Learned food preferences are thought to be a result of repeated associations between food sensory cues, such as flavour, and post-ingestive consequences. As a food is eaten, the sensory characteristics of the food item becomes associated with the resulting physiological consequences. Consumption of foods that lead to positive responses result in an increase in preference for and acceptance of that flavour. This form of learning has been demonstrated in numerous laboratory studies with animals involving the pairing of consumption of a novel neutral flavour (CS) with a nutritive solution (US) either added directly to the novel food or administered through intra-gastric infusion following
ingestion. The resulting association is referred to as flavour nutrient learning (FNL). Some
evidence suggests that even mildly aversive flavours can become associated with positive post-oral consequences, resulting in increased preference for and acceptance of a
previously disliked flavour (Rozin & Kennel, 1983). The same learning mechanism could be responsible for the development of aversions to foods. Flavours which become associated with negative post-ingestive consequences such as gastro-intestinal malaise will be
avoided and strongly disliked often resulting in a disgust response from individuals that have developed this association (Burnstein, 1998).
Recent studies involving rats, provides evidence that not only do preferences develop through FNL but that the acquisition of these preferences can develop very rapidly
(Ackroff, Dym, Yiin, & Sclafani, 2009; Revelle & Warwick, 2009). In a study examining how quickly rats acquired flavour preferences for flavours paired with the post-oral effects of glucose, Ackroff and colleagues found that the effect of glucose produced a learned
preference after just one trial (Ackroff et al., 2009). A study by Revelle and Warwick (2009) revealed similar effects using sucrose, with a learned preference being observed after only two conditioning trials. Their study also demonstrated that pairing novel flavours with carbohydrates, such as sucrose and glucose, resulted in a more rapid acquisition of preference than pairing flavours with fat, which took at least 6 conditioning trials. However, while flavour-nutrient learning is slower when associations are based on
pairings with fat, there is no difference in the level of conditioning once the association has been learnt (Revelle and Warwick, 2009). While both of these studies demonstrate the development of flavour-nutrient associations in adult rats, studies have also evidenced FNL in rat pups (Myers & Hall, 1998) as well as their propensity to develop these associations pre-weaning (Myers, Ferris, & Sclafani, 2005).
In addition to the speed at which flavour-nutrient associations can be learned, a study by Yiin, Dwyer and Sclafani (2005) showed that preferences developed through FNL show particular resistance to extinction. Interestingly a later study by Dwyer et al. (2009) found that flavour nutrient associations remained even when extinction of the conditioned hedonic reactions to flavour cues had occurred. This apparent resistance to extinction exhibited by preferences resulting from FNL has significant implications regarding the importance of FNL in the development of valuable flavour preferences and long-term healthy eating behaviours.
There are very few successful FNL studies involving human participants. Furthermore studies that successfully demonstrate FNL in adults are scarce. Brunstrom (2005) argues that this reflects the ‘plasticity’ of young children who are most responsive to
physiological cues. He proposes that it is during this early stage that the majority of dietary learning occurs making infancy a critical period in food preference development. A number of human studies have examined the role of hunger as a possible influencing factor in FNL and whether an individual’s energy requirements effect the associations they make between flavours and post-ingestive consequences. Appleton et al. (2006) paired novel flavoured yoghurts with two levels of energy density and asked participants to consume the yoghurts while in two states of energy requirement- low and high.
Participants that consumed the novel yoghurts in a state of high energy requirement were found to develop a liking for these flavours and this liking was found to increase. Liking for flavours consumed in a state of low energy requirement were not found to increase. Moreover rated pleasantness of yoghurt flavour when paired with high energy content
and consumed in a state of high energy requirement was greater after the conditioning period even when tested in a state of low energy requirement. Similar results were found in a study involving carbonated drinks paired with a sweetener. Participants that
experienced both conditioning and testing in a hungry state had significantly increased liking of this drink when compared to those conditioned and tested sated (Mobini,
Chambers, & Yeomans, 2007). Development of flavour preference in these cases is a direct result of associations formed between the flavour of the yoghurt or drink and the positive post-ingestive effects.
Evidence from animal and adult research suggests that FNL might be an effective strategy for promoting intake of target foods in children, however research in children is scarce. A study by Johnson et al. (1991) investigated the effect of energy density on children’s liking for yoghurt drinks. Children received eight conditioning trials with either yoghurt drinks of low or high fat content. Following conditioning children’s preferences were found to have increased for the high fat yoghurt drink but not for the low evergy version suggesting FNL had taken place. A subsequent study by Kern et al. (1993) replicated this study adding a mere exposure condition so that comparisons could be made between the effects of FNL and those of simple repeated exposure. Again the results showed significant increase in preference for the high energy yoghurt drinks in the high energy condition, however those in the mere exposure group increased liking for both high fat and low fat versions of the yoghurt drink. In addition Kern and colleagues found that the effects of FNL were reduced by satiety but that the increase in preference brought about by mere exposure were unaffected. This allowed Kern et al. (1993) to conclude that FNL produced by high fat content may contribute to children’s development of preferences for high fat foods but also suggests that in terms of effectiveness, mere exposure is highly effective in increasing children’s liking of target foods.
