Authors:

HENSON, R. & Kouider, S.

Reference:

Year of publication:

2007

CBU number:

6505

Abstract:

We investigated subliminal face priming using a sandwich masking paradigm (500ms forward mask; 34-50ms prime; 50ms backward mask; 500ms probe). Experiment 1 showed that fame judgment RTs for famous faces were primed by same or different views, even though prime identification was minimal and uncorrelated with priming. Experiment 2 found fMRI evidence of reduced haemodynamic responses in occipital and fusiform regions for primed faces, both famous and unfamiliar. Experiments 3 and 4 used EEG and MEG, and found "early" (~120ms) and "late" (~400ms) priming effects, with the early effect occurring for both famous and unfamiliar faces, and the late effect occurring only for famous faces. Subsequent regression analyses showed that the early EEG and MEG effects, as well as the fMRI effects, could be largely explained by a coarse pixelwise measure of "visual overlap" between prime and probe image (despite their different sizes, and even across different views), suggesting these effects reflect low-level visual properties rather than face identification per se. The late EEG and MEG effects however, which correlated with behavioural priming, were not attributable solely to this measure of visual overlap, raising the possibility of unconscious semantic processing of famous