Just an idea that came to mind: The way to get a higher score, in Tetris, is to prepare for a line piece and get a 'tetris' (clear 4 rows at once). Being interested in probability learning, I wonder, if we manipulate the probability of the line piece, or make it's appearance contingent on something else, is the participant (or rather, the gamer) going to learn it and adapt his strategy?
This version of tetris was coded by me in Pygame, making use of Numpy arrays. Supply your own sound files!