Right, only for the flanker-sentence combinations would the preceding flanker image matter. This would be the only combination that matters since it is made up of only experimental items. Then, the rest are fillers. Therefore, I think to make things simpler, it would be easier to have the latinMix function take three labels, flanker-sentence, flanker, and sentence. It would also cover the remaining combinations, though sometimes there might be 3 of the same tasks in a row, which shouldn’t be a problem.
Thank you again! Sorry I didn’t explain this experiment very well from the beginning.
Val