This framework could account for the strong concreteness effects observed at the “edges” of the temporal lobe (i.e., STG and PHG) in terms of their relative specialisations for verbal versus visual inputs, while predicting equi-modal activations in the centre (ITG and the vATL). Importantly, this framework assumes graded specialisation within a single functional system in the ATL, rather than an absolute Selleck NVP-BGJ398 division into separate processing modules. Finally, we consider C > A activations observed in other areas of the brain. As in previous studies and meta-analyses, we found C > A effects in angular gyrus, posterior cingulate and mid-PHG. As discussed

in the previous section, the activation of PHG most likely reflects retrieval of stored visual characteristics

of concepts, which is only possible for highly imageable, concrete words (D’Esposito et al., 1997 and Sabsevitz et al., 2005). C > A effects in the angular gyrus and posterior cingulate are harder to interpret. The role of posterior cingulate in semantic cognition is unclear, though it has been suggested that it may be involved in the interface between semantic knowledge and episodic memory (Binder et al., 2009). Stronger claims have been made about the function of MAPK inhibitor angular gyrus. Binder et al. (2009) proposed that the angular gyrus is critically involved in semantic representation Thalidomide and that concrete regions activate this region strongly because they have more detailed semantic representations. It is therefore interesting that the activation profile of angular gyrus diverged strongly from that of the vATL, for which a similar representational function has been proposed. There were three findings that suggest the function of the angular gyrus is very different to that of vATL. First, the angular gyrus was not activated in the main contrast of semantics over numbers; in fact, the contrast in this region slightly favoured the numbers (see Fig. 3). This suggests that, in addition to any

putative role in semantic processing, the angular gyrus is at least equally involved in the processing of numerical magnitudes. This was not the case for vATL. Second, angular gyrus showed a clear C > A activation pattern, while the activation in the vATL slightly favoured abstract words (though this effect varied elsewhere in the ATL region). Finally, angular gyrus and vATL showed very different activation patterns with respect to rest, with angular gyrus significantly deactivated by the semantic task while vATL, along with other elements of the semantic network, were positively activated. This result is consistent with the status of angular gyrus as a key element of the default mode network. Binder et al., 2009 and Binder et al.