I totally agree with this.
I am now finishing up a 4-year BS in Computer Science (from a Greek university) and waiting on my PhD applications. I scored 750 on the Quantitive GRE test while I could easily ace them at home. The problem is that a science student tends to look at the quantitive questions in an analytical way, understanding the problem and trying to solve it using the methodology through which he has solved countless similar problems before. However, during the quantative test, with an average available time per question of 1.5 minutes, with 30 seconds of that spent reading the actual question (so ending up 1 minute to solve the thing), you can not always have the time to properly solve the problem at hand (especially since most of the time, the parameters are such that an analytical solution without a calculator is not possible in the given timeframe). So you have to figure out the "trick" aspect of the problem, which can lead you to a fast solution. People who "train" just for the GREs focus their efforts around finding these trick aspects, and thats why there are instances of non-science students out performing the science people.
It basically boils down to this: Give a science student 1 + 1/2 hour instead of 45 minutes and he will be able to ace every possible Quant GRE test without breaking a sweat. Do the same thing with a non-science student, and you probably will not get the same result...
