As otherwise healthy adults age, their performance on cognitive tests tends to decline. This change is traditionally taken as evidence that cognitive processing is subject to significant declines in healthy aging. We examine this claim, showing current theories over-estimate the evidence in support of it, and demonstrating that when properly evaluated, the empirical record often indicates that the opposite is true. To explain the disparity between the evidence and current theories, we show how the models of learning assumed in aging research are incapable of capturing even the most basic of empirical facts of "associative" learning, and lend themselves to spurious discoveries of "cognitive decline." Once a more accurate model of learning is introduced, we demonstrate that far from declining, the accuracy of older adults lexical processing appears to improve continuously across the lifespan. We further identify other measures on which performance does not decline with age, and show how these different patterns of performance fit within an overall framework of learning. Finally, we consider the implications of our demonstrations of continuous and consistent learning performance throughout adulthood for our understanding of the changes in underlying brain morphology that occur during the course of cognitive development across the lifespan.