Abstract
Subjects with grapheme-color synesthesia report that a given number is always associated with a certain color (e.g., ‘5’ may be green). We have previously shown that synesthetes perform better than controls in various perceptual tasks, presumably due to their induced colors (Ramachandran & Hubbard, 2001). This suggests that synesthesia is a sensory effect, perhaps due to cross-activation between color selective occipital brain areas (V4v/V8) and a nearby brain area responsive to graphemes. To test this hypothesis, we conducted behavioral and fMRI experiments with four grapheme-color synesthetes and four non-synesthetic controls. Each subject participated in a perceptual grouping experiment in which a pattern of numbers was detected amongst a background of other numbers (e.g. a triangle of 3's against a background of 5's). The foreground and background graphemes were chosen to induce opposing colors in each synesthete. Performance was compared against an unselected pool of 20 non-synesthetes. Behavioral performance was highly variable, with two synesthetes performing significantly better than controls, and two synesthetes performing no better than controls. Next, we identified retinotopically organized visual areas using standard fMRI retinotopic mapping techniques. FMRI responses were then measured while subjects viewed letters or numbers alternating in a block-design with non-linguistic symbols while performing an upright vs. italic discrimination on these symbols. Synesthetes showed significantly greater fMRI responses than control subjects to letters and numbers in color selective areas (V4v/V8), but not early visual areas (V1/V2). Activity in retinotopically organized visual areas (V1, V2, V4v, V8), but not grapheme selective areas, was positively correlated with behavioral performance on the pop-out task. These results support the hypothesis that grapheme-color synesthesia may arise from cross-action of color selective areas by visually presented graphemes.