EEG and heart-rate variability reflect arousal level, not the specific source of load

What it is

Electroencephalography (EEG) and heart rate variability (HRV) are commonly used to infer cognitive load, but what these indicators actually reflect is a fairly general arousal level, not which part of the task the load comes from. The same arousal reading could stem from working-memory occupancy, time pressure, frustration, or plain nervousness — the indicator itself can't tell these sources apart.

Why it happens

The EEG components commonly used to infer load (frontal theta-band power, parietal alpha-band power changes) and the autonomic nervous system activity reflected in HRV are both linked to the activity of a more upstream, more general arousal-regulation system, and that system responds in a similar direction to many qualitatively different sources of pressure — whether the pressure comes from cognitive processing itself or from emotional or physiological tension, the trace it leaves on these indicators is, to a large extent, overlapping and non-specific. Physiological measures broadly capture an upstream, generalized arousal signal rather than the specific location of the load's source in the processing chain — pupil dilation being confounded by both lighting and emotional arousal at once is simply another expression of this same non-specificity.

Studying it

Studies using these indicators for load research commonly use a task-difficulty gradient as the independent variable and record how EEG band power or HRV metrics (high-frequency power, inter-beat interval variability) change as difficulty rises. Because these indicators are non-specific, merely observing that an indicator changes as task difficulty rises isn't enough to prove the change is specific to cognitive load — a control condition that separates out the arousal source is usually needed as well (an emotion-induction condition with comparable arousal but low cognitive demand, say) to rule out a purely emotional or physiological-tension explanation. The most common overinterpretation methodologically is labeling a change in some indicator directly as "cognitive load increased" without first ruling out that the change actually came from emotional fluctuation or plain physiological tension.

Where it stops holding

When the research goal is simply to find out whether users are in a more generally tense state during one part of an operation than another, these indicators are appropriate. But when the goal is to pinpoint which specific component of the task the load comes from — memory burden versus operating pace, say — these physiological indicators alone cannot answer that; locating the source requires pairing them with manipulations in the task design or with subjective follow-up questions.

Applying it

• When using physiological indicators for product evaluation, treat them only as a coarse screening signal for "is this segment generally more tense," and never state in a report that a change in EEG or HRV is evidence of a specific load source (such as "memory burden is too high").
• Once a segment shows a clear rise in physiological arousal, supplement it with a specific subjective probe or behavioral observation for that segment to determine whether the rise corresponds to cognitive demand, time pressure, or emotional factors.

Related

• Same group: A9.09.1 Pupil dilation grows with task difficulty, but is also confounded by lighting and emotional arousal · A9.09.3 Physiological measures usually don't interfere with the task itself, but are sensitive to equipment and environment · A9.09.4 Physiological measures suit tracking load's continuous change over time, while subjective scales suit an overall post-hoc evaluation
• Adjacent: A9.07 Subjective Load Measurement
• Search terms: EEG workload · heart rate variability · arousal · non-specific physiological index