Objective Functional MRI (fMRI) has great potential for unravelling mechanisms of functional decline in Alzheimer's disease (AD) and Mild Cognitive Impairment (MCI), but task–fMRI studies have produced conflicting results, partly due to failure to account for underlying morphological changes, and to variations in ability to perform the tasks. Resting–fMRI is promising because it does not require a task. We aimed to improve the understanding of how resting–fMRI relates to brain function.

Methods 80 elderly (25 controls, 25 MCI, 30 AD) underwent a combined multimodal MRI protocol including task–and resting–fMRI. Task–fMRI data were acquired during the execution of a memory paradigm designed to account for differences in task performance. Structural and physiological confounds were modelled for both fMRI modalities.

Results Successful recognition was associated with increased task–fMRI activation in lateral prefrontal regions in AD relative to controls; this overlapped with increased resting–fMRI functional connectivity in the same regions.

Conclusions Our results show that task–and resting–fMRI can reveal residual ability over the known changes in brain morphology and cognition occurring in AD, and suggest that resting–fMRI has a potential to measure the effect of new treatments.