If you look at neurons, these cells stop dividing early in our life. Subsequent growth is via dendrite and axon branching processess. These combined to make synapses.
One theory for aging is connected to cell division and the shrinking of the tail ends of the DNA during each cell cycle. Once the tail end get too short, the cell can no longer replicate, successfully, causing death. If this is true, the neurons are like eternal cells, since they have learned to stop replicating, preserving their DNA tails. Neurons can live for decades on atomic recycle. Logically, it is all the other cell types, which will periodically divide. These will break down first (lose tails), with the brain sort of a victim of the local and global attrition. The brain uses a lot of energy and needs substantial support. As the global processing plant, composed of temporal cells ages, this support suffers.
If you look at the brain and nervous system, nerve endings appear near most cells. Since nervous tissue is smart tissue, what we have in our body is smart tissue near nearly all the cells. The question becomes, why put smart tissue near most of the cells? Is it there for decoration? The answer is to do something smart, such as to be part of a control systen for temporary cell expression, via feedback control to the brain. This way you have your eternal cells in charge of the temporary cells setting the potential to make them last longer; approach eternal. By controlling cell cycles and by helping cells get more life from each cycle, life is extended. But there are many external factors which impact this ideal situation, many of which will cause the global processing plant to atrophy.
Since the nervous control system is everywhere in the body and can send feedback to the brain, a smart control design would compare the real time image of the current state to an ideal image of what it should ideally be. The potential between the ideal set point and the real time control, sets potential for feedback.
Say you used a youthful neural control image as the set point for real time differentiation control, but with the body being much older. There will be a huge difference between the ideal and the real (teen age set point image in someone 75). The elderly return to a second childhood because the brain set point image needs to regress to childood to help set a maximum control potential with the aged power plant. The trade-off is an extension of the power plant life, but at cost will be the neurons regressing to the control image; one will lose that mature edge.