Cognitive load theory, developed by John Sweller in the late 1980s, holds that working memory has a limited capacity and that learning fails when the total mental effort required to process new material exceeds that limit. The theory distinguishes three types: intrinsic load (the inherent complexity of the material itself), extraneous load (effort imposed by poor instructional design — confusing layout, irrelevant animations, redundant narration), and germane load (effort that directly contributes to forming long-term memory schemas). The practical implication for learning video is direct and consequential: every design decision that forces a learner to simultaneously read on-screen text and listen to a different narration, navigate a cluttered interface, or hold multiple disconnected concepts in mind at once adds extraneous load that displaces the mental capacity available for actual learning. Short segments reduce intrinsic load by limiting the amount of new information presented at once; chunking organises material into logical units to make relationships visible; worked examples lower load by providing a complete model before asking the learner to generate their own. For video specifically, the split-attention effect — where a learner must mentally integrate two sources of information presented in separate spatial locations — is a common and avoidable load source. Placing narration and relevant graphic in close proximity, or replacing redundant on-screen text with spoken explanation, are evidence-based mitigations. Engineers should understand cognitive load because it influences product decisions: forcing a learner to scroll between video and a side-by-side transcript increases extraneous load; in-video captions do not. The trade-off with cognitive load management is production cost — lower-load content requires more design work upfront.
