Please note, this is recovered content from the former website of the New York Obesity Research Center website.


Analysis of Food Cravings

Date: January 24th, 2008
Speaker’s Name, Affiliation: Marcia Pelchat, Monell Chemical Senses
Seminar Title: “Mechanisms of Food Cravings”

Dr. Marcia Pelchat of Monell Chemical Senses spoke at our first seminar of 2008 on “Mechanisms of Food Cravings.” Food cravings are common events reported by the majority of the population at some time in their lives. According to Dr. Pelchat, liking and craving are two distinct events that can be measured behaviorally. Her data suggest that fMRI (functional magnetic resonance imaging) can be used to take a brain image of these distinct events. Observing the brain while individuals are reportedly having food “cravings” can be a useful technique for improved understanding of the mechanisms of how food cravings work.

Liking, wanting, and craving have been defined in the literature as distinct concepts. Kent Berridge has published extensively on the experience of liking vs. wanting, but the definition of “craving” has been somewhat elusive. Dr. Pelchat defined “liking” as the pleasantness or hedonic or evaluative response to a stimulus. Wanting, on the other hand, was defined as the “desire” for a stimulus. “Craving” can be viewed along the same continuum as wanting, and was defined as a very strong or obsessive desire for a stimulus. By this, a “food craving” can be defined as an intense desire to eat a particular food.

One clear example of the dichotomy between liking or pleasure and craving is that humans often do not report cravings for all the foods that they like. In 1993, Robinson & Berridge reported that drug addicts “crave” the drugs they are dependent on, despite the fact that they experience little pleasure when they are on this drug. Pelchat & Schaffer (2000) published similar findings with foods, suggesting that food cravings and liking of pleasure for food are distinct events. Both studies reference a classic study by Heath in 1963, where patients were implanted with electrodes that would allow them to self-stimulate. Patients reported that they felt compelled to continue self-stimulation (desire) despite the fact that they reported that the stimulus was not pleasant! These distinctions are important because they suggest that craving and liking may influence behavior differently, and furthermore, may originate from different brain mechanisms.

The most commonly reported foods that people crave are chocolate and pizza. Other critical characteristics of craved foods are that they are typically highly palatable and may contain a lot of fat or calories. Often, they are “forbidden foods,” or foods that people would not allow them selves to have under typical circumstances. Young women more often report cravings for sweets (60%) and for savory foods (40%), but in older women, these proportions are reversed. Men more often report cravings for savory foods, as do older women.

How are food cravings measured in the laboratory? Food intake does not appear to be a good measure of craving, as most of the foods we eat are not craved foods. Simple self-report methods that have subjects respond to questions about their food cravings and list the types of foods they crave are often helpful ways to make assessments. This has the subject doing a vigilance task – e.g., pressing a button when a dot appears on the screen. The idea is that obsessive thoughts about food interfere with the task and make the reaction times to the dot longer. But it apparently does not work well in practice.

Reinforcement can also be measured to assess participants’ willingness to work for a food. Dr. Pelchat has been using fMRI, or functional magnetic resonance imaging, to capture changes in the brain that occur when subjects are experiencing cravings for foods. Subjects who were placed on a monotonous diet experienced cravings for foods that they liked, and these could easily be turned on and off during fMRI analyses. These analyses demonstrated that three areas of the brain, the hippocampus, insula, and caudate, all show activity during reported food cravings. Although fMRI is a nonverbal measure of food cravings, this measure must still be validated against self-report.

Dr. Halls Dr. Halls
Disappointingly, a lot page full of text, and the result is: MRI shows something when a person is craving a food.


Future studies in Dr. Pelchat’s lab are focusing on how food cravings are learned and determining whether you can produce cravings for foods that are disliked, perhaps even for healthy foods. Preliminary reports suggest that subjects can be conditioned to crave Boost, despite the fact that they do not report liking for this test food. In addition, future studies are also investigating further into the potential mechanisms behind food cravings.


Q. Where would you classify “need” compared to wanting and craving?
A. Needing may make you more likely to have a craving, but it is distinct from a craving.

