Friday, October 19, 2007

Classifying Raw Ingredients

One of the biggest challenges of modern cuisine is the proliferation of ingredients available to the average cook. Between gourmet supermarkets and ethnic groceries the aspiring chef has access to products from around the world, often from cultures far removed from eir own. What are we to do in the face of all these interesting ingredients with which we have no personal experience? We could shy away, stick with the familiar ingredients that our parents used. But we're not cowards, are we? We want to sail the uncharted vastness of the unusual and exotic, to plunge boldly into the bounty that our flat little world has delivered to us. Why? Well, why not? There's nothing to lose and everything to gain by experimenting. But, to extend our metaphor, we need some sort of a plan so that we don't get lost in the sailing. To that end I propose the following principle: The treatment of an ingredient is largely determined by that ingredient's structure and composition.

It follows from there that we can impose some order on the chaos by classifying ingredients according to their structure and composition. What we're going to do is develop an ontologic inventory of raw ingredients, a taxonomy of foodstuffs if you will, with the idea being that once we've accounted for the categories of things that are we can go on to consider the treatment of each category in depth.

According to our friends at the USDA and the National Public Health Institute of Finland, all foods are made up of the following:

  • Alcohol
  • Carbohydrate
  • Fat
  • Fiber
  • Protein
  • Water
  • Vitamins and minerals
For the time being we're going to ignore alcohol, vitamins, and minerals; alcohol isn't present in raw ingredients, and vitamins and minerals don't generally have much effect on how ingredients are used. Using the USDA and Finnish databases I've put together a spreadsheet of some common ingredients; its not exhaustive, and I suspect that some interesting ingredients have undoubtedly gone unnoticed, but it should prove sufficient for the purposes of developing at least a rough set of categories.

Perusal of my spreadsheet immediately turns up one stark division: vegetable products have fiber, while animal products have none. Duh, right? I'll grant its pedantic, but allow me a little bit of pedantry; I'm laying the groundwork for more subtle distinctions later on. The presence/absence of fiber provides us with a justification for the traditional distinction between plant and animal products. "Animal products" and "plant products" will serve as our most basic categories; we'll proceed by looking at animal products first.

Are there any obvious distinctions among the various animal products in terms of structure or composition? Animals produce meat, a solid, but they also produce milk, a liquid. I think its uncontroversial to assert that solids and liquids should constitute separate categories in our taxonomy, since the difference in state requires different treatments.

In addition to milk animals also produce other liquids: cream, honey, eggs, and blood, some (or all) of which belong in separate categories. Cream is like milk, for the most part, except that is has much more fat. You can do things with cream that are diificult, if not impossible, to do with milk on account of its relatively high fat content. Honey is the only animal product which consists primarily of carbohydrates, so it belongs in a separate category. Eggs, in aggregate, are high in protein and fat, but when you break them down into their constituent parts (yolk and white) you end up with a slightly different story. The yolk of an egg is high in both fat and protein; it actually contains a higher percentage of protein than the white. Egg white, however, are most similar to blood; both are high in protein but have very little fat or carbohydrates, suggesting that they should be lumped together. It remains to be seen whether this division actually makes sense in practice, but we'll accept it for the time being. I believe that we've exhausted the animal-derived liquids at this point, leaving us with a partial taxonomy that looks like this:

  • Animal products
    • Liquids
      • Blood and egg white
      • Cream
      • Egg yolk
      • Honey
      • Milk
      • Whole eggs
    • Solids
  • Plant products
We're starting to make progress, but we still have a way to go. Next up, the animal-derived solids.

Muscle tissue is, by far, the most commonly used solid animal product. Everyday cooking, especially in the West, tends to concern itself with the preparation of muscle in a variety of forms: steak, pork chops, chicken breast, etc. But there's more to an arbitrary animal than just muscle; it will also have viscera (aka "offal" or "organ meats") and a skeleton, all of which can be profitably used in culinary endeavors. Muscles are, technically, just another organ, but their structure is so distinctive and their use so prevalent that they deserve a category of their own.

Let's now take a look, at some length, at the category of muscle, since it's the most important of the animal-derived products. In analyzing this category I am, of course, tempted to say "chicken, fish, beef, pork"; these traditional divisions are not, after all, without some basis in fact. But arguing against these traditional divisions is the fact that beef and pork aren't all that different when compared to chicken. The lot of them, chicken, beef, and pork, look similar when compared to fish. And where do more exotic meats like ostrich and alligator fit into the usual scheme?

