| |
Visual areas in the macaque monkey
Figure 32.1 shows the estimated extent of visual cortex in the macaque and its relationship to other sensory modalities and to motor cortex. This is displayed on a surface-based atlas of the right cerebral hemisphere that was generated from a high-resolution structural MRI volume (Van Essen et al., 2002; Van Essen, 2002; see legend to Fig. 32.1 for details). The top panels show lateral and medial views, respectively, of the fiducial (three-dimensional) configuration; the lower left panel (Fig. 32.1C) is a flat map that shows the entire surface in a single view.
Figure 32.1..
Visual cortex and other functional modalities mapped onto a surface-based atlas of macaque cerebral cortex. The atlas was generated from a high-resolution (0.5 mm3 voxels) structural MRI volume generously provided by N. Logothetis (Case F99UA1, M. mulatta), using the SureFit segmentation method for surface reconstruction and Caret software for surface manipulation and flattening (Van Essen et al., 2001a). This atlas has many advantages over its predecessors, which include manually generated maps (Felleman and Van Essen, 1991; Van Essen and Maunsell, 1980) and surface-based atlases from a hemisphere that lacked corresponding structural MRI data (Van Essen et al., 2001a). A, Lateral view. B, Medial view. C, Flat map. Surface coloring represents different functional modalities, as identified on the flat map; darker shading represents cortex buried in sulci. D, Lateral view of the atlas spherical map, with latitude and longitude isocontours. By convention, the lateral pole is set at the ventral tip of the central sulcus. E, Latitude and longitude isocontours displayed on the flat map. Data for Figures 32.1 to 32.3 can be accessed via http://brainmap.wustl.edu:8081/sums/archivelist.do?archive_id=448857 (see color plate 11).
The dotted lines on the flat map represent the estimated boundary between regions dominated by different modalities, based on the analysis of Felleman and Van Essen (1991). These boundaries are not sharply defined, as there is considerable intermixing of function in transitional regions between modalities. The blending of colors on the maps in Figure 32.1 qualitatively reflects this functional overlap. Cortex that is predominantly or exclusively visual (blue shading) occupies about half (52%) of the cortical surface area (as measured on the fiducial surface rather than on the flat map, which contains significant distortions). This greatly exceeds the amount devoted to other modalities: somatosensory (green, 10%), auditory (red, 3%), motor (magenta, 8%), and olfactory (brown, 1%). Unassigned cortex (gray, 25%) is mostly cognitive or emotional in function but is not subdivided along these lines in the figure.
Figure 32.1 also shows latitude and longitude isocontours that are determined on a spherical map (Panel D) and projected to the flat map (Panel E). As with earth maps, spherical coordinates provide a concise and objective way to specify precise locations on the map (Drury et al., 1999; Fischl et al., 1999a, 1999b; Van Essen et al., 2001a).
The identification of distinct visual areas is generally based on finding reliable differences in one or more characteristics related to (1) architecture, (2) connectivity, (3) visual topography, and/or (4) functional characteristics (Felleman and Van Essen, 1991; Kaas, 1997). Using various combinations of these criteria, numerous partitioning schemes for visual cortex in the macaque monkey have been published over the past century. Figure 32.2 shows 10 such schemes that were mapped to the atlas using a surface-based registration method (Van Essen et al., 1998, 2001b; see legend to Fig. 32.2 for details). Panels A–C illustrate three schemes that encompass most or all of the visual cortex. The Felleman and Van Essen (1991) and Ungerleider and Desimone (1986) schemes are based on anatomical and physiological data from many sources. The Lewis and Van Essen (2000a) scheme is based on an architectonic analysis of multiple hemispheres, with the atlas map generated via surface-based registration of a particular individual map. Panel D shows several schemes that cover more restricted regions, including a connectivity-based analysis of occipital cortex (Lyon and Kaas, 2002) and architectonic analyses of temporal and parietal cortex (Seltzer and Pandya, 1978, 1980, 1986). Panel E shows two additional architectonic schemes covering temporal (Baylis et al., 1987) and parietal (Preuss and Goldman-Rakic, 1991) regions, plus a visuotopic mapping analysis of dorsal occipital cortex (Galletti et al., 1999). While far from an exhaustive compilation, Figure 32.2 includes most of the partitioning schemes in current use that have been reported in a format suitable for registration to the atlas.
