Abstract
In this fMRI study, we aim to investigate Greeble training effect, especially in early visual areas, in participants who were trained to recognize Greebles with either Gauthier97 (or Gauthier and Tarr, 1997) or Gauthier98 (or Gauthier et al., 1998) paradigm. A retinotopic mapping procedure was used to delineate early visual areas (V1, V2, V3, V4, and/or V3a). Then, these ROI masks were applied to the passive viewing task of faces, objects, and Greebles, both before and after training and in both Gauthier97 and Gauthier98, to evaluate respective BOLD changes in V1 to V4 ROIs associated with Greeble training. We compare brain activity by both univariate and multivariate analysis, which show quite different pictures: while univariate analysis (both ROI time course comparisons and GLM contrasts) showed no significant BOLD changes across early visual area (and no difference across both Gauthier97 and 98); multivariate analyses, including both ROI-MVPA and whole-brain searchlight, present significant classification (Greebles_before vs._after) accuracies across V1–V4 ROIs (via ROI-MVPA), and wide-spread training effects across many brain regions (including parietal and prefrontal regions, via MVP searchlight), once again across both training paradigms. To further verify that whether early visual area would be recruited by automatic attention, the bird expertise fMRI study in Yang et al. was also analyzed with retinotopic procedure, and was found that in passive viewing task (but not in 1-back identity), subject’s behavioral expertise also predict the involvement of early visual cortex. Taken together, these results suggest that while FFAs have been ‘sharpened’ by Greeble training; multivariate analyses reveal widespread changes in both visual and associative areas. In addition, early visual cortex could also be predicated by expertise-driven attention, consistent with Harel et al. (2010).