Designing Pairings — Joining Ingredients by Aroma

Advanced Chapter 11 touched on the sharing of key aromas as the starting point of flavor development. This piece sets that general theory in order and shows that it does not carry over to gelato as it stands. It then tries to translate it into the conditions peculiar to gelato — low temperature and milk fat. This is not physical chemistry (Approfondimento 7) but a question belonging to the craft of design.

The key-aroma hypothesis

The aroma of an ingredient rises as a mixture of many aroma molecules, the volatile compounds. Each ingredient has a few key aromas that characterize its scent. The theory of food pairing puts it this way. Ingredients that share their key aromas at a strength that can actually be perceived tend to sit well together. The shared molecules work as a bridge between them.

Many affinities long known by experience can be accounted for this way. That a certain fruit and a flower go together, or one roasted ingredient and another, is often because both hold a common aroma molecule. The value of the theory lies in showing, at the level of molecules, why they go together.

But the wording of the formulation should be taken exactly. What the theory says is: key aromas, at a strength that can be perceived. It does not hold that the more molecules are shared, the better.

The difference is not small. The cook who brought this theory into an actual kitchen has said plainly that sharing many components is no guarantee of a good match. The molecular profile of a single ingredient is itself extremely complex, and counting numbers will not give the answer. Food pairing, in his view, is a splendid tool to start from, but past that one must search, try, and go on tasting for oneself.

There is a reason on the side of perception as well. The perception of a mixed aroma is not the sum of its single components. Once four or more components are mixed each loses its individuality, and a new aromatic quality arises that no single component gives. Neurophysiological work has confirmed cells that respond to a mixture and not to its components alone. Analyze a strawberry and no molecule that smells of strawberry is found — fruity esters, a coconut-like lactone, caramel, green leaf, a hint of cheese. Their combination becomes strawberry.

So this piece treats sharing as a bridge. A bridge is a footing for crossing; it does not guarantee what lies on the other side. What the theory teaches is where a bridge can be built, not whether it is worth crossing.

Here, though, one has to stop. This theory was built with dishes and drinks at room temperature chiefly in view. Gelato is cold, and mostly holds fat. These two conditions change how a bridge works, from the ground up.

But gelato is cold

For an aroma to be felt, its molecules have to volatilize from the ingredient and reach the nose. As Approfondimento 7 showed, the lower the temperature the more volatilization is held down. In gelato served around −14 °C, much of the aroma that would rise at room temperature never reaches the space above it, the headspace.

This puts one condition on the key-aroma hypothesis. Unless the molecule that makes the bridge still volatilizes at low temperature, no bridge is built. A molecule that joins two ingredients at room temperature will lie asleep inside cold gelato if it is not very volatile. So in pairing for gelato the question is not merely whether many molecules are shared, but whether those shared molecules work in the cold range. An aroma that volatilizes readily, and is felt at a low threshold, can be relied on in the cold.

And there is milk fat

The other condition peculiar to gelato is milk fat. But how it works is not settled by saying that fat holds aroma down. What fat holds down is only what dissolves in fat.

Whether an aroma molecule prefers fat or water is carried by the oil–water partition coefficient K_OW (Approfondimento 7). Above 1 it prefers fat and is hydrophobic; below 1 it prefers water and is polar. And the two behave wholly differently toward fat.

• A hydrophobic aroma gives up the headspace as the fat goes up. The fat globule takes it in. The more hydrophobic the molecule the sharper the effect. Where K_OW is larger by orders of magnitude, a slight rise in fat drops it a great deal.
• A polar aroma hardly moves when the fat goes up. It remains in the water phase, and its scent is settled almost wholly by what stands there.

So milk fat is not a filter that weakens aroma evenly. It is a selective filter, concealing only what will dissolve in fat. To add fat is to lower the strength of the aroma and, at the same time, to rearrange its composition. The same ingredient put into a fat-rich crema and into a fat-free sorbetto will not raise the same thing to the front.

