Curling physics - solution approach?

I'm in the process of gathering parts for a Prusa and was thinking about the differential contraction/curling problem. I don't know if I understand the problem correctly, but it seems the issue is that each new layer pulls together the one below it, because it is deposited at a higher temperature than that of the lower layer and therefore contracts more, introducing residual stress and accumulated curling.

If this is correct, why is the focus on heated beds? It seems like heating the top of the build would be a lot more important to mitigating the problem, particularly the point where plastic is being deposited. Does introducing a substrate heating system onto the extruder introduce a lot more problems than it solves? I'm thinking a parabolic reflector with its focus just below the deposition point, with poor thermal contact to the rest of the system so that it's forced to get very hot and lose much of its heat radiatively.

I'm not much good at optics but if someone more knowledgeable thinks its possible I can do some studying.

If the topmost layers don't cool enough, it acts a bit like a wobbly jello mold. that's what happens when i print tall, thin things like on the cathedral spire. the skeinforge cool plugin forces the hot end to slow down or even move away from the freshly printed layer so that it has enough time to cool.

So, i don't think this is a great solution as you'd end up with messy prints. I also thought that a top mounted heater would be beneficial, but experience shows otherwise.