The use of video lectures has become a core feature of digital learning, but how the media diversity carried in videos affects learning experience has been rarely studied. Adopting a two-factor experimental design, this study used cognitive style questionnaires, brain wave detection, cognitive load scale, and post-test to explore the impacts of three commonly used algorithm-based video lectures on the sustained attention, learning engagement, cognitive load, and learning outcomes of verbal and visual style learners. The results show that cognitive style and video lecture type had a small effect on learners' sustained attention and learning engagement levels; and visual learners demonstrated significantly higher attention and learning engagement levels in the animation group than in the Tablet drawing and PPT groups. Similarly, with the increase of media diversity, cognitive load also increased, but the increase did not reach a significant level. The study also found that media diversity had a small impact on learning outcomes, but cognitive style not. It further proved that cognitive load caused by moderate media diversity didn't affect learning outcomes. This research provides a valuable reference for creating effective video lectures and significant support for researching video courses from neuroscience perspective.