Raw beef muscle, even tenderloin, is tough and hard to chew, but high-quality raw fish such as toro (fatty tuna, for all you non-sushi-eating folks) can be tender like butter. This is because, as outlined in an excellent article by L. E. Lampila, beef muscle and fish muscle have much different structures. At the macroscopic level beef muscle, in common with that of most terrestrial animals, forms elogated bundles that attach to the bone via tendon. Fish muscle, on the other hand, occurs not in elongated bundles but rather "W"-shaped units called myotomes stacked end-to-end along the length of the animal. The myotomes provide regular divisions which keep the average length of a muscle fiber relatively short (in the range of a few mm). Contrast this with the beef muscle, one long package without any natural division in which the muscle fibers can be centimeters long. Its easy to pull apart a fish filet at the junctions between the "W"s, contributing mightily to the relative tenderness of fish flesh, but you can't pull off that trick with beef. In addition to the presence/absence of these macroscopic divisions there are also differences in the overall amount of collagen present in beef and fish. Collagen, the connective tissue which holds muscle tissue together, is more prevalent in terrestrial animals than in fish, so a slice of fish is going to be (significantly) less chewy than a slice of beef of the same thickness.

All of which is an excellent justification for separating fish out from the terrestrial animals on the basis of muscle structure. But we're not done yet; we still need to figure out what to do with another popular animal category, shellfish. Should they be grouped with terrestrial animals, with fish, or should they be in a class by themselves?

The culinary category "shellfish" comprises two distinct groups of animals: members of phylum mollusca (clams, oysters, scallops, etc.) and members of phylum arthropoda, subphylum crustacea (shrimp, lobsters, and their ilk). For those of you who've been away from biology class for awhile recall that the phylae are second from the top on the taxonomic chart, right under the kingdoms. With such a distance between the two groups it could easily be the case that they have markedly different muscle structure, meriting different treatment.

I haven't had a whole lot of luck as far as finding information on the macroscopic structure of either crustaceans or molluscs, so I'm going to have to go on personal experience here. Shellfish muscle is sufficiently different from both fish and terrestrial animals to merit its own category in our grand hierarchy. As far as separating out crustaceans and mollusks goes we run into a bit of a dilemma. There's no doubt that oysters are different from shrimp, but both scallops and shrimp are amenable to the same treatment in many cases. That seems to be a distinction which arises from the chosen method of consumption: we eat whole oysters, but we generally only eat the adductor muscle of scallops. If oysters were bigger we might treat them like scallops, and if scallops were smaller we might treat them like oysters. So it seems that we should maintain shellfish as a category of its own and save further elaboration, if necessary, for later.

Ok, let's take a step back for a moment and see what our inventory looks like:

  • Animal products
    • Liquids
      • Blood and egg white
      • Cream
      • Egg yolk
      • Honey
      • Milk
      • Whole eggs
    • Solids
      • Muscle
        • Fish
        • Shellfish
        • Terrestrial animals
      • Viscera
      • Bones
  • Plant products
I think we've come to a good stopping point as far as muscles are concerned; go any further and the distinctions become more of degree than kind. Some muscles are fattier than others, some have more collagen, etc., all of which we can take into consideration when we talk about how the muscles are prepared.

I don't have much to say about bones or viscera. Bones are bones; the collagen they contain is useful for making stock. Chicken bones make your stock taste like chicken, beef bones make your stock taste like beef, etc. Since animals are all mostly the same on the inside there's a finite number of different types of viscera to think about; it'll be easier to solve that problem by exhaustion at a later date. And that, I believe, wraps up the animal products... on to the plants.

How are we going to tackle the plants? There's so many different kinds of plants exhibiting so much variation its hard to know where to start. With such a diversity in regards to form there aren't the same kinds of obvious natural divisions that helped us categorize the animals products. Without such natural divisions I think its appropriate to start imposing some of our own by examining the composition of various plant products.

Let's start by looking at water content. I sorted all the plant products in my spreadsheet by how much water they contain and then plotted the resulting, giving me the following:

Interesting, yes? Almost all of the plant products are either >70% or <30% water; the funny one hanging out in the middle is chestnuts. This bi-modal distribution gives us a natural division to work with.

With one exception (dates, which we'll deal with later) the low-water items are exclusively seeds in the botanical sense. From a culinary standpoint we commonly divide them up into "seeds", "nuts", and "grains", though it remains to be seen at this point whether those divisions are warranted. Looking at the fat content of items in this category there is again a bi-modal distribution, this time with respect to fat. Those items which we typically regard as "grains" are relatively low in fat content, while those items we regard as "nuts" and "seeds" are rather high in fat content. The distinction between "nut" and "seed" seems to be somewhat arbitrary though; for example, sunflower seeds and cashews have a similar composition. So it would seem that, at this point, we can justify two categories, "grains" and "nuts/seeds".

