What is the ultimate impossible experiment?

There's an old psychological theory that it takes 10000 hours of practice to become an expert in a field...and some argue that it's only practice, not natural talent, that makes you an expert. But that theory is completely untestable.

That's the conclusion of Union College psychology professor Christopher Chabris, who was asked by Scientific American to imagine an experiment that he would like to do, but the logistics of which make it utterly impossible. The idea itself goes back to psychologist Anders Ericsson, and it's gained renewed currency in recent years thanks to writer Malcolm Gladwell. Chabris explains the history of this idea:

"The prevailing theory in cognitive psychology, going back to Adriaan de Groot, who studied chess grandmasters, and later to Anders Ericsson, who studied other domains such as music and sports, is that expertise is all a matter of how much one practices, and that there's no such thing as a particular talent that will make it easier for someone to become an expert. If that's true, that's a positive thing - there's nothing holding me back from, say, becoming a professional basketball player."

It's an idea that really doesn't jibe with most people's intuitions - surely, there must be some amount of natural talent involved, and even 10,000 hours of practice isn't going to turn everyone into an expert? The thing is, the data necessary to really come to a conclusion about all this is woefully incomplete, and that's why an experiment would be hugely useful. And "huge" is about the right word to describe the thing, as Chabris explains.

In order to test this properly, you'd need to get lots of volunteers - maybe a couple thousand to control for all the variables between different people. That immediately presents problems, not the least of which is cost, as paying 2,000 volunteers to put in 10,000 hours of work - even at the low, low rate of $10 per hour - would still cost a massive $200 million.

And that's assuming you could actually get volunteers to sign up in the first place, or that those you could get would be well-suited to the exercise. As Chabris points out, that's an open question:

"You're volunteering 10,000 hours of your life, and imagine a situation where you're not happy with what you've been assigned - 'Congratulations, Mr. Smith, you've been selected to become a master purchasing manager. There are arguments that the younger you are, the more easily your brain soaks up skills, and I'm not sure in our society whether parents would really go along with randomly assigning kids to learn a skill."

Even if you could put the experiment together and keep it running for the decades needed to generate all the data, that still might not be all that helpful. After all, someone is going to need to go through and interpret all that data, and all the masses and masses of information might just prove too complex for people to really find anything useful or come to definitive conclusions.

For more on this, check out Scientific American.