Welcome to "AigoraCast", conversations with industry experts on how new technologies are transforming sensory and consumer science!
AigoraCast is available on Apple Podcasts, Stitcher, Google Podcasts, Spotify, PodCast Republic, Pandora, and Amazon Music. Remember to subscribe, and please leave a positive review if you like what you hear!
Dr. Tian Yu is a Computational Sensory Science Consultant at Aigora. Tian holds a Ph.D. in taste neuroscience, focusing on signal transduction. After a postdoctoral fellowship researching brain lipid metabolism, and an instructor role delving into sour taste sensing. Dr. Yu devoted herself to sensory and consumer science research to combine her love of neuroscience and expertise in analyzing data. Tian has published numerous peer-reviewed articles and one book chapter. She has presented her work at international conferences and received broad recognition for her work.
Transcript (Semi-automated, forgive typos!)
John: Tian, thank you very much for being on the show today.
Tian: Very happy to be here.
John: Yeah, it's great. It's really a pleasure and you're here, in part because many of our clients have requested that you'd be on the show. They wanted to learn more about you. And you really have been a blessing, I think, to our company. And I'm grateful for you every day joining our team. For the people who haven't had a chance to meet you yet. You can take us through your background and how it is that you ended up working at Aigora and being in sensory and consumer science.
Tian: After finishing PhD in taste and neuroscience and my postdoc, I started looking for roles that I can do in industry, actually. So it's kind of like accident that I know there's sensory science, it's my ignorance. I never know sensory science as this field was existing at the time. So one networking event, I met somebody and she was a sensory scientist and I talked to her and thought, "That could be a perfect role for me." So, I started looking into this field and I found your podcast, and that's how I started looking at and contact you.
John: Okay, and so what is your undergraduate degree? And maybe you talk about that if you go back, take us kind of through your educational background and you're always interested in taste, I assume, like at what point did you know you wanted to do something involving the senses?
Tian: Right. So my undergrad was Biology. In China, it was called biotechnology at the time, but it's more of like, biology, molecular biology that have more classical biology subjects. So at the time, I really fell in love in neuroscience. I was always in love with neuroscience. So when I was applying for graduate school, I was looking for neuroscience program and it just happened that the lab that I was in was doing taste neuroscience, and I found that one was fascinating. So I kept in that lab and I was doing taste research afterwards.
John: Was that at Utah State or where did you do PhD?
Tian: Yeah. That was in Utah State.
John: Right. Can you take us to that process? So you came to America then right after undergrad went to Utah State, and then you were studying taste neuroscience. So what were some of the experiments, what was some of the research you were involved in during your graduate career?
Tian: Right. So at that time, I was mostly doing fat research, fat taste research. So my lab, Dr. Tim Gilbertson, he is in University of Central Florida right now. He moved out of Utah State. But at the time, we were researching fat taste. So fat taste was not one of the five traditional tastes. More traditional we think it's a texture cue. But if you look at neuroscience aspect, there is a symmetry sensation part of fat taste. And there is a taste part of fat taste. It divided that part only because it's from different nerves. So we were looking at specifically how, like free fatty acid, which when you eat food, it will digest in your overall cavity into free fatty acids. And that will activate those pathways separately. So we're looking very specifically how that signal the pathways was.
John: I see and what are some of the key insights that you came across as you're connecting that research.
Tian: Right. So we were one of the first labs I found fatty acid can activate like single taste receptor cells and my research specifically can activate trigeminal neurons like free fatty acid can activate your neurons from the oral cavity. Before that, my mentor at the time were looking at molecular. We're all looking for the molecular receptors for fatty acids and we found that a group of G protein-coupled receptor could be served as long chain, mid chain, short chain fat assets like from food intake. So that was the major inside findings through that research. Yeah. They will activate both, like, taste pathways and some other things pathways.
