Abstract

Human occipitotemporal cortex (OTC) is organized into distinct areas that are selectively activated by the visual presentation of different categories, like faces (Kanwisher et al., 1997), bodies (Downing et al., 2001), tools (Chao et al., 1999), hands (Bracci et al., 2010), and places (Epstein et al., 1999). However, the precise role of these areas in processing their preferred stimulus category is still unclear. Moreover, the debate on how such brain organization originates is still unresolved. In order to investigate these issues, we focused on one of these selective areas: a region in left lateral-occipitotemporal cortex (lLOTC) that has been shown to be selective to both hands and tools and that is functionally connected to left intraparietal sulcus (IPS) and premotor cortex (PM), regions involved in action processing (Bracci et al., 2012). Although tools and hands are visually very different, they are both involved in object-directed actions. Thus, lLOTC might process both categories in order to efficiently communicate with IPS and PM. Alternatively, however, tool selectivity in lLOTC may simply reflect epiphenomenal association of tools with hands, for example related to mental imagery. Furthermore, the role of feedback projections between IPS and lLOTC in shaping tool selectivity in lLOTC has not yet been investigated. Using fMRI, we found that lLOTC was more strongly activated when participants processed action-related properties of tools (deciding on the typical hand action associated with a tool), as compared to when they processed contextual properties of tools (deciding on the typical location associated with a tool). Importantly, TMS over lLOTC led to a specific decrease in accuracy for action judgments but not for contextual judgments, supporting a causal role for lLOTC in processing tool actions. We investigated lLOTC-IPS connections by interleaving repetitive TMS with short fMRI scans. In a first experiment, we applied TMS over left IPS and found a decrease in BOLD signal in left IPS, confirming that TMS suppressed neural processing in the target area. Moreover, BOLD responses also decreased in left dorsal PM and right IPS, suggesting that these two areas receive projections from left IPS. However, no significant change was found in lLOTC, indicating that tool selectivity in lLOTC does not require feedback from IPS. In a second study, we stimulated lLOTC. In this study, however, no significant TMS-induced BOLD changes were found in the target region, possible reflecting insufficient statistical power. In conclusion, our results show that lLOTC is critical in processing action-related properties of tools, indicating that tool selectivity in this area reflects processes that are necessary for understanding tool actions. As a consequence, the finding that lLOTC is causally selective for both tools and hands, even if these two categories are visually very different, supports the hypothesis that the functional organization of OTC partly reflects non-visual organizational principles. Finally, the fact that we did not find significant back projections from IPS to lLOTC suggests that the role of lLOTC in processing tools does not arise as a consequence of activity in higher areas, but that lLOTC is relatively autonomous in processing action-related properties of tools.