An internal control is say in biochemistry, adding a known concentration of the agent/chemical [being tested] to the system and seeing if the reading/results obtained represent that known quantity. If not, there's something wrong with the assay system. For audio, it might be adding a known distortion, non-linearity, FR or phase response deviation, etc to the DUT/system that should be audible and see if in fact, the participants can correctly ID that problem. If you can't, then there's something wrong with the system. I'd also suggest that that "distortion, non-linearity, FR response deviation, etc." be at the borderline of detectability, not at the maximum levels.