Helene Hopfer - Using New Tools
Welcome to "AigoraCast", conversations with industry experts on how new technologies are transforming sensory and consumer science!
Dr. Helene Hopfer is the Rasmussen Career Development Professor and Assistant Professor of Food Science at Penn State where her research group studies product- and human-inherent factors that affect perception. Her research focuses on the human-food interface, and how food composition and food structure affect human perception and food choice, particularly, aroma and flavor. Combining human sensory and consumer science with analytical chemistry, materials science, and multivariate statistics, her recent and ongoing projects fall into three broad areas, (1) Sensory Mixture Effects for Improving Foods, (2) Linking Human Perception to Plant Genetics, and (3) Differences in Food Quality Perception.
Dr. Hopfer has degrees in polymer science & engineering, and chemistry, and studied under Hildegarde Heymann at UC Davis as a postdoc. She is also the program chair of Penn State’s graduate certificate in Sensory & Consumer Science, a fully online 12-credit degree program developed for working individuals to address the shortage of skilled sensory & consumer science professionals.
Transcript (Semi-automated, forgive typos!)
John: Helene, thank you very much for being on the show today.
Helene: Thank you, John, for having me.
John: This is great. I really enjoyed talking to you. Actually, you are one of those people I want to talk to, you give me energy and so I feel like it's a real pleasure to talk to you. So I think it would be good to start here if you talk about your path into sensory because I think you have a kind of a unique technical background that I think gives you a very strong base from which to operate. And I think it would be good to hear how did you end up in sensory science?
Helene: Sure. So, as you said, I have degrees in polymer science and engineering. And after I finished my Masters, I applied for a PhD position that looked at older active compounds and plastic polymeric packaging materials, and I thought that was really cool because I've never thought about the smell of plastics up to that point, despite me spending my undergrad and my master's thinking about polymer science and engineering. So that was the first time I got into contact with sensory science and also flavor chemistry and like combining my polymer science background and now bringing that human into the play and into the game. And I thought it was really, in retrospect is like I've never heard about human perception during and in the end, a lot of plastic materials are made for the consumer. So I'm like, that makes no sense. So fast forward, at the end of my PhD, I was looking around what I could do next and I was lucky enough to apply to a position to postdoc with Hildegarde Heymann and to this day I remember my 15 minutes with Hildegarde on Skype because, after 15 minutes, she just hang up on me. And I thought we had a brilliant and wonderful conversation. And now that I know Hildegarde so much better. I know that after 15 minutes she thought, you know, that's it. I'm going to offer the position to her. And so I ended up going to work with Hildegarde mostly on wine and on my polymer science friends were like, how do you do this? How can you move from polymers to wine? It comes down to using the same tools. You are trying to figure things out and you're building on what you know, and then you're learning new things and you're trying to incorporate those. And I think that's why what I appreciate so much about sensory science and consumer science is this cross-pollination, cross-learning, interdisciplinary aspect. So you have, you know, chemists and you have food scientists and you have psychologists and you have statisticians and they all come together to understand human perception. And I think that's the fascinating part and that's why I'm sticking around, at least for now, in the sensory and consumer science area.
John: Okay, well, I hope you stick around because I think you definitely are having a huge positive influence on the field. I do want to get into a little bit of your teaching because I think you do a great job preparing your students for a kind of life in the real world as a sensory scientist. It is interesting what you just said about using the same tools because when we think about computational thinking, one of the things one of the aspects of that is an abstraction, the ability to recognize when a problem is similar to structure to another problem which means now a tool that you would use to solve one problem can be used to solve the other problem. Can we talk a little bit about your experience working in so many fields and what advice do you have for people who are trying to incorporate ideas from other fields or they're trying to learn from other fields? Like what is your advice for taking a multidisciplinary approach to sensory science?
