It is my understanding that the higher the impedance the more closely the speaker will follow the incoming voltage. Something to do with it's resistance to electrical movement that it will track the incoming voltage with more accuracy? Can't find any immediate literature on it at the moment, but perhaps others can elaborate?
Luke,
The general principle you describe is not a real phenomenon, and falls in the snake oil camp.
However, like most wive's tales, there is some underlying physics that has been confused in the retelling:
A loudspeaker has a fundamental degree of electrical damping of it's oscillating behavior, normally quantified via the Theile/Small (T/S) parameter, Qes. If one raises the Re (DC resistance) of the voice coil, then Qes is raised in direct proportion.
An increase Qes equates to
lower inherent electrical dampening for the driver, which leads to less overall damping for the box/driver system. If the box driver system was designed with a specific example of Qes in mind, and the driver Qes is increased (e.g. by voice coil heating), the overall damping behavior of the box/driver system will move towards a higher Qt, or less total damping. This results in a progressively more underdamped response.