John: Yeah. Listening to that, there are people who think this world is a simulation. There's a simulation hypothesis in the tech world, right? It's really bizarre how every group of people manages to think that God is just like them, whether it's hunters, if there's hunters, they have a hunter God. And if it's farmers, there's a farmer God. And if your computer programmer, there's a computer programmer God who programmed the simulation. But when I listened to you talk about that, I think there's no way this is a simulation because it's way too complicated. There's no reason that it should be other than evolution, that it should be, maybe that you could say, oh, well, the programmers program the initial condition and let evolution happen. I don't know. That's getting pretty far-fetched. It is very complicated. And I actually think the experience of life. I don't know how much you've thought about this, but I think the idea that there are just five senses, that there's five discrete senses is really not actually true. I think the brain is just getting a lot of information and then making sense of it. Do you thought much about that? What's your opinion? Having studied this a little more closely.
Tian: Right. So when it comes to senses, especially chemical sense, you'll be surprised how little that people actually know about sense of smell and taste specific, no matter sensation and often not giving enough credit to them. But without those senses, you cannot really perceive the world as what we are doing today. And all of the evolution, as what you said, the evolution of chase were coming to where we are right now.
John: And it's so important because especially with the tech world now pushing this idea of the Metaverse, right? That we're going to go into some alternate reality. Of course, sight and sound are fair, I mean you talk to sound experts and they'll tell you the sound is actually not easy. Sound is fairly complicated because there's all sorts of bouncing around of sound waves and that it's actually very hard to recreate a true Sonic experience in virtual reality. But even compared to sound, right? The chemical senses are fantastically complicated and trying to recreate something like our experience of life in a virtual environment, a full immersion to me, I think, is going to be very difficult. So what are your thoughts on that? Are you optimistic that that's going to happen in our lifetime or having studied it more closely? Do you feel that we're a long way away from recreating chemical experiences in a virtual reality?
Tian: That's a very good question. It's definitely difficult, but it's also very important so without that chemical senses playing a part of the metaverse, I don't think the metaverse is complete. Obviously, it's not complete. So how far was it away? In my life, I probably will see progress, but probably not get to the point that it's going to sound and vision and to get that far that we can replicate the chemical sense in another kind of space.
John: Yeah. I think what's more likely is we're going to go into an augmented reality where it's not full immersion. But you just have a world that's got extra information. Maybe there's a sense of ownership in this augmented reality as well, and we can talk about that more later in the show. But, yeah, the chemical senses, that's a big, big problem. And it may be that full immersion is not fully solved until you get direct activation of cortex. It may not be possible to reproduce the chemical experiences through chemical signaling. Anyway, this is important. Do you have anything else you want to say on the topic before we go on to your postdoc?
Tian: I think it's good. At your point to activate the cortex, that part I believe people were looking in that direction a lot. But because I got cereal cortex, the smell is kind of deep. So that part is a little delayed. Then the vision, you can just do it on the cortex.
John: Right. Yeah and another thing is that which cells to activate. You're going to have to map everybody's brain individually, right? If you want to know what to activate to create certain experiences. Anyway, these are hard problems. Let's move on to your postdoc. I mean, they're important problems as you said, you won't have the full experience the metaverse without the chemical senses. But let's talk now about your postdoc. So then you finished up at Utah State, and then what happened?
Tian: Right. And then I actually went for a short postdoc at UT Southwestern. We were in a very interesting lab, like doing system biology. So we're trying to use mathematic approach to solve some of the difficult biological problems. Like how the cell grew with a hat-like, neutral fill cells that grow head and tail like that part of the problem. So that was a year of me in Texas. But I started to realize that for my background, it's probably not easy at the time for me to go to that field. So I want to move back to more of something that I'm more familiar with which is like monocular biology, those type of things. So I moved back to Colorado, not back. But I moved to Colorado and did the second postdoc researching brain lipid. So how lipids roll in brain which is a fantastic problem as well. Like, brain is actually lots of different types of lipid. But people know very little about how lipids roll in brain. What it will affect brain function. Yeah. So that's my second postdoc.
John: I see, and that was at Colorado State or where did you do that?
