Individual differences exist among learners. Among various individual differences, cognitive styles can strongly predict learners’ learning behavior. Therefore, cognitive styles are essential for the design of online learning. There are a variety of cognitive style dimensions and overlaps exist among these dimensions. In particular, Witkin’s field dependence/independence and Pask’s Holism/Serialism share some similarities. To this end, it is necessary to develop a framework to show overlapped behavior between these two cognitive style dimensions. To address this issue, this study used the Lag Sequential Analysis to examine the overlaps between these two cognitive style dimensions from the aspect of online learning behavior. The results from this study indicated that the overlaps mainly appear in comprehensive/local and dynamic/fixed approaches. Based on the findings of this study, we develop a framework that can support the improvement of instruction design so that the needs of different cognitive style groups can be accommodated. Accordingly, this study is an interdisciplinary work, which makes scientific contributions to three communities, i.e. human–computer interaction, digital learning and learning analytic.