Sandeep Robert Datta and Venkatesh Murthy: Why is Smell Such a Mystery to Scientists?

Lydialyle Gibson: Over the past year and a half COVID-19 has brought a lot of new attention to something that many people, perhaps most people, probably didn't think much about before, our sense of smell. Across the globe, the virus affected millions of patients' ability to smell, most temporarily. But for some, the loss has been long term. Besides making clear how strongly sense of smell affects a person's larger sense of emotional well-being, as patient after patient has described the strange and distressing feeling of smelling nothing, the pandemic also revealed just how much of a mystery our olfactory system remains to scientists. And unwinding that mystery is important for human health—even beyond COVID-19. Loss of smell is linked to depression and anxiety and can be an early warning signal for neural illnesses like Alzheimer's and Parkinson's disease. Cancer, studies show, has a specific odor. And for neuroscientists, smell holds the potential to unlock larger mysteries about the brain's inner workings. The oldest of all senses, it is intimately connected to areas of the brain that govern fundamental functions like memory, and fear processing. So how can something so seemingly basic still remain so unknown? What have scientists learned? And what are some of the important questions they are trying to answer now? And how does the pandemic fit into the broader landscape of olfactory research? To talk about all this, we're joined by two leading experts in the field. Sandeep Robert Datta is a professor of neurobiology who studies, in part, how smells help animals and people process information and make decisions and predictions about the world around them in ways that then help guide their behavior. Last year, he was lead author on a paper analyzing how COVID causes smell loss. The researchers found that it was the support cells surrounding the olfactory neurons, not the neurons themselves, that the virus was infecting. Venkatesh Murthy is the Raymond Leo Ericsson life sciences professor of Molecular and Cellular Biology, and the Finnegan family director of the Harvard Center for Brain Science. Part of his research investigates how the brain uses smells to help store memories, as well as the neural processes that help us identify individual odors in an environment cluttered with numerous different smells. Recently, he's also been studying how animals follow scent trails, which, like smell itself, is a deeper question than it at first appears. Professors Datta and Murthy, Welcome to Ask a Harvard Professor.

Sandeep Robert Datta: Thank you for having us.

Venkatesh Murthy: Thank you for having us.

Lydialyle Gibson: So I guess the first question I want to ask you is, why is the sense of smell such a mystery? And from a neuroscience perspective, why is it so important to understand?

Sandeep Robert Datta: I'm happy to kick us off. So I think, you know, perhaps it's worth thinking about the senses that we as humans really think matter. To better understand why I think for so long, we have underappreciated our sense of smell. So we as humans are obviously visual creatures, right? A third of our brains are devoted primarily to processing visual information, and we interact with the world around us primarily through our sense of sight. Consistent with that, you know, historically, neuroscientists, who have been interested in trying to understand basic principles of how the brain works, have really focused on trying to understand our visual systems to understand how cells in our brain encode information about the sights that are around us, and organize that information in a useful way to allow us to, for example, identify objects. So there's been basically a 60-year history since the founding of neuroscience, where we've paid attention to how sight works. And I think, you know, increasingly, folks are paying attention to how our brain processes sounds, of course, because audition and speech are so important to us. Touch is fundamental, so we think a lot about that. But olfaction, you know, for humans has long been kind of considered a bonus sense. An add-on, a convenient aesthetic pleasure, that sometimes, when we smell rotten eggs, you know, also causes aversion—that isn't really important to our daily lives. And so I think for this kind of sociological reason, it's tied to our neurobiology as humans. You know, we might, as scientists, have just historically paid much less attention to smell. But I think it's important to remember that for the vast majority of animals on this planet, the most important sense they have is their sense of smell. Most animals interact with the world by detecting volatile chemicals in the environment that signify things like food, predators, or mates, and use those cues—use their sense of smell to—usefully navigate the world in a way that enables their survival. And the fact that we don't use our sense of smell in exactly that way, I think, has biased the way that we think the brain works and has biased the research that we focused on.

