What Is Cognitive Load?

Cognitive load is the demand a task places on working memory during the interval in which that task is performed. It rests on a well-established asymmetry in human cognition: long-term memory is effectively unbounded, while working memory holds only a few elements at once and holds them for seconds unless they are rehearsed. When the elements a task requires a person to hold and relate exceed that capacity, performance degrades and learning fails, regardless of motivation or effort. Cognitive load theory, developed by John Sweller and colleagues beginning in the late 1980s, formalized this constraint as a design principle for instruction.

The theory's central claim is that the amount of knowledge a learner acquires depends on how efficiently the available working memory resources are used. Expertise changes the arithmetic: a chess position that occupies a novice's entire working memory is a single retrieved chunk for a master, because schemas stored in long-term memory are treated as single elements when recalled. Cognitive load is therefore never a property of a task alone. It is a property of the pairing between a task and a particular person's prior knowledge.

Intrinsic, Extraneous, and Germane Load

Cognitive load theory partitions demand into three components. Intrinsic load arises from the inherent complexity of the material, specifically the number of elements that must be processed simultaneously because they interact. Learning individual vocabulary items has low element interactivity and low intrinsic load; learning to balance a chemical equation has high interactivity, since each term constrains the others. Extraneous load comes from how material is presented rather than from what it contains: a split-attention layout that forces a reader to hold a diagram in mind while searching for its caption imposes demand that carries no instructional value. Germane load refers to the working memory resources actually devoted to building and automating schemas in long-term memory. The design implication follows directly: since intrinsic load is fixed by the learning goal and total capacity is fixed by the person, reducing extraneous load is the available lever.

Measurement

Quantifying load is the field's persistent methodological problem. Subjective rating scales dominate practice, and the NASA Task Load Index is the most widely used instrument, collecting self-reported ratings across six dimensions including mental demand, temporal demand, effort, and frustration. Its limitations are understood: it is administered after task completion, relies on recall, and yields a single retrospective figure rather than a continuous trace, a constraint examined in research on the psychometric properties of NASA-TLX and the Index of Cognitive Activity. Studies applying NASA-TLX to measure learners' cognitive load and comparing two mental workload measurement approaches in multimedia learning report that subjective and performance-based measures often diverge. Physiological methods add continuous signals: pupil diameter, heart rate variability, electrodermal activity, and electroencephalography, each with its own sensitivity to confounds such as lighting, movement, and emotional arousal.

Design and Instrumented Environments

The applied side of the field converts these findings into design rules: integrate text with the diagram it describes, remove redundant duplicated information, use worked examples with novices and problem solving with more advanced learners, and split narration and visuals across auditory and visual channels rather than overloading one. Virtual and augmented reality environments have become common testbeds because they permit precise control over stimulus and synchronized capture of physiological signals, an approach illustrated by an open platform for cognitive load research in virtual reality.

Applications

Cognitive load research has applications in a range of fields, including:

• Instructional design and educational technology
• Human-computer interaction and user interface evaluation
• Aviation, air traffic control, and flight deck design
• Automotive human factors and driver distraction assessment
• Clinical simulation and medical training
• Control room and process plant operator interfaces
• Adaptive systems that adjust task difficulty from sensed workload