I’m a postdoctoral scholar with Leanne Williams at Stanford in the Williams PanLab researching how cognition and psychopathology interact in the developing brain. I completed my neuroscience PhD at the University of Pennsylvania with Ted Satterthwaite, where I studied normative neurocognitive development.

My goal is to build the neuroscientific foundations needed to advance mental health care, and in parallel, to translate this research to meaningful real-world outcomes. I tend to focus on the interplay between mood disorders and cognitive dysfunctions, with specific expertise in how depression and cognitive functions emerge in adolescence. Neuroscientifically, many of these advances are rooted in hierarchical cortical organization and hierarchical cortical reorganization in development.

Out of respect for the scientific community and gratitude for previous generations of open-source scientists, I’ve created walkthroughs for all of my code underlying every analysis in my first-author papers since learning how to code. All of the code within these walkthroughs (and the walkthroughs themselves) have been independently verified to work as intended by particularly stalwart co-authors. For example, in this pre-print, we were able to build off of fantastic code bases for spherical registration of cortical surfaces (freesurfer) and for tracing the migration of progenitor cells on spherical zebrafish gastrulas to construct, validate, hypothesis-test with, and disseminate a pipeline for tracking and quantifying the hierarchical movement of cortical activity.

Outside of neuroscience, I’m a long-time mediocre improvisational musician, enjoy backpacking into the mountains with my soon-to-be-wife and friends, am an ex-personal trainer, and am active in the great American folk science of meat smoking. Thanks for visiting my page, and feel free to reach out for inquiries. Here are some findings I’d like to highlight:

Whether cognition is positively or negatively associated with symptoms depends on the class and magnitude of symptoms. Positive and negative relationships are robust to bootstrapping across modeling techniques at >99% statistical confidence, but co-exist within the same class of symptoms. These results can explain dozens of discrepant papers, often with strong methods and large sample sizes, arguing that cognition is either negative or positively associated with depression symptoms. If you look more closely at these studies, you’ll see the papers arguing for a positive association (higher cognitive scores = higher symptom scores) are almost all in subclinical samples, whereas papers arguing for negative associations (higher cognitive scores = lower symptom scores) are almost all in clinical samples. Both sets of papers are “correct”, but relying on linear models in samples that did not cover the full spectrum of mental health obscured that the association between cognition and internalizing symptoms depends on how well or unwell an individual is. Dashed lines represent borderline and clinical thresholds from the CBCL. From my JAMA psych paper.

Where in the cortex individual variability in functional arealization localizes is dependent on how granular your definition of functional networks is. More granular = more variability in higher-order cortices. From my nature comms paper.

Functional differentiation proceeds from lower-order to higher-order cortices in development, such that a single unimodal-to-transmodal gradient explains most (r^2=.71) of the developmental variability we observed across functional networks. Functional differentiation is associated with enhanced cognitive capacity, and de-differentiation seems to follow the same spatial trajectory years-to-decades later (still waiting to run this on full lifespan data). From my nature comms paper.

Hierarchical distance provides a parsimonious description of how individual edges in the functional connectome develop. This pattern has since been replicated in 4 independent datasets (Luo et al., 2024, Nat. Comms). Check out figure S9 from my paper (also Nat. Comms, 2022) to see how it lines up relative to other edge-level descriptions of neurodevelopment.

Optical flow can be used to delineate bottom-up and top-down hierarchical propagations in fMRI data. From my neuron paper.

Both bottom-up and top-down propagations are common in all individuals we tested. 2 of 5 verifications of this finding depicted in this figure. From my neuron paper.

Top-down propagations become increasingly prominent over neurodevelopment. Not pictured is the same results 100% holding after controlling for previously known properties of functional neurodevelopment. From my neuron paper.

Multi-shell diffusion weighted imaging can be leveraged to confer increased sensitivity to neurodevelopmental effects and decreased sensitivity to the confounding influence of head motion in neurodevelopmental studies. Schematic from my developmental cog. neurosci. paper.