In introductory economics, one usually learns about "Giffen goods," where people paradoxically consume more of it as the price rises, violating the law of demand. Your demand for an item is influenced by its relative value and your overall wealth -- substitution and income effects. This is formalized in the Slutsky equation, but the bottom line is that the effects can go in different directions under very unusual circumstances. For normal goods, the income effect is positive -- higher income leads to more demand -- but for some inferior goods like Ramen noodles and American cars, the demand decreases with greater income. For a select few inferior goods, the income effect is so large it overwhelms the substitution effect, making it a Giffen good. I suppose it's a similar model to how burning coal causes global cooling in standard models.
In practice, economists have used Irish potatoes circa 1850 as prototypical Giffen goods, but this example has pretty much been discounted as apocryphal. The best example now is wheat in China, and that's basically the only one.
But what about US Treasuries? Over the weekend, their value certainly declined as the S&P downgrade may have been wrong, but it didn't decrease anyone's default probability. Yet after a full day of trading, the 10-year US T-Bond fell 25 basis points. On a relative basis, the decline in US interest rates was greater than for Australia, Great Britain, Canada, Germany, Japan, or Switzerland. Meanwhile, the US equity market tanked, which suggests the Treasury markets were not rebounding from prior expectations of an even larger downgrade.
So the value of the US Treasury falls, which lowers its relative price via the substitution effect relative to other assets. But everyone is now poorer, as with $60T or so in present valued unfunded promises, the AA+ downgrade is about a 0.1% increase in our discount rate, and that's about a $1T drop in our net worth. This income effect is so large, the relative value of Treasuries increases because now other financial assets actually decline by even more, so much so the absolute value of Treasuries increase after a fall in their quality.
Monday's Treasury move didn't work directly via the income effect, but indirectly via the income effect's effect on the substitution effect, so it's not a traditional Giffen good, but a "Geithner good."