Visuospatial computations vary by category and stream and continue to develop in adolescence

Abstract

Reading, face recognition, and navigation are supported by visuospatial computations in category-selective regions across ventral, lateral, and dorsal visual streams. However, the nature of visuospatial computations across streams and their development in adolescence remain unknown. Using fMRI and population receptive field (pRF) modeling in adolescents and adults, we estimate pRFs in high-level visual cortex and determine their development. Results reveal that pRF location, size, and visual field coverage vary across category, stream, and hemisphere in both adolescents and adults. While pRF location is mature by adolescence, pRF size and visual field coverage continue to develop - increasing in face-selective and decreasing in place-selective regions - alongside similar development of category selectivity. These findings provide a timeline for differential development of visual functions and suggest that visuospatial computations in high-level visual cortex continue to be optimized to accommodate both category and stream demands through adolescence.

Figure 1.. Toonotopy experiment

(A) Toonotopy stimuli from Finzi et al. (2021) features a bar with colorful cartoon images of faces, words, bodies, places, and objects that change at 8Hz sweeps across a gray background at 4 angles (0°, 45°, 90°, 135°) each in 2 directions. Participants fixated at the center dot and indicated when the dot changed colors. (B) Population receptive field compressive spatial summation (pRF CSS) model. Left two panels show a single pRF with parameters of location (x,y) and size (σ) modeled by a 2D Gaussian followed by a compressive nonlinearity, used to model the voxel’s response. Middle right panel shows schematic of the pRF distribution within an ROI, and the rightmost panel depicts the visual field coverage of all pRFs in left V1 ROI in an example 11-year-old. (C) Phase, eccentricity, and size maps in an example adolescent (age 11) with V1 (purple), V2 (magenta), and V3 (gold) borders illustrated on the size map. All participants - Supplementary Fig. 2.(D) pRF size versus eccentricity relationship is similar across adolescents (dotted line) and adults (solid line) in early visual cortex (V1 - V3).

Figure 2.. Category-selective regions are modulated by the Toonotopy experiment

(A) Functional localizer (fLOC): Example stimuli of faces (children, adults; red), characters (words, numbers; blue), bodies and limbs (yellow), places (corridors, houses; green), and objects (car, guitar; black) from the fLOC experiment). (B) We identified in each participant category-selective functional regions of interest (ROIs) from the fLOC experiment, ROIs are labeled by preferred category and anatomical locations. 7 ROI were in the ventral stream (IOG-faces, pFus-faces, mFus-faces, pOTS-words, mOTS-words, OTS-bodies, CoS-places), 4 ROIs in the lateral stream (pSTS-faces, LOS-bodies, ITG-bodies, MTG-bodies), and 2 ROIs in the dorsal stream (MOG-places, IPS-places). Left: bilateral ventral ROIs in an example 17-year-old. Right: dorsal and lateral ROIs in the right hemisphere in an example 11-year-old. Left hemisphere dorsal and lateral ROIs are the same as the right hemisphere ROIs. (C) Violin plots of proportion of voxels with greater than 20% variance explained in each pRF in category-selective ROIs in the left hemisphere (light) and right hemisphere (dark) ventral, lateral, and dorsal streams in adolescents (a) and adults (A). Black circle: mean. Error bars: ± SE (standard error of the mean). Each dot is a participant.

Figure 3.. pRF properties in high-level category selective regions vary across category and stream.

(A) Top: pRF center polar plots for right hemisphere (dark) and left hemisphere (light) face-selective (reds; IOG, pFus, mFus, pSTS), word-selective (blues; pOTS, mOTS), bodypart-selective (yellows; OTS, LOS, ITG, MTG), and place-selective (greens; CoS, MOG, IPS) regions in the ventral, lateral, and dorsal streams in adolescents ages 10 - 17. Bottom: pRF center polar plots same as A but in adults ages 22 - 32. (B) Violin plots of y-position of pRFs in visual degrees for category-selective ROIs in the left hemisphere (light) and right hemisphere (dark) ventral, lateral, and dorsal streams in adolescents (a) and adults (A). ROI colors are the same as in A. Black circle: mean. Error bars: ± SE. Each dot is a participant. (C) Same as B but for eccentricity. (D) Same as B but for pRF size.

Figure 4.. Visual field coverage (VFC) of category-selective ROIs develops differentially.

(A) Average VFC of each ROI in the left hemisphere across adolescents (top row) and adults (bottom row) hemisphere with a gradient ranging from higher coverage in dark red and lower coverage in dark blue. VFC is calculated as the proportion of pRFs covering each point in the visual field for each participant and is then averaged across participants in the group. White asterisks with coordinates: average center of mass of the VFC. Black dotted lines: average full-width half max (FWHM) of the coverage. (B) Same as A but in the right hemisphere. (C) Violin plots of total FWHM in visual degrees for category-selective ROIs in the left hemisphere (light) and right hemisphere (dark) ventral, lateral, and dorsal streams in adolescents (a) and adults (A). Black circle: mean. Error bars: ± SE

Figure 5.. Category-selectivity differentially develops from adolescence to adulthood.

(A) Violin plots of average t-values for the category contrast for the preferred category of each ROI (e.g., faces > all other categories for the IOG-face ROI) in the left hemisphere (light) and right hemisphere (dark) ventral, lateral, and dorsal stream ROIs in adolescents (a) and adults (A). Black circle: mean. Error bars: ± SE. B. Linear relationships between total FWHM and mean t-value in each ROI in the left hemisphere (top) and right hemisphere (bottom). Each dot is a participant; adolescents are colored in lighter colors.