It's a really interesting question. I'm looking askew at the coffee cup in front of me. I suppose it makes the most sense for studies to begin prior to first exposure; to have data on subjects going back perhaps even as long as it takes for them to become tolerant. I suppose factoring out the effect of learning for a long-term study which involves cognitive tests is it's own challenge as well? I'm enjoying http://www.gwern.net/Nootropics#caffeine now :)