The metabolic fallacy

There is, or so one often hears (as I heard last weekend), one sure-fire, guaranteed, absolutely foolproof way for a fat person to lose weight: eat less. More specifically, first you figure out the average number of calories burned a day, then you eat less than that. You'll burn `excess fat' making up the difference. (Presumably, once you reach your target weight, you resume a normal intake.)

Only problem: this doesn't work. As anyone who either is, or knows someone who is, fat or thin and tried to lose or gain weight knows, `naturally thin' people can eat `all they want' and never gain and ounce, while `naturally fat' people can't keep their weight down longer than about five years.

Maybe you think the former cases -- the thin people -- just have a high metabolism, and the latter cases just are `psychologically weak'. Fat = psychologically weak is certainly a popular narrative in our culture. But that's an odd asymmetry: thin people are thin for physiological reasons, but fat people are fat for psychological reasons. And the `high metabolism' explanation doesn't work for thin people anyways -- just factor the metabolic rate or an estimate thereof into the calculation of the number of calories burned a day. Likewise with a `low metabolism' explanation for fat people. You'll see the same thing.

No, something is wrong in the line of thought above. And that something is the assumption that there's this number, called the average number of calories burned a day.

Now, in trying to calculate this number, of course you have to look over a period of time (previous week? previous one year? previous ten years?), and throw out exceptional days (that day you went on the 12 mile hike on Mt. Rainier, and that week you were too sick with the flu to do much more than shuffle to the bathroom and back). This requires making choices about what data to include and exclude, and these will be arbitrary but make a difference. If that's what I meant, then the easy reply would be to point out that this difference will, most likely, be negligible, or utterly irrelevant to the process by which you actually calculate your average number of calories burned a day (that is, by carrying around a little journal on a `normal day' and making note of what you do, how often you do it, and for how long).

So that's not what I mean. The average number business is confusing, so let's drop it. What we really assume when we assume there is such a thing as the average number of calories burned a day is that metabolic rate is roughly constant from day to day. Again, some daily variation is likely and expected, but typically things will stay within a fairly narrow range.

In particular, the line of thought from the first paragraph needs to assume that the metabolic rate is roughly the same for a given individual both when they are fat and when they are thin.

And this assumption was pretty thoroughly debunked. Nearly fifty years ago.

There is a reason that fat people cannot stay thin after they diet and that thin people cannot stay fat when they force themselves to gain weight. The body's metabolism speeds up or slows down to keep weight within a narrow range. Gain weight and the metabolism can as much as double; lose weight and it can slow to half its original speed.

You might still think the counting-calories method would work -- we just need to be able to measure the change in metabolism. But the change in metabolism isn't the only effect of dramatic weight loss.

fat people who lost large amounts of weight might look like someone who was never fat, but they were very different. In fact, by every metabolic measurement, they seemed like people who were starving.

Before the diet began, the fat subjects' metabolism was normal - the number of calories burned per square meter of body surface was no different from that of people who had never been fat. But when they lost weight, they were burning as much as 24 percent fewer calories per square meter of their surface area than the calories consumed by those who were naturally thin.

The Rockefeller subjects also had a psychiatric syndrome, called semi-starvation neurosis, which had been noticed before in people of normal weight who had been starved. They dreamed of food, they fantasized about food or about breaking their diet. They were anxious and depressed; some had thoughts of suicide. They secreted food in their rooms. And they binged.

The Rockefeller researchers explained their observations in one of their papers: "It is entirely possible that weight reduction, instead of resulting in a normal state for obese patients, results in an abnormal state resembling that of starved nonobese individuals."

This doesn't formally refute the line of thought from the beginning of the post: if you're fat you can lose weight by carefully monitoring caloric intake. And literally starving yourself. So the `fallacy' of the title is more polemical than logical. Still, this is a case of the `cure' being orders of magnitude worse than the `disease'.

Boy, not filibustering Alito sure worked out well, huh?

Too busy for a real post. I'm just dropping by to share the bad news.

The Supreme Court narrowly upheld a federal law today banning a controversial abortion procedure, giving the anti-abortion movement one of its biggest legal victories in years.

The justices ruled, 5 to 4, that a law passed by Congress in 2003 and signed by President Bush does not violate the Constitution by imposing an undue burden on a woman’s right to end a pregnancy. The majority said its ruling reflects the government’s “legitimate, substantial interest in preserving and promoting fetal life.”

Feministing has a nice assortment of links commenting on the general fuckeduptitude of this decision, why it's completely not a surprise at all given that Alito replaced Day O'Connor, and so on.

Is Singer's principle consequentialist?