Venkatesh Murthy: Yeah, to add to that very nice kind of a background to that, I would say that beyond this intuitive “So I can I live by sight, therefore it seems reasonable to study and think about that,” there's also this way of studying it. So imagine like, if you want to do research in sight or sound, you can reproduce sounds, you can record sounds, you can reproduce images, you can record images, all incredibly easily enough for some decades now. And very easily now. That's just not true for smells. To say that's an interesting smell coming out of this homemade bread that I just baked. Well, what are the chemicals? How do I record what the chemicals are? And how do I present it? So even when now we're in a place where it clearly is interesting, and I think hopefully, there's increasing interest in studying it, I think part of the limitation is also like, how do we actually study it? How do you present things? What do we present? So, it's kind of this two-layered thing that I feel is going to get. But as Bob said, in a funny way, this pandemic has shown how important this so-called background sense is to us, that if you don't have it, people, I guess I've heard descriptions, right, of people saying like, “Oh, that just feels like I'm smelling grayscale air.” I mean, it's like, there's nothing there, right?

Lydialyle Gibson: Okay. What are the questions that you all, in your own labs, are most interested in at the moment? Like, what do you wake up in the morning thinking about?

Sandeep Robert Datta: You know, I think a question that I've recently become very interested in—for which we have no answer, but it's like a super interesting question—is a question about perception. You and I both think that lemons and oranges smell similar to each other. And we both think, even though, you know, we're separated by some distance, and we haven't talked about this specifically—we both think that lemons smell really different from pizza, right? So we, in our brains, we have this kind of common map that describes relationships between all the odors in the world, right? And I'm really interested in understanding where that map comes from. Is it built into our genomes? Is it the consequence of our experience with odors? How does our brain modify the map? Is the map in my brain really similar to yours? Or is it fractured in interesting ways that reflect my own experiences and my own personality maybe? And so we've been thinking a lot in the lab about, you know, a seemingly simple question, like, “Why do lemons smell like orange is different from pizza?” And we have no answers. But I think it's a really interesting question.

Venkatesh Murthy: Excellent. So I think I can layer on that. So you have lemon, you have pizza, you have all these smells. But you know, the world doesn't present these smells in isolation to us, you know, unfortunately, or fortunately. So, I think my obsession has been this, for the last, I don't know, maybe eight years or so is like, how do you take a collection of molecules, chemicals, that in your head, you're grouping them into objects, you know, for it to fit the description? It may be, it may turn out that maybe is not the right metaphor, because we're borrowing a little bit from the visual system, right, I can see a face I can see a car. So here, we're thinking I can, I can smell a pizza, I can smell a lemon. But whatever it is, there is some way of looking for one object or one collection of thing—but nevertheless, being surrounded by all this other stuff. So this parsing, the olfactory scene, it just it seems to be fundamentally fascinating. And it's related to what Bob is saying. Because if things are all completely related to each other, and you mix them all up, maybe it's much harder to pull them all apart. Whereas maybe if you put things that are very different, distinct from each other, maybe it is, but again, we don't know the answers. But that's one thing that our lab is very excited by.

Lydialyle Gibson: One thing I wanted to ask about it, you know, thinking about how smell itself is like such a slippery thing to even sort of grasp, what it is, with the chemistry of smell, and what is a smell? And how compared to vision, this is not something that's sort of neurobiologically, you know, we're biased to understand, but how do you approach questions like these? How do you think about sort of, experimentally, when you think about these questions or questions like them, what does that work look like?

Venkatesh Murthy: Yeah, I can start off maybe with the easy thing, so Bob can tell you the more complicated ways. So I think, we tend to, so far, focus on presenting more or less stationary, you know, collections of chemicals that will mimic the specific odors. And then, you know, ask mice—we use mice in our lab mostly. And we ask them like, can you basically train mice to tell us, you know, is it this or that? Is it different from that? So you're giving kind of synthetic smells, if you will, and asking mice to tell them apart. There's many different flavors of it. But at the bottom of it, it's discrimination and recognition sort of indirectly, like in is it something that you saw before? What we lack is, what I know what Bob was getting at, which is the perceptual. It is very hard for us to ask experimental animals, like, what does it feel like? What does it sort of smell like? Typically, it is it does it smell like X? Does it smell like Y? You're comparing with something else. So that's where we spend most of our time. Moving forward, we're getting a little bit more natural behaviors where like, if you just ask an animal to track, as you said in your introduction, track a scent trail. I mean, it's just amazing when you think about it, it’s actually not clear what the actual strategy algorithm ought to be for an intelligent agent. So I think it's the seemingly simple behavior which animals do effortlessly. We’re interested in studying that a little bit more controlled in the lab. Sorry, Bob, I interrupted you.

