Thursday 15 January 2015

Chemical coffee

During the past few weeks, I have been at the food coalface, engaging in much hands-on flavour research. These flavours, many of which are soluble in alcohol, provide some of the key memory elements for this time of year. While providing an ad-hoc chemical analysis of one such flavour (Result: 2 parts vodka, 1 part vermouth, a drop of orange peel oil) I read a brief web article on why coffee smells and tastes good, the first part of a series on the chemistry of coffee. There was nothing unique about this article – it explained, correctly, that coffee is a mixture of many hundreds of different compounds, the vast majority of which smell not at all like coffee itself.

But the article was very similar in many ways to other articles, especially in the media or on the internet, that seek to explain preferences for food flavours or other odours in terms of their chemistry. Another recent article addressed the question of preference for the odour of bacon in the same way. Now, of course, in some sense this must be true – the chemical compounds in coffee produce an odour that is highly liked by many people. Without this mixture of compounds there is nothing to like.

This article went on to talk about odour-activity values, which is the idea that some odours are detected at much lower quantities (often parts per billion, or even trillion, in air or solvent) than others and therefore can have a disproportionate impact on what is being smelled. Hence, if we look at coffee, acetone is present at a higher proportion than methyl furanol, but even so most coffees (those served in hospitals and government departments excepted) do not smell like nail-polish remover, and it is this latter component that comes closest to coffee odour. Incidentally, this compound tends to be pleasant smelling.

So far, so (mostly) good. But … there are three underlying assumptions to linking chemistry and preference that have to be examined. The first of these is that some compounds, by virtue of their low thresholds, have a disproportionate impact above threshold. This could be true, but is not necessarily true, because thresholds are poor predictors of psychophysical function – that is, the growth of intensity with increasing concentration. A compound with a low threshold may have a low psychophysical function, meaning that above threshold its intensity (and hence, presumably, impact) grows slowly, or it may have a high psychophysical function. Conversely, there is no theoretical reason why compounds cannot be influential below threshold – this has been shown, for example, by the impact of sub-threshold tastes on odour quality [1]. Indeed, inferring much of anything from odour thresholds is tricky – they can, and do, respond to repeated experience with the odour [2].

The second assumption behind the chemistry of preferences idea is that, in the sort of mixtures that constitute most foods and drinks, individual odour qualities can stick up their hand and be noticed. Thus, in something that smells good, the pleasant odours in that mixture somehow force their way into our attention against the background of those odours that are less pleasant … or vice versa. Certainly, in the latter case, it is easy enough to spoil a pleasant odour by dropping in something nasty.

Here, too, though the science doesn’t back up this assumption. David Laing showed in the early 1990s that in typical mixtures of very many odours, the influence of individual odours rapidly disappears as the mixture approaches and then passes four components. This is true even if you are highly familiar with each of the components and it is also true even if you are an odour expert such as a flavourist or perfumer [3]. The implications of this work are profound and the results of Laing’s well-replicated experiments have met with some resistance from those who aim is to link chemistry of individual components to the smell of the compound made from those components.

In essence, then, a complex, pleasant mixture such as coffee really needs to contain no compounds that are either coffee-like or pleasant. The odour quality of the mixture is a gestalt – both more than the sum of its parts and mostly unpredictable from the contributions of its parts. So, how does this new quality emerge? This is as yet unknown, and to a degree, mysterious. How, for example, can a wine made from sauvignon blanc grapes smell like white wine (a gestalt), sauvignon blanc wine (also a gestalt), and also have notes that smell like cut grass, asparagus, cat’s pee (seldom present … ok, perhaps in New Zealand wines) and so on?

We can say that it is not about chemical interactions in the mixture. Laing’s experiments were done with mixtures of odours not odourants. Hence, the odour mixture property is perceptual - that is, a product of olfactory processing and interpretation in the brain. Beyond that, we know little. But this fact does help to explain how cognitive processes such as attention can facilitate or hinder (to a real, but limited degree) the blending of different chemical compounds into a distinct odour quality [4].

The final assumption underlying a chemistry of preferences is that some odours are inherently liked or disliked and that this is a property of the odourant compound. Like other negatives, it is impossible to prove that this is not the case, at least with some odours. Cadaverine, the product of rotting flesh, could well be a candidate as it appears to be universally repulsive. However, we can show that likes and dislikes for the majority of odours, and certainly food odours, are learned through repeated exposure, or pairing with pleasant tastes/experiences or metabolically useful nutrients [5].

These mechanisms of associative learning are well known, but we really only need to look at variations between individuals or across cultures to see this convincingly demonstrated. Let’s take one example. Sheepmeat (lamb; mutton) is well-accepted in many Western countries, where children begin to be exposed to its odour and flavour from a very young age. In those countries in which eating sheep is not the norm, there is considerable resistance to adding it to the local cuisine. This isn’t based on sympathy for the cute little lambs, but rather the presence in their meat of the compounds skatole and octanoic acid, the former fecal and the latter like a barnyard on a hot day [6]. To you these compounds are the inviting aromas of a roast dinner; to many others, they are ….. somewhat less so. But importantly, with repeated exposure, even these demonstrably initially unpleasant odours can come to get the gastric juices flowing.


The chemically-independent origin of preferences allows too for cognitive influences, explaining how we can be encouraged to like or dislike foods because of what we know about their origin or composition. You are much more likely to find the presence of trichloranisole found in some wines as unpleasant if you know about its origin as a taint from cork. Alternatively, it may be that many people show a liking for the vile vegetable du jour, kale, merely because they feel that it has health benefits.
 ____________________________________________________________

1.         Dalton, P., et al., The merging of the senses: integration of subthreshold taste and smell. Nature Neuroscience, 2000. 3(5): p. 431-432.
2.         Dalton, P., N. Doolittle, and P.A. Breslin, Gender-specific induction of enhanced sensitivity to odours. Nature Neuroscience, 2002. 5(3): p. 199-200.
3.         Laing, D.G. and G.W. Francis, The capacity of humans to identify odours in mixtures. Physiology & Behavior, 1989. 46: p. 809-814.
4.         Le Berre, E., et al., Perceptual processing strategy and exposure influence the perception of odour mixtures. Chemical Senses, 2008. 33: p. 193-199.
5.         Prescott, J., Chemosensory learning and flavour: Perception, preference and intake. Physiology & Behavior, 2012. 107, 553-559
6.         Prescott, J., O. Young, and L. O'Neill, The impact of variations in flavour compounds on meat acceptability: a comparison of Japanese and New Zealand consumers. Food Quality and Preference, 2001. 12(4): p. 257-264.