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. Curtis Luckett is an assistant professor of food science at the University of Tennessee. He got his M.S. and Ph.D. in food science from the University of Arkansas. Since joining the University of Tennessee, Curtis has built a research program in which food texture is a cornerstone. However, Curtis’ lab has many evolving research interests and does additional work on sensory methodology, cross-modal correspondences, and olfaction.
Transcript (Semi-automated, forgive typos!)
John: So, Curtis, welcome to the show.
Curtis: Thanks for having me.
John: Yeah. Thank you for coming. So, Curtis, we've talked before the call about texture, which is not something I know a lot about. So hopefully I'll just indulge me on this call and educate me as I ask you a lot of questions that I would firstly just like to know the answer to. So I have heard that texture is very difficult to study, and it's also extremely important when it comes to understanding the experience that people have when they're reading. So could you maybe start by just explaining to our listeners, why is it so hard to study texture and why is it so important?
Curtis: Yeah, I mean, I guess that's really important because it's really integrated into liking. So we just wrapping up a large survey finding about 94% of Americans have a food that they avoid due to its texture. Which is right up there with flavor. As far as almost every resident or every citizen, I should say, the countries is avoiding a food or a subset of foods because of its texture. So it's definitely a big driver when it comes to food choice and to properly purchase habits and those types of things. And then to kind of figure out why. That's another tough thing, too, right? Texture is very context dependent. So like a texture attribute that's good in one setting can be bad and another depending on culinary use or what type of food product you talking about. And foods contain several different textures. So even a simple liquid has more than viscosity and which makes it often hard to track down what changes in a certain texture actually do for liking because it's very hard to change a single texture without changing other texture attributes. You know trying to nail down the exact effect of liking or something like that, it's tough. You also have this idea that texture is not hierarchical, necessarily like flavor. So if you have the flavor of orange, right? It's also in a more broad sense, a citrus flavor. More broad it can be fruity. Textures has always follow those same guidelines. So if you're trying to make like a texture wheel or some type of lexicon, often have a hard time classifying textures that are either like or dislike each other. It's also just like a lot of other things, cultural differences, you know, so different cultures, different cuisines from those cultures have different texture attributes. And those can lead to, I guess, differences in expectation when it comes to different food products.
John: Okay, so I've got like a million questions to ask as a follow up on that. Alright. So one thing I was kind of curious about right off the bat is that so texture is very multi-dimensional then? If you're into PCE or something, you find many different kind of independent directions in the texture space. What does that typically look like?
Curtis: Well, there's one thing I haven't touched on yet, I guess, how this picture. So, yes, it's very multidimensional in that sense. And it also has a temporal aspect. So, like, if you take something like a baguette, right? It has a different texture and your first bite than it would on that last bite of the bullet before you swallow or before you prepare to swallow. And so I think it's important to kind of if you're looking at the multidimensionality of texture, it does change quite a bit across the eating process. And so adding its temporal aspect to that dynamic change in the texture is always important. You're trying to capture that multidimensionality.
John: Yeah, that's very interesting. Okay. Well, I want to learn texture data, so maybe you can talk because this is something that I want to learn more about. And so is it a situation where there's kind of a step function where people either accept certain textures or they reject them? Or is it really complexly interwoven into the whole experience?
Curtis: Yeah, it's a little bit more complex than that. And one of the things that we're kind of getting at, since there doesn't seem to be textures that are either good nor bad. The context dependent is the combinations and the contrasts of the food when it comes to texture. Texture contrasting combinations are pretty important for acceptability of a food product. So you think something about like a pizza when you get like chewy cheese, you have soft parts of the crust, hard parts of the crust. So that's an area I think that's really important to consider when you're thinking about what makes a good texture, what makes a good or bad texture. It is not necessarily, the attributes that are present, maybe, it's maybe how they relate to each other. Almost kind of like the flavor or how different flavor compounds need to have a balance or something like that. There's probably something along those lines in the texture.
John: That's fascinating. So you've got these different aspects of texture within some food. And then you need to have I guess you've got the contrast. If you've got a crispy outside, soft chewy inside, so many foods seem to have that. It's interesting cause it's kind of the opposite of like, I often think of sensory properties as having this kind of satiety curve or typically, like, sweeter is better for a while and then it's not good anymore. Salty is better for a while and then it's not good. But it sounds like what texture it's more about combining extremes. Is that the case, would you say? What do you actually find in practice?
