Lateral occipitotemporal cortex, extrastriate body area, and the somatic prior: disentangling feature and function

Is the extrastriate body area is best characterized as a narrowly defined, feature-selective region or as a node whose significance depends on its network? We argue that this opposition is misleading: feature-level claims (what a region responds to) and function-level claims (what its output is used for) answer different questions and cannot substitute for one another. Recent evidence indicates that the extrastriate body area's degree of feature tuning, including the species-specific tuning relevant to conspecific detection, itself depends on the network it is coupled to, and lateral occipitotemporal cortex fractionates into subdivisions rather than forming a single functional unit. Conspecific detection, on this evidence, is a property achieved by a distributed network, not a property the extrastriate body area supplies independently. We use this distinction to reframe several commentaries' apparent disagreements as answers to different questions rather than competing claims. We close by asking whether this feature/function conflation recurs across other category-selective regions, such as face- and food-selective cortex.