Figure 32.2..
Ten partitioning schemes for macaque visual cortex registered to the atlas and displayed on flat map views. See the abbreviations list for full names of areas. Data were registered to the atlas using a surface-based registration method in which geographic (gyral and sulcal) landmarks were used to constrain the registration (Van Essen et al., 2001b). Depending on the data source, this method was applied to computerized maps (spherical or flat maps) and to scanned images of manually generated flat maps or drawings of the hemispheric surface and of schematically opened sulci. Thus, there are substantial differences in the fidelity of the published representations and in the distortions and uncertainties involved in registering to the atlas, but visualization on a common substrate nonetheless remains advantageous. A, Felleman and Van Essen (1991) partitioning scheme for visual areas. Areal boundaries were originally charted on a physical model of an individual hemisphere (case 79O). They were transferred to a computerized surface reconstruction of the same hemisphere and then were registered to the Case F99UA1 atlas using spherical maps of each hemisphere. B, Ungerleider and Desimone (1986) scheme (their Figure 1; see also Figure 2 of Desimone and Ungerleider, 1989). Registration was achieved using geographic landmarks on a manually generated flat map. C, Lewis and Van Essen (2000a) scheme from Case 95DR (their Figure 14) registered from the computerized flat maps. D, A composite map of partitioning schemes for occipital cortex (Lyon and Kaas, 2002, their Figure 4), parietal cortex (Preuss and Goldman-Rakic, 1991, their Figure 4C) and temporal cortex (Baylis et al., 1987, their Figures 2 and 3). The Lyon and Kaas map was based on a visuotopic analysis of V1 projections as displayed on sections of physically flattened cortex. The Preuss and Goldman-Rakic maps were derived from their illustrations of schematically partially unfolded sulci; the Baylis et al. maps were based on cortex flattened using a straight-line unfolding technique. E, A composite map of dorsomedial cortex (Galetti et al., 1999, their Figure 17), and parietal and temporal cortex (Seltzer and Pandya, 1978, 1980, 1986), all illustrated on schematically unfolded sulci. Data can be accessed via http://brainmap.wustl.edu:8081/sums/archivelist.do?archive_id=448857 (see color plate 12).
Although there are many similarities, these partitioning schemes differ in many ways for reasons that reflect various technical impediments faced by cortical cartographers. (1) Subtle boundaries. The distinctions between neighboring regions are often subtle, even when evaluated with the most sensitive anatomical and physiological techniques available. (2) Internal heterogeneity. Many (perhaps most) visual areas are internally heterogeneous. This heterogeneity may be manifested by modularity (repetitive organization at a finer scale than the overall areal dimensions); asymmetries (differences between the representations of the upper [superior] and lower [inferior] visual fields); or internal gradients (gradual shifts in characteristics rather than discrete modularity). (3) Individual variability. The overall size of well-defined visual areas such as V1 and MT can vary twofold or more from one individual to the next (Maunsell and Van Essen, 1987; Van Essen et al., 1984). This is compounded by substantial variability in the exact pattern of convolutions and in the location of areal boundaries relative to gyral and sulcal landmarks.
A compelling case for areal identification entails finding region-specific characteristics that are robust, consistent across multiple approaches, and replicated by multiple laboratories. The diversity among partitioning schemes signifies that a consensus has been achieved for only a minority of the visual areas illustrated in Figure 32.2. Several well-defined areas (V1, V2, V4, and MT) are shown by different colors in Figure 32.2; others are included as part of a regional coloring pattern that includes six clusters discussed below.
