Conference on Cognitive Computational Neuroscience

Functional MRI has become a key tool in neuroscience, but conventional group-level analysis techniques fail to yield precise results in high spatial resolution settings. We show how functional alignment can help alleviate this problem. In particular, we use a method based on opti mal transport to align data from a recent 7T study on the Visual Word Form Area (VWFA) in bilingual subjects on the basis of functional localization. We then use aligned data to compute reliable group-level analysis, removing the need for manually defining individual regions of interest (ROI). Our alignment procedure involved selecting a reference subject on which all remaining subjects were mapped, and finally running a group analysis in this reference space. Eventually, we show that fine-grained word-selective patches can be found in the ventral occipito-temporal visual cortex and that stimuli increasingly similar to real words show increasingly anterior activation clusters in the VWFA at the group level.