Sandeep Robert Datta: Oh, no, that's great. Yeah. So I think, you know, one thing that both Venki and I do, is try, pretty explicitly, to relate the behaviors that are generated by mice in response to different odorants that we present in the lab. And, and as Venki was saying, these are, you know, generally, but not always, synthetic odorants. We buy from like chemical companies. Often there are purified components of things like cat pee, or fox poop. I will tell you that the cat pee odor smells really horrible when you open it in the lab. And you shouldn't spill it, which has happened. And it’s awful. So, you know, we're often working with these kind of these synthetic purified chemicals, or mixtures of these chemicals. And then using the tools of modern molecular biology or neurobiology to ask about how the circuits that convey information about odors from your nose to your brain—how the circuits operate. So we ask pretty basic questions about how, for example, the cells in your nose might encode information about what odor you're smelling. And we think about how the structures in your brain that respond to odors that are detected in your nose, how those brain centers might organize information. And our hope is, in the long run, to be able to relate that to how animals actually perceive odors. And Venki is, you know, exactly right. You know, humans, you know, we are, we are very self-critical in the sense that we think we have pretty bad senses of smell. It's not obvious that that's true, but, but it's likely that we have a worse sense of smell than say, a bloodhound. But the one thing we do have is language. And so, I can give Venki you know, something to smell—some random chemical off the shelf—and ask him what he's smelling. And he will tell me, “Oh, that smells like you know, cheese, actually a blue cheese,” and can go on and on with evocative descriptors of what the modern molecular odorant smells like that I bought out of a chemical catalog. But mice, mice lack language, they can't tell us what they're thinking or perceiving. All they can do, is kind of report back to us what their brains thinks they're perceiving via behavior. And so there's this kind of layer that sits between us and the animals, where we can't really ask them what's going on in their, in their perception. And we can only kind of see a shadow of it, as reflected by, you know, whether they lick or don't lick when we train them to discriminate two odors, or whether they turn left or they turn right response to a particular odorant. And that creates, you know, huge challenges for both me and Venki and understanding how activity in the brain is actually transformed into perception in a meaningful way.

Venkatesh Murthy: Yeah, I think one thing to maybe add to that Lydia, to go back to your original question, is that I think what drives us I think, certainly Bob and I and many of us, is to understand the brain basis of this, right? Because one can, one can be satisfied with understanding the behavior, right? You give it, this is what you got, and then leave the brain alone. But I think, for us, I think understanding how brains, neurons, circuits, do this is is probably a kind of a fundamental driving desire.

Sandeep Robert Datta: Yeah, but I will tell you that, you know, one of things that Venki is really spectacular at, and one of the things that we're working at in our lab, is kind of trying to unveil as much as we can by looking at animal behavior. I'm sure Venki will tell you, he's built these incredible systems for tracking odor trails. We've started doing a thing where we put little hats on mice, and those hats have cameras on them that actually can videotape the mouse's face. And so we're asking whether or not we can learn something about what the mouse is smelling based on how it grimaces you know, in response to the smell, in the same way that we grimace when we smell a really horrible smell. And we're wondering whether mice do the same thing and whether we can use that as a kind of trick to approach this problem. I don't know, Venki, do you want to talk about odor tracking a little bit? I mean, your work is so interesting.

Venkatesh Murthy: Yeah, I'm happy to but I think let Lydia maybe ask us. I think it's a great thing that Bob’s saying. I think, in a funny way, I think this just also illuminates that I think there's sort of overarching questions we're all excited by and there's so many wonderful different ways to kind of attack this. It sort of illustrates the richness of what needs to be done and the poverty of what is actually being done, you know?

Lydialyle Gibson: So, I have to ask, do mice grimace? They must.

Sandeep Robert Datta: We're, we're working on it. We think so. We’re working on it though.

Venkatesh Murthy: Yeah, they have complex facial expressions. And I think this maybe, I think, people who have mice as pets have known this, but I think that neuroscientists are discovering this as well.

Lydialyle Gibson: Well, I do love the tracking experiment. Professor Murthy, can you tell us? Can you describe for us a little bit about how that works?

