More on aroma perception by sniff and sip

I am again writing here about comparing orthonasal and retronasal olfaction – smelling things through the nostrils, compared with through back of the mouth. If having two modes of smelling is a new concept for you, you might like to refer to an earlier post on the subject where I explain it in more detail. I started to get interested in the subject when I realised that a number of studies on wine aroma perception only looked at orthonasal olfaction, i.e. sniffing the wine, and wondered how much difference it would make if retronasal olfaction was studied instead. One of those studies was the relatively well-known one where a red dye was added to white wine, leading tasters to describe it in terms of red wine aromas (Morrot, Brochet and Dubourdieu, 2001, The color of odors, Brain Lang, 79 , 309-320). Another was a series of studies with the conclusion that we can reliably identify a maximum of four aromas in a multi-aroma mixture. Reading the recently published books by Gordon Shepherd and Jamie Goode has given me a few new insights, which I shall share here.

The most general point is that the smell of a wine is never perceived in isolation. It is part of what we might call the wine’s overall flavour, and it is impossible to totally separate smell from what we perceive through other senses. With retronasal olfaction, we are simultaneously tasting the wine on our tongue, and also feeling it in our mouth as possible astringency and alcoholic heat. Orthonasally, the interference from other senses is maybe less obvious, but we before we sniff the wine we still normally see it, and that affects expectations. This was in fact the point of the Morrot et al experiment mentioned above. Additionally we can also experience chemicals in our nostrils through our sense of touch, for example as alcoholic heat, or the pungency of hydrogen sulphide. These non-smell sense modalities are not only confusable with smells, but they can also affect our sensitivity to true smells

We should also note that we never smell the wine itself, but the volatile molecules that escape from it. In the glass, however much we may swirl and sniff, we have relatively little control over how the volatile molecules escape and reach our nasal cavity. But once in the mouth, the processes acting on the wine can be complex, and vary a lot from time to time, and person to person, even if we are largely unconscious of what is going on – the wine is mixed with saliva, warmed towards body temperature, moved around the mouth to a greater or lesser extent, and allowed to coat the mouth and throat, and its molecules enter the nose when we breathe out.

Then, there are the effects of background smells. You can become so used to a background smell that you no longer notice it. Normally this is an advantage when it comes to wine tasting. It will, for example, filter out the smell on your hands of cigarettes or mildly scented soap. But what if that background smell also happens to be a component of the wine? IN that case your desensitisation to the background smell will affect how you perceive the wine. There is also the possibility of cross-adaption, where one odour affects your sensitivity to other one. The most obvious example is perhaps how the TCA of a corked wine mutes its fruity aromas. If background smells may be important from the environment, retronasal olfaction must be affected by non-wine smells inside your mouth. Apart from the possible remnants of lunch, coffee and cigarettes, there are also the normal odours of your mouth and throat to consider. Unpalatable to think about perhaps, but perfectly natural, and a potential source of desensitisation and cross-adaption when tasting wine.

There have, by the way been a number of studies that have concluded that odour detection thresholds are higher for retronasal that orthonasal olfaction – in other words that we are more sensitive to smells when sniffing through our nostrils. But we need to be careful about drawing false conclusions from these results. In the experiments, the samples are presented to subjects as gases containing an odour, through a tube that goes either to the nostrils or to the back of the mouth, so it is difficult to see what direct relevance these results have to wine tasting.

Finally I would draw your attention to a 2006 study, not discussed in the above-mentioned books, that pursues the question of how many flavour components we can identify in a mixture, but unlike earlier studies that have looked at orthonasal olfaction alone, this one concerns how we perform with a liquid in the mouth, and involves both odourants and tastants – things we detect using both the nose and taste buds. Thus, it is more applicable to the question of how many component we can detect when tasting wine. I won’t give a detailed description of the results here, but just mention that they were very much in line which the findings of the orthonasal olfaction studies. To quote from the authors: “In conclusion, the present study has shown that humans have great difficulty identifying the components of odor–taste mixtures when more than two components are present”.

In summary, I would say that there are many potential reasons why orthonasal and retronasal olfaction might lead to different perceptions in wine tasting, but it is still far from clear how important the differences are in practical terms.

