In some retinotopic occipital regions (for example, V1 and V2), there is one unambiguous way to parcel BOLD topographic data into discrete, coherent regions. The most parsimonious boundaries for these topographic regions are immediately obvious, and virtually all investigators accept these boundaries as likely boundaries of cortical areas. The opposite extreme applies in large portions of frontal, temporal and parietal cortex; these regions may be divisible into cortical areas by other metrics, but they are blank on currently available thresholded topographic maps. But what about regions between these extremes? In much of occipital and parts of frontal, temporal and parietal cortex there is evidence for topographic organization, but unambiguous locations for cortical area boundaries do not leap out of the data. When this occurs, can we parcel the difficult topographic data into likely cortical area locations? On which a priori expectations can we rely? How confident can we be in our conclusions?
We review the variety of conceptual approaches that different neuroimaging laboratories have applied to these basic questions. Because the topographic organization in some parts of the human brain may be complex, we also review some complex topographic organizations previously documented in nonhuman species.
Finally, we provide several novel tests that we have found helpful when considering various topographic parcellations as putative cortical areas. There are undoubtedly many cortical areas that have some topography but do not pass all of these tests; however, each test that a topographic parcellation does pass lends additional weight to the notion that it is equivalent to a cortical area. To illustrate each test, we show how it is passed in our human topographic parcellation V4. This parcellation accounts more parsimoniously for the available data than either of the two competing sets of parcellations, V4v/V4d-topo/V8 and hV4/LO1.