Venkatesh Murthy: Yeah, happy to. So I think the—I mean, it's a such a straightforward idea. Right? You see it in people, a lot of people in the US have dogs, so they, you know, take them to the park. They really just do this—any old dog will kind of sniff on the ground and kind of seem like they're following something. But of course, bloodhounds, which Bob mentioned, and animals, which are specialized for these kinds of tracking, they do this, how do they do that? So they know, some template, they say, “Okay, here's an odor that I need, a smell that I need to follow.” If it's a tracking dog, they're given a piece of clothing or something of a person. And then off they go. So there's remnants of this odor smell on the ground in some patchy way that we don't quite know. But we can sort of imagine them to be some long trail that's broken up in many ways, in many places. And the animals track that. So how do we study that? Because you know, that's a large area that’s sort of uncontrolled. So one trick that we borrowed from paper that was published by others, you know, a few years ago, is that if you put mice on a treadmill, right, they'll just run—they just like running on treadmill. But imagine on the treadmill, you're able to print, in a continuous way, odor trails, right? So you can take, let's say, an inkjet printer, which is what we did, and you just start putting on odor. So as the mouse is running on the treadmill, it's now encountering the odor trail, and you're mimicking kind of a long two-dimensional field, if you will, of trail tracking in a small arena, because it's a treadmill, it's kind of continuously moving. And mice do this spontaneously. We can reward them, we can encourage them to do this—do this meaning follow the trail—moreso by just giving periodic awards and chocolate milk or Ensure along the trails, so that they get more of the rewards if they follow the trail, rather than just wander off. So this gives us two advantages. One, it's a small, enclosed area, so you can video record the behavior of the mice continuously for long periods of time. Also, you can draw very long trails without running out of lab space. Because we're still in an arena. Like, it's not like a long corridor that you have to continuously kind of, you know, run after the mice chasing. So that's worked spectacularly well and so we have a lot of behavioral data on this. And the hope now is to look at the behavioral data, get some clues about like, how are they doing it? Like are they moving the head side to side across the trail? Fine, they can do that. But if they're off the trail, they seem to quickly get back to the trail—how? One hypothesis, which I don't think is true, is that they're smelling the trail from far. So they’re kind of basically triangulating going towards the trail. I know we don't know for sure, but that seems physically much harder, just to imagine. So it must be more that they have some memory of the shape of the trail. If they say, okay, look, I've been going along in this direction. So it's very likely the trail is, you know, pointing in that direction. And so I'm just giving you that as an example of just by watching the behavior, we can come up with ideas of how the behavior might be accomplished by an intelligent agent. But the beauty of mice is that we know a lot about their molecular biology, as Bob said, their circuits, the brain regions, we can now record neural activity and perturb specific brain areas to really ask how does the brain implement whatever the algorithm that we come up with, is being of course accomplished with the brain? And that will be the next stage of question. And so in a way, I'm excited by the questions that we can ask, rather than the answers that we already know. I mean, that will come. But it's just very exciting to just think of all these questions.

Lydialyle Gibson: Okay, my next question sort of tangentially relates to what you guys were talking about. But I wanted to ask about the intersection between artificial intelligence and olfactory research. AI technology plays a role in the work that both of your labs do. And Professor Murthy, I know you have a background in AI. So I was hoping you guys could talk a little bit about the ways you use AI and what you think are the possibilities for machine learning in this field. And will we see a robot nose that does something similar to computer vision?

Venkatesh Murthy: Yeah, I’ll start by saying, maybe a small correction, that I don't necessarily have a direct background in AI. I was an engineer who came into neuroscience because of the kind of the second boom in neural networks. This is back in the gosh, I'm dating myself, back in that late 1980s, early 1990s, was kind of the—what do you think—was the second wave of neural networks? And I think that that kind of got me, so I was interested more than in the background. But I would very briefly say that the two ways in which our lab, and I think Bob can weigh in. There are two ways. (20:12) One, you use the AI tools, yeah, tools, meaning all the modern machine learning and neural network tools, to essentially understand the data that our experiments are producing, whether it's behavioral data, neural data. So you think of it as a tool to appreciate the patterns in the data and uncover them. The other one is more that, can we actually take olfactory behaviors or behaviors inspired by olfaction to kind of think of new intelligent systems? Like is there something to be had just as visual neuroscience has helped a lot of this so-called deep neural networks, that are good for image recognition. So in some sense, it's more of an understanding of what it means to be intelligent and do intelligent behavior. So I would say we're doing a lot more of the former, which is use the tools to kind of parse the data and organize the data. So I think I'll stop there. And I think, you know, Bob has a couple of amazing stuff that they do with AI tools.