Q. Do we know where the electrodes were implanted in the experiment done by Heath, 1963?
A. I’m not really sure, but I think the basal forebrain, hypothalamus and cingulate cortex.

Q. For drug addiction, people take a drug to get rid of an unpleasant sensation. How does this relate to food cravings and addiction?
A. I will get to that later.

Q. For the drink (Sustacal) that you served to participants in the study, how did the liking on the final day compare to the liking reported on the first day?
A. It did not go down that much.

Q. Do women and men report craving different things?
A. On average, women crave sweets (60%) and savory (40%). For males, and elderly women, those proportions are reversed.

Q. Are there differences across the menstrual cycle for reported cravings?
A. We have not reported that, but other labs have.

Q. What about women that are hystorectimized at a young age? Do they experience variations in cravings across the cycle?
A. We had an idea to do that for an RO1, but it was not funded.

Q. What about dieting? How does that affect craving?
A. Some find that dieting and restraint increase the reported level of craving.

Q. Does anyone report craving for things like olive oil or butter (straight fats)?
A. Most people crave food combinations, not isolated nutrients.

Q. Have you examined personality characteristics or novelty seeking?
A. We actually haven’t done that yet.

Q. What do you mean when you say you collected information on food cravings? How was that assessed?
A. We have questionnaires that we give about mood, appetite, exercise, and cravings. We ask people “did you have any food cravings in the past 24 hours,” and “if so, what foods were these cravings for.”

Q. During the monotony condition (when participants are drinking Vanilla Sustacal) because the test drink is sweet, did anyone report cravings for the Sustacal?
A. No one reported cravings for “sweets” during the conditions where they got the Sustacal.

Q. When you gave participants the chicken flavored Sustacal, they weren’t salt-deprived, were they?
A. No, they shouldn’t have been. The supplement drink had sufficient amounts of all the micronutrients (sodium included).

Comment: Did you look at BMI as a function of reported cravings? Hashim and VanItallie demonstrated that if you give Sustecal to obese individuals, they will lose weight, but lean subjects will maintain weight on this diet.

Q. If you measure intake, is there a general correlation between intake and the level of craving?
A. That’s a good question. There are some people that do not eat craved foods, but there are individual differences on this factor (across different people).

Q. Do you distinguish between resistance when target food was available versus when they have to drive to get something that is craved?
A. This was a structured interview setting, so we weren’t able to assess that. In older adults, they tend to substitute one food for another, rather than driving to get a food they are craving.

Q. How did you get the information on what people were craving while they were undergoing the fMRI?
A. We interviewed them afterwards. We can’t really monitor them while they are having a craving. According to Dana Small, even when people know they are being monitored, it changes their results. The process of thinking or making a decision can alter the results of the fMRI.

Q. Is there a reason why the data look lateralized?
A. That’s an interesting question. They were not all right handed, but it’s possible the data are lateralized.

Q. Isn’t it common for the sensory stimuli to be lateralized?
A. Other people here probably have much more experience with that than I do.

Q. How many subjects did you have and did you control for BMI?
A. We had 10 in each group (control vs. craving group), split equally by males and females. We did not control for BMI.

Q. Were the subjects fasted?
A. We made very sure that they were not hungry when they came in to do the test, as hunger might increase the likelihood of a craving.

Q. Did you measure dietary restraint?
A. Yes, we used the Dutch Eating Behavior Questionnaire. I like the externality score on that questionnaire better than the TFEQ.

Q. Were all of your subjects na?ve to Boost before the study?
A. We enrolled most people into the study, and did not have a lot of exclusion criteria.

Q. Did some of your participants start drinking Boost after the study?
A. We are trying to call them up and find this out.

Q. If you had two foods that were equally liked, but only had a craving for one of them, which food do you think people would choose if given a choice?
A. I would predict that people would choose the food they liked at the time.
Craving and liking are separate things, and should be able to be measured, but we have not figured out how to do that yet.

Q. Did you screen for smokers?
A. We did not do that. We only screened out people who said that they didn’t have any cravings.