We have a pretty good handle on plants products with low water content, but we obviously still need to work on those with relatively high water content. This is a large category containing, I would imagine, the majority of plant products. How best, then, to start slicing this huge, amorphous mass into something more manageable?

Is there anything to be gained by looking at relative carbohydrate content? Here's a graph of all of the products with high water content; note that I've included dates and chestnuts as well:

The divisions here are less obvious; carbohydrate content is more of a continuum than water content, but there is a dicontinuity around %14 which might be meaningful.

The cluster of high-carbohydrate products to the right of the discontinuity contains a lot of what are typically regarded as "starchy" vegetables, things like corn and potatoes. But the group also includes grapes, rose hips, bananas, garlic, and water chestnuts, a motley assortment if I've ever seen one. Potatoes and grapes may have the same sugar content, but they're little alike in texture or traditional treatment. What accounts for this difference?

An obvious distinction is the type of carbohydrate present in each plant. In a potato the carbohydrates are stored mostly in the form of starch, while in a grape the carbohydrates are divided roughly 50/50 between glucose and fructose. Starch is a very large, branching polymer which is responsible for the characteristic texture of plants like potato and yucca. Glucose and fructose, on the other hand, are dinky little, single-carbon-ring monomers. They don't really provide much in the way of structure. The culinary treatment of the two classes of foods is different as well: foods containing fructose/glucose are often eaten raw, while high-starch foods are more pleasant (and more digestible) when cooked. These considerations justify the creation of a category "starchy vegetables" to separate out those products which contain a high percentage of starch\. I'm going to lump one of our odd products, chestnuts, in with this group on this basis. We should also note at this point that starchy/sugary is not a strict dichotomy: a plantain is different from a banana because the former is mostly starch and the later is mostly simple sugars, but both products contain some of each.

Now, before we go any farther, I want to spend a moment talking about dates. They're hanging out at the far right of the carbohydrate graph, way up there at ~%63 by weight, annoying me with their refusal to fall into line. What we want to find out is whether dates are an anomaly or if they represent a class of super-sweet plant products.

The date palm (Phoenix dactylifera) is a member of Arecales, an order that contains just a single family, Palmae. So right away we know that we're potentially dealing with an aberration, since we've got a single plant family sitting out in taxonomic space all by its lonesome. Furthermore, when you look up the genus Pheonix you find out that its hanging out by itself as well, so its unlikely that there are closely-related genera which produce date-like fruits. Straying fslightly further afield we find that there are a number of palms producing such useful products as oil, betel, and coconuts, but none of these as is high in sugar content as the date. Additional confirmation that dates are unique is given by the USDA's list of food products ranked by carbohydrate content; the only item with more carbs than a raw date is sweetened, condensed milk. Looking down the list the next sweetest, raw plant product is mangos, coming is at about a quarter of the carb content of dates. So I think its justified to put dates into their own category.

Consideration of dates, however, leads us to another category of plant products worth considering: saps. Many palms, when tapped, yield a sweet (10-20% sugar) liquid often referred to as "date honey" or"palm syrup". But palms are not the only trees which produce a noteworthy sap; for example, the maple tree produces sap which is boiled down to produce maple syrup. There are also gums/resins such as pine and mastic which are used as flavorings. These products also deserve their own category since they're liquids rather than solids.

We're developing quite an interesting hierarchy; here's what it looks like so far:

  • Animal products
    • Liquids
      • Blood and egg white
      • Cream
      • Egg yolk
      • Honey
      • Milk
      • Whole eggs
    • Solids
      • Muscle
        • Fish
        • Shellfish
        • Terrestrial animals
      • Viscera
      • Bones
  • Plant products
    • Liquids
      • Saps/resins
    • Solids
      • Dates
      • Grains
      • High-starch vegetables
      • Nuts/seeds
      • Everything else
At this point we've considered water and carbohydrate content so, being the systematic folks that we are, we should consider protein content next.

There aren't any big surprises with respect to protein content. At the top of the list are a bunch of different types of beans though, interestingly enough, soybeans don't top the list despite their much-touted reputation. The humble lentil, of all things, has the highest protein content of them all. Once you put the beans into a separate category you are left, once again, with a big, undifferentiated stretch of vegetables.

