HOW WE LEARN TO LIKE FLAVOURS

It is important to spend a little time defining the word ‘flavour’. It is a simple word that we all use in everyday language, but we usually use it incorrectly or at best vaguely (Gibson, 1966). We taste our food to see if we like the flavour, we talk of wines with the taste of oak, raspberries or spice, and of course we all know where we experience the flavour of our food – in our mouth. The reality is that much of our information about food derives from our sense of smell detected high up in our nose; it is estimated that as much as 80% of the information about our food depends on olfaction (Murphy et al., 1977). The olfactive epithelium is in fact a direct extension of the brain and is the only part of it that protrudes from the skull directly into the outside world.

Even though we are not conscious of its location (behind the eyes and midway between the ears), it is from the receptor cells in this organ that signals are sent to the brain giving information about the changing pattern of volatile molecules being released from food during eating. The olfactive receptor cells are continually renewed during our lifetime and are among the fastest growing cells in our body. All individuals differ in their olfactive sensitivities and there are well-reported cases of genetically linked anosmias; one of the best known is the differing ability to smell the volatile steroid androstenone (Wysocki et al., 1989). Repeated exposure to an olfactive stimulant can also induce enhanced sensitivity to that odorant so that an individual’s ability to smell and distinguish different volatile compounds can be trained.
It has also been reported (Dalton et al., 2002) that such induction of sensitivity is observed to a much greater extent with women of reproductive age than is the case with men. This supports the many anecdotal reports that women show a much higher sensitivity to environmental odours than men; such heightened sensitivity would probably have major advantages for mother–infant bonding and for the recognition of safe food sources. We may sniff our food before putting it in our mouth as a preliminary and cautious step to checking its quality, but it is during eating, drinking and swallowing that the volatile components of the food are transported in the breath to the olfactive receptors, and this is what provides most of the information about our food, which we consciously perceive as flavour.
The fact that we cannot ‘taste’ our food when we have a cold (another example of the inaccurate way we use these words) means that the volatile molecules cannot get up into the nose and we lose the olfactive signals. Of course, even with a blocked nose we can in fact still ‘taste’ our food because our sense of taste is truly located in our mouth but limited to just five informational components – sweet, sour, bitter, salty and umami. Our perception of taste is very important to the overall flavour as will become increasingly apparent, but it is the olfactive signal that gives the brain the information that allows it to discriminate and recognise what it is that we are eating. Up until the 1990s, several research groups had studied flavour release from food in the mouth and this work is well reviewed (Overbosch et al., 1991).
However, it was not possible at that time to study the fine details because the analytical instruments available were too slow to record dynamic changes or too insensitive to measure the quantities of flavour molecules being released from food. As already mentioned in the introduction, the breakthrough came a few years ago at the University of Nottingham in England (Linforth et al., 1998) when an atmospheric pressure chemical ionisation mass spectrometer (APCI-MS) was modified to allow it to make real-time analysis of flavour molecules in the breath of a volunteer while eating. Work using this new dynamic analysis of flavour release could be linked to sensory perception.

