Smart Drugs

Jesse Walker | August 2, 2004

The Telegraph reports:

Research to be published tomorrow by academics at University College London has found that those who even drink only one glass of wine a week have significantly sharper thought processes than teetotallers.

The benefits of alcohol, which are thought to be linked to its effect on the flow of blood to the brain, can be detected when a person drinks up to 30 units of alcohol -- about four to five bottles of wine -- per week.

The researchers were unable to test the effect of higher levels of alcohol consumption, although drunkenness probably negates any positive effects on the brain.

"Probably" negates? Surely this calls for further research. If anyone needs me, I'll be in the kitchen.

Comments

|8.2.04 @ 1:00AM|#

"drinking half a bottle of wine a day can make your brain work better, especially if you are a woman."

The writers of the article (and the scientists, if they have claimed the above) have made a grievous error in their interpretation of the data of this study. The scientists are quoted as saying that there is an association, a correlation, between drinking and improved neural processes. As my high school teacher drilled into our heads, in observational studies, correlation is not causation! As joe astutely implies, from this study, it's impossible to know whether drinking makes your brain work better or those who have well-functioning brains tend to drink, or if there's some lurking/confounding variable. In order to conclude that A causes B, you need to perform a controlled experiment. That is, get a random sample of people, randomly assign each person to drink a certain amount for a period of time, then administer the test.

|8.2.04 @ 1:23AM|#

Paul- A fair point, but also such an obvious one that I can't imagine they'd publish this unless they controlled for other factors as much as possible.

Jesse Walker|8.2.04 @ 1:45AM|#

From the article: "The effects were apparent even after the results had been adjusted to take into account factors such as physical and mental health." Which is admittedly vague, but it does suggest that it would silly to write this off without first reading the actual study.

|8.2.04 @ 1:48AM|#

Jesse, never drink in the kitchen. It just makes it more likely you'll eat food instead.

Cheers.

|8.2.04 @ 1:50AM|#

It's not a question of controlling for variables; in this case, in order to establish causation, you need to impose consumption of alcohol, in varying levels, on a randomly selected group of people. This study seems to have consisted of "psychometric tests [administered] to more than 6,000 civil servants." No matter how well-written the test, all that can be concluded from it is that a correlation exists between the variables.

Incidentally, I think it would be great if the alleged findings of this study were proven to be true by a full-fledged experiment.

|8.2.04 @ 1:53AM|#

Wine good, water bad!

Jesse Walker|8.2.04 @ 2:04AM|#

Again, Paul, I'd hold off on dismissing the study until I see (a) what exactly the tests were and what the data say, and (b) just what the researchers' speculations consist of. (I doubt the discussion in the Telegraph article is exhaustive.)

Also, until I've had some more drinks.

|8.2.04 @ 2:24AM|#

I love how peer-reviewed studies with statistical caveats are dismissed out-of-hand on this forum because somebody comes up with an idea off the top of his head that sounds better than what he read in the press's synopsis of a long research article.

OK, I haven't read this study in particular yet, but on the occasions when I've gotten to read studies critiqued on this forum, I've usually found that the authors put in all sorts of statistical fine print and caveats, and spent some time looking at alternative hypotheses. Usually the study is much more thorough than a journalist's interpretation of the abstract.

I think I'm going to save this reply to my hard drive so the next time a study is dismissed on this forum I don't have to retype it.

|8.2.04 @ 2:46AM|#

If I'm reading this correctly, one glass of wine will offset each joint. Problem solved!

|8.2.04 @ 2:56AM|#

And I love it when somebody who is clearly ignorant about statistics assumes the same ignorance in the object of his contempt.

I do not dismiss the entire study. In the first sentence of my first post, I acknowledge that the Telegraph makes the erroneous claim. The quotes from the scientists suggest that they understand the distinction I'm trying to make here. They say they have found an association, a relationship, between the two variables, but mention nothing about causation. That an association exists is an interesting finding, and I do not dismiss it. However, you cannot determine causation without performing an experiment.

Contrary to thoreau's beliefs, the distinction between causation and correlation, observational studies and experiments, is not "an idea off the top of my head." Look at any introductory statistics textbook, and you'll see that it's a pretty important distinction.

