Response threshold models are an important tool to model division of labor in social insects and to investigate the underlying principles of self-organization. In this article response threshold models which incorporate dynamic environments with varying demand for work and their influence on division of labor are studied. In their natural habitats, social insects are always exposed to dynamic environments, however, the effect that such environments have on response threshold models has rarely been investigated. In the course of this article it is shown that overworking and underworking, i.e. working more or less than the ideal amount, over a certain time is a colony-size dependent effect in dynamic situations. By adjusting the number of possible learning steps, which correspond to changes in the maximal threshold values relative to a colony's size, the performance of colonies in dynamic environments can be increased. A setup inspired by repeated migration behavior is also investigated. It is shown that these different learning rates affect a colony's ability to maintain an activity onset for a reappearing task.