Hemispheric asymmetry for face recognition: some effects of visual masking, hemiretinal stimulation and learning task.

Five overlearned target faces were presented hemiretinally and identified by name on the recognition trials (Experiment 1). A post-exposure pattern mask was projected by the opposite eye to either the same (MASK) or the opposite (DISTRACTOR) hemisphere as the targets. Target exposure durations yielding 50% accuracy showed significant left-hemisphere and temporal hemiretinal superiorities in DISTRACTOR, but not in MASK. In Experiment 2, recognition accuracy for six faces at a constant exposure duration, under DISTRACTOR and no-mask control conditions, replicated the left-hemisphere advantage. Manipulation of the information accompanying targets during the initial learning task, to encourage either "social" or "physical" encoding of the faces, produced a "crossover" pattern of contrasting hemispheric asymmetries under the two encoding conditions. The overall results suggest that within this design, face recognition can be accomplished in three different ways, producing either no hemispheric asymmetry, undirectional asymmetry, or complex, contrasting asymmetries. These effects appear to have different functional loci. Hemiretinal differences and individual differences in encoding strategies also contributed to complex patterns of hemispheric asymmetry.