Objects and categories: feature statistics and object processing in the ventral stream

Abstract

Recognizing an object involves more than just visual analyses; its meaning must also be decoded. Extensive research has shown that processing the visual properties of objects relies on a hierarchically organized stream in ventral occipitotemporal cortex, with increasingly more complex visual features being coded from posterior to anterior sites culminating in the perirhinal cortex (PRC) in the anteromedial temporal lobe (aMTL). The neurobiological principles of the conceptual analysis of objects remain more controversial. Much research has focused on two neural regions-the fusiform gyrus and aMTL, both of which show semantic category differences, but of different types. fMRI studies show category differentiation in the fusiform gyrus, based on clusters of semantically similar objects, whereas category-specific deficits, specifically for living things, are associated with damage to the aMTL. These category-specific deficits for living things have been attributed to problems in differentiating between highly similar objects, a process that involves the PRC. To determine whether the PRC and the fusiform gyri contribute to different aspects of an object's meaning, with differentiation between confusable objects in the PRC and categorization based on object similarity in the fusiform, we carried out an fMRI study of object processing based on a feature-based model that characterizes the degree of semantic similarity and difference between objects and object categories. Participants saw 388 objects for which feature statistic information was available and named the objects at the basic level while undergoing fMRI scanning. After controlling for the effects of visual information, we found that feature statistics that capture similarity between objects formed category clusters in fusiform gyri, such that objects with many shared features (typical of living things) were associated with activity in the lateral fusiform gyri whereas objects with fewer shared features (typical of nonliving things) were associated with activity in the medial fusiform gyri. Significantly, a feature statistic reflecting differentiation between highly similar objects, enabling object-specific representations, was associated with bilateral PRC activity. These results confirm that the statistical characteristics of conceptual object features are coded in the ventral stream, supporting a conceptual feature-based hierarchy, and integrating disparate findings of category responses in fusiform gyri and category deficits in aMTL into a unifying neurocognitive framework.

Figure 1

Temporal Signal to Noise Ratio (TSNR) around perirhinal cortex is sufficient for detection of BOLD activity. Colour bar shows group mean TSNR, where a minimum of 40 is needed to detect BOLD activity. Slice positions are reported in MNI coordinates and shown as dotted lines on the axial section.

Figure 2

Brain activity associated with picture naming. Top: contrast of basic-level naming vs. scrambled images overlaid on the ventral cerebral surface. Activation specific to naming meaningful objects (controlling for verbal output) was found along the anterior to posterior extent of the ventral stream. Bottom: activity explained by the visual model, where effects were focused in the bilateral occipital poles, with weaker effects extending to the posterior parts of the fusiform and inferior temporal gyri (see text for details). Colour bars represent voxel t- and F- values (degrees of freedom).

Figure 3

Sharedness of object features modulates BOLD activity within the fusiform gyrus. Top: objects with relatively more shared features were associated with greater BOLD activity in the bilateral lateral fusiform gyri, regions previously associated with activity for animals (Chao, et al., 1999; Martin, 2007), consistent with the greater number of shared features in animals than tools (voxel-level threshold p < .01, cluster-level threshold p < .05 FWE). Bottom: the a priori prediction that Sharedness would differentially modulate the medial and lateral fusiform gyri was tested within an anatomically-defined fusiform region of interest without cluster-level thresholding. Objects with more shared features (orange) produce activity in lateral fusiform and with fewer shared features (blue; corresponding to tools) produce activity in the bilateral medial fusiform gyri. Slice positions are given as MNI coordinates and colour bars represents voxel t-values (degrees of freedom).

Figure 4

Activity patterns within the fusiform gyrus for the contrast of living vs. nonliving objects closely track the correlation with Sharedness, but not the correlation with CxD. Linear ROIs (centre column; see Imaging Analyses) traversing the fusiform gyri at y = −48 mm (A), −57 mm (B) and −66 mm (C) were used to extract activity values (t-values) in successive voxels from left to right. The resulting plots (left and right columns) confirm region-specific similarities in activation between the living vs. nonliving contrast (black) and the correlation with Sharedness (solid grey) but not the correlation with CxD (dashed grey). Regions responding preferentially to living things (relative to nonliving things) also respond to concepts with relatively more shared features, and regions responding preferentially to nonliving things (relative to living things) also respond to concepts with relatively fewer shared features. In contrast, these regions are not modulated by the requirement for feature integration (CxD). The fine and course dashed horizontal reference lines indicate t-values corresponding to p < .01 and .001, respectively.

Figure 5

The feature statistic Correlation × Distinctiveness modulates BOLD activity in anteromedial temporal cortex. Top: Objects with lower Correlation × Distinctiveness values – indicating relatively weakly correlated distinctive features requiring more complex feature integration processes for their unique identification – were associated with greater activity in the anteromedial temporal cortex including the left perirhinal cortex at voxel-level p < .01, cluster-level p < .05 FWE. Bottom: At voxel-level p < .01, uncorrected cluster-level p < .05 in voxels with intermediate TSNR of 40-100, bilateral perirhinal cortex activation was seen. To maximize anatomic localizability of the clusters with respect to the PRC (Pruessner, et al., 2002), clusters are shown on the average participant brain. Slice positions are reported as MNI coordinates and the colour bar represents voxel t-values (degrees of freedom).