The pathogenesis of Alzheimer disease (AD) remains poorly understood and controversial. While a widely adopted framework has proposed that protein aggregates of β-amyloid (Aβ) and tau play key roles in initiating and propagating the disorder,1 many other processes, including cerebrovascular disease and inflammation, have been suggested as etiologically important.2 However, despite these controversies, there has been substantial agreement that AD pathology somehow converges on the synapse. Clinicopathological studies going back almost 3 decades established that synaptic loss is the strongest correlate of cognitive decline measurable in postmortem AD tissue.3,4 Because neural energy is largely expended on signaling,5 and because of associations between glucose utilization and synaptic markers,6 it has been widely believed that positron emission tomography (PET) measurements of glucose hypometabolism with fluorine 18–labeled fluorodeoxyglucose (18F-FDG) reflect synaptic loss in AD. However, until recently, there has been no in vivo method to specifically examine synapses in the human brain.