recode responses?

Hi,

I´m trying to do the following: I would like to show several pictures on the screen and participants task is to choose the right picture (almost like in a memory game, they saw a picture-picture pair before and are now presented with possible picture pairs and have to remember the correct pairing). The difficult thing is that I need data telling me more than just if they picked the correct picture or not. I need data that tell me for instance if they picked the wrong picture but a picture with the same (or the opposite) valence as the correct picture...the position of the pictures should be randomized for every trial...how can I do this?