Helene: So I can only speak to what I'm trying to do. I usually I'm a very curious person, and so I usually get into new fields by talking to people or by reading something somewhere or by hearing people giving a presentation at the various conferences or going out for a beer with a colleague that is studying chemical ecology and like wait for a second, you're basically doing what I doing. You just using a different, using insects instead of humans. But you're still interested in getting how a certain chemical is triggering a certain behavior or reaction.
Helene: I think curiosity and just like willingness to listen and to learn, I think is really important. The one thing that I've learned since I use a lot of multivariate statistics and I'm teaching the sensometric course and a lot of the stuff that at the very beginning when I started getting into sensometrics drove me really nuts is that in different fields, the same tool, the same method has different names, right? So like what we call I think it's multidimensional scaling. People in microbiology might call Principal Kwatinetz analysis and unless you really understand what the data looks like that goes into the process and what the method does, it just looks like it's a completely you know, it's like a different name. So different names, different methods. The idea is that there's this family of like fact analysis and there's so many tools that fit underneath and they're similar in the way how you can interpret them and what kind of conclusions you can make and how you apply them. It's just understanding that the data structure might look differently, right? You might not have continuous variables. You might have discrete variables, whatever, but basically doing the same thing. And I think in my opinion, as a sensory scientist, you don't really need to do the matrix algebra yourself. Right? That's what computers are for. But getting to that point where you like, look at something, and even if the name that it is introduced isn't the name that, you know, at least you can recognize, oh, wait a second, that looks very similar to what I've learned before. I think that is another big aspect of building upon what you have and then trying to make connections. I don't know if that makes sense.
John: It makes a lot of sense because for one thing it reminds me of the skill and it is a skill of learning to ask your questions to Google in such a way that you can get answers, right? But when you're writing a computer program that over time what you evolve is like, okay, suppose I wanted to manipulate some matrix. I'm going to take it from this form to that form or whatever, and I'm going to ask some questions and I might say how do I rearrange the matrix so that the factor levels or whatever? And I can ask the question so that Google will give me the answer. But I think that that is a kind of meta-level skill to understand what are the structures and how do people talk about these structures so you can even find the right tool for your problem.
Helene: Yeah. In sensometrics when I started teaching it, I thought, you know, I Google all my stuff. I am a very bad programmer from the start with I think I might have written two or three functions and they're not even worth talking about, but I'm really good at finding the stuff and modifying it. So it works for my needs and luckily, we don't deal with big data sets where the time to process is a big issue. Even if it's slow, it will get done. But this idea that just Googling it, you have to kind of understand what you're Googling for. And how to phrase it in a way that other people have phrased it. I'm still struggling with that. How do I communicate that? How do I provide tips and tricks? How to get people to Google the right things, I mean, there are really good resources out there. So many. We are using are in my class. Everything is open-source. But finding the stuff and then finding it in a way that you understand what's going on and then you can modify it, I think that's the next step. But as I see students progress through the course from the beginning of the semester to the end of the semester, they're definitely getting towards that. Oh, this is how I find it. And sometimes they're excited and then come back and they're like, I found it and then I will say, okay, great. You are just a geek as I am getting super happy when you find the one right answer or the answer that you getting further.
John: Right. Well, maybe. Let's talk a little bit about that journey. First off, that you've been personally learning to code and then how you take your students on that journey, because I think we're in agreement that really sensory scientists should have some basic level of coding that if you aren't able to code, it's kind of like when reading got invented, if you didn't learn how to read, you are at a disadvantage. Now, you've got coding, you have no coding stuff. So can we talk about your journey and then also this journey you take your students on?
