Sulcal Morphology of the Lateral Prefrontal Cortex Predicts Individual Differences in Cognitive Development

The human cortex folds in distinctive and predictable patterns of gyri and sulci (LaGuen et al., 2018). Deep primary sulci emerge early, while comparatively shallow tertiary sulci appear later and deepen through development. Due to their morphological variability and protracted development, a classic, but largely untested, hypothesis situates tertiary sulci as functional landmarks for cognition in slower developing cortical regions (Sanides, 1964). Here we extend this hypothesis to cognitive development and predict that morphological features of tertiary, but not primary, latPFC sulci will be associated with reasoning ability in children, as reasoning is a cognitive process known to heavily tax the latPFC. Cortical morphometric analyses were performed on high resolution T1‐weighted MPRAGE anatomical scans in 65 participants ages 6–18, randomly sampled from an existing dataset (see Wendelken et al., 2017). This dataset was further divided into Discovery (N =37) and Replication (N = 28) samples. Cortical surface reconstructions were generated using Freesurfer.14 sulci in LatPFC were manually defined from recently proposed definitions (Petrides, 2019). 1,829 labels (14 labels per hemisphere, per subject) were assessed across both samples for mean sulcal depth, maximum sulcal depth, and cortical thickness. Each participant also performed a matrix reasoning task. To investigate the relationship between sulcal morphology and cognition, we employed a data‐driven approach using nested cross‐validation to relate sulcal depth to reasoning performance in the Discovery sample. We first conducted a LASSO‐regression as a regularization technique to select only sulci that are relevant to reasoning performance. Only two tertiary sulci showed a significant relationship with reasoning skill. Using a leave‐one‐out cross‐validation procedure, we then predicted individuals’ reasoning performance from the depth of these two relevant sulci, with age included as a covariate in the model. This model explained 50.4% (RMSE: 2.32) of the variance in performance on the matrix reasoning task, which was higher than models with either a) age as the sole predictor or b) the depth of both these tertiary sulci and the surrounding primary sulci as predictors. Finally, we used the model defined in the Discovery sample to predict reasoning performance in our Replication sample and found that we can predict performance with high accuracy (R 2 cv = 0.55; RMSE = 2.99) in this independent sample. Our findings demonstrate for the first time that tertiary latPFC sulci can be identified reliably in children, and provide strong evidence for tertiary sulci as sources of individual variability in cognitive performance. The results corroborate prior research on the importance of latPFC development for higher cognition, and demonstrate that precise anatomical parcellations in individual subjects can provide valuable insights into relationships between cortical anatomy and cognitive development. Support or Funding Information National Science Foundation (Grant BCS1558585). Jacobs Foundation Research Fellow Health (R01 NS057156). NIMH (R01 MH091109). National Institute of Child Health and Human Development (R01HD067254).