We developed the Scaffolding English E-assessment Learning (SEEL), where instant feedback and scaffolding hints were provided, to facilitate students to acquire the knowledge of English grammar. On the other hand, an empirical study was conducted to investigate how cognitive styles (i.e. Holists vs. Serialists) affected learners' reactions to the SEEL, including learning performance, learning perception and learning behavior. The results indicated that Holists and Serialists demonstrated similar learning perception. Nevertheless, they showed different learning behavior, which corresponded to the characteristics of their cognitive styles. Additionally, Holists obtained better post-test scores and gain scores than Serialists. This might be because Holists could make the best use of various hints and benefit from feedback provided by the SEEL while Serialists needed additional support. The findings could provide guidance for incorporating personalization into the SEEL so that the needs and preferences of Holists and Serialists could be accommodated. Nevertheless, such findings are correlational in nature and, we, thus, leave open questions concerning causality. Implications for instructional design are also discussed in this study.