Helene: Yeah. So my sensometrics and multivariate data analysis journey started with me taking Hildegarde's sensometrics class. So my sensometric course is largely based on Hildegarde's course offering. With a couple of differences. When I started taking the course with Hildegarde, she's like, I want you to learn R. I don't want to use SAS. I know SAS. I want you to learn R, and you know, she was my boss, she was paying my salary. So I was like okay. At that point, I had some basic programming. In my polymer science engineering degree, I learned Java. Nobody uses Java anymore. But at least it got me a little bit into this mindset of how do you speak computer? And I do think it's a language. It's a language like any other language that you have to kind of understand the syntax and how to express yourself and how to put it together so it runs. So it's very similar to any other language. I do admit, R is not the easiest language, but once you kind of get into the groove and it's getting so much easier with the whole tidy verse and all of those things. It makes it really easier for people to use once you get over that initial hump. And then it was just like learning by doing and I think that is really important so that's a lot of the stuff we do in sensometrics is here is a sensory data set. I'll show you how you can analyze it with this particular method. I give you some example code. That doesn't mean that's the only way, but at least gets you started so you can modify it and then I give you a new data set and you apply to modify it, analyze it, interpret it because I think that's really important. It doesn't really matter what kind of statistical program you use. You can get a lot of stuff out and it might not be correctly applied. So I'm really trying to emphasize, what are you trying to understand? What is the question? What is an appropriate method? Now you get some numbers out, what does that number mean and how can you turn back to what you initially wanted to solve or figure out? I mean, I've used Excel stats and all of these other ones, and they give you a lot of stuff and it might not be appropriate. At least, R in my opinion forces you to know a little bit more. You have to ask for what you want to get back out. And I think that helps in this kind of process of making sure that what you do is really informing what you wanted to do? I don't know if that's really true, but I feel that it forces you a little bit more to think about a little bit more about the process and yes, there's a lot of black box in between, but at least you can like okay and I run a PCA, I do need to understand, do I need to scale to unit variance or not? What is the difference? And I encourage my students to really play around. To kind of see what kind of an effect does it have and what kind of different conclusions you may take from that result. Because I do think there is no ultimate truth, right in the end. And I think that's often frustrating to the students as they go through. There is no truth. It really all depends on what you ask and how you interpret it.
John: Exactly. Yeah. I mean, PCA map, right? And you're going to interpret it. It matters how it was created. But if you scale or you don't scale, they both are the data and visualizations, they just described the data visually. And so then the question is what does it mean? And that the arguments, the choices that are made for everyone. What are the things that you help me to understand in one of our previous conversations that I think was really insightful? Is this idea that coding is closer to doing the work by hand that when you are in, say, GUI- based program and you're just pressing buttons, there's an element of automaticity to that that doesn't really involve thinking. Whereas in the code, even when you do call a function and that function is a bit of a black box, there's still a feeling of doing something more like on paper. So can you talk a little bit about that to our audience? I thought that, yeah, some of the comments you made about that were really insightful.
Helene: Yeah. What I generally think is that coding, in contrast to a GUI interface is really forcing you to step through the things at a time that you realize what are the different steps that you have to do to get to the result. And I think once you step through this, you might also be more aware of where things could go wrong and what influence each of those steps might have. Right? To be honest, in my experience, most of the stuff that doesn't work is when it comes to R coding is getting the data in the right format into R studio, whatever you use, and then have it in the correct format before you do. Because I mean, a lot of like a senso-minor package in sense R all of these sensory-specific packages, they're really easy to use, but they do require you to have your data already properly set up and that might be a little tricky, but I also know that myself I've so many times, and I'm getting better, but so many times just shows that I'm a little bit learning resistance here. I look at my data, do some basic checks, a method, and then suddenly, I don't know, look at the degrees of freedom table and I have nine treatments and suddenly my degrees of freedom is one for the treatment and I'm like, something's wrong here. And then I forgot to set the thing to factor or whatever. And those things are now quite naturally coming to me. But that's also something that the students kind of need to learn and so many times they sent me little snippets and then the error message might not even make sense. It doesn't come until like or there's no error, right? Because it's still running.
John: That's worst.
