Functional MRI responses to naturalistic stimuli are increasingly typical across early childhood

Abstract

While findings show that throughout development, there are child- and age-specific patterns of brain functioning, there is also evidence for significantly greater inter-individual response variability in young children relative to adults. It is currently unclear whether this increase in functional "typicality" (i.e., inter-individual similarity) is a developmental process that occurs across early childhood, and what changes in BOLD response may be driving changes in typicality. We collected fMRI data from 81 typically developing 4-8-year-old children during passive viewing of age-appropriate television clips and asked whether there is increasing typicality of brain response across this age range. We found that the "increasing typicality" hypothesis was supported across many regions engaged by passive viewing. Post hoc analyses showed that in a priori ROIs related to language and face processing, the strength of the group-average shared component of activity increased with age, with no concomitant decline in residual signal or change in spatial extent or variability. Together, this suggests that increasing inter-individual similarity of functional responses to audiovisual stimuli is an important feature of early childhood functional brain development.

Fig. 1

Mean inter-subject correlation (ISC) across the group. Group mean correlation maps were derived from a linear mixed effects model with crossed random effects (Chen et al., 2017), controlling for total number of censored volumes per pair. Colour gradient indicates the ISC value at each voxel in units of Pearson’s r. Clusters were formed at a voxelwise threshold of p < 0.0001 and a cluster forming threshold of α = 0.01. As this analysis is sensitive to small effects due to the large number of pairs, for visualization purposes, the map was further thresholded at a voxelwise r = 0.05.

Fig. 2

The association between ISC and average age of pair. Colour gradient indicates the β value of average age of pair (in years) at each voxel in units of Pearson’s r. Warm colours indicate a positive relationship, where a pair with a greater average age show greater synchrony to one another, and cool colours indicate a negative relationship, where a pair with a younger average age show greater synchrony with one another. Clusters were formed with a voxelwise threshold of p < 0.0001 and a corrected cluster p-threshold of 0.01 (Cox, 1996). Significant positive clusters are found throughout much of the parietal, occipital, and temporal lobes, as well as some clusters in the frontal lobes. Negative clusters are found in the very posterior of the bilateral occipital poles.

Fig. 3

The association between ISC and absolute difference in age. Colour gradient indicates the β value of age difference at each voxel in units of Pearson’s r. Cool colours indicate a negative relationship, where a pair closer in age show greater synchrony with one another compared to pairs farther apart in age. Clusters were formed with a voxelwise threshold of p < 0.0001 and a corrected cluster p-threshold of 0.01 (Cox, 1996). Significant negative clusters were found in the left precentral gyrus, left supramarginal gyrus, right cerebellum, left cerebellum, right supramarginal gyrus, bilateral superior precuneus, and the right inferior lateral occipital cortex.

Fig. 4

Age associations for temporal response properties in ROIs. The top row shows the scatter plots for each ROI between age and the strength of the β value of the group average time series in predicting the time series of a left out individual. The bottom row shows the scatter plots for each ROI between age and the standard deviation of the residuals after regressing out the group average time series from each individual. The rho on the plots indicates the partial Spearman correlation between age and the metric of interest, controlling for sex and head motion, p-values are one-tailed based on directional hypotheses. Trend lines are included for visualization purposes.

Fig. 5

The relationship between age and the percentage of stimulus-evoked GM voxels at p ≤ 0.05. rho indicates the partial Spearman correlation between age and percentage of GM associated with the group average signal, controlling for sex and head motion, while the p-value was determined through permutation tests with a null distribution computed from 10,000 iterations, and are one-tailed based on directional hypotheses. Trend lines on the models are included for visualization purpose.