I take it Stanovich is doing a lot of experiments where he controls for IQ, or compares performance within and across IQ groups. Here is my concern... there is always measurement error, and the more error in his measure of IQ, the more it will appear he's measuring something distinct from IQ which he terms "rationality."

That said, I also agree that IQ, and G, are often reified. The point is, I'm not sure Stanovich has succeeded in carving cognition skills at their joints, but I don't have anything better to offer.