Helene: That's even worse and communicating all this is like these peculiar aspects of that thing. And again, I have realized that not every student that takes the sensometrics course will go out and analyze the data for the rest of their lives. Some of them really like it and they will they run with it. And some of them really you can see that they want to learn. They want to get through that course. They struggle and at the end, they are like, okay, I've done this. I know how theoretically could do it. I may choose to do or not to but at least I think it still moves them forward because at least when they have a statistician in their company and so many big science companies now have statisticians that analyze sensory data for the people, but at least they can now talk to the statistician. And I had some of my working professional students told me, like, I'm not a new best friend of our statistician because I told him I'm taking this course and then I told him what we're doing. And he was like, that's just awesome if you ever need our help, let me know. We can work on this together. And I was like, this is great. And so, you know, I do think they may not use it themselves, but at least they are aware of where the challenges might be and what decisions to make, and they are now starting to talk a language that a statistician might easier understand. And I do think that. When it comes to statistics, this connecting it back to the research question and a statistician is great to have, but that statistician also needs to work with the person that is super close to that project to make this without that context, it's really hard to do any statistics I think.
John: Right. Yeah, I totally agree and I would also just to build on what you said, say that knowing which problems are going to be easy and which one is really hard for a computer or for a programmer to work on, is really important, right? Because a little bit of programing teaches you that looping things is easy. Right? And so you think, okay, great. Well, here is my little code. Maybe a programmer can loop it for me and the problem is resolved. Whereas other problems, you know, I think it's funny, but I think this is more of a paradox that things that are easy for humans are often hard for computers and things that are easy for computers are hard for humans. That a lot of times we think, oh, this is easy, but the task is not well described and we're not well defined. So getting a computer to do it is almost impossible. Even if a human activities, I mean you're all networked help with this when there's a lot of data, but we don't usually have that much data.
John: Okay, well I have a lot of questions still for you. So I want to make sure we get to some of your research topics. So what are some of the research initiatives that you're working on right now that you're most excited about and then kind of building that next few years? What do you see yourself working on?
Helene: Yeah, so we do a couple of really cool things. Right now with just finished up a follow-up study on some of the sweetener, bitterness, antagonism through cross-modal and taste test interactions and how we could actually use that to reduce added sugar content. We specifically started with fluid milk because, it's the food, but it's still fluid so we don't have to deal with melting and all of that stuff and translated it from two ingredients like sugar and vanilla to three ingredients, cocoa powder, sugar and vanilla, and trying to like basically get to chocolate milk that is has less added sugar and still maintain perceived sweetness and liking, although liking is a more robust measurement than, you know, people definitely see that the sweetness goes down, but they still like it and like really trying to test and to identify these specific ingredient interactions that lead to an enhancement of sweetness and a suppression of bitterness. I mean, yes, you can just do it benchtop trial. But, you know, we want to understand the kind of the more structured and mechanistic point of view. The other thing we're doing is we're looking at a novel starch-based encapsulation system and one of the applications is in chewing gum, so doing these temporal methods, trying to understand how when you chew on chewing gum, how does your saliva, amylase degrade these encapsulation systems and freeze the encapsulated flavor and does that actually lead to continuous and prolonged flavor delivery and also taking into consideration that there's huge variability with regards to saliva, human saliva, both composition and volume, and amylase activity, protein content, and how all of these factors also affect the partitioning of the volatile like we encapsulate menthol and so suddenly decides that, okay, so you might get it out of your starch cage, but then it's dissolved in the saliva. And then if you have a lot of saliva or if you have a lot of protein that binds to that then it's still not partitioning into the nasal cavity and it still does give you the menthol smell impression. So it started off as a very material science-heavy project, but it's now bringing in this human component, trying to really understand factors that affect that. And we're doing a lot of chocolate and cocoa. So both understanding how plant genetics affect cocoa and chocolate composition, both from flavor compounds, but also from a fat cocoa butter point of view. Because if you think about cocoa bean is 50 percent fat, dark chocolate like 80 percent cocoa is then 40 percent cocoa butter. And suddenly you are definitely affecting release. Volatile release. Right? So you might have certain volatiles present in the cocoa, but they might again get delayed in their release out of this fat matrix and then this is melting behavior, the whole textural component is very important as well. So we're working with a couple of the Penn State plant science, plant biologists on that aspect. We're also looking into, again, you know, foods are complex mixtures and so how can a chemically complex mixture induce a unique sensory perception and how, you know, simple mixtures might get very sincerely complex impressions, and then that's a lot of the stuff that I'm thinking about, like, you know, strategically and more broadly. This connection between composition and perception, then that lots of different chemicals present in the compounds in your food does not necessarily mean it's sensorially complex and the other way around. So, yeah, that's it in a nutshell.
