There are times when I wished I worked with animals. Human subjects can be incredibly frustrating; they are inattentive, often don't follow directions, and won't let you poke around in their brain or screw with their genetics. Seriously though, trying to get reasonable data from human subjects can be incredibly difficult, especially when those human subjects are undergrads who are doing your experiment just to earn credit. As much it would simplify matters, the IRB won't let me just take out their brains and look at them. Psssh, ethics.
Other times (and these are rare) the subjects are so good that they completely change the way you look at your experiment. I had a subject like that this week. In the post-test questionnaire he or she (I'm maintaining strict confidentiality here in case anyone who cares reads this. They don't. But let's pretend.) described their strategy for completing my task. Not only were they very articulate (a rarity, I'm discovering) but their strategy was explicitly different than the one I assumed people would use. This is an interesting development and is probably the first exciting part of this experiment so far.