The differences among partitioning schemes can be categorized along four lines. (1) Terminological equivalence. Some differences are essentially terminological in that different labels are assigned to what is fundamentally the same visual area. For example, area 17 is equivalent to area V1; MT (Fig. 32.2A–C) is equivalent to V5 (Shipp and Zeki, 1985); and PO (Fig. 32.2A–C) is equivalent to V6 (Fig. 32.2F). (2) Lumping versus splitting. Some regions are considered a single area by some investigators (the lumpers) but as two separate areas by other investigators (the splitters). This is an issue, for example, for subdivisions V3d/V3v, V4d/V4v, and LIPd/LIPv. In some cases, it can be helpful to use the term zone to describe a consistently identifiable region whose status as a distinct area is uncertain, ambiguous, or contentious (Lewis and Van Essen, 2000a). The term subdivision can then encompass both zones and well-defined cortical areas. (3) Boundary uncertainty. Even when there is a consensus on the identity of an area and the main criteria for its identification, the location and extent of any particular area on the atlas map vary substantially across schemes (cf. V2 and MT in Fig. 32.2A–D). These differences may reflect (a) the choice of criteria or emphasis among different criteria available for delineating boundaries; (b) variability in the size and location of areas in the individuals studied; (c) inaccuracies in portraying areal boundaries on a summary diagram in the original publication; and (d) inaccuracies or distortions arising when registering data to the atlas. (4) Genuine incompatibility. In some regions, partitioning schemes differ more profoundly than can be explained by the factors just mentioned. For example, in inferotemporal cortex (blue) there is rather little correlation between the areal boundaries in panel A versus panel E; hence the schemes are largely incompatible.
Occipital Visual Areas
V1 and V2 are both large, well-defined areas, with V1 occupying 13% and V2 occupying 10% of the total cortical surface area (12.6 cm2) in the right hemisphere of the atlas. V1 and V2 have a mirror-symmetric visuotopic organization, with the vertical meridian represented along their common boundary, and with upper fields represented ventrally and lower fields dorsally in each area. In addition, both areas have prominent internal modularity related to the processing streams that course through these areas (Felleman and Van Essen, 1991; Van Essen and Gallant, 1994; Chapter 34). Interestingly, there are significant asymmetries between dorsal V2 (V2d) and ventral V2 (V2v): V2v is notably wider than V2d (Fig. 32.2A–D), and the two subregions differ in the global pattern of thick stripes, thin stripes, and interstripes (Olavarria and Van Essen, 1997).
Area V3 was originally identified as a strip of cortex that adjoins V2 and has mirror-symmetric visuotopic organization (Zeki, 1969). Subsequent studies confirmed this visuotopic organization but revealed that the dorsal and ventral subdivisions are physically separated from one another by the interposition of area V4 (cf. Fig. 32.2A–C). More significantly, evidence for pronounced dorsoventral asymmetries in connectivity (especially with V1), myeloarchitecture, and physiological properties prompted the classification of these dorsal and ventral subdivisions as distinct visual areas: VP (or V3v) and V3 (or V3d), each containing a partial representation of the visual field (cf. Burkhalter et al., 1986). In contrast, Lyon and Kaas (2002) reported relatively symmetric projections of V1 using more sensitive tracers, and they support Zeki's original proposal that V3d and V3v are subdivisions of a single area V3. A sensible interim strategy is to designate these subdivisions as zones V3d and V3v while awaiting additional data on the overall magnitude and nature of V3d/V3v asymmetries, along with insights regarding the asymmetries encountered in other regions and other species (see below). Independent of this terminological debate, it is noteworthy that V3d and V3v vary considerably in location and size among the various maps shown in Figure 32.2.
V4 is a moderately sized area whose lower-field (V4d) and upper-field (V4v) representations are contiguous with one another, in contrast to the physical separation of V3d and V3v. The visuotopic organization of dorsal V4 appears to be more complex and variable across individuals than that of ventral V4 (Boussaud et al., 1991; Gattass et al., 1988; Maguire and Baizer, 1984; Van Essen et al., 1990). Because analogous but more pronounced dorsoventral asymmetries have been reported for the corresponding region of human cortex (see below), Tootell and Hadjikhani (2001) have suggested that macaque V4d and V4v might themselves constitute distinct visual areas.
Five additional areas adjoin or closely approach V2 along its dorsomedial and ventromedial boundaries. The dorsomedial cluster (magenta in Fig. 32.2) includes areas V3A, PIP, and PO, each of which includes an upper-field as well as a lower-field representation (Colby et al., 1988; Felleman and Van Essen, 1991; Lyon and Kaas, 2002). Area PO, as charted by Colby et al. (1988), corresponds to area V6 of Galletti et al. (1999). The ventromedial cluster (brown in Fig. 32.2) includes areas TF and TH (near the bottom of the flat map), both of which receive visual inputs (see Felleman and Van Essen, 1991). The posterior portion of TF, identified as VTF by Boussaud et al. (1991), is visually responsive but has at best a crude visuotopic organization.