Having your subjective cake and objectifying it

Is wine appreciation is subjective or objective? Professionals can often be difficult to tie down on that issue. For example, they often agree that opinions on wine quality are totally subjective, while at the same time claiming to be objective in their judgements. In that way they can encourage everyone to say what they think about a wine, and yet not lose their special role as experts. The problem with subjectivity is that it can mean everyone’s opinions are equally valid, which then leaves no space for connoisseurship and meaningful dialogue about wine. If there is no right and wrong, what is left to discuss when it comes to wine? So they say. Some people at least. Sometimes.
Personally, while I may occasionally use the language of objectivity for convenience, I am very consistently on the side of subjectivity in this debate. Here is how I see it… The objective aspect of a wine is represented by its chemical and physical structure. Using our senses, mainly smell and taste, and the brain to integrate sensory input, we create there a perceptual model, and then interpret that model to draw conclusions about the wine. In other words, in the sense that it is created within ourselves, flavour is literally subjective. It is not only subjective, but the perception of flavour can vary greatly from person to person according to our genetic makeup. So flavour is totally subjective: it simply does not exist independently of an observer. In addition to this relatively straightforward flavour perception, some of us are also interested in aesthetic aspects of wine such as balance, harmony and elegance, and that adds yet another level of subjectivity. Even if we can agree on the principles of wine quality, those principles will always be rather arbitrary, and our interpretation of the principles subjective. However, unlike others, I do not see subjectivity in wine as a cause for concern. There is still room for wine expertise in my opinion, and with the acceptance of subjectivity I see even more opportunity for interesting discussions about wine quality.

While this strongly subjective stance does not seem to be popular amongst wine people, it is the orthodox view in perceptual science, and I find it difficult to understand how anyone could possibly begin to seriously disagree. However, Barry Smith (of the Institute of Philosophy, School of Advanced Studies, University of London) manages just that. Based on the objective chemistry on the wine, he proposes that there are emergent flavour properties. The flavours might vary with time, but they are objective in the sense that they depend solely on the wine itself. According to him, tasters do not experience the chemical and physical structure of the wine directly: only indirectly, through the flavour properties. Here, there is still subjectivity, but the subjectivity comes into play only when the objective flavours are sensed by a taster to create flavour perceptions. At that point, the person-to-person variations I mentioned above again become relevant, and net result is again a mix of different perceptions. However, according to this theory, they are regarded as more-or-less imperfect perceptions of the objective and true wine flavours. Less imperfect for expert tasters, and more so for novices.

My impression is that Barry Smith has spent a lot of time talking to wine experts, and flavour technologists, and what they say seems to have led him to believe the objective flavours are necessary. Jamie Goode, in his recent book I Taste Red, seems to like the theory, largely because it accommodates the idea of flavours being both objective and subjective, and leaves the traditional role of wine expert intact. However, I see no need for the existence of these objective flavour properties whatsoever. In fact, the only reason I am writing about them in this blog post is because I fear the idea might be gaining traction as a result of Jamie’s new book.

Barry Smith himself, writing in Issue 50 of The World of Fine Wine, identifies what for me is the major problem with his objective flavours: They give us two tough tasks instead of one. The first task is to establish the relationship between the wine chemistry and its emergent objective flavours. The second task is to establish the relationship between the objective flavours and flavour perceptions. Well how are those tasks progressing? Not very well I suggest. Meanwhile however, quite a lot of scientific progress has been made with the single task that replaces Barry Smith’s two, showing how chemicals interact directly with the tongue and nose, and how the resulting signals are processed, and routed to and through the brain, where they are integrated to create flavours.

Whether these mooted objective flavours exist or not matters not a jot as you sit with a glass of wine. However, it does matter to those who write about wine, or buy and sell it. It matters at a philosophical level, and livelihoods based on the myth of objectivity could be at stake. But it does not need to be like that. There is money to be made in the subjective world too. And fun to be had.

(If you would like to check that I have not been misrepresenting Barry Smith, or would simply like to read more about his ideas, in addition to his The World of Fine Wine and sections in Jamie Goode’s book, both mentioned above, you might like to see his essay in the book “Questions of Taste: The Philosophy of Wine”, his Nature Outlook article, or this article I found online. For a scientist’s view of flavour perception I can recommend Gordon Shepherd’s book Neurogastronomy. The same author has also written a more recent book, Neuroenology, which might well be of even more interest. Finally, for a fuller argument in favour of subjectivity, may I again draw your attention to the online version of my article in The World of Fine Wine?)

How smell is like vision, and what that means for wine

Considering the very different impressions that vision and smell make on us, there are surprising similarities in how the two senses are processed before they reach the brain. And it is quite possible that these similarities may throw some light on how we describe the aromas we find in wine.