Tian: That was in University of Colorado. And then I moved to an instructor role, back to sour taste and then I started to look at sour taste for some time. So that's a wrapped up of my life before joining Aigora.
John: I see. And so when you're doing that work, you really didn't know about sensory science? And then you went to a networking event and found out about sensory science. I'm curious, how was it described to you? So what have you met somebody? And they said, what do you do? And they said, I'm a sensory scientist? How did that happen?
Tian: At those events, I explained them I was doing taste research and things like that and that person just happened to be, she was a sensory scientist. So she was, like surprised, "Oh, you're researching like taste neuroscience. That's so interesting." And she was like describing what she was doing as a sensory scientist and then I was like, okay, that was the thing. Because in academia, we really don't talk to each other. I guess that's totally my ignorance. We didn't talk to food science people that much especially when you are in medical campus maybe.
John: Right. Interesting.
Tian: I never really know there's a sensory science, that's you know, sensory and consumer science that is a field that neuroscience people can definitely play a part there and go into future probably play an important role.
John: Very important, yeah. That's great. Okay, so then you started listening to AigoraCast and do you remember what were some of the episodes that you were most interesting? What kind of got you with any particular episodes early on that kind of got your interest and got you hooked on the field and what you might be able to contribute?
Tian: Yeah. I was listening pretty much the early episodes. I listened to a lot. The one with Alex, actually, I remember that one.
John: Oh, Alex Pierce-Feldmeyer.
Tian: Yeah. I remember the one with your dad.
John: Oh, yeah, of course.
Tian: Yeah, that one and that one with Danielle. The earlier ones, I was trying to learn more about what exactly is sensory science doing. I'm trying to figure out how I could play a role there. Use my knowledge, that's why I always think, like, the neuroscience will have somewhere to combine with sensory science. These two fields together will help the whole like the metaverse, you know chemical metaverse to move forward. So one of the sensory science or sensory neuroscience itself won't really move it that far. So I was always trying to find a gap to bring these two together. That was my initial intention of moving to sensory science world.
John: Right. But you weren't coding at that point, right? Do you remember when you first started to write computer code?
Tian: Right. So I have to say, I was always a little bit interested in it. So back maybe five years ago, I started like Coursera course on Python. I just followed the whole course of Python. So I know a little bit of how code work, but starting was fully your influence. So you were, and some of the AigoraCast and...
John: AigoraPlus
Tian: AigoraPlus, yeah, in that event, you were saying the sensory science need like doing R code. So that's why I look into it. So I did another Coursera course on R before I start seriously on the data science book. That's how I started.
John: And I have to say, Tian it has been amazing how you went from really no experience in R to I would say, very high level of professional competence in R and maybe a year and a half, I think that experience would be really valuable for the listeners to hear because the question I get all the time. I was invited to present at UC Davis office hours for Becky Bleibaum, I did that earlier this week. One of the questions I was asked is, do you need to learn to code? And the thing about that question is, while you technically do not, I don't think in order to make contributions, you definitely need to make sure everything you're doing is related to coding I think, to have any real future right now. However, it really is better if you can learn to code. And I think that it's much more accessible than most people realize. So can you take us through your experience? What did you find as you were studying R and as you were starting to apply it first and you did an internship with us, and then you started to work on our client projects. Can you take us through that process, please?
Tian: Right. So I went through that R for Data Science book, and I started notice that, yes so when you are doing R with data, your views of data are a little bit different. Before I was looking at data most of the time, when I was doing my previous research, I was trying to find the significance. I was trying to find the different groups. Is there any significance on what I'm measuring? That is like, the important things.
John: Interesting. Okay.
