CLASSIC 27: The Universal Eating Monitor for Studies of Human Intake
KISSILEFF, HARRY R, KLINGSBERG, GARY, and Van ITALLIE, THEODORE B.
Universal eating monitor for continuous recording of solid or liquid consumption in man.
American Journal of Physiology 338:R14-R22.
With an interview of Harry Kissileff by Kathleen L. Keller, PhD, Department of Nutritional Sciences, Department of Food Science, The Pennsylvania State University (January 19th, 2022)
This classic paper from Harry Kissileff and colleagues is part of a series of studies that was centered around the validation of the Universal Eating Monitor (UEM). The UEM and an accompanying novel food mixture that could be served in either solid or liquefied form was designed to translate animal models of ingestive behavior to humans. The primary advance of the UEM over methods that were available at the time was its ability to measure intake of food in a variety of forms (i.e., solid, semi-solid, and liquid), continuously across a meal or eating bout. As noted by Booth (1), “Measures of intake alone will not provide evidence for the control of intake.” If the goal is to understand mechanisms, one must measure the ingestive acts or behaviors that facilitate food intake. The UEM was a critical advance in the ability to test psychological and physiological mechanisms of food intake control in humans.
Since its inception, the UEM has been applied to answering fundamental questions about how food intake varies with food form (2), palatability (3), sex (4), obesity and disordered eating (5), hormonal manipulations (6), and, most currently, bariatric surgery.
I had the pleasure of speaking with Harry Kissileff about his classic study. A summary of our interview follows. My questions are in bold, followed by Harry’s responses in plain type (edited or summarized in some places).
Can you talk a bit about how you developed the UEM? How long did it take? Where did the idea come from, and how did it differ from the devices that were currently being used in the field?
I came to St. Luke’s Hospital (Now Mount Sinai Morningside) in New York in July 1976. I had previously been acquainted with Hashim and Van Itallie’s feeding machine, which required participants to press a button to receive 7 ml of formula diet. I did not think this captured the nature of normal eating by people. First of all, it was a liquid, and not a particularly palatable one at that. Second, it required the intervention of an artificial response for small reward. Third it was not particularly portable. It did not take long to figure out how to create a device that could be used for solids or liquids and for which eating could be executed with ordinary utensils. I think we had the design completed in less than a year, and an abstract was submitted shortly after that. I had a programmer, Rich Whiffen, develop both the hardware and software for the application.
I got the idea to use load cells from the animal research of Cliff Baile in sheep and Jaques Le Magnen in rats. The idea of covering up the scales came from the Stanley Schachter’s lab (see Classic #16), via Kathy Porikos, as did the use of cover stories to keep the participants from responding based on knowing what the experimenter expected. We simply told our subjects we were interested in their reactions to foods and beverages (e.g., “We have developed some diets for improved nutrition, and we would like subjective evaluations of them.”). Henry Jordan at the University of Pennsylvania and Volker Pudel at the University of Göttingen, Germany, were using devices that required liquids. Jordan’s device was a graduated cylinder with a reservoir in the cylinder and a tube that came from the bottom of it. I don’t recall there being any valves, so I’m not sure what subjects had to do to get the liquid. Pudel’s device was similar. It was a graduated reservoir, and they could either have the subject look at the device, or not look at it. They even played around with the rate to make it look like the liquid was going down faster than it actually was.
Where did the name Universal Eating Monitor come from?
I called it universal because it could be used both with solids and liquids. I’m happy with the name, but my colleague, Xavier Pi-Sunyer, was not. Xav always thought we should just call it an Eating Monitor. He thought “Universal” was over the top! But I think it’s accurate. You can also use it with different utensils, with a straw or a spoon.
One of the challenges we have in our field is that we don’t have a standardized or universally accepted food that we can use to study human eating behavior. How did you go about selecting this food? If you were to go back in time and design the study again, would you stick with this same food?