Dorsal Temporal and Posterior Parietal Cortex
The dorsal stream of visual areas (Chapter 34) includes cortex in and near the intraparietal sulcus, plus the dorsal part of the superior temporal sulcus. Area MT, the most extensively studied area in the dorsal stream, contains a complete visuotopic map, but with a bias toward lower versus upper visual fields (Maunsell and Van Essen, 1987). Though MT is well defined by several criteria (including a high incidence of direction-selective cells), its precise location and extent on different summary maps vary significantly (cf. Fig. 32.2A–D; see also Desimone and Ungerleider, 1986). Adjoining MT on its lateral side is a narrow strip that has been identified as V4t (Felleman and Van Essen, 1991; Ungerleider and Desimone, 1986) or MTc (Lyon and Kaas, 2002). Within V4t, architectonically distinct zones V4ta and V4tp have been described (Lewis and Van Essen, 2000a).
A major target of MT is the MST complex (orange in Fig. 32.2), which lies dorsal and medial to MT and is implicated in higher-order motion analysis. MST is heterogeneous both anatomically and physiologically, but it remains unclear whether there are two subdivisions (MSTd and MSTl of Komatsu and Wurtz, 1988a, 1988b, and Felleman and Van Essen, 1991; MSTc and MSTp of Boussaud et al., 1990) or three (MSTdp, MSTm, and MSTl of Lewis and Van Essen, 2000a) and whether these constitute distinct areas or zones within a larger MST complex. Area FST lies ventral to the MST complex and medial to MT, but its position differs markedly in different partitioning schemes (Fig. 32.2A versus 32.2B, C).
Posterior parietal visual cortex (yellow in Fig. 32.2) includes several elongated areas in and near the intraparietal sulcus, plus additional areas in more posterior regions that are represented in some schemes. The most fine-grained scheme (Lewis and Van Essen, 2000a; Fig. 32.2C) is based on immunohistochemical as well as cyto- and myeloarchitecture and includes seven distinct areas (7a, LIPd, LIPv, VIP, MIP, MDP, and DP) plus additional architectonic zones. In the region previously described as area LIP (Blatt et al., 1990), subdivisions LIPd and LIPv differ from one another in architecture, connectivity, and physiological characteristics and are thus likely to constitute separate visual areas (cf. Lewis and Van Essen, 2000a, 2000b). Area VIP, along the fundus of the intraparietal sulcus, contains lateral (VIPl) and medial (VIPm) architectonic zones, but these are not considered distinct areas on the basis of current evidence (Lewis and Van Essen, 2000a, 2000b). Areas LIPd, LIPv, and VIP of Lewis and Van Essen (2000a) largely correspond with POa–e, Poa–i, and IPd of Seltzer and Pandya (1980, 1986), except that they are more restricted anteroposteriorly. The relationships with other schemes are somewhat confusing because of “cross-talk” in the terminology: area VIP* of Ungerleider and Desimone corresponds to LIPv of Lewis and Van Essen; areas LIP and VIP of Felleman and Van Essen correspond to LIPd and LIPv/VIPl, respectively, of Lewis and Van Essen.
A second group of areas, near the posterior end of the intraparietal sulcus, includes areas MDP, MIP, and DP, plus zone LOP (Fig. 32.2C). Area MDP corresponds to V6A of Galetti et al. (1999). Zone LOP may correspond to the cIPS subdivision identified physiologically by Taira et al. (2000) based on surface orientation selectivity.
Ventral Temporal Areas
Cortex ventral to area MT and anterior to V4 is generally considered part of the ventral processing stream (cf. Chapter 34). This region includes a cluster of inferotemporal areas (blue in Fig. 32.2) that are implicated mainly in pattern recognition and form analysis (Desimone and Ungerleider, 1989; Tanaka, 1997). Another cluster of ventral-STS areas (green in Fig. 32.2) includes polysensory regions and regions involved in both form and motion processing (Cusick, 1997; Oram and Perrett, 1996). The five schemes in Figure 32.2 subdivide ventral temporal cortex along both anteroposterior and dorsoventral axes, but the number of identified subdivisions ranges from 4 to 10 and the location of areal boundaries varies widely as well.