Tian: Yeah. That's what our findings are usually tied to. But when I was going for R data science, I figured there's a lot of data visualization. There's a lot of ways I can look at data. I think that is like a revolutionary for me. It's like not to just find the significance, but to know how that data is arranged and what insights you can get from data because data tell you story. It's your job to find what the story is. It may not be just like a point of five significance level, but there will be a story about it. So it kind of changed my view and then using R, I think it just kind of connect me closer to the data because I really is the driver of how I want the data to be kind of not want the data to be, but how to look at data. I can do like magic things about it and I see what the analysis and why the analysis is there. So about the distance, those type of things previously, it was just like one button, one test to do when I was using elsewhere, and then the figure will be there. The significance will be there but now I understand it a little deeper. I know why I won't do that and why this test is appropriate. The coding itself, it's definitely accessible but you need a little persistence. That's my only advice I guess if somebody wants to learn R. It will work if you do it a little bit every day. That's how I did it. I try not to stop, even though I'm busy, retired or whatever, but I try to do a little bit every day. So I opened R studio. I tried to code a little bit. After I started, I grow obviously, I have to code a lot. That helped at times to speed up my learning process.
John: Yeah and then what about finding a mentor? Because I know I mentored you quite a lot and I think that you've had mentors internally, for example, Kuba and Arkadi, you work with other team members, So how important has that been for you like having a mentor?
Tian: Yeah. That is very important, at least for my learning experience. At first, of course, there are ten different ways to solve one problem in coding like Kuba was telling me.
John: We should mention, Kuba is our head of AI at Aigora.
Tian: Exactly, he is supreme in R. He was telling me that I try to remember that every time I code is that if you encountered a problem, try to solve it the best way and not to move it too fast just to solve the problem. You try to make it right. Doing like the right, even though in the beginning will take a long time because we have errors. You don't really know how it works. But the learning curve is like you go slow slow slow and then you have a jump. So after that jump, it's like all of the things start to make sense.
John: Yeah.
Tian: So yeah, I felt that jump only like this year, to be honest. So even at the beginning of the year, I still suffer from making some functions and the mutate, like the multiple columns, all those things. But I forgot when, maybe some earlier this year, I felt that jump happens when I write code start to work. Like, I don't need to debug a lot to find out.
John: It's a lot like snowboarding. I don't know if you're a snowboarder, actually, but..
Tian: I'm a skier.
John: Yeah, skiing is different. With skiing, it's fairly easy to get to a novice level, make it down intermediate or whatever, even make it down black. But to become a truly advanced skier is a tremendous amount of work. Whereas snowboarding, you pay an enormous price. The first three or four days, you're slamming your chest into the snow every 30 seconds, and you are in incredible pain. But after about three days, it gets to be fairly easy. And then you have that jump, and it's a different learning curve. And so I think that awareness that when you're struggling at the beginning, that is normal and you just have to hang in there. It will get better, especially if a mentor who can help go through your code with you teach you some patterns. It can be done. It's not something that is just for the people in Silicon Valley. I think it is accessible to anybody with a scientific background.
Tian: Definitely. So with a little help from the mentors that tell you initially which part you can improve what is the better way of coding it? This is like a positive feedback loop. Right? So after the loop going on for a bit, you will be learning a lot. But initially, yes, you do need to hang in there like persistent, like doing it day by day. After a certain amount of time, you will feel that snowboard jump.
John: You won't be falling every 30 seconds. So now what about for people who don't have a mentor, what are some other resources that you found to be helpful if you get stuck on a problem, maybe people aren't available or actually, it's a good idea to try to solve it yourself first. What are some of the problem troubleshooting techniques that you rely on?
Tian: Right. So initially, I think R for data science is indeed a very good resource. So a lot of the problems I had the web page part of the book. That one is at my bookmark, so I would check. I would search the problem that I was looking for, and then I just read that little part and then try to apply that part to whatever I'm writing. The other thing, obviously, just Google it. Google the problem. You have to Google the right problem. That is like a skill that you will slowly get. It's like Kuba needs to check to Google as well. But he can find the right word combination and lead him to a good solution. So it's hard to say, but what is that stock flow?
John: Stack overflow.
Tian: Yeah, stack overflow. People have questions there. I don't post questions, but I will look at people's solution. Nine out of ten times, you will find a solution in there. You just need to change your keyword for searching a little bit if you don't find the right one.