That’s a great question. I had a master’s student from Columbia’s Teacher’s College, Gary Klingsberg, who is now a successful D.O. I said “Gary, your task is going to be to create a diet that will be identical in composition, but can either be eaten as a solid or a liquid. And, it should be well-balanced and similar in composition with the typical American diet.” And he came up with this formula for all those items that are in it. You can either have subjects chewing the items that were in it…the apples, the bananas and soy nuts. Or, you can blend it up. Bananas and apples are sweet, so it was reasonably palatable. And because the apples weren’t peeled and the nuts were hard, it required enough chewing to simulate a normal solid meal. I think it got 5.6 to 8.8 on a 9-point scale. I think for this purpose, it was a good choice. I would certainly use it again if I designed the study today. The benefit of a blended food is that people don’t tend to pick out individual components. That was a problem that we had with the pasta meals. People would often pick out the macaroni and leave the beef behind. The downside is that it had to be mixed up. Later we went to easily prepared liquid yogurt shakes, which simulated well the liquid meals used in animal studies.
There have been some newer approaches that have built on the UEM. I’m thinking in particular of the Smart Dining Table developed by Miguel Alonso-Alonso (7). The main advantage of this approach is that you can measure multiple items at once. What do you think of this approach?
Here is the problem with measuring multiple items at the same time. It’s not a problem technically, but I think it’s a problem theoretically and analytically and conceptually as to how you’d frame the respective outcomes, and what types of questions you are going to ask. We had the same sort of problem when we originally tested binge meals in collaboration with Tim Walsh (8). One of the things we looked at was what order they consumed the items in. We did find that the patients and the controls differed in the order in which they consumed the items. If you have questions that can only be answered by using multiple foods, then it’s definitely useful to have devices that allow you to address those questions. But if you are just measuring intake, in order to elucidate underlying physiological controls you are probably better off with single food items. If you are trying to simulate, in a laboratory, what people do in real life, a multiple item meal makes sense. However, trying to explain the underlying controls of intake in real life was not the problem we were most interested in at the time. Physiological controls of eating in real life may be secondary to other sorts of controls, such as context, time of day, social controls, as well as composition, both physical and chemical, of the food.
From your writing, it seems that you view cumulative intake as representative of physiological signals on the control of intake. Do you think that cumulative intake patterns are more affected by “state” (i.e., meal context) or by “traits?” Phrased another way, under identical meal conditions and physiological states of deprivation, would we expect individuals to exhibit similar cumulative intake curves across several visits?
You know we did this study in 1982 (4). We tested some subjects for 4 days and other subjects for 8 days, and the percent of variation was about 15% from day-to-day. So, it’s pretty constant. And whether it’s a trait or a state? I would say it’s probably a habit. I don’t know whether that makes it a trait or a state. Habits are malleable, but they are also persistent. I don’t know that attempting to characterize it as one or the other is appropriate, but yes, these behaviors are reliable. That doesn’t necessarily mean it’s a trait or a state. And, that also doesn’t mean that it is representative of physiological signals. That’s a whole other question. I would say that when I went into this, I was sure that intake was controlled by physiological signals. After 50 years of research, and a lot of ups and downs, finding that some parameters are influenced and others aren’t, I would say that it’s quite complex. A lot of it has to do with the consistency of the diet. How quickly are people able to swallow? We were very disappointed that CCK didn’t seem to affect the rate of (eating) deceleration which we thought was an indicator of the rate of onset of satiation (9).The rates were the same under hormone and non-hormone conditions. They just stopped eating sooner. That suggests that there is some threshold that brings the meal to an end, rather than a gradual development of satiation that has to come to some point where it interacts with facilitation. That whole notion that I developed in 1982, 40 years ago; I’m not so convinced of its applicability nowadays. Physiological feedbacks may account for a percentage of the variance in how much is consumed (maybe 20 or 30%), but I think a lot of the variance is accounted for by how much people like the food and what the consistency of the food is.
Another thing that I didn’t really get into in the initial paper (2), but mentioned in the 1982 paper (4) is that the coefficients in the cumulative intake curve are correlated, which means they probably do not represent separate components, as we initially thought they did. However, we were able to separate these components of the curve by manipulating palatability in Bobroff and Kissileff (3). Another thing that is of great interest is that the solid meals tend not to be as decelerated as the liquids, they just start off eating more slowly and eating rate is more constant. If they are actually eating at a slower rate with the solid (as the liquid) why do they end up eating the same amount? We found in the first paper that the duration was significantly different, but the amount eaten wasn’t, and that the rate was the same with liquid and solids. That goes against what Robinson (9) reported and what we found in another paper. We address some of these complexities in a recent review (10). It may have to do with the fact that if you are going to measure rate, you have to measure it under the same conditions and with the same food. That’s the take home message.
