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. Delarue’s research focuses on methods to measure sensory perception and preferences and on their effective use in food design. He seeks to develop high validity consumer-oriented methods. In particular, his research explores the role of context in hedonic measures using immersive environments and digital technologies. He also works to develop and validate rapid and flexible descriptive analysis methods with applications to new product development and consumer research. Using these methods, he seeks to understand the sensory determinants of food preferences and to find levers to drive healthy and sustainable food behaviors.
Formerly an Associate Professor at AgroParisTech in the food science and technology joint research unit with INRA and Université Paris-Saclay, he has served as the Chair of the French Society for Sensory Analysis (SFAS) and of the European Sensory Science Society (E3S).
Metaverse article Julien mentioned
Transcript (Semi-automated, forgive typos!)
John: Julien, welcome to the show.
Julien: Hi, John. Thank you.
John: You have a great voice for this too, so I think this is especially a good episode for our listeners.
Julien: Thank you for that.
John: Okay. Great. So, Julien, let's talk a little bit about your background. I would expect that our listeners have encountered your research in their training at some point or maybe just recently in professional development. But can you take us through a kind of arc of your career, maybe starting as an undergraduate. What got you interested in food science and sensory and take us all the way up to the present.
Julien: Okay, well, I was always interested in food as a kid, but also interested in human behavior, animal behavior, too. So I think that was the start of all of it. And so I was trained as a food scientist and as an undergrad, in France, we have a lot of internships with companies. So I did my last six months internship with Nestle in Switzerland and work on the development of sensors to detect coffee flavor, coffee aroma. So they were unprofitably named electronic noses. But that was the start for me and that was more about flavor chemistry and sensory. But I was in connection with sensory science. So after that, I decided to go to graduate school. Go for Masters first, while did some flavor chemistry and then I moved to Ph.D. and to sensory and consumer science.
John: Okay. Where did you do your Ph.D.? I actually don't know where you went.
Julien: That was also at AgroParisTech. At the time it was named ENSIA and I did that in collaboration with Yoplait, the yogurt company and at the time was a French company. It was bought by General Mills. Again in France, very frequent to do that with Ph.D. in collaboration with the industry. So I was hired by the company but could do my research at the same time. So it was a great experience.
John: Right. Yeah, that's interesting about France. We both know Ben were so quite well, my former colleague and now your colleague. I guess he's not at UC Davis, but he's in Davis. And I think he did his French military service by going to graduate school or something like that.
Julien: Exactly.
John: Great. Okay. So then can you take us into the present, please? So you were at AgroParisTech, and then recently you joined UC Davis. When did you officially start as a professor at UC Davis?
Julien: It's about two years ago, less than two years ago. But of course, with the covid and the pandemic, it was very complicated. I had in transfers and teach online, which was interesting. And then I moved to California in January this year.
John: Right. That's right. And for people who don't know, Julien and I are working on a book together with Thierry Worch on Introduction to Data Science for Sensory and Consumer Science. I guess for the last two years, working with you regularly Julien, I've heard this saga and I would have to say, I don't recommend changing countries in the middle of a pandemic. It doesn't sound like fun. So good job and I'm glad you survived and your family. Okay, so let's talk about your kind of research interest now, because obviously the new technologies, the use of all these new digital capabilities to expand sensory research is a shared interest. So can you take us through some of the highlights of your work in that area, please?
Julien: Yeah. That's one of my research topics is to work on hidden testing and how we can make consumer tests, in general, more valid, more predictive of the reality of real life. I already have the feeling that we are missing something when we do consumer testing, sensory booth, or in very controlled conditions that they should be artificial. And because of that, I think it's nice to either move to real life and try to do some tests, for example, in experimental restaurants. But it might not be the perfect solution because you cannot control everything. You cannot play with different scenarios. So it's much more convenient to try to recreate real environments or natural environments in the lab. And technologies or immersive technologies offer a lot of possibilities for that.
John: Right.
Julien: Yeah, so that's the idea, so to play with special effects, very large screens, wind, heat. You can play with the heat or cooling system. You can even play, of course, sounds, ambient sounds, or other smells to recreate an experience. So that's the type of technique I'm using.
John: What are you finding? What are the key differences that you would, is a sensitivity that you're finding? What are the kind of key, I guess, metrics that you're looking at?
Julien: So there are different types of results. One of them for my first experiments I connected back in France that was on hidden evaluation appears and we found out that we could level off the effect of the session timing using this type of system. So we had people tasting beer in the morning in the winter when it was freezing outside, and we had this recreated nightclub or tropical beach, and people completely forgot about the outside condition, and their responses were just very similar to afternoon sessions or to other situations. So that was a way to easily control the situation and also to screen which deal, in that case, which product would match a given consumption situation. So I think it's a way to explore consumer experience into that. In other experiments, I tested the evaluation of gourmet dizers to see what kind of favor would better adapt to different types of situations. So, again, it could be a way to screen or to investigate how some consumption situations would match different versions of your product that you're trying to develop for instance.