Curtis: Yeah. Again, it's kind of food dependent but worst. There is psychology there's this kind of idea of the peak and endural, where you're kind of your recall or your hedonic experience is based upon the peak or the highest intensity. And also how it ends. So if you look at texture as an active process, right, the oral processing is going from intake all the way to swallowing. I think that there's something that we can probably draw from that or probably it really depends on what was the most prominent or in sensory terms it is being called most dominant. It is also kind of how in how it finishes, you know, mouthfeel and residual mouthfeel is really important texture driver for a lot of foods. So when you see the term grittiness, a lot of times that's in reference to residual properties.
John: This is fascinating. Okay, well, this kind takes me into like one of my favorite topics, because I think that more of these variables should be involved in machine learning models. I mean, what do you do? You click this texture data. I suppose you also have, like more standard sensory data that you've collected in terms of flavor and other maybe more traditional, less highly targeted measurements. Well, what does a typical analysis equipped for you in your lab then? How do you use this information?
Curtis: Yeah. So we really do like a correspondence analysis as a kind of visualization. So when you get more of textures are present, then start by time. So that's a good way to kind of explore the data and kind of seeing what texture should be associated with. What are the products made in subsets of consumers that have different texture preferences or even oral processing strategies. And so I think if you're talking about building big models, that's definitely the variable that's probably the most needed. I would say would be mastication and oral processing data from individual consumers. Texture almost has this emergent quality where if you don't ever bite into a food, a lot of times you'll never understand it's texture complexities. You just throw a piece of candy or cook infection in your mouth. How you process it orally is going to be a huge determinant of what texture attributes you perceive in like classic chocolate, right? Like, if you don't ever bite the chocolate, you just throw a Hershey's kiss or something into your mouth you perceive this melty like kind of sensation. So doesn't have that dynamic contrast, like if you actively processed the food orally like using your molars and that type of stuff?
John: Right. So there is really this like the texture is, yeah, if all you were to do is take these texture measurements, but you didn't measure something about the people, you wouldn't make any sense out of it?
Curtis: I mean. Yeah, I think that's pretty general. What I would say is that if you're not considering the human aspect of it in multiple ways, you're not really understanding texture as a sensory property. And it is an actively sensed sensation. You know, it's not passively perceive and especially they're trying to relate it to, you know, some type of measure of consumer liking, you know, you have to take into account a lot of factors. Yeah, it's kind of fun and the problem with it.
John: Yeah. That's fascinating. Clearly speaks to me because you know there's always this discussion about do we use foods to measure people? Do we use people to measure foods? But, you know, it's both. It's always like this interaction in any kind of sensory exercise, I think that, you know, food doesn't have flavor, right? It only adds flavor when someone tastes it. So it's the same way. Yeah. Sounds like it's even more true with texture. Okay, well, that is really interesting. So now I'd like to kind of transition into the current situation so depending on when you're listening to this podcast, this podcast is recorded during the coronavirus pandemic and so we were not able to be around people as easily. And so I know there are some limitations in texture research typically like, can you described kind of classically? What does it look like when it comes into your lab and you're gonna evaluate the way that they're chewing the food, what does it look like? What are the measurements you make and what are the instruments that you use for that?
Curtis: Yeah. So when I first started this kind of line of research, like actually back my PhD, I would actually measure jaw muscle contractions through using electrodes basically and quantifying mastication parameters from that, you know. Since then, we've kind of transitioned into a little bit more modern approach where we can use video and basically create a two dimensional plane and kind of track your movements in a little bit more detailed setting so we can get like you can pick up on patterns, especially when it comes to horizontally. So a lot of mastication and the nuance of mastication is kind of horizontal, it's not vertical. Like if we only chewed vertically, we wouldn't have, we would really do a good job of making sure the food is between our molars so we can maxillary process it. So the horizontal subtle movement is kind of where it makes a kind of difference being a good chewer or a bad chewer or somebody can aggressively, you know, build a bolus and somebody who's a little bit kind of slower to do that. And so right now, what we do maybe come into our lab, they would sit in front of a webcam, basically, and they would put on a modified welding mask that has cut out for their face but it has markers so the software can pick up the location of their jaw in reference to their head. So if they turn their head a little bit or something like that, it's not mistakenly perceived this jaw movement.