What do you think were your biggest limitations at the time you developed the UEM? Were they technological or computational? Can any of these limitations be overcome by the more advanced statistical and computational approaches of today?
I think the biggest limitations are getting people to participate in the study, and whether or not you can generalize what people are doing in the lab with what they would do outside the lab. I don’t think we had any technological or computational limitations.
Recently it seems like a new crop of investigators (me included) is integrating measures of meal microstructure into their studies. What do you think will be the future of the UEM in our attempt to understand mechanisms of human eating behavior?
One thing I would say is that if you are going to measure meal microstructure, it’s important to use units of food that are uninform enough so that the behavioral effects are not contaminated by unit size. Speaking of unit size, I think this is an important consideration that has been understudied. There are a couple studies by Paul Rozin on the effect of unit size (11,12). We did a study with our computerized virtual creation size task (13) where we showed patients who had bariatric surgery pretzels that were either sticks or knots. We also used chocolate kisses and chocolate mini-morsels similar in composition, but differed by unit shape and size. We got big differences depending on the shape of the unit of food that was presented. If you’ve got any students that want to work on a project, I’d be happy to send you those data! It would make a great thesis project!
Thank you so much, Harry, for your time and for the insight you provided to help us put this historical paper into context.
1. Booth DA. How to measure learned control of food or water intake. In: Toates FM, Rowland NE, editors. Feeding and Drinking. 1st edition. Amsterdam: Elsevier Science Publishers B.V. (Biomed, Division); 1987. p. 111-49.
2. Kissileff HR, Klingsberg G, and Van Itallie T.B. Universal eating monitor for continuous recording of solid or liquid consumption in man. Am J Physiol. 1980;338:14-22.
3. Bobroff EM, Kissileff HR. Effects of changes in palatability on food intake and the cumulative food intake curve in man. Appetite. 1986;7:85-96.
4. Kissileff HR, Thornton J, Becker E. A quadratic equation adequately describes the cumulative food intake curve in man. Appetite. 1982;3:255-272.
5. Guss JL, Kissileff HR. Microstructural analyses of human ingestive patterns: from description to mechanistic hypotheses. Neuroscience and Biobehavioral Reviews. 2000;24:261-268.
6. Näsland E, Gutniak M, Skogar S, Rӧssner S, Hellstrӧm PM. Glucagon-like peptide 1 increases the period of postpandial satiety and slows gastric emptying in obese men. Am J Clin Nutr. 1998;68:525-530.
7. Manton S, Magerowski G, Patriarca L, Alonso-Alonso M. The “Smart Dining Table”: Automatic behavioral tracking of a meal with a multi-touch-computer. Frontiers in Psychology. 2016;7:142-51.
8. Kissileff HR, Walsh BT, Kral JG, Cassidy SM. Laboratory studies of eating behavior in women with bulimia. Physiol Behav. 1986;1;38:563-70.
9. Kissileff HR, Pi-Sunyer FX, Thornton J, Smith GP. Am J Clin Nutr. 1981;34(2):154-60.
10. Kissileff HR. The universal eating monitor (UEM): objective assessment of food intake behaviour in the laboratory setting. In J Obes. https://www.nature.com/articles/s41366-022-01089-0
11. Geier AB, Rozin P, Doros G. Unit bias: A new heuristic that helps explain the effect of portion size on food intake. Psychological Science. 2006;17(6):521-5.
12. Rozin P. The meaning of food in our lives: a cross-cultural perspective on eating and well-being. J Nutr Educ Behav. 2005;1;37:S107-12.
13. Hamm JD, Dotel J, Tamura S, Schechter A, Herzog M, Brunstrom JM, Albu J, Pi-Sunyer FX, Laferrere B, Kissileff HR. Reliability and responsiveness of virtual portion size creation tasks: Influences of context, foods, and a bariatric surgery procedure. Physiol Behav. 2020;223:113001.