Yesterday I had a letter published in the Observer where I gave essentially the same argument from Singer's principle for universal health care that I used in the post from a few days ago. This was in response to a column the previous day by a much less pleasant libertarian than Andrew Bailey.

I had some compliments from an officemate and a few students, which was nice, but several people said they were surprised to see me say something so consequentialist.

But I don't think Singer's principle is consequentialist. To see why, let's first remind ourselves what Singer's principle says.

If we can prevent something very bad from happening by doing X, and if we can do X without sacrificing something of comparable moral worth, then doing X is morally required.

Singer is, of course, a consequentialist, so in the sense of the origin of the principle, it would be fair to call it a consequentialist principle.

But that's kind of a silly reason to call it consequentialist. Just because a consequentialist was the first person to formulate a principle doesn't mean that principle is verboten to ethicists of other theoretical camps. When we say a principle of ethics is consequentialist, we mean only a consequentialist can consistently endorse it. And, since I said Singer's principle was as close to an analytic truth as one would like, if the principle was consequentialist in this sense, it would seem that consequentialism is analytically true.

Which, while many consequentialists would stand up and applaud here, is probably not right. What I want to claim is that any ethical theory, not just consequentialist ones, can endorse Singer's principle.

Let's look at a classical example: suppose we can save the life of a skilled and influential doctor by harvesting the kidneys from a homeless person. Crude consequentialism takes Singer's principle as a major premiss, takes as its minor premiss the claim that the life of the doctor is of more moral worth than the life of the homeless person, and concludes that the sacrifice of the life of the homeless person is morally required to save the life of the doctor.

Now, no-one (at least, not very many ones) thinks this conclusion is acceptable. The problem is in the minor premiss: crude consequentialism gets things wrong when it starts saying that the life of one person is of more moral worth than that of another person. That is, the problem is not with the idea that different goods have different relative moral worth (as in the major premiss, Singer's principle), but instead with the particular assignments of different relative worths made by the theory (the minor premiss).

Every theory of ethics and politics is going to need to give an account of which goods are more valuable than others. This is the concrete difference between different theories, and possibly the only concrete difference. When the Aristotelean says that X is part of our natural function and Y is not, while the consequentialist retorts that Y will promote more aggregate good than X, is the argument really over anything but which of X and Y is of greater moral worth?

Thus, Singer's principle is an absolutely general -- and analytic -- principle of ethics, and one that, in some form or another, will be endorsed by every theory of ethics. (Perhaps communitarians will dislike the language of `sacrifice', and Aristoteleans might not like the language of `requirement', but they will still agree that some goods are to be favoured over others.) What the principle does not tell us is how to do the comparing of goods, how it is to be applied to a given situation. That is where different approaches to ethics will disagree.

Finally, I note that those disagreements do not always come up in every situation. I think universal health care is a good example: pretty much every serious approach to ethics is going to say (at the very least) that maximally unrestricted property rights are less morally important than individual lives. (Non-collectivists should note that I have said `maximally unrestricted property rights', not `all property rights per se'.) The ethical judgements that we take to be obvious and irrefutable serve as data for good theories. For example, if a theory says that organ harvesting is morally required (as in crude consequentialism), then we consider that a reductio ad absurdum of the theory. `Common sense' or pre-theoretical ethical reasoning is useful for getting at these ethical data. Hence, to the extent that there is a substantive domain of obvious and irrefutable ethical judgements, we don't need highly articulated ethical theories to actually do ethics. This is the level at which I generally try to write both on this blog and in my letters to the Observer.

Why all the attention on Walter Reed?

If all you knew was what you saw on teevee, you'd think the combination of terrible care at an outrageous cost is only a problem in a single hospital. In reality, of course, that's an accurate characterisation of our entire health care system in this country. The question we (and, in particular, the journalists amongst us) should be asking is not `How did Walter Reed get so bad, and stay so bad for so long?' (to paraphrase some guy who just said something on The News Hour), but instead `How did our entire health care system get so bad, and stay so bad for so long?'

Incidentally, Mark Shields is a lameass for not asking this question in The New Hour's followup segment on the Walter Reed story.

Of course, I already know the answers to both of my questions. The blame for Walter Reed can be laid squarely at the feet of the VAA, and fixing it is just a matter of replacing some bureaucrats, maybe closing the hospital down and doing some superficial reorganising and rebudgeting. It doesn't require the radical overhaul of the entire medical system.

And, more importantly, fixing Walter Reed doesn't require rich people to pay higher taxes. Which is directly related to the reason health care is so fucked up in the US today: for 25 years now, rich people have cared more about their money than the lives of poor people, and in our system, money is power.