Visual events are then mapped onto words, based on their probabilistic distribution. This is the same technique used in machine translation to translate words from different languages.
