With the increasing prevalence of age-related dementia and other neurodegenerative disorders, there is a growing demand for new brain health maintenance strategies, and for novel interventions targeting cognitive decline. Pharmaceutical management of cognitive deficits remains largely ineffective, and generally inferior to multimodal functional and cognitive interventions. In this context, computerized cognitive training interventions have emerged as promising alternatives. Neurofeedback (NF)-based brain training protocols, whereby patient are provided with feedback indexed on ongoing measurements of brain activity, have also undergone a renaissance in recent years. The affordable and ubiquitous computational resources of tablets and smartphones combined with new sensor technologies have sparked considerable interest from both the general population and industry. Yet, amongst a flood of NF implementations and promising outcomes, pervasive methodological shortcomings have clouded to what extent NF-specific effects can account for the positive findings. My vision is that before NF approaches to cognitive decline can be properly translated to clinical and wearable technologies, they must adhere to established principles of learning psychology, must target evidence-based neurophysiological markers, and must be rigorously tested against placebo effects. I hypothesize that a personalized NF training protocol explicitly based on learning psychology principles, and framed as a cognitive skill acquisition process, will increase the efficacy of computerized cognitive training interventions. Furthermore, I hypothesize that targeting complex functional connectivity measures will improve learning and increase transfer to activities of daily living.