Fig. 5

A Variance in current cognitive performance (leftmost bar) is driven by a number of contributing factors. B If the variance explained by an adverse factor (e.g., hippocampal atrophy, pathology, etc.) is regressed out, the remaining variance is largely the same as the current cognitive performance. C A large portion of current cognitive performance is explained by premorbid cognitive performance. The remaining variance can be interpreted as “change in cognitive performance” compared to expected. D Variance explained by an adverse factor can be regressed out of this “change in cognitive performance,” but what remains is highly correlated with the original “change in cognitive performance” measure. E Variance that remains in current and past cognitive performance can be explained by a host of known and to-be-discovered genetic, environmental, and lifestyle factors and pathologies, as well as measurement noise. Ultimately, our goal is to understand what contributes to this variance and reduce error in our model of cognition. F These models can be used to predict cognitive state or forecast cognitive decline. The more comprehensive our models of cognition, the better our individual levels of prediction will be. With better models for cognition, we shift our focus to simulating how modification of a pathological or resilience factor might influence maintenance of healthy cognition