As cells degenerate and their numbers are reduced in circuitry, other mechanisms become engaged at a cell and systems level beyond those of the molecular level. Examples include the sprouting of new synapses in response to nearby cell loss and the sprouting of dendrites in neighboring cells. In addition, whole networks respond in terms of altered processing using somewhat redundant, but maybe initially suboptimal strategies. These, oftentimes, can increase the time for cognitive processing, but will still accomplish the task. There is also a growing body of literature indicating that, in the course of brain aging, more of the brain has to be involved in a task that would normally require only minimal activation of circuits; thus, the circuits are working much harder to accomplish the same task. This would indicate that use-dependent change and practice effects, together with appropriate pharmaceuticals, might have a rational basis for cognitive rehabilitation.

In conclusion, there are several principles that appear to be evolving in the field that are in need of additional testing:

• Brain aging is not a linear process; the aging process passes through phases.

• The initiation phase can compromise neuronal function and is probably reversible as it represents functional homeostasis.

• Interventions include antioxidants, use-dependent plasticity (behavioral/physical/cognitive stimulation), regulation of inflammation, estrogen replacement therapy, etc., and are most effective in this phase.

• The propagation phase is initiated through a series of molecular cascades driven by accumulating failures and compensation mechanisms and is less readily reversible.

• Interventions may be phase-dependent, and effective interventions at one phase may be inappropriate/inadequate at others.

Strategies must and can be developed to identify weak molecular linkages and to assist cells in correcting them prior to irreversible losses and the development of cascades.

• There is a clear significant and major gap in supported research in the essential hierarchical areas, and circuit-based analyses at a systems level are needed.

• Transgenic animals offer great promise, but there is a great need for