Rational inattention theory models human decision-making as an optimal information-processing problem akin to rate–distortion theory. This talk will motivate and introduce this framework, now a workhorse model of cognitive economics, which seeks a principled explanation of seemingly irrational behavior as arising from cognitive constraints on information processing. I will outline the basic problem and its known structural properties, then present a multi-attribute generalization based on the recently introduced multi-attribute Shannon entropy. The talk will discuss ongoing work with the goal of highlighting open computational challenges—particularly how to solve these models at scale—and will explore how ML methods may offer promising avenues for progress.