Growing evidence suggests that a more complete understanding of cortical function requires developing cognitive models that are predictive of multivariate neural behavior (e.g., Raichle, 2000; Shulman et al., 1997). Towards this end, one approach is to interpret population-specific activity in cortex from the perspective of capacity theories of selective attention (e.g., Handy, Hopfinger, & Mangun, in press). In brief, the model is founded on the ideas that (1) processing capacity is limited and (2) different processes may draw on different capacities (e.g., Boles & Law 1998; Polson & Friedman, 1988; see Kramer & Spinks, 1991). Testable hypotheses are then based on whether manipulations of process-specific load will lead to negative or positive covariation between different function-related clusters of activation in cortex—the predicted pattern depends on whether or not the clusters in question are assumed to share a common processing capacity. Expanding on these ideas, the current article addresses several recent issues that have arisen in the effort to apply capacity theory to the study of cortical function.