Fig 1. Experimental design.

A: Attention task during subliminal encoding. Participants saw a flickering stream of black-and-white pixel masks. Subliminal stimuli were presented between masks. The top left depicts one encoding trial containing twelve repetitions of one subliminal stimulus. Four encoding trials constitute a condition block in this fMRI design. On the top right, a section of an encoding trial is highlighted with indicated presentation durations. To the lower left, the used fixation screens are displayed with their respective frequencies of appearance. Each encoding trial contained one response slide (either a vertical or horizontal line segment). To the lower right, we display the three stimulus categories that belong to the three experimental encoding conditions (from left to right): Face-Occupation Pairs for associative encoding, Faces Alone for single item encoding (non-associative baseline) and Contour for a non-encoding baseline (not discussed in this paper). Portraits belong to the FERET database [15]. B: Unconscious-conscious retrieval interaction with indicated presentation durations. A former subliminal face is briefly presented to cue the unconscious reactivation of previously formed face-occupation association. Next, a portrait of a celebrity comes up for the conscious retrieval of the celebrity’s occupation (actor or politician) Participants were required to recognise the famous person and to indicate his occupation by button press. Each condition block contained four trials. Fig 1B illustrates a trial of the associative retrieval condition Incongruent and a trial of the Old Faces baseline condition, where no unconscious-conscious interaction was possible. Celebrities’ portraits were taken from Wikimedia Commons (http://commons.wikimedia.org/wiki/Main_Page). Berlusconi: public domain; DiCaprio: Siebbi (http://www.ipernity.com/home/siebbi).