The simplest scheme (Ungerleider and Desimone, 1986; Fig. 32.2B) includes a visuotopically organized area TEO that contains a lower-field and an upper-field representation (see also Boussaud et al., 1991), plus three areas (TEc, TEr, and TEm) that lack clear visuotopic organization. The Felleman and Van Essen (1991) scheme, based mainly on topographic and connectional data, includes three areas (VOT, PITd, and PITv) in the region corresponding to TEO of Ungerleider and Desimone (1986). VOT contains only an upper-field representation, whereas PITd and PITv are reported to each contain a crude topography with lower-field and upper-field inputs (Van Essen et al., 1990). Areas CITd and CITv of Felleman and Van Essen (1991) together correspond approximately to TEc of Ungerleider and Desimone (1986) but are distinguished from one another by differential connections. More anteriorly on the atlas map, there is reasonable correspondence between AITd and TEr in the two schemes but poor correspondence between AITv and TEm.
The Seltzer and Pandya (1978) architectonic scheme (Fig. 32.2E) includes five inferotemporal areas (TEa, TEm, TE1, TE2, and TE3), plus four additional areas in the superior temporal sulcus (TAa, TPO, PGa, and IPa). The Baylis et al. (1987) scheme (Fig. 32.2D) uses the same criteria and terminology as Seltzer and Pandya (1978), but the areal boundaries differ significantly on the atlas map. The Lewis and Van Essen (2000a) scheme, like that of Cusick (1997), identifies three subdivisions within TPO (TPOc, TPOi, and TPOr). In inferotemporal cortex, the Lewis and Van Essen (2000a) scheme does not discriminate among TE1, TE2, and TE3 but does identify dorsoventral subdivisions TE1-3d and TE1-3v. The latter correspond approximately to subdivisions TEad and TEav of Yukie et al. (1990) and Saleem and Tanaka (1996).
Frontal Areas
Felleman and Van Essen classified areas 8 (FEF) and 46 as part of visual cortex. Areas 8 and 46 have both been partitioned into multiple subdivisions, including eight zones (8A, 8Am, 8Ar, 46d, 46v, 46dr, 46vr, and 46r) by Preuss and Goldman-Rakic (1991) and zones 8Ac, 46p, and 46v by Lewis and Van Essen (2000a).
It remains unclear which of the aforementioned schemes for inferotemporal and parietal cortex best reflect the underlying neurobiological reality. Indeed, alternative possibilities worth bearing in mind are that much of inferotemporal and frontal cortex might contain gradual gradients of properties rather than sharply defined areas, or else irregular clustering whose arrangement varies markedly from one individual to the next.
Figure 32.3 shows a composite map in which all of the 10 schemes represented in the preceding panels are superimposed and displayed in lateral (Fig. 32.3A), medial (Fig. 32.3B), and flat map (Fig. 32.3C) views. Besides the individually colored areas, the composite map includes the six clusters discussed in preceding sections: the dorsomedial occipital complex (DMOcx), posterior parietal complex (PPcx), MST complex (MSTcx), ventral superior temporal complex (VSTcx), inferotemporal complex (ITcx), and TF/TH cluster.
Figure 32.3..
A–C, Lateral, medial, and flat map views of a composite map obtained by superimposing the data from Figure 32.2. Although the specific areal assignments at each location cannot be ascertained by viewing this image, they can be readily obtained using a node-identification option in the Caret visualization software. See the abbreviations list for full names and Figure 32.1 legend for data access information. (See color plate 13.)
How Many Visual Subdivisions Are There?
It is naturally of interest to have a current estimate of the total number of cortical areas and subdivisions associated with visual processing. Given the diversity among partitioning schemes, it is impractical to arrive at a definitive single number. Altogether, there is substantial evidence for more than 40 anatomically and/or functionally distinct subdivisions of visually responsive cortex, but for some the evidence does not warrant identification as separate areas. A conservative lumper can argue that the number of convincing visual areas is only about two dozen. A more generous splitter can make a credible case for at least three dozen distinct areas. The number of areas that show clear visuotopic organization is also difficult to determine precisely but is in the range of 10–12, depending on the designation of zones versus areas.
Coping with Multiple Schemes, Areal Uncertainties, and Individual Variability
Several general observations emerge from the preceding discussion. First, a multiplicity of competing partitioning schemes will likely persist for years, because many of the issues remain difficult to resolve with currently available techniques. Second, it is becoming increasingly important to have objective methods for quantifying and visualizing the uncertainties associated with charting areal boundaries and the variability associated with individual differences in cortical organization. This leads to the notion of probabilistic surface-based maps of cortical organization. The composite map in Figure 32.3 constitutes one step in this direction, as it includes many different schemes in a common coordinate system. Another important aspect of a probabilistic approach is to incorporate information about individual variability by registering maps of many individuals on the atlas (Van Essen et al., 2001a).
