Ultra-high resolution imaging of laminar thickness in face-selective cortex in autism

Abstract

Gray matter cortical thickness (CT) is related to perceptual abilities. The fusiform face area (FFA) (Kanwisher et al., The Journal of Neuroscience: The Official Journal of the Society for Neuroscience, 17, 4302-4311, 1997) in the inferior temporal lobe is defined by its face selectivity, and the CT of the FFA correlates with the ability to make difficult visual decisions (Bi et al., Current Biology, 24, 222-227, 2014; McGugin et al., Journal of Cognitive Neuroscience, 28, 282-294, 2016, Journal of Cognitive Neuroscience, 32, 1316-1329, 2020). In McGugin et al. Journal of Cognitive Neuroscience, 32, 1316-1329, (2020), individuals with better face recognition had relatively thinner FFAs, whereas those with better car recognition had thicker FFAs. This opposite correlation effect (OCE) for faces and cars was pronounced when we look selectively at the deepest laminar subdivision of the FFA. The OCE is thought to arise because car and face recognition abilities are fine-tuned by experience during different developmental periods. Given autism's impact on face recognition development, we predicted the OCE would not appear in autistic individuals. Our results replicate the OCE in total FFA thickness and in deep layers in neurotypical adults. Importantly, we find a significant reduction of these effects in adults with autism. This supports the idea that the OCE observed in neurotypical adults has a developmental basis. The abnormal OCE in autism is specific to the right FFA, suggesting that group differences depend on local specialization of the FFA, which did not occur in autistic individuals.

Keywords: Autism; Face recognition; Individual differences; MRI.

Fig. 1

Behavioral performance across object categories and groups. Behavioral performance on memory tests (left; Cambridge Face Memory Test (CFMT), Cambridge Car Memory Test (CCMT), Novel Object Memory Test (NOMT)) and matching tests (right; Vanderbilt Face Matching Test (VFMT), Vanderbilt Car Matching Test (VCMT), Novel Object Matching Test (NOMaT)) for typical development (TD) and autism (AUT) groups. The NOMaT is scored as d prime; all other tests are scored by accuracy. Box plots show the range and mean of behavioral performance within and across groups. Asterisks denote significant group differences (p <.05 with Bonferroni correction for multiple comparisons)

Fig. 2

A. For a representative participant, functional data from the face localizer is overlaid on the anatomical T1-weighted scan. The right rFFA2 is identified as the contiguous cluster of voxels in the middle fusiform gyrus that is more activated by faces compared with common objects (hot colors are face-selective). White boxes show the strategic alignment of our ultrahigh-resolution slices perpendicular to the activated fusiform cortex (causing a rightward tilt to the ultrahigh-resolution coronal slices in B). B. Top: One ultrahigh-resolution slice is shown for a representative participant with autism. The inset zooms in on the right lateral fusiform gyrus, where this participant’s functionally defined rFFA2 falls between the occipital temporal sulcus (OTS) and the middle fusiform sulcus (MFS). The cortical boundaries of the rFFA2 are marked in yellow (superficial) and red (deep). Bottom: The line plot (blue line) represents the mean of all traces from the superficial border to the deep border (and vice versa) that fall between the borders of the rFFA2. A fourth order polynomial (black line) is fit to the mean. Points of inflection (red circles) denote changes in signal intensity and are used to isolate the middle layer from the superficial and deep laminar subdivisions. C. Box and whiskers plots show the means (x) and individual variability of thickness of the rFFA2 and its deep laminar subdivision, separated by group. The FFA was thicker in AUT compared with TD (t = 6.6, p <.001). This pattern was consistent in the deep (t = 6.12, p <.001) laminar subdivisions but not present in the anatomically defined right fusiform gyrus region of interest

Fig. 3

Scatterplots for typical development (TD; top row) and autism (bottom row) groups, showing correlations between face recognition ability (filled squares, solid line) or car recognition ability (hollow circles, dashed line) with cortical thickness. Correlations are shown for rFFA2 total CT (left panel), rFFA2 deep laminar subdivision (middle panel), and the anatomical rFG (right panel). The oppositive correlation effect (OCE) indices are bold if significant (p <.05). Bar graphs between scatterplots compare the OCE across groups. *Significant group difference (p <.05) was observed only for rFFA2 total CT and rFFA2 deep layers