John: Do you ever find that there's a tension between, sometimes you can kind of go too far with making things realistic. Let me give you an example from some research that I've been involved with Chris Simmons. I'm sure you know him quite well and his graduate student, Ashley Soldovini, and I should mention Hamza Diaz and our team has been involved in this research. So there we were looking at again, immersive environment, but with voice-activated surveys. and one of the things that was interesting was the feedback from the panelists was that it was more natural in an immersive environment to use a traditional phone-based survey rather than a smart speaker because they reached a point with a smart speaker where it was so close to talking to somebody. It felt very strange because it was almost real, but not real. So it's the uncanny valley. Are you familiar with this concept where the technologically supported experience is close enough that your brain starts to compare it with real life, but it's not close enough to be real life, and you end up with this tension? Have you experienced any of that in your research where this tension between the fact that someone knows it's an experiment and also you're putting them in immersive environments? So what are your thoughts on this kind of, I think it's going to be more and more relevant, actually?
Julien: Very interesting. I wouldn't put it this way but that's true that participants could be maybe sometimes over distracted by the effects that you're trying to implement, so that's one thing. Sometimes very surprisingly, we have negative effects of the immersion and that's when we try to recreate office environments. And we were to investigate the snacking behavior with the office environment, it turned out to be quite negative, which makes sense, right? So that's one thing. Also, I noticed that sound, since you were commenting about speakers, the sound could be very disturbing and plays a very important role in the immersion. At least as strong as the screens and the vision I would say.
John: That's interesting. Yeah. Well, we've had some experts on audition on the show, and they've talked about just how hard it is o recreate sound because we have all sorts of really fine-tuned sensors in the auditory system where when the waves come at different angles into the ear, the timing, it's very hard to recreate the virtual auditory environment. So they talk about the challenges that are rising.
Julien: And also the environment that you're recreating is kind of fixed. It's one environment, so you cannot really customize it. And that's probably where also the mismatch comes from because each consumer has his own experience, right? But then if you create, like a coffee house or I don't know, a kitchen, then it's one specific kitchen. It's not everyone's kitchen. It kind of makes it artificial, I guess. But still, I still think there's value into it. But I agree that we should be careful about this type of biases maybe or effects.
John: It's interesting because there's kind of two, there's kind of like this square, right? Where you would have completely ecologically valid natural behavior by a consumer that is not experimentally controlled at all, right? And then you have the other extreme, right? Where you have a completely controlled environment that's really anesthetic that is not ecologically valid. And it seems like there's kind of two pathways between the one square and the other. One is the path that you're on which is to have a more controlled environment, like bringing ecological validity into the lab, and then the other pathway is to have measurements that are made in the real world that might bring more of the experimental design or the more of the experimental control into the real world. So what are your thoughts on that? Do you see promise for that other path as well? Or do you think that's just too hard of a problem to really?
Julien: Yes, I think that is definitely the fact that it's being much easier now to record a lot of data. We work with any device. It's not only pencil and paper, but you could use watches, smart watches, phones, cameras. I think that it's been easier. Of course, you don't want to be overwhelmed by data that are not always very insightful. Another thing that I think that is under-investigated is the task. I had a Ph.D. student working on that in France. But the task because you can work in a very natural environment, very ecologically valid. But if you ask tons of questions to consumers and completely biased or stole their response because the way you ask the question is not natural, then you're killing everything. You don't want that either.
John: Yeah. That's an excellent point I would say. With smart speakers, we've definitely found that on a smart speaker you don't want to ask, how sweet is this from one to five, right? That isn't a natural question to answer out loud. You do want things like, Is it sweet? Yes or no? Is it very sweet? You have to adapt. Yeah. So that's quite interesting.
Julien: I was reading yesterday an open ad in the Atlantic from Superman, I don't know if you read that but it was about the Metaverse. Really hot topic now, and he's commenting about a history of metaverse and second life and virtual environments like that. That's very interesting. I said, well, there's been so much evolution in technology. I mean, the technology has improved, but it's still very far from the real world, and we're just trying to make artificial words that look nicer than the real world. So it's kind of critical but also interesting to read. I recommend that.
John: Okay. I'll check that out. I'm very interested in that, actually, as you probably know. So along those lines, how optimistic are you that the chemical senses will ever be put into the Metaverse? Because it seems to me to be a much, much harder problem. Sight and sound, okay, but the chemical senses, it seems like an almost hopelessly hard problem. Honestly, like you've worked on e-nose. What are your thoughts on that?
