Contextual Priming in Grapheme-Color Synaesthesia

Grapheme-color synaesthesia is a neurological phenomenon in which particular graphemes, such as the numeral 9, automatically induce the simultaneous perception of a particular color, such as the color red. To test whether the concurrent color sensations in graphemecolor synaesthesia are treated as meaningful stimuli, we recorded event-related brain potentials as 8 synaesthetes and 8 matched control subjects read sentences such as ‘‘Looking very clear, the lake was the most beautiful hue of 7.’’ In synaesthetes, but not control subjects, congruous graphemes, compared with incongruous graphemes, elicited a more negative N1 component, a less positive P2 component, and a less negative N400 component. Thus, contextual congruity of synaesthetically induced colors altered the brain response to achromatic graphemes beginning 100 ms postonset, affecting pattern-recognition, perceptual, and meaning-integration processes. The results suggest that grapheme-color synaesthesia is automatic and perceptual in nature and also suggest that the connections between colors and numbers are bidirectional. Although somewhat bizarre to most people, statements such as ‘‘The southern California sky is almost always a beautiful shade of 2’’ are quite meaningful to people with grapheme-color synaesthesia. In these individuals, achromatic graphemes, such as the numeral 2 or the letter P, automatically elicit the concurrent perception of a particular color, such as bright blue or a metallic shade of green. Grapheme-color synaesthesia typically begins in early childhood and displays a remarkable degree of consistency over time (Cytowic & Wood, 1982). Although researchers have studied synaesthesia for well over a century (Galton, 1880), key aspects of this phenomenon remain unclear. One basic question is whether synaesthesia is simply a set of overlearned associations that have persisted since childhood (Cytowic & Wood, 1982), or whether it has a perceptual basis grounded in altered cortical connectivity (Ramachandran & Hubbard, 2001). Further, although the phenomenology of synaesthesia is unidirectional—that is, graphemes induce color sensations, but colors do not induce the sensation of their corresponding graphemes— some researchers have suggested that connections between colorand form-processing areas in synaesthetes might be bidirectional (Cohen Kadosh, Cohen Kadosh, & Henik, 2007; Knoch, Gianotti, Mohr, & Brugger, 2005). Although the neural substrate of grapheme-color synaesthesia has been elegantly studied with functional magnetic resonance imaging (Hubbard, Arman, Ramachandran, & Boynton, 2005; Nunn et al., 2002) and diffusion tensor imaging (Rouw & Scholte, 2007), few studies have addressed the real-time processing of graphemes in this population (though see Sagiv & Ward, 2006; Schiltz et al., 1999). In order to explore the extent to which grapheme-color synaesthesia invokes perceptual versus conceptual processes, we employed electrophysiological techniques to examine grapheme-color synaesthetes’ brain response to graphemes embedded in sentence contexts. Whereas previous researchers used mental arithmetic to invoke the concept of an inducer (e.g., ‘‘7’’) and showed that it affected the processing of its corresponding color (e.g., yellow) in a color-naming task (Dixon, Smilek, Cudahy, & Merikle, 2000), in the present study we utilized linguistic context (e.g., ‘‘The sky is’’) to invoke the concept of a color (‘‘blue’’) in order to examine its impact on event-related potential (ERP) responses to an inducer grapheme. In particular, we focused on the N400 contextual-priming effects observed for meaningful linguistic stimuli (color terms), nonlinguistic stimuli (color patches), and color-inducing graphemes presented in the same sentence contexts. A negative-going wave that peaks approximately 400 ms after the onset of a contextually relevant stimulus, the N400 is sensitive to the degree to which a meaningful stimulus is primed by the preceding context (Hagoort, Hald, Bastiaansen, & Petersson, 2004; Kutas & Hillyard, 1984). The N400 is elicited by all words, whether spoken, written, or signed, though different stimulus modalities alter its scalp topography in a way that suggests the contribution of different neural generators. The size of the N400 Address correspondence to David Brang, Cognitive Science Department 0515, 9500 Gilman Dr. 0515, La Jolla, CA 92092-0515, e-mail: [email protected]. PSYCHOLOGICAL SCIENCE Volume 19—Number 5 421 Copyright r 2008 Association for Psychological Science elicited by words is reduced by the presence of congruous context. Consequently, the N400 to a word such as cat is smaller when the word is preceded by a related word (dog) than when it is preceded by an unrelated word (sugar), and is also reduced in a congruous sentence context (Growing up, every kid should have either a dog or a cat) relative to an incongruous sentence context (I take my coffee with cream and cat; Kutas & Hillyard, 1980). Meaningful nonlinguistic stimuli, such as line drawings, photographs, cospeech gestures, and even environmental sounds, also elicit N400 components whose amplitude is similarly affected by contextual priming (i.e., reduced by the presence of a congruous context; Wu & Coulson, 2005). In contrast, musically incongruous stimuli typically elicit a positive-going response in the ERP (Besson & Macar, 1987). This response has been linked to both the P300 (Donchin & Coles, 1998), a general response to unexpected events, and the P600 (Patel, Gibson, Ratner, Besson, & Holcomb, 1998), which is often elicited by ungrammatical language. These positivegoing ERP responses to incongruities are thought to reflect arbitrary or overlearned rules, and differ from semantic-integration processes indexed by the N400 (Besson & Macar, 1987). To test whether grapheme-color synaesthetes’ concurrent sensations of color when perceiving graphemes are subject to contextual integration effects, we recorded ERPs as participants read sentence frames that suggested a particular color term as the sentence-final item (e.g., ‘‘Looking very clear, the lake was the most beautiful hue of . . . ’’). These sentences were completed with congruous or incongruous color words (‘‘blue’’ or ‘‘yellow’’), congruous or incongruous color patches (blue rectangle or yellow rectangle), or graphemes known to elicit congruous or incongruous colors for a given synaesthete (e.g., in the case of 1 of our synaesthetes, ‘‘2,’’ which appears blue, or ‘‘7,’’ which appears yellow). If synaesthetic experiences are meaningful stimuli, then the congruity manipulation in the grapheme condition would be expected to give rise to N400 effects. Alternatively, if synaesthetic percepts are overlearned associations, then incongruous graphemes would be expected to elicit a positive-going wave.