The processing of visual shape in the cerebral cortex of human and nonhuman primates: a functional magnetic resonance imaging study

Abstract

We compared neural substrates of two-dimensional shape processing in human and nonhuman primates using functional magnetic resonance (MR) imaging in awake subjects. The comparison of MR activity evoked by viewing intact and scrambled images of objects revealed shape-sensitive regions in occipital, temporal, and parietal cortex of both humans and macaques. Intraparietal cortex in monkeys was relatively more two-dimensional shape sensitive than that of humans. In both species, there was an interaction between scrambling and type of stimuli (grayscale images and drawings), but the effect of stimulus type was much stronger in monkeys than in humans. Shape- and motion-sensitive regions overlapped to some degree. However, this overlap was much more marked in humans than in monkeys. The shape-sensitive regions can be used to constrain the warping of monkey to human cortex and suggest a large expansion of lateral parietal and superior temporal cortex in humans compared with monkeys.

Figure 1.

Stimuli. A-D, Intact and scrambled grayscale images (A, B) and drawings (C,D) of familiar (A, C) and novel (B, D) objects. Same stimuli as Kourtzi and Kanwisher (2000). E, Orientation change of fixation target in high acuity task superimposed on a familiar object (bar not to scale). F, Original and distorted grayscale stimulus (familiar object). G, Familiar object with and without a portion dimmed (stimulus-dimming task). Unlike what is shown in E-G, a grid was always present in the stimuli presented to the subjects.

Figure 4.

A, B, Shape sensitivity on flattened full hemispheres of human (A) and monkey (B) atlas maps. Red-to-yellow shading indicates surface nodes that lie in voxels (resampled to 1 mm³ for human and 0.5 mm³ for macaque) with significant activation (p < 0.0001 uncorrected, random effect, group, in A; p < 0.05 corrected, fixed effect, group, in B). Green nodes correspond to surface nodes lying immediately adjacent to significant voxels (within an additional 0.5 mm for the macaque, within 3 mm for human); this provides a reasonable approximation of the limited resolution and spatial uncertainties associated with the fMRI mapping process [coregistration of functional MRI with structural MRI plus volume-based registration of individual brains to SPM template (A) or M3 brain (B)]. Blue nodes on the macaque map indicate locations where an activation that mapped to one bank of a sulcus was even stronger on the opposite bank of the sulcus (as in the IPS) or in a neighboring sulcus across a narrow white matter gap (as in the arcuate sulcus in prefrontal cortex). This method of compensating for spatial blurring of the fMRI activations was applied to the macaque but not the human data, because the volume-based registration process is less effective at compensating for individual variability of convolutions in humans compared with macaques. Data sets are accessible for visualization or downloading via http://pulvinar.wustl.edu:8081/sums/archivelist.do?archiveid = 665489 (human colin atlas) and http://pulvinar.wustl.edu:8081/sums/archivelist.do?archiveid = 665817 (macaque case M3).

Figure 11.

Registration of macaque to human cortex using a standard set of landmarks likely to reflect homologies across species and additional landmarks derived from shape-sensitive regions. A, B, Landmarks displayed on right hemisphere of monkey (A) and human (B). C, Shape-specific activation on the macaque atlas map, with boundaries of selected visual areas from the Lewis and Van Essen (2000) scheme superimposed. D, Shape-specific activation on the human atlas map, with boundaries of selected visual areas from Hadjikhani et al. (1998) and Tootell and Hadjikhani (2001). E, Shape-specific macaque activations after deformation to the human flat map using standard landmarks, with boundaries of selected human visual areas superimposed. Arrows 1-4 indicate fMRI activations that are not in register between the human and deformed macaque maps. F, Same as in E; shape-specific macaque activations after deformation to the human flat map using standard and additional landmarks. In A and B, landmarks include areas V1, V2, MT (posterior and anterior borders), A1 (posterior-lateral border of primary auditory cortex), boundary between areas 3 and 4 along the fundus of the central sulcus, medial and lateral borders of the hippocampal complex (HC), FEF, olfactory/gustatory cortex (Olf./Gust.), orbital sulcus (Orb S.), and posterior and anterior boundaries along the corpus callosum (CC-post., CC-ant.). In D-F, the region identified as LOC/LOP by Tootell and Hadjikhani (2000) has been labeled LOS (see Results). Data sets are accessible for visualization or downloading via http://pulvinar.wustl.edu:8081/sums/archivelist.do?archiveid = 665489 (human colin atlas) and http://pulvinar.wustl.edu:8081/sums/archivelist.do?archiveid = 665801 (macaque F99UA1 atlas).

Figure 2.

