Just out of curiosity, would you be able to break up your large single input file into a group of smaller ones to run? It seems like you have some natural break points in the trials with ~instruction that occur between blocks of picture presentation. If you have MediaLab, you could use it to handle the presentation of all of the stimuli. I had this thought after rereading the statement from your first email about how splitting the file in half did not seem to invoke the "style not found" error.