Confusion is an emotion that is likely to occur while learning complex information. This emotion can be beneficial to learners in that it can foster engagement, leading to deeper understanding. However, if learners fail to resolve confusion, its effect can be detrimental to learning. Such detrimental learning experiences are particularly concerning within digital learning environments (DLEs), where a teacher is not physically present to monitor learner engagement and adapt the learning experience accordingly. However, with better information about a learner's emotion and behavior, it is possible to improve the design of interactive DLEs (IDLEs) not only in promoting productive confusion but also in preventing overwhelming confusion. This article reviews different methodological approaches for detecting confusion, such as self-report and behavioral and physiological measures, and discusses their implications within the theoretical framework of a zone of optimal confusion. The specificities of several methodologies and their potential application in IDLEs are discussed.