Experimental detection of photons emitted during inhibited spontaneous emission

We present an experimental realization of a "sudden mirror replacement" thought experiment, in which a mirror that is inhibiting spontaneous emission is quickly replaced by a photodetector. The question is, can photons be counted immediately, or only after a retardation time that allows the emitter to couple to the changed modes of the cavity, and for light to propagate to the detector? Our results, obtained with a parametric downconverter, are consistent with the cavity QED prediction that photons can be counted immediately, and are in conflict with the retardation time prediction.