Mapping of Extrastriate Visual Cortex

      The cerebral cortex is a rather disappointing tissue at first glance. Its surface appears as a convoluted, gray sheet of tissue with no recognizable boundaries. In cross-section, architectonic borders can be identified with stains for cells and myelin, but only between a few areas. In a classic work, Korbinian Brodmann compiled a complete histological survey of the cerebral cortex, dividing it into 47 distinct areas. Sadly, most of the boundaries he drew have not stood the test of time.

      For example, in the visual cortex Brodmann recognized three regions: areas 17, 18, and 19. Area 17, the primary visual cortex (V1), is easy to identify because it contains a prominent myelin band, the stria of Gennari, and a precise representation of the contralateral field of vision. However, the boundary between area 18 and 19 is invisible, and it is clear that there are far more than 3 visual areas in the primate brain. After processing in area 17, visual information is distributed for further analysis to at least a dozen surrounding visual areas. These areas are known collectively as “extrastriate visual cortex”. Mapping the boundaries, organization, and projections of these visual areas is a difficult challenge, in part because the primate cerebral cortex is highly folded into sulci and gyri. To facilitate anatomical studies of the macaque cortex, we have developed a method for flattening it prior to preparation of histological sections. The flatmount technique allows one to obtain single sections of the entire cortex, greatly facilitating the analysis of anatomical data. Figure 9 shows an example of a flattened hemisphere from a macaque monkey:

Fig. 9) Flattened right hemisphere from a macaque monkey. A full description of the flatmount technique is provided in: Sincich LC, Adams DL, and Horton JC, Visual Neuroscience, 2003.

Function of Extrastriate Cortex

      Why does the brain contain so many cortical areas for vision? A generation ago, visual neuroscience was in the throes of a debate about how information is disseminated from the primary visual cortex (V1) to higher visual areas in the primate brain. Traditionalists believed that the visual image is first digested in V1, and then passed essentially intact through a series of higher cortical areas for further processing to extract perception. Revisionists proposed that images are broken down by V1 into various components, such as color, form and motion. According to this scheme, these separate components are then distributed via parallel projections to extrastriate areas specialized for their analysis. The matter appeared to be settled by a series of studies conducted by Livingstone and Hubel (1984, 1987). First, they identified three compartments in V1 corresponding to color, form and motion. The color compartments were equated with the cytochrome oxidase patches in layers 2/3, the form compartments were designated as the interpatches in layers 2/3, and the motion compartment was assigned to layer 4b. Second, they demonstrated that each compartment in V1 sends a segregated projection to a distinct compartment in the next visual area (V2). Finally, they showed that the compartments in V2 are also divided by color, form, and motion. Contemporaneously, other investigators traced specific connections from V2 compartments to higher visual areas (V4, MT). Figure 10 summarizes this proposed pattern of connections:

Fig. 10) Compartmentalization of visual information from the lateral geniculate body to extrastriate cortex, according the scheme proposed by Livingstone and Hubel (Science 1988).


      Livingstone and Hubel’s work furnished powerful support for the idea that color, form, and motion can be assigned to distinct cortical compartments in V1 and V2. However, we have undertaken a re-examination of the projections from V1 to V2 and discovered a pattern of connections that conflicts with Livingstone and Hubel’s original description. The discrepancies highlighted by our new anatomical data cast doubt on their account of the projections from V1 to V2 and cast doubt on the notion that color, form, and motion processing are well segregated in the early visual cortex.

      In area V2 (the second visual area), the cytochrome oxidase stain has revealed three parallel compartments, organized into a system of repeating thick-pale-thin-pale stripes that stretch from the V1 border to the V3 border:

Fig. 11) Cytochrome oxidase section showing parallel stripes in macaque V2, organized into a repeating sequence of pale and dark stripes. The dark stripes alternate between thick (large arrows) and thin (small arrows). From: Horton JC, Hocking DR. J. Neuroscience, 1996.

      Rather than arising from three compartments, we have found that projections from V1 to V2 originate from only two separate compartments: cytochrome oxidase patches and interpatches. They terminate in different compartments in V2. The patches send projections to thin stripes, whereas the interpatches send projections to thick stripes and pale stripes. The projections are not equal in strength: the strongest output from V1 terminates in pale stripes (Sincich LC, Horton JC. J. Comparative Neurology, 2002).

