Previous studies have identified a number of word-sensitive regions in the ventral occipitotemporal cortex. In this study, we used fMRI adaptation and multi-voxel pattern analysis to investigate their functional properties such as orthographic and lexical sensitivity. Three categories of Chinese character-related stimuli were adopted - real characters, pseudo characters, and false characters. They were structurally matched but differed in their lexical and orthographic properties. Real characters were common left-right structured characters. A pseudo character was created with two parts according to the orthographic rule. A false character was also formed of two parts but they occupied incorrect positions thus breaking the orthographic rule. During the fMRI adaptation experiment, target probes from one category were briefly presented amidst prolonged exposure to adaptors from either the same or different category of stimuli. Comparing the within-category and between-category adaptation effect for a region could reveal its orthographic and lexical sensitivity. Three word-sensitive regions were functionally localized, one in the occipital lobe, another near the occipitotemporal sulcus (i.e., the VWFA), and the third located further anterior to the VWFA. fMRI adaptation results show that real and pseudo characters were differentiated in the VWFA and the more anterior ROI, since they had greater within-category adaptation effect than cross-category adaptation effect. However, adaptation to false characters was less category selective, possibly reflecting that the ROIs under consideration were less sharply tuned to the false characters. Within these ROIs, information carried in the multi-voxel pattern of response could support successful classification of all three types of characters. Interestingly, adaptation to real characters significantly reduced the correlation-based classification performance in the VWFA and the more anterior ROI. Thus, while the orthographic information is likely registered earlier in the pathway, the lexical difference between real and pseudo characters are represented once the information reaches the VWFA and thereafter.
Meeting abstract presented at VSS 2014