John: I see.
Curtis: And then we would put a small marker, like a small dot, if you will, underneath or on the chin, basically. And that dots movement in reference to the frame of the face is kind of how we would start extracting chewing parameters through analyzing the video.
John: I see. And this is software that, where's the software developed then?
Curtis: Yeah, it's actually a software developed out of the, trying to think of the name the Center for Plant and Food Research in New Zealand. So Aaron Wilson is the researcher that I work with there, and he's got the software. It's been very helpful for us. It's not commercialized or anything, but it's really, really good. And one of the cool things that it's been able to do is classify different chewing cycles into the shapes that they represent. And so I got ahead of myself a little bit there. But chewing cycles tend to follow three different forms, either a crescent, a crossed meaning, like a figure eight and then a circular, which would be kind of like the classic like a cow chewing its cud, it's kind of, it's oscillating. So it's able to classify each individual chew as one of those three types. Helpful or picking up, you know, individual differences in how people are perceiving the product.
John: So is there a measurement then that each chew, I guess is a time point, chew happened here and it was chew of the following type. That's the kind of data you've got.
Curtis: Yeah. So we have we extract, I think, around 15 different parameters from the eating process. So all the way from like the percent of chews that were crossed versus crescent shaped. We also have how wide somebody is opening their mouth, how quickly they're closing it. How much horizontal movement there is in the average chew, those types of things.
John: That's fascinating. Yeah. And I can imagine that. Okay. So I do want to ask you about how this is, you know, how you're coping with the fact that maybe it's harder to make these measurements without bringing people into the lab. I also want to get to thoughts on using facial recognition. I suppose there's facial recognition, some sort of facial recognition built into the software. Is that something you're seeing kind of improving over the course of your research?
Curtis: So this is something that my collaborator and I were talking about recently is integrating some facial feature type of extractions into the software. We can start to maybe even address hedonics. So then, you know, obviously the close examples infants, right? People have certain hedonic facial responses to food stimuli. And I think I'm not sure where's it at with that. It's definitely working on integrating that into there, and then with others need to do validation steps and stuff to see if where we can not only pick up mastication patterns, but maybe pick up mastication strategies that are associated with disliked foods and like foods. The ideas you could possibly chew a food differently based upon how pleasurable it is.
John: Right. If you're trying to avoid eating it.
Curtis: They kind of found that stuff with flavor, I think on some I can't remember the exact citations, but people have different eating behavior with food than aversive flavor versus something special.
John: Right. Yeah. with my four year old son, I can definitely relate to that. Like someone's eating something they don't want to eat. Yeah. Well, let's come back around now to that whole kind of coronavirus topic. So what sort of adjustments are you having to make? What are you learning during this pandemic that you might be able to use, you know, into the future? Is it you know, how is it going?
Curtis: Yeah, it's challenging. It's just challenging not being able to deal with the traditional central location, sensory test and trying to address some basic factors. So when it comes to a little processing, I think our biggest concern is going forward is how can we get rid of the helmet, if you will. I talked about that modified welding helmet. And how can we somehow get an estimation of somebody's jaw motions in order processing without accounting for head movement? And so there are methods out there coming from animal pose behavior type of research where there's different AI's that have been developed, emerge camera angles, sorry, multiple cameras from different angles. It allows you to kind of account for different, in our case, head movement. And also, one of the cool things that allows you to do is have three dimensions tool. Right now the thing we're doing is two dimensions. And so I think that that would be a really big step forward for us to be able to really start to extract as much data as we can when it comes to oral processing behavior.
John: And is this something you think you could get subjects to do remotely? I mean, you could train people to do.
Curtis: Yeah, I think so, because, I mean, almost every laptop now has a webcam, you know. And so if we were creative as far as having like maybe webcams or maybe even as something like a phone booth where people can come in and have a snack and we can extract some or processing information that way, too. But yeah, I mean, right now we're still kind of the brainstorming wave. But, I mean, the actual software to integrate and get a three dimensional estimation of an animal, including us, their eating behavior and that type of stuff is out there. So I think it's up to us now to capitalize off of other fields progress.
