I wondered the same, netwolf. Thinking about it I wondered if the mirror's "core" is still well above ambient while the "surface" has cooled to almost that. The inference here is that the cooling system is sufficient to keep the surface at a reasonable temperature while the "core" energy is being released but as soon as the cooling effect is removed the surface once again begins to heat up relative to the ambient air temperature.
I'm probably wrong - it wouldn't be the first time.
