Operations Near the Edge of cognitive decline: Behavioral and Neural Correlates of cognitive load in Mild cognitive impairment
Abstract
ABSTRACT Motivation Early detection of cognitive strain is essential for identifying Mild Cognitive Impairment (MCI) and its progression to more severe neurodegenerative diseases. This study investigates cognitive strain in MCI by examining both behavioral and neural correlates during a real-world task. Methods Twenty-four older adults (M=69, 14 male, 10 female), including 11 with MCI and 13 healthy controls (MMSE >22/30), performed a spatial navigation task under varying memory load. We recorded brain activity using electroencephalography (EEG) and a mismatch negativity (MMN) paradigm to capture error-related potentials (ErrPs), which reflect cognitive load and processing errors. Results Both groups showed a decline in working memory performance as cognitive load increased (β = -41.39, SE = 10.73, t(90) = -3.86, p < .001). MCI participants, however, exhibited more efficient use of non-intrinsic (allocentric) reference frames under high cognitive load (β = 2.47, SE = 0.95, t(82) = 2.61, p = .009). MCI individuals also displayed larger ErrPs in central brain regions, indicating greater neural resource allocation. Additionally, MCI participants showed a decrease in parietal-to-frontal theta dominance under high load (β = -0.68, SE = 0.32, t(90) = -2.12, p = .037), and reduced fronto-parietal connectivity. Discussion These findings suggest that MCI participants experience greater cognitive strain, evidenced by increased neural resource allocation and strategy-shifting under load. This heightened strain may serve as an early marker of MCI progression, offering potential for early detection of pathological changes.
Fields of Study & Subjects
Dementia and Cognitive Impairment Research
Functional Brain Connectivity Studies
Intensive Care Unit Cognitive Disorders
Publication details
- Journal / source
- bioRxiv (Cold Spring Harbor Laboratory)
- Publication date
- 5 January 2026
- Document type
- Preprint