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


Our Knowledge About Human Dietary Learning

Date: September 7th, 2006
Speaker’s Name and Affiliation: Jeff Brunstrom, University of Bristol
Title: “What do we know (and need to know) about human dietary learning”
Dr. Jeff Brunstrom began our 2006 seminar series with a discussion of human food learning, what we currently know, and what we need to know. Animals have long demonstrated the ability to show preferences for a flavor that has been “conditioned” or matched with a nutrient stimulus. This flavor-nutrient learning has also been demonstrated to some extent in humans, but an almost equal number of studies fail to find such an effect.

In addition to flavor-nutrient learning, flavor-flavor learning is a more immediate process whereby a flavor is presented along with a flavor that already has a strong preference in the animal or human.

In flavor-flavor learning, a novel flavor is the conditioned stimulus (CS) while the already liked flavor is the unconditioned stimulus (US). The result of this learning is that the subject forms a preference for the stimulus that is similar to the already liked US.

Dr. Brunstrom hypothesized in his talk that flavor-flavor learning might be particularly important to humans, who often consume meals with a variety of flavors presented at once.

Part I of the talk addressed the issue of flavor-preference learning, and asked whether or not there are variations across humans. To begin to answer this question, data were presented on an experiment in which participants were trained with a flavored dessert (mandarin or chocolate).

One of these was formulated with a high energy density (1182 kJ) and the other was formulated with a low energy density (227 kJ). Following six days of training, participants were tested with both mandarin and chocolate presented in a medium energy dessert. Results suggest that liking ratings for the high and low-energy desserts do not differ across all subjects, but if you separate participants into unrestrained and restrained, the results tell a different story.

Unrestrained eaters show increased liking for the flavor presented in a high-energy formulation during the test phase, while restrained eaters show limited evidence of flavor-nutrient learning [Brunstrom JM & Mitchell G.L. (Accepted)].

The next experiment presented in the talk asked whether or not restrained eaters show evidence of flavor-flavor learning. In 42 females, 21 with low restraint and 21 with high restraint, 3 different unconditioned stimuli were used, milk chocolate chips, sugar coated puffed rice, and raisins.

Each was paired with one of three novel flavors (CS), apple and ginger, red zinger, and pineapple and coconut. Three contingencies ratios (ratios of presentation of CS to presentation of the US) were used in this experiment, 90%, 50%, and 10%. Unrestrained eaters show a preference for the flavor paired with the US 90% of the time, where as the unrestrained eaters fail to show a preference for any of the flavors. Again, what this suggests is that restrained eaters show a decreased ability to learn from food cues that are presented to them in an experimental context. Whether or not this might impact on their food intake control is not known.

In summarizing this portion of the talk, Dr. Brunstrom discussed the issue of awareness in dietary learning behavior. This interesting theoretical issue has not been addressed by many research laboratories, so little is known about its role in dietary learning. In certain contexts, the speaker posits that lack of awareness may enhance dietary learning.

Part II of the talk discussed how dietary learning may affect energy intake. In many situations, people simply eat what they find on their plate, and thus portion size becomes an important cue for determining intake at a meal. Measuring differences in portion size selection might provide an important window into learning about human food intake control, and Dr. Brunstrom’s lab has been studying this behavior with a unique approach. In some experiments, participants are asked to “Imagine you are about to prepare a meal. Which one of the following portions would you select?”

Accompanying this question, participants are shown a variety of pictures of portion sizes of test foods. The question asked and foods shown can be changed depending on the purpose of the experiment, but the tool might provide useful information for researchers wishing to learn about individual differences in food cue reactivity.

There are, of course, many other factors that influence energy intake. Dietary restraint, palatability, and cost are just a few of these. Expectations about how filling a food will be is another factor that has received little attention from the literature. In order to assess individual differences in expectations about the satiating potential of food, Dr. Brunstrom has developed several novel techniques.

One example might be to show participants a series of pictures of different foods, all of which contain the same calories. Participants might be asked to rate how filling each food would be on an analog scale. However, when comparing intra-individual ratings from one week to another, it appears that these ratings are not that reliable. One suggestion is that various problems with ratings can be addressed using procedures based on classical psychophysics. Dr Brunstrom went on to report various results illustrating how this might work in practice.

In summary, while we have made much progress in understanding how humans regulate their energy intake, we still have a long way to go. Individuals differ in their sensitivity to dietary learning paradigms. In particular an important modulating variable appears to be dietary restraint. Issues relating to contingency awareness are likely to be critical to our understanding of dietary learning. Despite this, the role of awareness in basic laboratory paradigms has remained unclear.

Finally, rather than influencing intake directly, novel approaches to the study of cue reactivity led Dr Brunstrom to speculate that flavor-nutrient associations might mediate expectations about the satiating quality of food.


Q. What is the consistency of the desserts that you use in your test-meal?
A. They can be scooped with a spoon. They are semisolid.

Q. But nutritionally, are they identical?
A. Yes.

Q. They don’t require chewing?
A. No, minimal chewing is required.

Q. Was the low energy product made with a non-nutritive sweetener?
A. No, just maltodextran.

Q. If you just used one flavor and varied the energy density would you expect the subjects to be able to discrimintate?
A. No, we wouldn’t.

Q. What do you mean by contingency ratio?
A. The contingency ratio is the ratio of the presentation of the CS (conditioned stimulus) followed by the presentation of the US (unconditioned stimulus).

Q. Did you collect baseline preference measures for your subjects?
A. I looked at the liking ratings for subjects who were given the US, but didn’t find any differences between restrained and unrestrained eaters.

Q. Have you tried (in humans) to condition learning to something that is unpreferred, in general?
A. That is a very interesting question, but I’m not sure that it has been tried?

Q. Is your subject pool mixed in gender?
A. That is a good point. All the subjects I am presenting on are female?

Q. Do you have your own research question to distinguish between dieting and restraint?
A. Yes, we asked if subjects were currently dieting to lose weight.

Q. Did you ever ask your subjects which food has the most calories (not the food they preferred the most)?
A. Yes, and restrained eaters tend to distinguish better?

Q. What scale do you use to rate pleasantness?
A. I think the question we ask is “How pleasant is this item?” and subjects respond on a 100 mm analog scale anchored with “not at all” and “extremely.”

Q. If you did the same rating twice, how similar would individuals rate within-subjects?
A. We didn’t do that, mainly because of the cost and feasibility, so I’m not sure.

Q. Do you have any idea how long it takes before an individual has a conscious idea of what types of food they like?
A. No, not yet, but I will present some data about that issue in the future.

Comment: There is also a hedonic contrast literature, and it suggests that with greater awareness, you have greater learning.
A. Great, I’d be interested to take a look at that.

Q. Have you looked to see if sensitivity to bitter tastants, such as PROP, affects the ability to learn flavor preference?
A. We haven’t at this point, but that adds an additional variable, and the results might be very difficult to analyze.

Q. Have you only used sweet tastes in your flavor learning experiments? Do you get the same results if you use a stimulus such as a soup?
A. Certainly, the appropriateness of the food used may have an effect.

Q. What you are describing in terms of cue-reactivity is actually disinhibition? Have you tested that?
A. You are correct, and I could have that in the list of important variables as well.

Q. Can you speculate whether your methodology might clarify some of the Barbara Rolls studies where she has found that a mixed food or meal is more filling than drinking water alone, along with the food (without the water)?
A. I do believe that adding water to foods might make it more difficult for participants to learn about the energy density of the food, so this may partially explain those results.