To illustrate my point: Perhaps the upbringing of those who drink tends to foster neural development better than the upbringing of those who don't. Perhaps a genetic predisposition to drink exists, and these genes also cause heightened neural ability. There are many ways to explain the association. The article jumps to the conclusion that consuming alcohol makes your brain work better, period, and according to the fundamental precepts of statistics, this claim is not proved by the study.

|8.2.04 @ 3:20AM|#

The idea from "off the top of someone's head" was Joe's suggestion that smart people drink wine because they're smarter.

And I'm not ignorant of statistics. I know as well as anybody that correlation does not equal causation. But correlation is often an important step in an investigation. You start with an anecdote, as somebody observes that X and Y seem to coincide a lot. So then you do a correlation, to find out whether X and Y do in fact coincide a lot. If they are in fact correlated, and this correlation holds even after controlling for several different variables, then you start looking deeper. The correlation is simply the first tool in a long process, but it's an important tool because it can be used to discard a plethora of bad ideas, and send the remaining ones on for more thorough scrutiny.

Also, remember that this study occurs in the context of a much larger body of research on alcohol and improved health. A variety of evidence exists to show that alcohol consumption may improve cardiovascular health under certain conditions. (I don't know if there have been previous studies indicating mental health benefits from wine consumption.) These researchers are simply pointing out that possible mental health benefits may derive from wine consumption so it's worthy of further study.

Finally, you yourself acknowledged that the researchers put some caveats on their work and know the difference between correlation and causation. So why are you being so harsh on them, and on me?

|8.2.04 @ 3:28AM|#

Uh, yesh, Paul, I wazh bein' ashstute. Hic!

|8.2.04 @ 3:40AM|#

I assumed the "off the top of my head" comment and others of the same tone were targeting me since I read joe's post as a joke (albeit with an implicit argument), hence me being harsh on you. I didn't mean to be harsh on the scientists. I said that if they claimed causation as the article did, shame on them.

|8.2.04 @ 3:46AM|#

Paul-

In all fairness, I probably went a little overboard too. This forum has a habit of jumping all over scientific studies with "Oh, those idiots didn't control for this variable, OF COURSE they're wrong!" Usually, however, it's done when the study reaches a conclusion that people find politically unacceptable. A study touting health benefits from alcohol is the sort of thing that Libertarians would actually enjoy :->

|8.2.04 @ 3:50AM|#

"I think I'm going to save this reply to my hard drive so the next time a study is dismissed on this forum I don't have to retype it."

Uh-uh, thoreau, that's cheating. You have to pound out your ceaseless drivel just like the rest of us do, that's the rule. No boilerplate. Look it up.

|8.2.04 @ 5:19AM|#

thoreau:

The forum doesn't have habits. You do.

|8.2.04 @ 5:48AM|#

The above is true, to wit: I got my wife drunk in order to date her, and she ended up marrying me. Since this was obviously a fantastic choice, the booze must have improved her brain function. hmmm...

They should study whether different types of tipple correlate with different increases in brain function. I would bet that the more expensive the booze was, the greater the brain function "boost"...

|8.2.04 @ 6:58AM|#

Jamie, you're missing a point here: if you didn't also get drunk with her, maybe your brain functions weren't up snuff there -- maybe you coulda done better, pardner...

|8.2.04 @ 11:11AM|#

The things I do for medical science....

|8.2.04 @ 11:52AM|#

Maybe this is the excuse I need to finally start homebrewing.

|8.2.04 @ 12:22PM|#

"those who even drink only one glass of wine a week have significantly sharper thought processes than teetotallers."

Well duh. They were sharp enough to realize that drinking wine is good.

|8.3.04 @ 4:29AM|#

This study seems to have consisted of "psychometric tests [administered] to more than 6,000 civil servants."

Well, heck, with civil servants the IQ points have nowhere to go but UP!

Kevin

|8.3.04 @ 5:14AM|#

The interpretation of these results is making a clear correlation/causation mistake. The obvious counter-theory too "booze makes you smarter" is "smart people like to drink more than dumb ones". To know what the truth is, we'd need a very different study.