Then what of the account, often repeated, that aroma held by fat is released slowly and stretches the length? This depends on the conditions. The time an aroma takes to escape a fat globule grows in proportion to the square of the globule's radius and to K_OW. Turn that around. In the fine globules of a homogenized mix, even a highly hydrophobic molecule escapes in seconds or less. This never becomes the step that sets the pace of the aroma. Slow release truly tells only where the globules are large, or K_OW is extreme, or the fat phase is too viscous to let molecules move. That fat holds aroma and draws it out is not a truth without conditions.

Crema or sorbetto changes the affinity

Lay the two conditions over each other and an important consequence follows. The same pair of ingredients does not come off the same way in a crema and in a sorbetto.

In a water-based sorbetto there is nowhere for an aroma molecule to hide. With no fat globule to shelter in, a hydrophobic aroma loses its place in the water. Water dislikes hydrophobic molecules, so they cannot stay in the liquid and are pushed out into the gas phase. Which is to say, a fat-free sorbetto is the extreme at which hydrophobic aroma rises most. It is the far end of what Approfondimento 7 showed: the less the fat, the more strongly a hydrophobic aroma comes through. Conversely, a polar aroma is at home with water and would rather stay in the liquid. Get this direction the wrong way round and every later design is inverted.

That citrus and herbs nonetheless take the lead in a real sorbetto is because another axis is at work. Volatility — how readily a molecule turns to gas. The chief aromas of citrus and herbs are light molecules, the head notes that rise first in the language of perfume. A heavy molecule like vanilla sits with the base notes that remain to the last. Polarity and volatility are separate axes. Light aromas come forward in a sorbetto not because they are polar but because they are light.

And the water phase has no fat to buffer it. The molecules that make the bridge join plainly, and they answer directly to the acidity as well (the design of acid is Advanced Chapter 9). To have nowhere to hide means that everything comes out, for better and for worse.

In a crema, which holds fat, the map of the aroma is rewritten. Roasted ingredients, nuts, cacao, vanilla — these aromas, at home in fat, are taken up by the milk fat and reach the gas phase less. Here lies an irony waiting for the designer. The very aromas one wants to lead with in a crema are the ones the crema's fat conceals. The bright polar aromas, meanwhile, are hardly affected and rise as they are.

The irony has a sequel. Books on pairing give a name to the method of deliberately assembling ingredients that hold similar aromas. Chocolate, caramel and coffee laid over one another share the roasted, caramel and nutty families. The view is that this yields a finer complexity than throwing strongly contrasting elements together. But set them side by side and it is plain: that family is, exactly and entirely, the crowd of aromas at home in fat. So the method is the one most tempting in a crema, and the one that works least well in a crema. The method is not at fault. It only needs to be built on the assumption that they will sink, allowing for how far they sink.

As a result, left alone, the aroma of a crema drifts toward the polar side of its own accord. Only the hydrophobic aromas sink and the polar ones do not move, so the ratio between them collapses if nothing is done. Designing a crema, then, is a matter of choosing as the lead an aroma at home in fat. Then comes putting the extra onto that lead alone. Not strengthening everything evenly — adding, to the aroma that sinks, only the amount by which it sinks. When Advanced Chapter 11 said to design a little stronger than it tastes at room temperature, that was an answer to the suppression by cold. In a crema a second answer is laid on top of it, against the selective suppression by fat.

Guides for the design

To put the above into working guides.

• Select the bridge by cold and by fat. Not only that many aroma molecules are shared, but whether they volatilize at the serving temperature, and whether they prefer fat or water. Look not at the width of the bridge but at the conditions under which it stands.
• Settle the base, then think about affinity. With the same two ingredients, a sorbetto can use a bright polar bridge as it is. A crema weakens the bridges at home in fat. The choice of base settles which bridges are usable.
• Add the amount only to the aroma that sinks. The suppression by cold falls on every aroma, but the suppression by fat falls only on the hydrophobic ones. Strengthen a crema evenly and only the polar aromas go over.
• Sharing is not the only road. Against the harmony of joining by shared aroma there is also the contrast of deliberately colliding opposite aromas. The shared-aroma hypothesis is a starting point, not a terminus.

The general theory of shared aroma molecules becomes usable in gelato only after it has passed the two filters of low temperature and milk fat. What remains is where a design of affinity that has come through this translation lands in an actual cup. The whole course of that thinking is traced as a case study in Approfondimento 12, "The Plan for One Cup".