John: Yeah, now, it's fascinating. So you're working on, I think, very interesting fundamental questions in sensory and food science. You know you got these questions from material science. I mean, for example, the starch encapsulation. That project sounds fascinating. I mean, and then there's a human component. I mean, at some point we're going to get, I think, some sort of personalized food to an extent people kind of self-select the products that work for them. But at some point, we may be able to make measurements of people and suggest, hey, or even the same product could potentially be personalized for based on things that are known. So, yeah, this is very exciting. I would say if I was starting my career, I would love to be in your life. I think it would be a great place to be working on these products.
Helene: Penn State is a great place to work at. Especially the Food Science Department, it's a very collaborative environment and department and so all of our students, I think it's a benefit, but they're at least officially advised by another faculty member. And, you know, with John Hayes and other colleagues of mine, it's really fun because you don't have to fight for yourself. You can always find someone again to, like, cross-pollination and interdisciplinary talking. Like, how do you connect sensory perception to health and nutrition benefits, right? This idea is that you can have your cake and eat it, too. I think there is a common ground and where you can have delicious chocolate. And that's also good for you. I mean, a colleague of mine, Josh Lambert, is a nutritional chemist and we do a lot of work on chocolate. And he found based on lots of epidemiological studies that any chocolate is good for you. So I just start eating chocolate. It's good for you. Significant decrease in cardiovascular disease death. Yeah, just eat chocolate. And so I do think that I see this as a big area within like you said, personal nutrition and healthy foods, we don't have to force people to eat healthy foods. We can make foods healthy and delicious. That's where I see food science going, especially with cell-based agriculture and cultivated meats and suddenly stuff that. You know, we thought we had figured out. Now you need to, again, bring it back to the fundamentals, so what does make a meat a meat a piece of steak? What structure is necessary or components are necessary? How do you need to arrange them? I think it's fascinating to be in food science right now with all the new technologies. And then, in the end, you know, you may be a biotechnology PhD from Stanford, but you have not so much idea about food science. So that's where food science really brings this interdisciplinary aspect and sensory sciences just the same.
John: Now, I totally agree, and I look forward to collaborating also more with the kind of sight and sound research that is happening in the UX space, because I think we're going to start to live in a more and more kind of augmented reality and the headphones, keep getting smaller and smaller, we're going to have sounds interacting with tastes. I mean, it's a very exciting time to be in sensory and I think you're doing a great job helping students get ready. So last thing, how can people get in touch with you if they want to apply to either and maybe join your lab or to join this program?
Helene: So I have a LinkedIn profile. I can also be easily found by Googling my name and Penn State. That usually works. I'm in the Penn State Food Science Department. So that's also a good way to contact me. And then, yeah, I'm in LinkedIn. I'm on Twitter as well, so you can tweet me.
John: And the last bits, I would like to conclude with a bit of advice, so what advice would you have for the young sensory scientist?
Helene: Continue to be curious and listening and just be open. I think there's so many connections to sensory science that we may not be even aware of yet, but looking into other fields and just be open and curious and talk to people. I think it's always interesting when I say that I'm a sensory scientist people always share some weird stories with me. And I think it's fascinating because it often gets me thinking, oh, interested. Why is that?
John: Interesting. Okay, that sounds great. Well, Helene, this is really a pleasure. Thank you so much for being on the show.
Helene: You're welcome, John. Thanks for having me.
John Ennis: Okay, that's it. 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!