For the first time the actual release of flavour molecules from food could be related to the conscious impression they created, and through this came the understanding of how the nose and brain transduce the information and give us a conscious impression of it. Most importantly, it was realised that our perception of odour strength of a volatile molecule did not depend on the absolute concentration of it in the breath but rather that it responded to changes in its concentration (Linforth et al., 2000) – a discovery that was to lead to a paradigm shift in understanding how to add flavouring materials more effectively to food; rapid bursts of flavour are more important to perception than a steady rate of release.
However, flavour does not rely solely on taste and olfaction; the mouth also sends information via the trigeminal nerve, one of the main nerves leading from the face to the brain. It is through this that we know whether our food is hot or cold, whether it is spicy (as with chilli or pepper) or cooling (as with menthol), and what texture it has. Our mouths are incredibly good at assessing the textural characteristics of food via our sense of touch, but it is only in recent years that we have come to understand that texture also plays a key role in what we recognise to be flavour.
Using this new ability to measure flavour release, the researchers at Nottingham went on to study how food texture could affect it and hence change its perception. It is well known that above a limiting viscosity the perceived intensity of sweetness and flavour of a thickened drink is reduced as the viscosity of the drink increases (Baines & Morris, 1989). The Nottingham group confirmed this finding (Hollowood et al., 2000), but contrary to expectation, found that the viscosity had no measurable effect on the actual concentration of flavour volatiles that were released to the nose. Irrespective of viscosity, the concentrations of volatile flavour molecules in the breath during consumption of the drinks were identical within the experimental errors.
Clearly, the change in viscosity was affecting the perception of flavour intensity through a mechanism that did not directly involve olfaction. These results could not be fully understood when they were presented at an American Chemical Society (ACS) meeting in March 2000 (Hollowood et al., 2002). At this same meeting, Dr Paul Breslin from the Monell Chemical Senses Center in Philadelphia showed that the sense of taste could directly affect olfactive perception. In this now classical paper, the authors (Dalton et al., 2000) showed that the presence of subliminal sweetness in the mouths of tasters was able to modify the threshold level for their olfactive detection of benzaldehyde.
In other words, a taste signal was shown to be directly modifying an odour signal whereas in the Nottingham experiments it was the texture of the drinks that was affecting odour perception. Whether this latter result was directly through the tactile receptors or because of reduced sweetness perception is still not completely resolved, but since that time we realise that in both sets of experiments what was being described were effects of crossmodal perception. A further important finding came out of the Nottingham work, which concerned the intensity rating of flavour in chewing gum. According to volunteer tasters, the perceived peppermint intensity of a gum was found to reduce to insignificant levels after about 5 min of chewing even though the measured intensity of menthol and menthone on the breath was found to be as high as it had been at the beginning of the test (Davidson et al., 1999).
This research further showed that the perceived intensity of the peppermint flavour did, nevertheless, correlate well with the sweetness of the gum measured by an analysis of sucrose present in the saliva; however, there is nothing inherently sweet about peppermint, the plant is not sweet so our association of its aroma with sweetness is entirely learned and is dependent on the way we use its flavour in our food. The fact that we have learned to associate peppermint aroma with sweetness probably starts from the time we first brush our teeth with sweet, peppermint-flavoured toothpaste. The way in which we learn to associate specific olfactive stimuli with a particular taste is an example of multisensory learning and clearly plays a very important role in our judgement of flavour acceptability. It is only with uncooked food that our accepted standards of normality are decided by nature.
We expect a lemon to be sour and a mango to be sweet, but many volatile flavour molecules are only paired with specific tastes through the cuisines and cooking techniques that create them. Whether we relate the flavour of tea or coffee to sweetness or bitterness is entirely consequent on how we have trained ourselves to accept these drinks. Nor is perception of flavour determined only by olfaction, taste and touch; recent work shows that vision is also involved. Food scientists have long known how much colour can influence our appreciation of food, which is one reason why taste panels are often conducted under lighting conditions to reduce colour bias; however, it seems that colour is also an important factor in flavour learning.
A recent report discusses how a group of experienced wine tasters were asked to describe the quality and characteristics of a selection of white wines, which they did expertly; when the same wines were again presented but now coloured red by the addition of a nonvolatile food dye, the experts were quite unable to correctly identify the same wines and could only describe them in terms of quite inappropriate red wine descriptors (Morot et al., 2001). The fact that we have great difficulty in recognising flavours that are presented in association with inappropriate colours is not just a party trick; it is because our brains have been trained and moulded by experience and consequently have great difficulty in processing mismatched stimuli.