      The projections from V1 arise from layers 2/3, 4A, 4B, and 5/6. Only layers 1 and 4C do not project to V2. To determine if segregated (but physically intermingled) interpatch populations supply either pale stripes or thick stripes, we made paired injections of different retrograde tracers into V2. We succeeded in placing tracer injections into adjacent pale and thick stripes. Up to a third of the retrogradely filled cells in V1 were double-labeled, indicating that many interpatch cells project indiscriminately to both pale and thick stripes. Figure 12 below shows our revised diagram of projections from V1 to extrastriate cortex.

Fig. 12) Bipartite pattern of projections from V1 to V2. Cells in patches project to thin stripes, whereas cells in interpatches project to thick and pale stripes. Cells in layer 4B also project directly to area MT. ( Sincich LC, Horton JC. Science, 2002).

      Our studies have identified numerous projections from V1 to V2 that were previously unknown. They include outputs from layer 4B to pale stripes, layer 4B to thin stripes, and layers 2/3 to thick stripes. This new account of the V1 to V2 pathway is difficult to reconcile with the prior model (Fig. 10) based on the idea that only three projections exist from V1 to V2. The old model proposed that each projection carried information about color, form, and motion to thin, pale, and thick stripes respectively. Instead, we find that each V2 stripe type is richly supplied by all output layers of V1 (Fig. 12).

      Previously, it was reported that layer 4B is a magno-dominated layer that projects exclusively to thick stripes. Although it is true that layer 4B receives strong input from magno-recipient layer 4Ca (Fitzpatrick et al., 1985), it also receives major parvo input (Yabuta et al., 2001). Moreover, our data show that layer 4B projects to all stripe types in V2. In light of these facts, it is inaccurate to construe layer 4B as simply a magno channel for conveying motion information to extrastriate cortex.

      Movshon and Newsome (1996) used antidromic stimulation to characterize the physiological properties of layer 4B neurons that project directly to area MT. The units were oriented and highly direction-selective. Thus, it appears that at least one subset of neurons in layer 4B does convey information about motion direction to area MT. Do these same units project to area V2? To address this issue, we have injected MT and V2 with different tracers and examined the retrogradely filled cells in layer 4B (Sincich LC, Horton JC. J. Neuroscience, 2003). Only about 5% of cells were double-labeled. This result means that distinct populations of cells in layer 4B supply areas V2 and MT, underscoring the specificity of intercortical connections.

      The pattern of projections that we report from V1 to V2 vitiates a sharp division of form, color, and motion/stereopsis by CO stripe class. Localization of form, color, and motion perception certainly occurs in extrastriate cortex, at least in some primates. Achromatopsia in humans, for example, provides compelling evidence that color is localized to the fusiform and lingual gyri. Our point is that form, color, and motion should not be equated with pale, thin, and thick V2 stripes respectively. Moreover, it is misleading to associate parvo and konio with the “ventral” pathway and magno with the “dorsal” pathway. It is likely, given that interpatches project to both pale stripes and thick stripes, that parvo and magno inputs are distributed to both V4 and MT.


Bypassing V1: A Direct Geniculate Input to Area MT

      Historically, the extrastriate cortical regions adjacent to the primary visual cortex were defined as “higher” because they were not thought to receive direct geniculate input. In humans, loss of V1 devastates eyesight by cutting off the flow of visual information from the LGN to extrastriate visual cortex. Curiously, patients affected by such lesions manifest some residual visual perception. This perception is optimal for moving stimuli, which occurs either consciously (Riddoch syndrome) or unconsciously (blindsight). The preservation of some vision following loss of the primary visual cortex has engendered considerable controversy because it defies conventional ideas about the organization of the visual system.

      The simplest explanation for blindsight phenomena is that a visual pathway exists that bypasses V1 to reach area MT, a region in the superior temporal sulcus implicated in motion perception. We have discovered a direct projection from the lateral geniculate nucleus to the motion-selective area MT, a cortical area not previously considered “primary”. (Sincich LC, Park KF, Wohlgemuth MJ, Horton JC. Nature Neuroscience, 2004). The constituent neurons are mostly koniocellular, send virtually no collateral axons to primary visual cortex (V1), and equal about 10% of the V1 population innervating MT. This pathway could explain the persistence of motion sensitivity in subjects following injury to V1.