Apparently the relationship between price and demand seems to be a slippery concept. Elizabeth Kolbert (in an otherwise interesting piece) on obesity in the New Yorker has this to say about the argument that part of the reason for the increase in obesity is that the cost per calorie has gone down:

The correlation between cost and consumption is pretty compelling; as Finkelstein notes, there’s no more basic tenet of economics than that price matters. But, like evolution, economics alone doesn’t seem adequate to the obesity problem. If it’s cheap to consume too many calories’ worth of ice cream or Coca-Cola, it’s even cheaper to consume fewer. [my emphasis]

Oh well!