Sandeep Robert Datta: Yeah, just to pickup on Venki was saying, I totally agree, there's like, on the one hand, there's the availability of modern tools, and machine learning and artificial intelligence, which, as I'll tell you in a second, really are changing the way that we can analyze data about chemistry and neurobiology and behavior in a way that's really incredibly powerful. There's also this interesting dialogue, where things that we're learning about how the olfactory system works, might actually help us build better intelligent systems. So a mutual colleague of Venki and mine, named Chuck Stevens, was inspired by the architecture of an insect olfactory system and used that to build a new kind of network that lets you tell whether two patterns are similar or different. That’s a very simple example, but it gives you a sense of how the brain itself can inspire new ways of thinking about artificial intelligences. And new ways of building tools that can allow computers to identify patterns and data, right, which is essentially one of the main problems that your brain is solving. So, you know, in the lab, we've really been excited by and have tried to take advantage of, kind of the modern revolution both in computing power and in machine learning. You know, now Venki and I can basically just log into a terminal, and be kind of instantly connected to either supercomputers that live on the Harvard campus. Or if those aren't big enough, God forbid, we can even you know, pay 50 cents and gain access to supercomputing at Google or at Amazon. And that capacity, which really wasn't present even five years ago, allows us to generate enormous datasets that capture important features of what's going on in terms of the firing patterns in hundreds or thousands of neurons in a mouse's brain or, or what's going on in terms of the behavior of the animal. And to search and sift through those data to try to find underlying patterns that might help us understand basic features of how smells are processed. So one thing that my lab has done a lot of is we're really interested in how odors change an animal's behavior, right? So one way you might think about the olfactory system is as a kind of interface between the odors in the world and actions that you generate. So if you're going to study that problem, we need some way of really thinking about how the behaviors that a mouse generates, even when it's just in its own cage, how are those behaviors organized? How do those behaviors evolve? How can we identify what the animal is doing? And so my lab is taking advantage of artificial intelligence to try to identify the components and structure of mouse body language. And we're using those tools to ask how odors change a mouse's body language, in an attempt to take advantage of that information to infer something about how the brain processes information about smells. And so I think these tools are, you know, they're going to be they're pervasive and powerful, and are really poised to revolutionize neurobiology. And I'm also excited to see, as Venki was mentioning, you know, what we can learn in neurobiology that might help this AI revolution as well.

Venkatesh Murthy: Yeah, I think to pick up on the last point, I just want to maybe make that even more elaborately clear, if you will, that the way in which the neural circuits are organized, even on the front end of the olfactory system, at the face of it seem very different from the visual system. Where visual system, spatial position means a lot. Like when you look at the image, you know, parts that belong to the same object are close together in the same pixels, right? And so everything is grouped by kind of positions in the image. And the nervous system is organized based on those rules, in something called a topographic mapping—meaning things that are close by kind of maintain their closeness as they go through the brain. But the olfactory system, that's kind of broken up very early on, and there are lots of good ideas of why that's almost inevitable, that, you know, the chemical world is kind of almost arbitrary, and anything can come together in any fashion. So you’ve got to be ready for that—ready to put this thing together. But that gives potentially a fundamentally new way of thinking about algorithms wherein something called a random projection. I mean, there's mathematics behind this, that people, as Bob mentioned, Chuck Stevens. And there are also other more mathematical ways of thinking about it. So that kind of inspiration where we can use this architecture to solve problems, picking out similar things out in the world, or doing something else. Or searching for objects. If you have a whole collection of things, I want to say, look, I want to give an image that's most similar to the one that I have. Right? So instead of going and searching one by one, there may be other clever ways of using the architecture from the olfactory systems, that, ah! Quickly I pick something that's very similar. So those are examples of things that one might imagine, could be inspired by understanding the olfactory system in the other direction for the AI.

Lydialyle Gibson: Ok. But maybe a ways down the road, it sounds like, or maybe not?

