I've only done ADR once. So I probably have more questions than answers. But if I'm understanding this right, then can't you just bus the audio signal of the guide track (I assume this is a moveable region of audio containing 3 beeps signalling the actor when to start his/her lines?) simultaneously to the cue send as well as the speakers? This way, the actor gets his cue to start whether or not he his headphones on.
I assume he wants the headphones off at times in order to get into the moment. I also assume that you're not letting the scene play out over the speakers otherwise, you'd be recording what's coming out of the speakers. But if you select the region that's being recorded on one track after the guide track plays its signal from a separate track and going to the speakers in that talent's room as well as to the headphones via cue send, then you shouldn't have to bounce back and forth at all.