The olfactory bulb (we actually have two of them) is an elongated protuberance lying close to the underside of the brain, but attached only at the back end. The surface of the human olfactory bulb has on it some 6,000 spherical bundles of cells called glomeruli, each one being connected by neurons to several thousand olfactory receptors in the nasal cavity. When odorous compounds enter the nasal cavity, each glomerulus is activated to a greater or lesser extent, creating a pattern of activity on the surface of the olfactory bulb that is a representation of the odours detected. That pattern can be regarded as analogous to the pattern of activity on the retina of the eye when an image falls on it. In fact the similarity does not stop there, because just as the image on the retina is further processed to facilitate detecting edges and motion, the activity pattern in the glomeruli is also enhanced by subsequent layers of cells in the olfactory bulb. Examples of smell images, reproduced from Gordon M Shepherd’s book Neurogastronomy (reviewed here), can be seen below – click to enlarge and make the text legible.smell images

In the same book, Shepherd proceeds to speculate that the smell images created by glomeruli activity are similar to visual images of faces. He suggests that this explains why smells, like faces, are difficult to describe in words but relatively easy to recognise. As a result, if asked to describe a smell we need to resort to comparisons with the smells of well-known objects. Also, neither smells nor faces are processed as the sum of distinctive component parts – we tend to recognise both of them holistically, not so much by the detail as by a general impression, and the relation between the parts of the image. Only occasionally can we recognise a face if we only see a small part of it, and usually only for faces we are very familiar with.

This speculation of Shepherd’s can be plausibly taken even further, and related to how we recognise and describe wines. Regardless of whether we are nosing a complex wine or sniffing a single chemical compound, at one level in our perceptual system the result is a glomeruli smell image. I would propose that, in the case of wine, certain aspects of that smell image may remind us of the smell images of other objects – blackcurrant maybe, or lemon – which then become the descriptors we use for the wine. In some cases, the aspects of the smell image that cause us to identify other objects in wine may arise from chemical compounds in common, but this need not necessarily be the case and similarities might be coincidental. The aspects in common may be as simple as discrete fragments of the smell image, or possibly with their root in common relationships between different parts of each image. To continue with the face analogy, the identifying of blackcurrant in a wine could be like saying that a baby’s face has his grandfather’s eyes – the eyes need not be identical, but there is however something that seems somehow similar, perhaps the relationship between the eyes and the nose. Something else that has a counterpart in wine is the idea that if we are very familiar with a face is it easier to recognise it from a partial image. In a smell image of wine, presumably the types of fruit etc. we may recognise in it are only partially represented, and that could explain why it we are more likely to recognise the aromas we are more familiar with, either from the experience of the actual fruits or from other wines.

I totally accept that most of this is speculation on the part of Shepherd and me, but nevertheless I think how we experience and describe wines is consistent with the idea of smell image recognition, and an interesting way of conceptualising it. Only time and more research will be able to refute or support these ideas.

Gestalt perception of flavour

coffee_cuppingGive your Man On The Street a glass of wine and ask him how it tastes, and you could well get the response “it tastes like wine”. Nothing wrong with that – it is what might be described as a gestalt perception. Wine contains hundreds (if not thousands) of chemical compounds, many of which will individually give rise to different flavours, but overall the impression is clearly recognised as wine. In the same way, when someone is shown a chair they will immediately recognise it as a chair – they won’t say there are four near-vertical sticks of wood supporting a horizontal square board etc.

It is only people who are relatively skilled in wine appreciation that will dive into a more detailed description like “blackcurrant with a hint of coffee on the finish” – and if we are honest “it tastes like wine” is arguably more useful in many circumstances. Also note that, while we might be feeling superior for spotting the nuances in our glass, we are saying one of the flavours is coffee. We have an idea of a generic coffee flavour, in the same way as many have a generic wine flavour. But coffee is another complex drink that experts analyse and describe in terms of other flavours. Worse than that: wine is a descriptor that is sometimes used in coffee tasting notes, in the same way that coffee is used for wines. So in an extreme case we could have a wine that happens to taste like a coffee that tastes of wine!

Joking apart, the idea that the human perceptual system can sometimes regard aromas that are chemically complex as a single entity, and sometimes analyse them into component parts, is an interesting one. I mentioned in an earlier post that, when considering the number of identifiable aromas in a wine, a chemically complex recognisable aroma behaves like an aroma that is due to a single chemical compound. In fact, the researchers reference other work showing that the two aroma types seem to act in very similar way even at a level as low as the olfactory bulb. But surely that cannot tell the full story. It is undeniable that, for some at least, wine does not merely taste of wine. Neither does coffee merely taste of coffee.