This review found only one study that has investigated the effectiveness of FNL in
developing preferences for vegetable flavours. Recruiting children aged between 7 and 8 years, Zeinstra et al. (2009) asked participants to consume fresh vegetable juices of two
energy levels. A high energy version was achieved by pairing pure vegetable juice with forty grams of maltodextrin producing 150kcal difference between the low and high energy juices. Zeinstra et al. were unable to find any evidence for FNL with preference for and consumption of juices remaining unchanged from pre-test to testing post-
conditioning. They argued that this could be accounted for by the low intake of the vegetable juices during conditioning and concluded that this was likely to be a
consequence of the high level of taste intensity of the juices and the possibility of the flavours becoming aversive past a certain level of intake. A recent study by Boulhal et al. (2010) examined the effect of adding sugar, salt and fat to foods and children’s
subsequent intake. They found that neither the addition of sugar or fat, both of which would increase the energy density, had any positive effect on intake of the foods. 1.3.1.i Flavour Nutrient Learning and Conditioned Satiety
In much the same way as learning can determine what foods are consumed it also plays an important role in determining when and how much of these foods are eaten.
Traditionally theories around food intake have focussed on biological cues and internal signals for the initation and termination of feeding. However, research has shown that through repeated experiences with food, associations can also be formed with cues in the environment and that these external cues can exert control over eating behaviour. A review by Birch (1987b) suggests that these ‘learned controls’ can be observed as early as during pre-school years in humans.
As well as influencing liking and acceptance of a food, the post-ingestive consequences following consumption can also influence the quantity of the food that is consumed. Repeated experiences with a food mean associations develop between the sensory characteristics of that food and the experienced satiety following ingestion. These associations allow individuals to anticipate how ‘full’ they expect to be after eating and help to influence the portion sizes they select and overall food intake.
During infancy children rely almost completely upon their internal biological signals, specifically those indicating energy depletion, to initiate and terminate feeding. Studies have shown that even very young children are sensitive to energy density cues and are capable of regulating their own energy intake, reducing intake when consuming formulas with a high energy density and increasing intake of low energy formulas (Fomon, 1974, cited in Birch and Deysher, 1986). Several studies have demonstrated that children receiving low or high energy preloads prior to testing are able to compensate by increasing or reducing their subsequent food intake (Birch & Deysher, 1985; Birch & Deysher, 1986; Birch, McPhee, Bryant, & Johnson, 1993; Birch, McPheee, Shoba,
Steinberg, & Krehbiel, 1987). However FNL studies have shown that conditioning involving the pairing of specific flavours with a high or low energy density can produce associations that are able to somewhat ‘override’ this self-regulation by allowing the individual to predict how filling they expect a food to be and to adjust their consumption accordingly. For example in the study by Birch and Deysher (1985), they found that the children continued to eat more following the flavour associated with the LED preload than that associated with the HED preload even when the preloads administered were isocaloric during extinction trials.
Unlike the studies using child participants studies of conditioned satiety in adults have shown much more varied results. In a study by Yeomans et al. (2005) participants were given two versions of the same breakfast cereal which differed in flavour and calorie content. Liking for the cereals were measured prior to testing and participants were then permitted to eat as much of each cereal as they liked during the first test session. During the subsequent training days participants received fixed portions of either the high or low energy cereals, alternating on each day. After conditioning trials participants were again allowed to consume as much as they would like of each cereal, receiving differing energy content on each day. Prior to conditioning ad libitum intake of the cereals showed no difference between conditions as both cereals were novel to participants. However, following conditioning, intake of the LED cereal was significantly higher than the more energy dense version suggesting that the LED cereal did not leave participants sufficiently
sated following breakfast and so they had learned to compensate for this by consuming more of this version. However, Yeomans et al. (2005) also report that liking of the low energy cereal had significantly increased following conditioning, another possible explanation for the increased intake contradicting the idea that we learn to develop preferences for high energy foods and regulate intake accordingly. A similar study by Wilkinson and Brunstrom (2009) looked to assess conditioned satiety and ‘fullness expectations’ brought about by a novel dessert of two different energy contents. The results of this study showed that expected satiety did increase in the HED condition but that this did not affect subsequent intake suggesting participants had not learned to regulate their consumption based on energy content. Later studies have suggested that individuals’ regulation of intake is more complex than merely learning to compensate for calorie intake and that liking for a food is as much of a strong predictor of amount
consumed or portion size than how ‘filling’ the food is expected to be (Brunstrom & Shakeshaft, 2009; Yeomans, Gould, Leitch, & Mobini, 2009).
The evidence discussed here suggests the possibility that while young children show a reliable ability to self-regulate intake of food based purely on energy consumed, adults’ control of intake is subject to other influential factors such as liking for the foods being consumed. This is perhaps a result of the many different associations and forms of learning that are acquired regarding food and eating during an individual’s development from childhood to adulthood which are able to influence and disrupt the internal
biological cues to which children are so responsive.