Another important issue involves the options for accessing and extracting information from surface-based atlases. The fundamental challenge is that existing atlases already contain far more information than can be easily gleaned from static pictorial images such as those shown in this chapter. Moreover, the amount of information represented on such atlases is likely to grow exponentially over the next decade. An attractive alternative is to use computerized visualization software to access atlas data more efficiently and flexibly. For the atlases illustrated here, this can be done by downloading specific data sets (using hyperlinks included in the figure legends) and viewing the maps using the freely available Caret software (Van Essen et al., 2001b; http://brainmap.wustl.edu/caret). Existing visualization options in Caret include zooming in on regions of interest; rapid toggling between different partitioning schemes; laying one scheme directly over another; listing areal identities for multiple partitioning schemes when clicking on locations of interest; specifying location by latitude and longitude (cf. Fig. 32.1E); encoding uncertainty limits for each boundary; and encoding variability by combining maps from different individuals.
Human Visual Cortex
The analysis of human visual cortex has benefited greatly from the advent of structural and functional MRI methods that can be routinely carried out on normal subjects. In the top half of Figure 32.4, the extent of visual cortex and other functional modalities is charted on the right hemisphere of a surface-based atlas (the Human.Colin atlas) that was generated from high-resolution structural MRI of a particular individual (Holmes et al., 1998; Van Essen, 2002; Van Essen et al., 2002). The surface displays include lateral and medial views of the fiducial surface (Fig. 32.4A, B) and the inflated surface (Fig. 32.4C, D), plus a flat map (Fig. 32.4E). The assignment of different functional modalities was estimated mainly from functional neuroimaging data that were mapped onto the atlas using a combination of methods applied to a number of published data sets (see the legend to Fig. 32.4). Based on these provisional assignments, cortex that is predominantly visual in function occupies about 27% of the total extent of cerebral cortex (950 cm2) in the right hemisphere of the atlas. By comparison, about 8% of cortex is predominantly auditory, 7% somatosensory, and 7% motor. The remaining half (51%) includes major domains associated with cognition, emotion, and language.
Figure 32.4..
Visual cortex and other functional modalities mapped onto a surface-based atlas of human cerebral cortex. The atlas was generated from a high-resolution structural MRI volume provided by A. Toga (Case Human.colin) using SureFit and Caret software (cf. Fig. 32.1). This atlas (Van Essen et al., 2002) is higher in quality and supersedes the Visible Man surface-based atlas previously published (Van Essen and Drury, 1997; Van Essen et al., 2001a). Modality assignments were based on (1) stereotactic mapping of fMRI volume data onto the atlas surface from the studies of Corbetta et al. (1998), Lewis and DeYoe (2000), and Lewis et al. (2001); (2) stereotactic projection of published Talairach coordinates of activation focus centers (Ishai et al., 1999; Kanwisher et al., 1997; Van Essen and Drury, 1997); and (3) manually transposed boundaries related to local geographic and/or functional landmarks from surface maps published by Burton and Sinclair (2000), Press et al. (2001), and Tootell & Hadjikhani (2001). Data for Figures 32.4 and 32.5 can be downloaded via http://brainmap.wustl.edu:8081/sums/archivelist.do?archive_id=449900 (See color plate 14).
Human area V1 (area 17) has a well-defined architectonic boundary that runs near the margins of the calcarine sulcus but with considerable individual variability (Rademacher et al., 1993). The estimated surface area of V1 on the atlas map (charted using the most likely boundary in the Rademacher et al. study) is 21 cm2, or 2.2% of cerebral cortex. This is about one-sixth of its fractional occupancy on the macaque atlas.
