This exercise program will consist of walking or running on a treadmill or elliptical trainer, or spinning on a stationary bicycle. The goal of the program will be for participants to exercise at a moderate to high level of intensity, defined as 70-80% (American College of Sports Medicine guidelines) of heart rate reserve (HRR), for 45-60 minutes 4 days per week. At the start of each training session and following a 5-minute warm-up, subjects will exercise at 50% HRR (0.5[HRmax-HRrest] +HRrest) and intensity will gradually be increased to the individualized target heart rate training zone. Exercise facilitators will use a pre-specified computerized program. This program provides guidelines for the individualized progression of exercise based on age and resting heart rate. Subjects will wear a digital heart rate monitoring device for the duration of the training session to ensure they are exercising safely at the specified level of intensity.

The low-intensity activity program will consist of an individualized and organized series of stretching and balance activities for the whole body, specifically designed for older adults. Consistent with the high-intensity protocol, subjects will complete the prescribed 45-60 minute stretching routine 4 days per week at the exercise facility. All stretching routines will include warm-up and cool-down activities, and will be within each subject's range of motion. Each stretch will be held for 20-30 s and repeated 5-10 times. Subjects will wear a digital heart rate monitoring device to ensure they are stretching safely and at an intensity below 35% HRR. The activity log completed during each stretching session will include HR, stretching duration, and mean HR during stretching. Subjects will also have the option to participate in structured pre-approved (by the exercise facilitator) stretching classes at the exercise facility when available.

The cognitive variables identified above, by domain, will be subjected to separate treatment group X treatment month (3, 6) multivariate analyses of covariance (MANCOVA), with baseline scores and age serving as covariates. When the MANCOVA proves significant, separate post-hoc analysis of covariance (ANCOVA) will be conducted on the relevant dependent variables.

Cognitive tests include measures of executive function and working memory.

Multiple regression and correlational (MRC) analyses will be conducted to evaluate whether treatment-induced change in insulin sensitivity and in other blood and cerebrospinal fluid biomarkers predict cognitive performance among subjects in the aerobic group.

Whole-brain averages will be computed, reflecting characteristics of the entire system. Whole-brain averages will be compared with random networks generated with the same node number and degree,48 and will be compared between intervention groups using general linear regression models. Regions of interest (hippocampus, posterior cingulate/precuneus, prefrontal cortex) will be extracted using WFU Pickatlas and regional degree, Eglob, or Eloc will be compared between groups using the general linear model.