Analyzing students' activities in their learning process is an issue that has received significant attention in the educational data mining research field. Many approaches have been proposed, including the popular sequential pattern mining. However, the vast majority of the works do not focus on the time of occurrence of the events within the activities. This paper relies on the hypothesis that we can get a better understanding of students' activities, as well as design more accurate models, if time is considered. With this in mind, we propose to study time-interval patterns. To highlight the benefits of managing time, we analyze the data collected about 113 first-year university students interacting with their LMS. Experiments reveal that frequent time-interval patterns are actually identified, which means that some students' activities are regulated not only by the order of learning resources but also by time. In addition, the experiments emphasize that the sets of intervals highly influence the patterns mined and that the set of intervals that represents the human natural time (minute, hour, day, etc.) seems to be the most appropriate one to represent time gap between resources. Finally, we show that time-interval pattern mining brings additional information compared to sequential pattern mining. Indeed, not only the view of students' possible future activities is less uncertain (in terms of learning resources and their temporal gap) but also, as soon as two students differ in their time-intervals, this di↵erence indicates that their following activities are likely to diverge.