John: Alright. Just to connect the dots, I suppose, because, I mean, I definitely see well, you know, like those people, the machine learning community I like and I see all these demos that people are doing. You know, I guess they're just a lot of times it's like solution as people develop something cool and then they go try to find the problem that it solves. But it does seem like you have a problem that somebody probably has made some progress towards trying to figure out different angles from just one camera.
Curtis: Yeah, I think it's important. I agree. And I think that's the good thing about all the progress that's been made in this field, is that it's probably out there and it is finding it. There's always going to be a little bit of a hiccup in tailoring it to your application, you know. But there has to be made. And I think some of the problems with that is, it doesn't this type of research doesn't fall within a natural funding mechanism like a lot of other types of more fundamental sensory research.
John: Interesting.
Curtis: And so to get them in these projects off the ground later, take an industry that has a really an industry partner that has a lot of interest in this type of topic or a kind of a very creative approach to getting your USDA, MIFA or something like that to take an interest.
John: Interesting. Okay. What are the product categories? I mean, do you notice that in some product categories, texture is more important than others? Like have you seen or maybe you're working food and beverage, I suppose primarily, right? There's not really any other options here. But what are the product categories where you think is most important to be measuring texture?
Curtis: Yeah, I think texture definitely starts to become more prominent when flavor takes a backseat. So I know, for example rice, rice is a great example of how texture can be a very prominent factor in acceptance or rejection. The meat is a great example as well, especially poultry products. It's nothing against poultry producers with some chicken breast doesn't have a lot of flavor. So the red shading factor is kind of its texture. You have things like, you know, woody breast and texture issues that arise through either genetics or animal stress, those types of things. So those would be the two that come to mind. I think anything with a dynamic texture also so meaning it changes as the eating process goes. Those are always kind of high up there and how important texture is, because when things are changing in your mouth and you tend to notice it more. So something like chocolate.
John: And where are you when it comes to engineering textures? I mean, the flip side, you are suppose to make measurements you find there, okay, we want like some things are better than others. How far along are you in the kind of the food science side of this producing certain texture?
Curtis: Yeah. So I didn't have a background in carbohydrate chemistry, so I'm kind of dangerous enough to make lab stimuli to maybe elicit the response that I'm trying to go forward to try to assess or a processing parameter or something. But I'm not nearly as advance as some of the more ironed out hydrochloride groups I know, like Dr. Sacar at the University of Leeds, that has a really interesting stimuli that are really able to kind of address questions that I can't. But I think one of the problems you're always going to run into is you really can't change one texture attribute without changing others in most foods, especially foods of consumer importance. You could be able to do something in a model gel or something like that. But I mean, if you're talking about, you know, any major type of consumer facing food product, usually you're always looking at changing at least two things with every switch in formulation.
John: Right. Yeah. I mean, that's a kind of product customization that is always issue, right? That we do these local optimizations or we say, okay, the following features are most important if you change this one feature, you can expect this kind of benefit. But it is hard to take account of the fact that that change is going to lead to other changes. So, yeah. That's I mean, there are some approaches for that for handling that problem. I can definitely relate to that, that kind of dilemma. So let me ask you then about kind of going back to the kind of the big picture of this, because sometimes there's an aspect of consumer education. I mean, so people just have individual differences in their chewing patterns. Is that right? I mean, so there's not any kind of if you found any variables. I mean, there's culture to different cultures to differently. What are the other variables that might predict someone's doing better?
Curtis: It's actually a paper come out recently from Marco Steger's group, I believe. And they found that the number one predictor of or processing was somebody's body weight, I think, or some general measure that could BMI, if I'm not mistaken. So I know that there was a recent study as well that looked at different facial morphologies and those types of things, and they didn't really come up with much. I think a look so across cultures. I think it's a kind of a confounded variable because not everybody used the same utensils to eat. So, like you thinking about maybe the East Asian culture that would be using chopsticks versus us here in United States or typically using forks and spoons. It's kind of creating a confounding factor with how big your bite is, which is a huge thing about downstream effects of more processing. That's the number one factor. So I think it's tough. So we've done some cross-cultural stuff looking at texture contrasts and combinations, using Singapore as a good foil for the United States because they have the same language, which is helpful. You don't worry about translation issues and language issues. And there are some differences when it comes to preferred textures. So like we found the people in Singapore like actually prefer a more texturally diverse food than people in the United States. So it talked about contrasting combinations. In general, I think the Singaporean population is reporting the desire to see a little bit more, you know, combinations of textures. But as far as our processing goes, I think there's a lot of work that still needs to be done, you know? But it's gonna be hard to control for other confounding variables.