Shape-sensitive regions in humans. Statistical parametric maps (T-score maps) indicating voxels (colored yellow to red) significantly more active when viewing intact than scrambled images in the group of 17 subjects superimposed on the flattened (FreeSurfer) left and right hemisphere (A) of subject S1 and on the right hemisphere of subject S2 without task (B) and with task (C). Threshold in A is p < 0.0001 uncorrected (random effect); p < 0.05 corrected in B and C. The labels of the eight shape-sensitive regions (Table 1) are indicated in black. InA, the white outlines indicate the motion-sensitive regions (Orban etal., 2003): full lines, p < 0.05 corrected; dashed lines, p < 0.0005 corrected (random effect; 30 subjects). The small letters indicate the local maxima of the motion-sensitive regions: a, hMT/V5+; b, hV3A; c, LOS; d, lingual region; e, VIPS; f, POIPS; g, DIPSM; h, DIPSA; i, PIC; j, fusiform. In B and C, selected motion-sensitive regions (only hMT/V5+, hV3A, and VIPS; p < 0.05 corrected) are also indicated by white lines. In B and C, the black lines refer to the retinotopy: full lines, horizontal meridian (HM); stippled lines, vertical meridian (VM); dashed line, central 1.5°. These were obtained in the subtractions: HM minus VM, VM minus HM, and central minus peripheral stimuli. The major sulci are indicated in green: LatS, lateral sulcus; OTS, occipitotemporal sulcus; CollS, collateral sulcus; POS, parieto-occipital sulcus.

Figure 3.

Overlap between shape and motion sensitivity in human LOS. A, C-F, SPMs indicating shape-sensitive voxels (red), motion-sensitive voxels (green), and voxels with both sensitivities (yellow) on coronal sections of four human subjects (S2, S4, S10, S13) scanned at 1.5 and 3 T (S13). B, Activity profiles, plotting change in MR activity relative to fixation in the four shape conditions (red) and relative to static in the two motion conditions (green), of the three parts of LOS of S2, as indicated. Thresholds in A and C-F, p < 0.001 uncorrected; y indicates the level of the section; dashed white oval, LOS region; L, left; R, right.

Figure 5.

Shape-sensitive regions in monkeys. A-C, Statistical parametric maps (T-score maps) indicating that voxels (colored yellow to red) are significantly more active when viewing intact than scrambled images in the group of four subjects superimposed on the flattened (FreeSurfer) left and right hemispheres (A) of subject M3 and in the right hemisphere of subject M5 without task (B) and with task (C). Threshold is p < 0.05 corrected (fixed effect) in all panels. The labels of the nine shape-sensitive regions (Table 2) are indicated in black. The white and black lines indicate retinotopic organization (of M3 in A, of M5 in B and C); full white lines, horizontal median; dashed white lines, vertical median; dashed black lines and stars, central visual field (1.5°); U and L indicate upper and lower visual fields, respectively (Fize et al., 2003). Blue labels indicate retinotopic regions and MT/V5 identified from motion localizer. Green labels indicate sulci; LaS, lateral sulcus; OTS, occipitotemporal sulcus; IOS, inferior occipital sulcus; LuS, lunate sulcus; POS, parieto-occipital sulcus.

Figure 6.

Shape-sensitive regions in sections through monkey brain. A-C, Statistical parametric maps (T-score maps) indicating voxels (colored yellow to red) significantly more active when viewing intact than scrambled images on sagital (A), coronal (B), and horizontal (C) sections of individual monkeys (M1, M4, and M5). The white labels indicate shape-sensitive regions. x, y, and z indicate levels of section (stereotaxic coordinates). L, Left; R, right. Vertical white line in C indicates coronal level, y = 0. Other conventions are the same as in Figure 5.

Figure 7.

Early shape-sensitive regions. A-F, Activity profiles of monkey V3d (A), V3v (B), V4 (C), human hV3A (D,E), and human LOS (F). All profiles were obtained in the group analysis of the scrambling effect, except in E where the local V3A maxima of the motion localizer of each single subject (ss) was probed and profiles of subjects averaged. Open and hatched bars indicate intact and scrambled images, respectively. G, Grayscale images; SG, scrambled grayscale images; L, line drawings; SL, scrambled drawings. Vertical bars indicate SEM (over all volumes). For significance of the main effects and interactions, see Tables 1 and 2.

Figure 8.

Occipitotemporal regions. A-D, Activity profiles of monkey mSTS (A) and TEda (B) and of human post-ITG (C) and mid-FG (D). All profiles from group analysis. For each type of stimulus, the first bar indicates average (A) data, the second data for familiar (F) objects, and the third data for novel (N) objects. Same conventions as in Figure 7.

Figure 9.

Motion effect [100 × (motion - stationary/stationary), full line] and scrambling effect [100 × (intact - scrambled/scrambled), dotted line] as a function of position in the lateral bank of IPS of monkey (A-C, coronal levels, y = 2 to -2) and of human (F, G, coronal levels, -85 and -87). The sampled positions along the bank of M5, right hemisphere (y = 0) and S2 left hemisphere (y = -85) are indicated on coronal sections in D and E, respectively. Coronal level, y = 0, is indicated in Figure 6C. A-C, Filled arrows, LIP. F, G, Filled arrows, VIPS. White numbers in D and E indicate position of local maxima.

Figure 10.

Shape sensitivity in passive and task conditions in monkey. Activity profiles of monkey M3 (A, E), mSTS (B, F), TEda (C,G), and LIP (D, H) are shown. A-D, Data obtained in experiments comparing passive viewing (P; white bars) with high acuity task (HA; green bars). E-H, Data obtained in experiments comparing passive viewing (P; white bars), stimulus-dimming task (DS; light blue), and fixation-dimming task (DF; dark blue). A-D, Data from a group of two monkeys. E-H, Data from M3 (lower MR activity is attributable to an exceptional batch of MION that was less efficient). Vertical bars indicate SEM (over epochs). Other conventions are the same as in Figure 7.