Abstract
Sleep-dependent memory consolidation relies on slow oscillations (SOs) that coordinate large-scale brain dynamics during non-rapid eye movement (NREM) sleep. Aging disrupts SO properties - reducing amplitude, density, and slope while altering the spatiotemporal patterns of slow-wave propagation - yet the circuit-level mechanisms linking structural brain changes to these disruptions remain poorly understood. Here we present a multi-scale, whole-brain thalamocortical network model incorporating biologically grounded human connectivity derived from diffusion MRI tractography, comprising over 10,000 cortical columns per hemisphere with spiking pyramidal and inhibitory neurons and an anatomically differentiated thalamic module. Simulating progressive synaptic loss, we find that selective degradation of recurrent excitatory connectivity, but not excitatory-inhibitory projections, reproduces empirically observed age-related SO changes. These results suggest that aging selectively disrupts the temporal structure of slow-wave sleep critical for interference-free memory consolidation, providing mechanistic insight into cognitive decline in the aging brain.