Can the stimulus processing assumptions of the sometimes-opponent-process (SOP) model explain instances of contextual learning?