Different analysis solutions for different spatial resolutions? Moving towards a mesoscopic mapping of functional...

This comment challenges the dichtotomy that Kriegeskorte and Bandettini (this issue) propose to exist between "activation-based" and "information-based" approaches to fMRI analyses and argues that multi-variate analyses are just a special case within the overall repertoire of methods for analyzing paradigm-related BOLD signal variations. Moreover, this comment argues that using multi-variate approaches comes at a price, trading-off spatial resolution for sensitivity, and thus partially cancels potential benefits from high-field fMRI. Paradoxically, this comment thus concludes that pattern analyses provide a powerful complement to existing methods but not the complement that will actually permit to map functional architecture at mesoscopic resolution, i.e., one of the most interesting applications of high-field fMRI.