Generation Z often faces challenges such as divided attention and difficulty in retaining knowledge when learning in fast-paced media environments, necessitating both theoretical and practical solutions. To address issues like short attention spans and information overload, this paper constructs a learning model termed "Multi-modal Stimulation Micro-unit Delivery Instant Feedback" (MSM-MD-IF). It designs a phased learning process centered around short videos: first, electroencephalogram (EEG) and eye-tracking data are collected to identify the media elements that most effectively capture attention; second, knowledge points are broken down into micro-units no longer than two minutes, delivered via multi-modal content (images, audio, interactive bullet comments); third, quiz and interactive modules are embedded to adjust the delivery pace and repetition frequency based on the learner’s real-time performance. A four-week experiment involving 320 university students aged 18–24 from Generation Z shows that the experimental group achieves an average attention increase of 0.15 and a mean knowledge recall accuracy of 91.4%. The study demonstrates that the MSM-MD-IF pathway effectively extends the focused learning duration of Generation Z and promotes knowledge retention.