Julien: Very good question. I've been contacted four years by people who want to develop systems to communicate about the chemical senses. And I think that's much more complicated if you think of the sense of smell in particular. Like coffee, the smell is based on about 800 if not 1000 compounds. Of course, not all of them are very important. Maybe you can go down to 20 compounds or something like that to make a realistic coffee aroma but still, it's a lot, and it's only for coffee. So I'm not so optimistic unless you have the possibility to place electrodes or chips directly on your olfactory bulb or olfactory epithelium. I don't think you can replay with chemical stimuli. That's the difference between chemical senses and other senses, I guess.
John: Yes. Well, it's a dimensionality reduction problem, right?
Julien: Right.
John: That when you transmit visual or auditory information, the waveform carries, like waveforms are carrying the information. And so it isn't really that many parameters and especially when you have an image where you've got correlated pixels and you can compress and decompress or whatever. That's a reasonable problem. But with the chemical senses, you would have to have a very small number of chemicals which are going to be basically your dimensions from which you're going to reconstruct the original experience. Yeah. Well, people are working on that problem, and I wish them luck.
Julien: Yeah and Flavor and Fragrance companies have worked on systems to create or to help like, for example, flavor is to create flavors. Some instruments that would blend the compounds of flavor blocks but it's very difficult. The risk is to become very careful to something that would smell artificial, and also one person would defer to from another. That's another story.
John: Well, that's right. That actually I really want to talk to you about the individual differences, but just to kind of wrap up what you're saying, there's definitely a risk of the uncanny value here with the chemical senses where it's closed but it's not really correct. I think if you're in one product category, suppose you're talking about vanilla fragrance or something. It's probably vanilla a very simple example but if you've got a very constrained set of possible things that might happen, I think it is reasonable that you could have a small number of chemicals that recreate the experience. But that's not the way the real world is, right? The real world is very often...
Julien: Yeah, and actually, vanilla is a very good example, because if you simplify the vanilla flavor of vanilla smells too much then that may smell weird and, for instance, here I'm surprised that people tend to use the vanilla fragrance in their bathroom which for me, is a total nonsense. For me, it's only for food, right? So it's a complete mismatch.
John: That's interesting. Okay, let's talk about individual differences because it's obviously extremely important as well. And you were talking about, well, maybe just talk to us a little bit about how individual differences have come up in your research. I think that's one of the things that's really underappreciated in our field and one of the reasons why data scientists have trouble fitting sensory into consumer data.
Julien: Yeah. For me, individual differences should be the foundation of sensory science. That's why we're here. It's because people differ. They don't perceive the same thing. They don't like the same thing. I think people are aware that we don't like the same things and that's very fortunate. But about perception, it's not always the case because we do not communicate so much about our perceptions. We are not aware that we don't actually perceive the same things and that's true for all senses, especially for chemical senses, but for all senses, in fact and I think that's for us, sensory and consumer scientist, that's a real challenge. That's a challenge for data scientists, but also for us because we need to find ways to measure individual responses and take these differences into account. But on the other hand, we also need to make our data actionable, because most of the time we need to make decisions based on this data. So if we consider that each person is different from another, that could become a problem to summarize this data and make it something like actionable. So what decision should I make? Should I launch this product or not? Should I go with this flavor or this other flavor? So that's the type of question we always face, even if we're not always aware of it.
John: Yeah. I definitely see that with machine learning, that you have to really try to make predictions at the individual level for your models to be useful.
Julien: And in terms of measurement, I think there's been a lot of changes and that's very nice during the past, maybe 15 years, so that would cover almost all my career. So we moved from expert or train panelists to more consumer-oriented approaches, and I think trained panelists are good. We provide you with robust and stable, accurate data, but they do not recover in terms of differences. So other methods that are maybe less accurate, but that allow you to measure these differences are also very important and very interesting.
John: Right. I think Chris Finley, who is actually the very first Aigoracast guest, and he's now, I believe retired, actually, he made the point that oftentimes we try to train the differences out of the panelists. And actually, when you do that, you're getting rid of important information because people were different, and now you're getting them to all be the same and you've lost valuable information because that variability would have been important to know, right? And you try to train away something that was potentially useful.
Julien: Yeah. I completely agree with that and that's interesting, too. If you look back at people who first attempted to do descriptive analysis, let's say, with consumers, their goal was to show that you can get reliable results, meaning results that are similar to what you would get for the train panel. And nowadays that would be more like the opposite. Can you get results with the train panel that would allow you to capture consumers diversity and consumers response without installing it so much?
John: Okay. Yeah. It's a great field because of all these, this is the psychological side of what we do.
Julien: Absolutely.
