Roles of Category, Shape, and Spatial Frequency in Shaping Animal and Tool Selectivity in the Occipitotemporal Cortex

Abstract

Does the nature of representation in the category-selective regions in the occipitotemporal cortex reflect visual or conceptual properties? Previous research showed that natural variability in visual features across categories, quantified by image gist statistics, is highly correlated with the different neural responses observed in the occipitotemporal cortex. Using fMRI, we examined whether category selectivity for animals and tools would remain, when image gist statistics were comparable across categories. Critically, we investigated how category, shape, and spatial frequency may contribute to the category selectivity in the animal- and tool-selective regions. Female and male human observers viewed low- or high-passed images of round or elongated animals and tools that shared comparable gist statistics in the main experiment, and animal and tool images of naturally varied gist statistics in a separate localizer. Univariate analysis revealed robust category-selective responses for images with comparable gist statistics across categories. Successful classification for category (animals/tools), shape (round/elongated), and spatial frequency (low/high) was also observed, with highest classification accuracy for category. Representational similarity analyses further revealed that the activation patterns in the animal-selective regions were most correlated with a model that represents only animal information, whereas the activation patterns in the tool-selective regions were most correlated with a model that represents only tool information, suggesting that these regions selectively represent information of only animals or tools. Together, in addition to visual features, the distinction between animal and tool representations in the occipitotemporal cortex is likely shaped by higher-level conceptual influences such as categorization or interpretation of visual inputs.SIGNIFICANCE STATEMENT Since different categories often vary systematically in both visual and conceptual features, it remains unclear what kinds of information determine category-selective responses in the occipitotemporal cortex. To minimize the influences of low- and mid-level visual features, here we used a diverse image set of animals and tools that shared comparable gist statistics. We manipulated category (animals/tools), shape (round/elongated), and spatial frequency (low/high), and found that the representational content of the animal- and tool-selective regions is primarily determined by their preferred categories only, regardless of shape or spatial frequency. Our results show that category-selective responses in the occipitotemporal cortex are influenced by higher-level processing such as categorization or interpretation of visual inputs, and highlight the specificity in these category-selective regions.

Keywords: animacy; category; fMRI; gist statistics; multivoxel pattern analysis; ventral visual pathway.

Figure 1.

Sample stimuli in each of the Category (animals vs tools), Shape (round vs elongated), and Spatial Frequency (low vs high) conditions. The contrast of HSF images here was increased for illustration purpose, whereas the contrast of the HSF images used in the experiment was decreased to equate visibility between LSF and HSF images. A total of 16 animals and 16 tools were used, each with 16 exemplar images.

Figure 2.

Pairwise dissimilarity (squared Euclidean distance) of gist statistics of low and high spatial frequency images showed significant differences between elongated and round shapes in each category, but no significant differences between animals and tools for each shape.

Figure 3.

Category-selective ROIs of a representative participant. The ROIs were defined within each participant using an independent localizer in which animals and tools with varied shapes were shown to participants. Animal-selective ROIs included left and right LOC, and left and right lateral FG; tool-selective ROIs included left medial FG, and left pMTG. L, left hemisphere; R, right hemisphere.

Figure 4.

The nine hypothetical dissimilarity matrices (models) for items across the Category, Shape, and Spatial Frequency conditions. For each participant, a total of four items was randomly selected from a pool of 16 animals or 16 tools for each of the eight conditions in the experiment.

Figure 5.

Univariate results in animal-selective and tool-selective ROIs. Averaged beta values of BOLD responses are shown for each of Category × Shape × Spatial Frequency conditions. Error bars indicate 95% confidence intervals of the three-way interaction.

′Figure 6.

Whole-brain univariate results of the contrasts between animals and tools, between round and elongated shapes, and between low and high spatial frequencies in the main experiment (q < 0.05, FDR corrected for each contrast; warm colors: animals, round, LSF; cool colors: tools, elongated, HSF). L, left hemisphere; R, right hemisphere.

Figure 7.

Averaged classification accuracy between Category (animals vs tools), Shape (round vs elongated), and Spatial Frequency (low vs high) in animal-selective and tool-selective ROIs. Error bars indicate SEM. Asterisks show significantly higher classification accuracy than the 50% chance level (q < 0.05, FDR corrected in each ROI). The black lines indicate significant pairwise comparisons among conditions, with higher classification accuracy for Category compared with Shape or Spatial Frequency in all animal-selective ROIs and in the tool-selective pMTG (q < 0.05, FDR corrected in each ROI).

Figure 8.

Each of the 32 × 32 matrices revealed averaged neural representational dissimilarity across Category, Shape, and Spatial Frequency, with four items included in each of the eight conditions, in the animal-selective and tool-selective ROIs.

Figure 9.

Correlations of the averaged neural representational dissimilarity matrices with the hypothetical dissimilarity matrices in the animal-selective and tool-selective ROIs in the main experiment. Asterisks indicate significant correlations (q < 0.05, FDR corrected); error bars indicate SEM. The gray bars represent the noise ceilings.

Figure 10.

Correlations of the averaged neural representational dissimilarity matrices with the hypothetical dissimilarity matrices in the animal-selective and tool-selective ROIs in the follow-up experiment. Asterisks indicate significant correlations (q < 0.05, FDR corrected); error bars indicate SEM. The gray bars represent the noise ceilings.

Figure 11.

Whole-brain searchlight results for significant correlations between neural pattern responses and the Animal, Tool, Category, Round, Elongated, Shape, LSF, HSF, and SF models, respectively, in the main experiment (q < 0.05, FDR corrected within each model). Critically, the significant clusters for the Animal and Tool models were found within the clusters for the Category model. L, left hemisphere; R, right hemisphere.

Figure 12.

Whole brain searchlight results for significant correlations between neural pattern responses and the Animal, Tool, Category, Round, Elongated, Shape, LSF, HSF, and SF models, respectively, in the follow-up study (q < 0.05, FDR corrected within each model). L, left hemisphere; R, right hemisphere.