Great question. I also wonder if the unused speakers will be taken out of the room, which would be essential IMO.
Are the various speakers all of similar sensitivity? What will the rest of the system be?
It is likely too time consuming to shuttle speakers in and out of the room. They need Toole's Spin-a-Rama!!
Still-one is right, though, our acoustic memory is not that good if switchover takes too long. Doing 4 speakers, as opposed to 2, considerably dilutes the effectiveness of such an audition. I would suggest, though, if all speakers are in the room, each unconnected speaker should have its inputs shorted together to prevent cone motion.
Actually, the very most essential thing is to accurately calibrate beforehand by mike or SLMeter measurement each speaker's sensitivity so that playback levels are as perfectly matched as can be. That is just not possible by ear alone. I do not think you can simply rely on mfr's specs. It should be measured in the actual room with the actual speakers.