Age-related fornix decline predicts conservative response strategy-based slowing in perceptual decision-making

Aging leads to increased response latencies but the underpinning cognitive and neural mechanisms remain elusive. We modelled older and younger adults response time (RT) data from a 2-choice flanker task with a diffusion drift model (DDM) and employed multi-shell diffusion weighted magnetic resonance imaging and spectroscopy to study neurobiological predictors of DDM components thought to govern RTs: drift rate, boundary separation and non-decision time. Microstructural indices of fractional anisotropy (FA), diffusivities and the restricted signal fraction (FR) from the Composite Hindered and Restricted Model of Diffusion (CHARMED) were derived from white matter pathways of visuo-perceptual and attention networks (optic radiation, inferior and superior longitudinal fasciculi, fornix) and estimates of metabolite concentrations [N-acetyl aspartate (NAA), glutamate (Glx), {gamma}-aminobutyric acid (GABA), creatine (Cr), choline (Cho) and myoinositol (mI)] were measured from occipital (OCC), anteri- or and posterior cingulate cortices (ACC, PPC). Ageing was associated with increased RT, boundary separation, and non-decision time. Differences in boundary separation but not non-decision time mediated age-related response slowing. Regression analyses revealed a network of brain regions involved in top-down (fornix FA, diffusivities in right SLF) and bottom-up processing (mI in OCC, AD in left optic radiation) and verbal intelligence as significant predictors of RTs and non-decision time (NAA in ACC, AD in the right ILF, creatine in the OCC) while fornix FA was the only predictor for boundary separation. Fornix FA mediated the effects of age on RTs but not vice versa. These results provide novel insights into the cognitive and neural underpinnings of age-related slowing.