Keeping the lecture alive: strategies for supporting sustained attention during online lectures and their impact on memory

<p dir="ltr">When watching online lectures, previous studies have shown that university students pay better attention and recall more of the content when the lectures incorporate strategies such as breaks, quizzes that prime students to engage with future topics (pre-testing), or quizzes that assess their knowledge of recently covered topics (post-testing). Using students’ eye-gaze patterns as a proxy for attention, this study directly compared the influence of lecture breaks, pre-testing and post-testing on students’ attention and recall. A standard lecture where content was summarised was also included as a control condition. Participants watched a 1-hour Zoom lecture while wearing eye-tracking glasses that captured the durations their eyes fixated on on-task content. After a 30-minute delay, recall was tested using a short answer response and multiple-choice question (MCQ) exam. Bayesian analyses, which enabled testing for evidence towards the null or alternative hypothesis, showed that all conditions were effective at sustaining attention. Specifically, pre-tests, post-tests and standard lecture formats produced similarly high rates of on-task fixation durations that were maintained throughout the lecture. However, breaks resulted in higher on-task fixation durations than the standard condition in the latter half of the lecture. Recall scores on the short answer response and MCQ did not statistically differ between conditions. The findings suggest that a simple attentional break may be more effective for promoting lecture engagement than pre-testing, post-testing or summarising the lecture material, but that all strategies produce the same outcomes for learning. The results of the study challenges previous claims that lectures are ineffective at sustaining attention and offer practical implications for educators. The research also contributes to theories on how patterns of attention unfold over longer periods of time in dynamic visual environments.</p>