Neuroimaging studies have revealed numerous visuotopically organized extrastriate areas in human occipital cortex in an arrangement that shows many similarities to the pattern found in the macaque. Corresponding areas that have fundamentally similar visuotopic organization in the two species include V1, V2, V3 (V3d), VP (V3v), V3A, and V4v (DeYoe et al., 1996; Hadjikhani et al., 1998; Sereno et al., 1995). In addition, tests for motion-related activation have consistently demonstrated a prominent focus in or near the posterior inferotemporal sulcus. This focus is generally presumed to correspond to macaque MT plus part of the MST complex and has therefore been identified as human MT+ (Tootell et al., 1995) or as V5 (Watson et al., 1993).
Figure 32.5 includes many visuotopic areas that were mapped onto the human atlas using surface-based registration of an individual case from Hadjikhani et al. (1998). For technical reasons these maps do not include the representation of the fovea or far periphery, so the extent of the extrastriate areas is likely to be modestly underestimated. Interestingly, the sizes of V3d, V3v, and V3A relative to V1 and V2 are all substantially greater in humans than in the macaque, suggesting a marked evolutionary divergence in the relative sizes of nearby cortical areas. In the region adjoining dorsal V3, Smith et al. (1998) and Press et al. (2001) found evidence for two representations, V3A and V3B, instead of a single V3A, with V3B being located more posterior (lower on the flat map). Area V7 (Press et al., 2001; Tootell and Hadjikhani, 2001) is a visuotopically organized area that lies anterior to V3A; its location and extent are displayed on the atlas using dotted boundaries to signify greater uncertainty.
Figure 32.5..
Identified visual areas on the Human.colin surface-based atlas. See the abbreviations list for full names. The boundary of MT+ on the atlas map was estimated by projecting the motion-related activation foci tabulated by Van Essen and Drury (1997) onto the atlas surface. The visuotopic map from Figure 5A of Tootell and Hadjikhani (2001) was then registered to the atlas using the V1/V2 boundary and the posterior boundary of MT+ as landmarks. Additional areas LO and V7 were drawn manually onto the atlas, based respectively on the studies of Tootell and Hadjikhani (2001) and Press et al. (2001). A, Lateral view. B, Medial view. C, Inflated lateral view. D, Inflated medial view. E, Flat map view. The red dotted contour represents the main cluster activation focus center involved in face and place analysis (see text). (See color plate 15.)
Human area V4v lies anterolateral to VP/V3v and contains a mirror-symmetric upper-field representation. Interestingly, cortex lateral and dorsal to V4v, identified as LO (Van Oostende et al., 1997) or as the LOC/LOP complex (Tootell and Hadjikhani, 2001), evidently lacks a complementary lower-field representation equivalent to that described for macaque V4d. Instead, the LOC/LOP region represents both upper and lower fields, but with an irregular and inconsistent map of the polar angle and an apparent discontinuity in the transition between central and peripheral eccentricities. LO as described by Grill-Spector et al. (1998a,b; 2001) includes much of LOC/LOP but extends further ventrally and anteriorly (down and to the right on the flat map). Area V8 lies anterior to V4v and contains a complete visuotopic map with a foveal representation near its anterior margin to the right on the flat map (Hadjikhani et al., 1998; but see Wade et al., 2002).
Tests using a variety of stimuli and behavioral paradigms besides those just discussed have revealed several additional regions of functional specialization. Ventral occipitotemporal cortex contains a region preferentially activated by object versus nonobject stimuli. Based on published stereotactic (Talairach) coordinates, foci involved in analysis of faces, houses, and chairs are concentrated in a region (red dotted outline in Fig. 32.5E) that partially overlaps V8 on the atlas map but extends more anteriorly (Haxby et al., 1999; Ishai et al., 1999; Kanwisher et al., 1997). This region has been proposed to include a fusiform face area (FF) and a parahippocampal place area (PP) (Epstein and Kanwisher, 1998; Kanwisher et al., 1997) and is accordingly identified as FF/PP in Figure 32.5E. Other studies suggest that these may not constitute distinct areas but rather a functional gradient within a larger subdivision (Haxby et al., 2001).
In the parietal lobe, visually related activations have been reported in a large swath of cortex in and near the intraparietal sulcus, dorsal and anterior to the known visuotopic areas. Paradigms effective in activating this region include shifting visual attention, target-directed eye movements, visual motion, and spatial analyses (Corbetta et al., 1998; Haxby et al., 1994; Lewis and DeYoe, 2000; Lewis et al., 2001). This region presumably includes a number of functionally distinct areas or subdivisions that have not been clearly discriminated using existing paradigms and analysis methods.
| |
