Contributed Talk Sessions | Poster Sessions | All Posters | Search Papers

Poster Session C: Friday, August 15, 2:00 – 5:00 pm, de Brug & E‑Hall

Intrinsic dimensionality of brain activity manifolds across tasks and development

Erica Lindsey Busch¹, Nicholas Turk-Browne¹; ¹Yale University

Presenter: Erica Lindsey Busch

To investigate cognitive processing in the brain, researchers often focus on the localization or amplitude of fMRI activity. Intrinsic dimensionality – the degrees of freedom needed to characterize complex signals – may offer additional insights into where and how computations occur across the brain. Here, we investigated how tasks and development alter the intrinsic dimensionality of brain activity. We used a nonlinear manifold learning method to estimate the intrinsic dimensionality of fMRI activity in whole-brain searchlights from infant (3–24 months), child (3–12 years), and adult datasets during resting and naturalistic fMRI tasks (i.e., movies or narratives). These analyses revealed two principles of intrinsic dimensionality in the human brain: task processing reduces the intrinsic dimensionality of task-irrelevant brain regions relative to rest and development increases the intrinsic dimensionality of task-relevant brain regions. These findings hold implications for understanding information processing in the brain and studying the development of adult-like brain functioning.

Topic Area: Methods & Computational Tools

Extended Abstract: Full Text PDF