around around there and use this to calibrate this and give us an approximate measure of the actual flow rate during the high speed sniffing it works reasonably well and we can see that the animals do significantly elevate their sniff velocity the actual speed of airflow into their nose when they're actively engaged in this task and that this should help reduce the effect of diffusion and I can go over this in more detail in questions if you want okay so all of this has been done basically with the tracing experiments of course was done with this scale model because of the potential effects of diffusion ham wanted to really have a direct measurement of is there again a spatial difference in the neural responses to odors coming from different sources and so he developed this really elegant preparation in which he was able to insert in the living well recently dead this is essentially EOG preparation and where he was actually able to use two electrodes coming from different sides recording from both the medial and lateral side of the dorsal myatus and deliver odorants in front of the animal varying the position and look at the EOG responses to the two electrodes and what he found was that if you deliver odorants from right and left then the amplitude of the EOG responses are nearly identical in the two cases but if you are delivering odorants so sorry I should point out here that the red line is for the response on the medial side the blue line is the response on the lateral side so if the odors come from right and left you might think that that would be the one that would give naively you would think that would be the one that would give the biggest left-right difference in the nose if there was one and there isn't really any difference that we can see whatsoever but if you use the result from the tracing experiment and test the dorsal ventral axis then you can see there's a swap so when the odor comes from the dorsal side the EOG is strongest on the lateral side when the odor comes from the ventral side the odor is strongest on the medial side the response is strongest on the medial side but it's about a 10% effect and we think this is mostly due to the effects of diffusion interestingly if you change the sniff rate in this model and again this is just done through the tracheotomy you can see that the information is present and discriminable at relatively what would correspond to the fast sniffing and it really gets wiped out if the animal is just doing baseline breathing speed so the sniff speed does seem to be an important factor in the ability to preserve this kind of information okay so just in the last couple of moments that I have here I want to engage in some speculation where we're sort of going with this work and again I want to emphasize we don't have the direct proof of any of this but I think if I don't tell you this you won't necessarily I think see what we hope is the complete picture here and that is what is really happening on the downstream neural hardware side and as Peter and others have mentioned prior to this you know that the multiple sensory neurons and the epithelium converge on to glomeruli there tend to be one or more medial glomeruli lateral glomeruli for almost all of the different odent receptors and it's been and that's maybe shown here this is like again some of the views that Peter showed and this is from one of his lab's papers where you can see here the lateral glomerulus the medial glomerulus it looks like two glomeruli there lateral glomerulus there and an interesting phenomenon is that if you actually look at the sensory neuron axons and again this is a dorsal view down from the top there's actually essentially a line of bifurcation where the more lateral neurons project to the lateral glomerulus and the more medial neurons project to the more medial glomerulus and so this actually has all of the ingredients necessary for preserving that lateral medial split which you remember is for actually be for coding the dorsal ventral direction of odors this actually maybe gives some insight into this actually shows that in a sense olfaction is a lot like the barn owl instead of just having the two nostrils and making two measurements there are essentially four measurements being made and that's because you have essentially two glomeruli lateral and medial on each side and that gives you actually enough information that you should be able to triangulate the position of a notar indeed we actually think that the mystery suspect this is pure speculation there's further than that and there's a really beautiful system sometimes called superficial or external tufted cells that receive input from a single glomerulus and then they send their axons deep and then into the other hemi bulb so staying within the same hemisphere but projecting to their hemi bulb their axon terminals lie directly underneath the same glomerulus on the opposite side and we've not really had a clear idea of why there are these two glomeruli receptor type and why these cells project directly to the opposite side but Cam's hypothesis is that this is essentially a differential amplifier because of the effects of diffusion these signals, these sort of 10% differences are relatively modest and like many other examples of contrast enhancement in color vision and other areas like that there's essentially a differential amplifier a push pull here that one side the stronger, more strongly activated on the opposite side will suppress the activity on the opposite side and so the overall picture again is that the nose is a lot more like vision or an audition than we thought there's a physical apparatus in the nasal cavity itself that seems actually a lot like the lens of the eye and that different positions in the external world map to different positions on the nasal epithelium and like the barn owl you get actually four measurements coupled together with what might be some clever circuitry to help you identify the source of a notar again, I just want to acknowledge this is really the brain child of Cam Kang he got some excellent help from Mike Okuma and we collaborated on the EOG experiments with Haicheng Zhao at Johns Hopkins and I will now take questions