As mentioned in a previous response, Cooks Illustrated did a test some years ago (2003, I think), where they concluded that the preferred vanilla in a taste test was some cheap artificial vanilla from a local drugstore's generic section or something. They have since done further tests (such as this one in 2009), and real vanilla sometimes edges out the cheap artificial competition (which here came in a close second), particularly for situations where the vanilla is uncooked and generally added in at the end (e.g., custards). Somebody over at Chowhound tried a similar experiment and agreed that artificial vanilla clearly won in baked goods. I don't have access to the full Cooks Illustrated article, but over here is another interesting claim -- that is, if you want to beat out all of the commercial extracts (both real and "fake"), just make you own at home.
Anyhow, there have been other similar tests over the years, but I find Kenji Lopez-Alt's test over at Serious Eats to be the most interesting. He tried blind tasting of vanilla sugar cookies, cooked vanilla ice cream, and simply stirred vanilla into a eggnog recipe. Like Cooks Illustrated, he found that in the cooked/baked applications, tasters couldn't tell the difference.
It was only in the (uncooked) eggnog application that the "real stuff" edged out the competition, but here Kenji went one step further and asked about the reason -- and it was just the "booziness" of the real stuff that people liked. By spiking the eggnog with a small amount of vodka to make up for the missing alcohol in the artificial vanilla, the "fake" stuff actually performed about as well as the real stuff. (Incidentally, the use of real vanilla beans actually suffered a similar problem and was declared as inferior to extract in some taste applications, since it didn't have the "booziness" element of extract.)