Flavour is therefore a synthesis of all the associated sensory signals received during eating and drinking, including the internally generated emotions associated with the occasion. Gibson (1966) argues that a better word for the composite sensation we call flavour would be palatability. Whatever term is used, it describes what is arguably the most multisensory experience we have, and it should now be obvious why it is so important for the experimental psychologists who are studying multisensory perception to be talking to the flavour scientists who are studying taste and olfaction. It should also be acknowledged, however, that almost 100 years ago the psychologist E. B. Titchener (1909) wrote of the ‘curiously unitary character’ of the senses when eating a peach:

"Think, for instance, of the flavour of a ripe peach. The ethereal odour may be ruled out by holding the nose. The taste components – sweet, bitter, sour – may be identified by special direction of the attention on them. The touch components – the softness and stringiness of the pulp, the puckery feeling of the sour – may be singled out in the same way. Nevertheless, all these factors blend together so intimately that it is hard to give up one’s belief in a peculiar and unanalysable peach flavour."

He was remarkably close to the modern description of a multisensory experience which is learned through our eating experiences wherein sensory signals that cause neurons to fire together on a repeated basis permanently change their neural connectivity, thus moulding the way the brain will process future sensory inputs. Put very simply, when we eat a particular food on a regular basis, our brain will ultimately combine all the signals received from the different senses during each of our eating experiences and create the memory we recognise as the ‘flavour’ of that food. The acceptability of this food is therefore a fundamentally personal thing and will depend very much on how familiar it is to us and our previous histories of eating it.
If we eat a food and its flavour is even slightly different from what we expect, then our brain will tell us to pause and consider what it is that we have in our mouth; if the flavour is associated with a previously bad eating experience or is totally different from anything eaten before, then almost certainly our reaction will be to spit it out. Flavour is Nature’s way of letting food communicate with us in order that we know what it is that we have in our mouth and can decide whether or not we continue to eat it.
Flavour has been described as a psychological construct of the brain by Prescott (1999) and he makes the point that although olfactory signals are detected in the nose, the illusion of olfactory quality of food appears to originate in the mouth. Prescott also proposes that this olfactory illusion (of it being in the mouth) is important because of the high survival value of correctly identifying safe food. He says that ‘since the mouth acts as a gateway to the gut, our chemical senses can be seen as part of a defence system to protect our internal environment’ (Prescott, 1999, 2003).
Some years earlier Gibson (1966) made the point that the sensory qualities within flavours were unlikely to be independent, and that there was more and more evidence showing how we learn to associate the different sensory aspects of flavour and integrate them into a total memory experience. According to Frank and Byram (1988), certain food odours such as strawberry and vanilla enhance sweetness. These authors suggested that perceptual similarity between a taste and an odour might arise through a history of association of these qualities in foods. There is also evidence that the most pronounced multisensory interactions occur when the sensory stimuli are presented close to their threshold levels, exactly the situation reported in the Monell work.
A region of the brain known as the superior colliculus shows little activity when dim lights or quiet sounds are presented in isolation, but its neurons fire at a dramatically increased rate when the two stimuli are presented simultaneously (Stein et al., 2003). Such crossmodal reinforcement of sensory signals close to their threshold levels would clearly have had enormous evolutionary advantages, since the integration of signals from all the senses would allow the recognition of opportunity or danger below the limit of detection of any one sensory channel. However, such integration is not always positive and there are also examples of one sensory input suppressing another (Stein & Meredith, 1993), which in certain circumstances could also have been advantageous, e.g. in focusing concentration on a specific task.

Without going too deeply into all the work which has now been reported on the way the brain can integrate the different sensory inputs it receives, it is clear that from now on we have to think about the perception of flavour as a daily multisensory experience. This new understanding of flavour is less than five years old, and at present there are very few food scientists who think in these terms. Inevitably, we are also brought to a conclusion that since the multimodal perception of flavour is learned, how this comes about will play a determining role in what foods we enjoy eating in later life. Let us now discuss the moments in our lives when we are first exposed to new eating sensations and what it is about them that will influence us to remember them as delicious or disgusting.

By Anthony A. Blake in the book "Flavor Perception", edited by Andrew J. Taylor and Deborah D. Roberts, published 2004 by Blackwell Publishing Ltd UK, excerpts from page 177 to 182. Adapted and illustrated to be posted by Leopoldo Costa.