Essentially there's not a lot we can do about that... what we CAN do, we do do...
Use Lithium AA cells which dont drop off at cold temps (http://data.energizer.com/PDFs/l91.pdf - OK to -40deg - very expensive at £1 each roughly, but worth every penny)
Give the camera quite a lot of work to do (hence the 4gb card, I intend to fill it on a schedule of 3hrs) thusdumping the 9 whrs of energy from the two cells about as rapidly as possible into the camera which will render itself as heat.
Insulated as well as I can in foam (although the fact that the lens 'opens' means it's not even slightly sealed around the front)
And a 'get up and get down' flight plan to avoid lingering at altitude.
The first, third and last of those has been enough for similar flights in the past.
I feel sorry for the GPS logger on top of the foam -it's only going to have a layer of duct tape between it and the hell outside
BUT - thermally speaking, the camera will be a warm glow underneath.
Doing something as simple as a pair of 9v cells dumping to a thin film heating element is just too much complexity and mass. If we mature the project to go for high altitudes or lingering, then I will be aiming to disable the camera's motor-driver lens deployment, keep it stuck out, and match foam around the lens to better insulate.
Of all the things that could go wrong - the camera getting too chilly isn't top of the list by a long long way. DOing it this way has worked for lots of other projects, so I'm happy to go that way. Next time around maybe we log temps internally on the camera, externally on the foam and then in free air under the balloon - using a Picaxe kit or similar.
It's worth mentioning though, that for perhaps 10 minutes, maybe half an hour, this whole schbang will be at 10mbars or less, at -50 degrees, with lots of UV radiation.
Remind you of anywhere
Doug