John: Yeah, I'm just wondering what can be done in terms of marketing, maybe telling a package instructions, take a big bite. That kind of thing. You know, like trying to control the way that someone or at least influence the way that someone's going to interact with the product.
Curtis: Yeah, that's great. I yeah, that's a great idea. I think, that's something to take into account more my own research interest is that looking about how, because obviously how you're processing a food orally is going to change how you perceive, just the hedonic of the texture. You can, you know, nudge somebody in a certain direction through surface appearance or packaging, something like that to do what you want them to do from an oral processing standpoint. Yeah, that's something that definitely help you increase consumer acceptance.
John: Yeah, that's interesting. Okay, great. Well, amazingly, Curtis, we have 25 minutes here with very little effort. So, I could keep talking to you for quite a long time, but we have to wrap it up here. So thank you very much for this education first off, and let me ask, you know, as we kind of wrap this up, any final points you'd like to make? I mean, I'd like to get your general advice, but, you know, to young sensory scientists but is there any last points you'd like to make to get a chance to get to?
Curtis: I think it's just always be vigilant of texture. You know, I think all the times, even in traditional descriptive analysis or something, it gets pushed by the wayside. And I think start looking at the language people use to describe oral and tactile sensations. You know, there's a lot of cool things out there. You can start to comb through different data sets that are online and as far as food reviews and that type of stuff and start to think we need to do a better job of looking at how consumers talk about texture as a kind of a window into how they think about texture. And so in things like natural language processing, what I know covered at earlier in your podcast with Jacob Lahne. You know, I think that there's a lot to be learned from that to try to kind of more assess how the consumers are how they're thinking about texture.
John: Yeah. Excellent. Alright. Well, I think I will call this episode, "Texture Matters" because it does. You've definitely convince me texture matters.
Curtis: That's kind of what my goal is here. So I'm happy with that.
John: That's great. Alright. And so then, just wrap up, what advice do you have for young sensory scientist, someone just starting their career, what should they be thinking about right now in terms of like, you know, hot topics, research areas that you see are promising?
Curtis: Yeah. I think just having as a young sensory scientists, I think having the ability to utilize large data sets because they're so much easier to collect data than even when I started like 2010 in the sensory, that the data's out there. And so, you know, whether it's a programming language or whatever. Being able to get that data and to make it into actionable recommendations I think is probably a great skill to get because it's so versatile.
John: Yeah, I totally agree with that for sure. I mean, personally, I think everybody should learn to code. And it's not like the most popular opinion. That's good. So for someone who listen to this and they want to, you know, maybe join your lab or reach out to you. How can they get in touch with you?
Curtis: So I think sensory.tennessee.edu is our lab website. It's got all my contact information on it. And it's kind of some examples of things that we can do for the food industry and research. So if you're looking to go to grad school or something like that, there's some good examples of what we do. From a research standpoint as well. And then on Twitter, @crluckett.
John: And you are on LinkedIn?
Curtis: Yeah, LinkedIn.
John: Okay, great. And so yeah. So for as far as industry goes, so you do run consumer studies. So to someone listening to this, they want to do consumer study, what will be kind of typical study you would run then?
Curtis: So we do a lot of just general CLT stuff with consumer acceptance. We do a lot of targeted texture, consumer work as well. But yeah, I think we do the same things most standard sensory labs do. And then we can always expand it more to make it a longer project. So we do everything from the simple CLT went off all the way to long term, you know, deep dives into the product categories, texture and, you know, even cross-cultural things. We have a really close collaborator in China that we work with to kind of see how products are differentiated from different consumer groups.
John: Okay, that's great. Excellent. Curtis, thank you so much. It has been great. And I really appreciate you being on the show today.
Curtis: Yeah, thank you.
John: 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.
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!