Developmental tuning of functional manifold dimensionality across the human brain

Area of Science:

• Neuroscience
• Developmental Neuroscience
• Cognitive Neuroscience

Background:

• Neural representations exhibit varying complexity across brain regions and tasks.
• The developmental trajectory of this representational complexity is not well understood.

Purpose of the Study:

• To investigate how neural representational complexity changes during human development.
• To understand the mechanisms underlying functional specialization in the brain.

Main Methods:

• Utilized T-PHATE, a nonlinear manifold learning technique, to estimate intrinsic dimensionality.
• Analyzed five naturalistic functional magnetic resonance imaging (fMRI) datasets from 781 participants aged 3 months to 53 years.

Main Results:

• Adults show higher dimensionality in task-relevant brain regions compared to task-irrelevant regions.
• Task-related modulation of representational complexity was absent in infants, emerging in childhood and strengthening through adolescence.
• This modulation resulted from a selective collapse of dimensionality in task-irrelevant regions, not an expansion in task-relevant regions.
• Dimensionality compression evolved from global and nonselective in infants to local and precise in adults.

Conclusions:

• Selective compression of neural activity is a key mechanism driving functional specialization during development.
• The brain achieves specialization by reducing complexity in non-essential areas rather than solely increasing it in task-relevant areas.