John: Okay, so let's just talk about the next, [we're almost out of time, actually] to talk a little bit about your kind of current and immediate research focuses the next few years. What do you think are the most exciting topics in sensory? What are you looking forward to working on yourself? What do you think other people should be working on? Where do you see sensory over the next two to three years, developing the most?
Julien: Well, first of all, it's not something very new, but I think that it's becoming more and more important. Sensory is central or must be central in product development, especially with food product development. If you look at all those companies and new products that are being developed because you want to make products more sustainable or that are healthier. This has a lot of constraints. There's a lot of reformulation going on, and you cannot be successful at it if you don't take consumers voice into account. So I would say that sensory science is more important than ever.
John: I totally agree.
Julien: But it also means that we as sensory scientists should be able to be more part of the product design and product development process. For me, it has always been a frustration that we always come at the end of a project as a validation of what has been done. I think we should be there to lead product development with our data, with data different consumers and sensory data. So that's an important point. There are many questions behind that. For example, if you think about reformulation, you could have different standpoints. You could see reformulation of a product from the product developers standpoints. So you start with your product and you want to change the recipe and see how the changes in recipes will affect sensory properties of the product. But you can also see it from a more consumer oriented perspective where consumers don't really care maybe about only one product or only one brand. They care about a product category, and they could do alternative. So it's also very important to monitor the market, have sensory market research, and then use that information to drive reformulation and find maybe opportunities. I think there's been very little work on that. I could see many opportunities, I would say that sensory science would probably be progressing. If we manage to have a broader perspective and embrace the consumer's points of view.
John: Right.
Julien: I think that's something very important.
John: I totally agree with that. I think actually, the tech companies have helped us with this because they put such an emphasis on UX, right? For example, Airbnb largely is successful because of their emphasis on UX that they had, I believe two of the three founders were design students, and they really push design. If you go to the Airbnb website, it's a beautiful website. It's easy to use. It's intuitive. And together with some of their exploitation of Craigslist, I don't know about the history of Airbnb, but they did a great job on design, and that was one of the I think, kind of shining lights in the tech world of why users experience, why design is so important. I think that's helping us, especially the food industry, to say, look, we're doing user experience. Sensory is the user experience of food, right?
Julien: Absolutely.
John: And we have momentum on our side in that way for sure. I think if you will bring the chemical senses into the Metaverse, sensory will become super important because then the whole experience of the metaverse will be the covenance of sensory so that's another topic for sure. Okay, well, that's good. Other thoughts, what about research questions, are you going to be working...
Julien: Just one quick thing since you were mentioning design. I think that sensory shouldn't also focus too much or only on food sensory science. There's so much work being done in the non-food industry or non-food applications. It's very important and then we can learn from that. I mean, as a scientist, we can learn from what people do in non-food applications.
John: I totally agree with that. In fact, Thierry Lageat, you might know.
Julien: Yeah, of course.
John: He was on the show, and they did a lot of work on private jets and cars and things like that. The feel of the leather, and it's all design work, really. But it's sensory-informed design work. So, yeah, fascinating. Okay, and so then research topics, for just a few minutes here. So what are the kind of initiatives you have lined up?
Julien: Well, the research topic would be on the methods precisely that we can develop and not only just the measurement methods but also the general approach that we can recommend or design to help developers to find optimal pathways for sensory product development. So that's one very important topic for me. And the other one, of course, is to keep working on the ecological validity of sensory tests of consumer tests in general, playing with these immersive technologies. But not only it's also about how to better engage consumers in the test, to better take user experience into accounts.
John: Right. Okay, that sounds good. Let me know if you want to do anything with smart speakers, we would be happy to collaborate with you on something. So if you have any use for that, let us know we'll be happy to.
Julien: Okay. I have some ideas.
John: Okay. That sounds good. So it's all handcrafted so whatever you want to do, we'll program it. Alright, so let's wrap up now with how can people contact you? What is the best way for them to reach out if they may want to apply for a spot in your lab or want to collaborate with you on the topic?
Julien: The best way would be to show up on campus, but it's not possible. You can email me at jdelarue@ucdavis.edu. I'll be very happy to discuss.
John: Okay. So we can put a link to your lab website then.
Julien: Sounds good.
John: Alright, and let's wrap up them with advice for the young researchers, what advice would you have?
Julien: Well, be curious. Be open to other fields, connected fields. I've collaborated with researchers from very different backgrounds, and that's where probably I learned the most, you know collaborating with economists, with nutritionists, you will always learn so be open, be curious, that would be my best advice.
John: Right. I think that's definitely great advice. Yeah, pretty close to what I would say so I agree. So, Julien, it's been a pleasure. It's always nice to talk to you.
Julien: Pleasure is mine.
John: Thank you very much for being on the show.
Julien: Thanks.
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.
John Ennis is a leading expert in sensory science and consumer research, with extensive experience in statistical analysis and product development methodologies.
