Yep, they've trained my neuronet that they always show the images in a 3-2-1 or a 3-1-1 sequence, so now even if shows me something weird, I know I need to select the number of most likely things they want me to. This way I almost always finish it on first try.