People often plan hierarchically. That is, rather than planningover a monolithic representation of a task, they decompose thetask into simpler subtasks and then plan to accomplish those.Although much work explores how people decompose tasks,there is less analysis of why people decompose tasks in theway they do. Here, we address this question by formalizingtask decomposition as a resource-rational representation prob-lem. Specifically, we propose that people decompose tasks ina manner that facilitates efficient use of limited cognitive re-sources given the structure of the environment and their ownplanning algorithms. Using this model, we replicate severalexisting findings. Our account provides a normative explana-tion for how people identify subtasks as well as a frameworkfor studying how people reason, plan, and act using resource-rational representations.