Moving on, let's look at fat content. As a general rule vegetables are fairly low in fat barring, of course, some notable exceptions. We've already singled out nuts and seeds for their low water content and high fat content. But we've got another statistical outlier, the avocado, which needs to be dealt with. Avocados are ~19.5% fat, but what's most unusual is that the fat isn't locked up in a kernel but rather present in the flesh of the fruit, leading to such colloquial names as "vegetable butter" or "butter pear". We're going to play the same game we played with dates and see if the avocado is a freak of nature or if it's a representative of a larger class of plant products.

The avocado is a member of Lauraceae, which contains other culinarily useful plants such as sassafras and cinnamon. Most members of this genera seem to produce small fruits which don't look particularly appetizing, though apparently a "medicinal aromatic fat" can be obtained from the fruit of the bay laurel tree. This suggests that other members of the laurel family produce fruits with high fat content as well, though actually getting your hands on any might not be worth the effort.

As long as we're on the topic, are there any other high-fat plant products that might have escaped our notice? The only one which comes to mind which hasn't already been categorized is the olive; olive flesh has a high fat content. Its a toss-up whether olives should be included in the same category as avocados, since avocados are usually eaten raw (or with minimal processing) while olives are almost universally pickled prior to being eaten. After consideration, however, it seems reasonable to create a category called "high fat vegetables" to hold these and any other veg we might find with similar characteristics.

At this point we've looked at all of the major constituents, but we're still left with a large "everything else" category. After reviewing the remaining items in my spreadsheet the following categories have suggested themselves:

  • Alliums
  • Cruciferous vegetables
  • Fruits
  • Fungus
  • Leafy vegetables
Let's review each of these and see if they merit inclusion in the final hierarchy.

Alliums such as garlic and onions definitely deserve a category of their own. This family of plants produces the sulfoxide alliin which, when combined with the enzyme allinase, produces their characteristic odor. The various species produce different types of alliin and allinase, thus accounting for the subtle (and not-so-subtle) differences in terms of flavor and aroma.

"Cruciferous vegetables" is a botanical category that doesn't seem to mean much from a culinary point of view. It includes plants such as broccoli, cauliflower, and cabbage, all of which are similar enough to possibly merit their own category. But, in addition to the examples which are usually cited, it also includes watercress and rutabagas. Watercress, broccoli, and rutabaga all merit markedly different treatments, so the cruciferous vegetables don't really make a good category

Fungus (mushrooms and the like) are in a kingdom all by themselves; they're significantly different from other vegetables in that regard. They're also a good source of glutamate, one of the reasons why everything tastes better with mushrooms. This suggests to me that "fungus" is a useful category.

Leafy vegetables seem self-justifying, since the leaf is a common natural structure that lends itself to a particular set of uses.

The distinction between "fruit" and "not fruit", in a culinary as opposed to botanical sense, seems to be somewhat arbitrary. It doesn't correlate well with sugar content; carrots have more carbohydrates than lemons or cranberries, but the former is a vegetable while the latter are fruits. Nor is "fruitness" analagous to "juicyness"; a bannana is a fruit, but it not particularly juicy. It doesn't necessarily even have anything to do with a particular class of tastes; cucumbers taste a lot like watermelons, but one is a vegetable and one is a fruit. The concept of "fruit" seems to correlate most closely with a sweet taste but, as in the case of the cranberry and lemon, even that seems to be negotiable. So the traditional category "fruit", while somewhat useful, doesn't hold up well to scrutiny.

We're almost there but, in thinking about the alliums, I realized that I'd excluded two other groups of plants with interesting chemical properties, capsicum and zingiberaceae. Capsicum is, of course, the hot peppers. Zingiberaceae includes all sorts of plants which are typically used as seasonings such as ginger, cardamom, and tumeric. More broadly, however, they represent a conceptual class of plants that is used to add flavor rather than substance. We'll call this category "herbs/spices" and leave it at that.

I think we're done. Here's our final inventory:

  • Animal products
    • Liquids
      • Blood and egg white
      • Cream
      • Egg yolk
      • Honey
      • Milk
      • Whole eggs
    • Solids
      • Muscle
        • Fish
        • Shellfish
        • Terrestrial animals
      • Viscera
      • Bones
  • Plant products
    • Liquids
      • Saps/resins
    • Solids
      • Alliums
      • Dates
      • Fungus
      • Grains
      • Herbs/spices
      • High-fat vegetables
      • High-starch vegetables
      • Leafy vegetables
      • Nuts/seeds
      • "What's Left"
"What's Left" is obviously a big category; if anyone can think of any substantial divisions which I've neglected please let me know. As a whole, however, I think that the list we've developed serves as a good starting point for further discussion. As I have time I'm going to write about each category in more depth. Stay tuned, same bat-time, same bat-channel.

No comments: