Learning Cue-Invariant Visual Responses