Venkatesh Murthy: Yeah, I don’t know, Bob, what do you think? So it’s like these things sometimes can snap like, you know, you kind of putter along, putter along, you know, and there's an insight. And then you know, all of a sudden, somebody just clever made a discovery. And there it is.

Sandeep Robert Datta: Yeah, that's what's so exciting about doing this kind of work.

Lydialyle Gibson: So the last thing I would ask you, maybe kind of feeds into that, is, I want to ask you guys a couple of questions about COVID related questions. One thing, how did the pandemic change the terrain of olfaction research as a basic science pursuit? Like what's different for you all now than it was in February 2020.

Venkatesh Murthy: I don’t know, Bob? I think you have been much closer to the ground on this?

Sandeep Robert Datta: Yeah, and my sense is that this is an important moment, for olfactory science. We're in this kind of very unusual, perhaps unprecedented situation, where many, many millions of people across the globe are losing their sense of smell. And in some significant portion of cases, you know, maybe five to ten percent of cases of people who get infected with COVID, that smell loss is long lasting, longer than six months, in many cases longer than a year. There are definitely people who are infected early in the pandemic, who still haven't recovered their sense of smell. And so for all the reasons you mentioned at the beginning, which is, you know, it's clear that your sense of smell grounds you emotionally in the world, we think that's, in part, because the sense of smell is so ancient, and anatomically, your nose is connected to parts of your brain quite directly, that they're responsible for things like memory and emotion. So, you know, I think our nascent understanding of olfaction is that as humans, when we just walk around in the world, we're constantly smelling things, and that kind of emotionally centers us, and when you lose your sense of smell, that sense of emotional centering is lost. And so there's a huge and well-defined risk of becoming depressed or having other kinds of psychiatric disturbances that are associated with new-onset anosmia, the sudden loss of your sense of smell. And, you know, typically, that happens. under non-COVID circumstances with, you know, traumatically, in a motorcycle accident, you hit your nose, you lose your sense of smell. Sometimes, rarely, viruses can cause you to lose your sense of smell. But never before have many millions of people experienced this all at once, with many people having this kind of lingering loss of a sense of smell. And I think we're only now really beginning to appreciate how many people fall into that category and how devastating it's been for them. And the truth is that, as scientists, and for my friends who are clinicians, we don't really have a whole lot to offer people. We don't understand a lot about the basic science of smell. We don't really understand how odors are detected in the nose, how the cells that make up the olfactory epithelium in your nose, that senses odors really works. I think Venki and I both agree that we're both working super hard. But we don't understand really basic principles of how the olfactory system links to perception. And so we're in this kind of weird space, where suddenly, there's essentially an olfaction crisis, a global olfaction crisis, and we're racing to catch up. So I would say that as a field, you know, I think many of us I'm sure, this is true for Venki, and for me, are taking this as a kind of call to action, to really think about what we can do to help patients now. And we're plotting out a research course, that I think will give us a much deeper understanding of the system in the years to come so that when the next pandemic comes, we'll be better prepared.

Venkatesh Murthy: Just to add to that very eloquent, essentially, sort of a plea for like, the importance of this is that I think we should really make use of—that sounds like a terrible thing to say—but we're, we're presented with this difficult circumstance of this large number of these people. So there is basically a huge category of people that one can do interesting research on to understand what is going on, right, both long term and short term. So I think we really need to start planning so we really make the best of all the available subjects, if you will. And the other one is that this needs to say like okay, if there are people who are don't have this sense of smell long term, what does it mean? Like is there a way—can you imagine sticking deep-brain electrodes somewhere and stimulate some pattern and they will get some, you know, arbitrary smell images in their head, and that's good enough? Even if it were, we haven't the foggiest idea. Like, what patterns should we do? Where should we be doing this? Right? And I think that is what Bob is saying, like, it's a call to action, to even begin to imagine what might be done. Not that we're going to actually do it. But what does it mean to have a—what would you call it?—an olfactory prosthetic, I guess, to kind of stimulate. And so I think those two things—like one, making sure that all these people who have with that we really, you know, study them carefully and thoughtfully, and ethically, obviously. We were presented with this, and we really need to think, I guess, take advantage of it. If that doesn't sound too crass.

