4CAPS is a cognitive architecture of interest to both cognitive science and AI. 4CAPS is of interest to cognitive science because it supports models of neuroimaging data collected using fMRI and PET. 4CAPS should be of interest to AI because it organizes human information processing in an optimal and mathematically tractable form. This paper focuses on the adaptivity of 4CAPS models in the face of changing task demands and fluctuating resource availability. It illustrates this adaptivity in the domains of problem solving, spatial reasoning, and sentence comprehension. It also identifies new forms of adaptivity ripe for future research.