While this large dataset on the LATL is still compatible with many definitions of “semantic processing”, the robustness of these findings and their generality across multiple methodologies presents an opportunity for a systematic investigation of the computational details of this activity. One step towards sharpening our understanding involves recent MEG results on language production (Del Prato and Pylkkänen, 2014), where the modification of object denoting nouns with color adjectives (blue cups) engaged the LATL,while numerical quantification of the same nouns (two cups) did not. Given that both of these combinations involve semantic composition, these data are incompatible with a general semantic composition account of the LATL. Instead, they suggest a narrower computation, perhaps better characterized as a type of “conceptual combination”, a label employed in the concepts and categories literature for a host of cases where, intuitively, the combination of two concepts serves to form a more complex one, typical examples being adjective–noun and noun–noun combinations. Given that in phrases such as two cups, two does not add a feature to the concept denoted by cup but rather enumerates the number of tokens in a set of cups, such caseswould, by hypothesis, fall outside the definition of conceptual combination that is relevant for the LATL. Related evidence for the conceptual nature of the LATL include the sensitivity of its combinatory response to conceptual specificity (Westerlund and Pylkkänen, 2014; Zhang and Pylkkänen, 2015) and the correlation between the LATL activation elicited by specific concepts like boy and the product of the activations for their constituent concepts (i.e., male and child) (Baron and Osherson, 2011).