Sandeep Robert Datta: Yeah, I think just to pick up on something that Venki said right at the beginning of this conversation, you know, Venki was talking about the basic challenges we have in terms of studying the olfactory system. And he was pointing out that one of the challenges is that we don't really understand smells very well. So you know, we can digitize images, or I can send them over a wire, so you can reconstruct them somewhere else. We can digitize sounds, obviously, in this podcast, that's precisely what we're doing. We have no way of digitizing spells, there's no real smell-o-rama. I can't like send Venki some set of instructions over the computer will cause him to have a particular olfactory percept. So even though that seems like a kind of abstruse problem to solve, and like, you know, really rooted in the basic science, if we're ever going to help people who've lost their sense of smell, we have to understand how to do that if we're ever going to build an olfactory prosthetic, if we’re ever going to build really an electronic nose, right, we have to understand how all of that works. And we have a long way to go.

Lydialyle Gibson: So understanding smells themselves is fundamental to the whole project.

Venkatesh Murthy: Yeah, yeah, I think to add to that, you had asked earlier, we sort of didn't respond. And Bob and I responded about the robotic nose, I think. One of the things is that even if we imagine that we get to the point where—and that I'm that I'm very hopeful of—you know, some sense of the algorithms involved. Like I said, if I give this mixture, and once I know that pattern of activity in my nose, you could probably figure out how is that interpreted with the brain. Let's say that that's going to happen in the coming decade or so. But we still for the artificial nose, we have no idea how to take the chemical world and make that into a pattern of activity of anything. There's no camera for chemicals, right? So that's a chemistry, material science, engineering problem for making electronic nose. So I almost would put it as two different aspects. One, the actual sensor, how do you take essentially a camera right for smells? Reversible, quick, easy-to-use camera? And then there's the whole software like, how do you make sense of it? Do you have the patterns? I'm more optimistic about the second. I think the first one seems very daunting at the moment, to make a physical device.

Lydialyle Gibson: Ok, and the very last thing I wanted to ask you all, and you maybe answered this already, but sort of what avenues of research your own or just sort of in the field, are you sort of see the most promise in or feel the most excited about or hopeful of right now?

Sandeep Robert Datta: You know, I think olfaction is interesting, because it's so mysterious, right? It feels like a fundamental thing for most animals. It feels like something that's incredibly pleasurable for humans. And we simply don't understand it. And as kind of Venki was just saying, one of the exciting things as a scientist, you know, who studies olfaction, is the fact that understanding it requires us to think about science in all sorts of different ways. We have to think about engineering, right? And how you would engineer sensors. We could think about how nature might have done that engineering. We have to think about biology, to think about neurobiology, we have to think about behavior. And so it gives us an opportunity to think about actually brain function really holistically, from lots of different perspectives. I think that's one of the most interesting things about olfaction. And one of the reasons why I'm so excited to be working in the field.

Venkatesh Murthy: Yeah, completely agree. Maybe another way of sort of phrasing that is that it almost feels like we have access to, in experiment animals, (33:48) to the entire chain of things. So we have the some idea of like, what is it that something even though we don't know what the chemicals are, we kind of know, and as Bob said, you know, fox poop, or cat pee, or in mouse secretions. Like there's something—so you go from the thing that the animal’s interested in, we can watch the behavior, we can actually watch the neural activity in almost all the stages—from the nose from the early parts of the brain, to kind of the action part of it. So it feels like you have this entire chain from the sensation to the action, right? In a sense, that is mysterious. So anything we learn is exciting, but also in animals that really care about this, right? Mice and in other live animals that that we do. So I don't know if there's any one direction to sum up. It feels like all of it’s exciting, right? The entire process of understanding the behavior in the brain is wide open and exciting.

Sandeep Robert Datta: Yeah, and I'll say it's such it's not just exciting. I think actually, this is a moment where, Venki, I think you feel this way too, where it's kind of hopeful. Yeah, like we don't know, there’s a bunch of mystery, we don't know a bunch of answers. But honestly, the tools are great. And it just feels like we're gonna make progress pretty soon. And so it's an exciting space to be in.

Venkatesh Murthy: Yep. Excellent, excellent point. So I think you know, you don't want to just chase something that just seems hopeless, but it feels it feels tractable and hopeful now.

Lydialyle Gibson: Wonderful. Well, Professor Murthy, Professor Datta. Thanks very much for joining us.

Sandeep Robert Datta: Thank you.

Venkatesh Murthy: Thank you for having us.