John: Yeah, exactly and one thing I would add to what you said about R for data science is that for me, I've found that typing every line of code in a book. If I'm working through a book that's got coding in it, the best thing I can do is make an RStudio project for that book and have one script per chapter and type every line of code. Would you agree with that? Have you found that approach to be useful? Have you applied that in your own learning?
Tian: That is definitely very helpful, especially when you're starting from zero to get you comfortable. The other thing that is very helpful is like the sensory book. I find that teach you a lot of sensory science per se, but it's actually a lot of help if I can rewrite the code because the code is a little bit old. But the idea is if I can try to use my own way to rewrite that code that helps a ton.
John: Right. For everybody, that's Sensory Analysis in R, I think it's called with, it's by Sebastien Le and Thierry Worch, two friends of ours. I don't know if you know Sebastian, but you know Thierry. I mean, if you want to learn sensory analysis, that book is great. Okay, that's great. Tian, we are almost out of time. Okay, so let's talk about what you're most excited about. What are the projects you like working on Aigora, obviously can't mention clients, but just the sorts of when you're working on things, what is most interesting to you? What are you most excited about? What do you look forward to in terms of your job?
Tian: Well, of course, a lot of times I'm interested in the specific questions that our clients are solving. That itself to me as a taste scientist, I'm interested in that. And then I found our machine learning tools can help people find out the insights into it which they often not knowing or ignore or just not paying attention to by putting that data itself. So I find that part to be very intriguing. So that is very exciting. Of course, we do the automated processes. It's very nice to see how people are satisfied with. It's very satisfactory for me, I guess, to feel the happiness when those things can be automated for them, so they don't have to do it every day.
John: Yes. If you spent hundreds of hours like I have at times in my life doing work more or less by hand and then it gets automated. It's magical. It's truly magical to be able to push a button and get a report. It's crazy. Yeah. So it is very satisfying. I think the longer you've been in the field, the more you've suffered, the more relief you feel when something gets automated. So I definitely relate to that. Okay, I didn't mean to interrupt you, so please continue.
Tian: I think that's pretty much.
John: What about smart speaker surveys, are you enjoying working on it?
Tian: Oh, yeah. Smart speaker surveys. Yeah, it's new. We were gaining experience on that, but I see people. How do I say it? I don't want to say hate, but sometimes it doesn't work smoothly, but people really love it. Like as a new handsfree device to collecting data. But we have to kind of be patient and let the technology improve to catch up with what we want it to be and yeah, just hang in there. Eventually, I think that's the future will play a role in our metaverse.
John: Yeah, voice-activated technology is a huge thing. I mean, that's definitely coming. As we've seen, it opens a lot of doors with new ways to collect data. So it's very exciting and then, of course, we're getting into NFT's and Blockchain so that's fun. You're learning about that. I think hopefully it'll be great for Aigora to make some contributions to chemical senses in the metaverse so I look forward to seeing what you're going to do there. Okay, so before we wrap up, how can people get in touch with you and what advice do you have for a young scientist?
Tian: Well, people can get in touch with me on LinkedIn. So that will be on the AigoraCast description, I guess. For the advice for young scientists, I guess don't limit yourself and don't set the limit for yourself and there's a lot of potential you can do R for sensory scientists, you can do R for computer science people. You can apply your skills to a lot of the areas, not just writing some software so there are a lot of things that we can do.
John: Yeah, that's great. Excellent advise. So Tian, this has been great and actually learned some things even I talked to you every day. It's nice to talk to you in a different format and learn some new things. Anything else you want to say?
Tian: That's it. I enjoy talking to you.
John: It's funny, you started off to listening to AigoraCast and then here you are on the show so thank you very much.
Tian: Thank you.
John: Okay, that's it. I hope you enjoyed this conversation. If you did, please help us grow our audience by telling your friend about AigoraCast and leaving us a positive review on iTunes. Thanks.
That's it for now. If you'd like to receive email updates from Aigora, including weekly video recaps of our blog activity, click on the button below to join our email list. Thanks for stopping by!
コメント