Physiological measures usually don't interfere with the task itself, but are sensitive to equipment and environment

What it is

Compared with the secondary-task method, a notable advantage of physiological indicators (pupil, heart rate variability, EEG) is that they usually require no extra action from the participant, and don't occupy attention or change operating style the way a secondary task does. But the cost shifts to the other end — the collection equipment and the environment it operates in are much more prone to contaminating the data.

Why it happens

Physiological signal collection passively records changes happening in the participant's own body and doesn't require actively allocating attention to comply (unless staying still or looking at a fixed point is required) — this is its advantage over the secondary-task method on the "doesn't interfere with what's being measured" front. But physiological signals are themselves very weak and easily affected by external factors: EEG is readily contaminated by muscle activity (blinking, chewing, talking), poor electrode contact, and electromagnetic interference; HRV is sensitive to postural changes, breathing rhythm, and ambient temperature. These sources of interference come from the collection environment and equipment itself, a completely different kind of cost from the secondary-task method's problem of the measurement act disturbing the participant's own behavior.

Studying it

When using physiological indicators for load research, the experimental design usually needs a dedicated artifact-detection and cleaning step — discarding EEG segments contaminated by blinking or muscle activity, discarding heart-rate segments distorted by postural change — and this data-cleaning workload is often as heavy as the formal load analysis itself. A point easily overlooked methodologically is that the proportion of data discarded as artifact should stay roughly balanced across experimental conditions — if a high-difficulty condition happens to lose more data because participants fidget more under it, the remaining data may no longer represent the true distribution for that condition.

Where it stops holding

This class of methods causes the least interference and yields the highest data quality in lab settings where participants can stay relatively still in a controlled environment. In real-world usage scenarios where participants can move around freely, talk, and change posture, the equipment itself and environmental noise substantially raise the artifact rate, and data reliability drops noticeably — a collection protocol validated in the lab cannot be assumed to maintain the same data quality when carried over unchanged to a real-use scenario.

Applying it

• Before planning to use physiological equipment for product testing, assess whether the target use scenario allows participants to maintain the relatively stable posture the collection requires — walking around with a handheld mobile device, for instance, makes EEG nearly infeasible, though it may tolerate a wrist-worn heart-rate device.
• Before formally analyzing the data, check whether the proportion of data discarded as artifact is roughly comparable across conditions; treat load comparisons drawn from conditions with clearly uneven discard rates with caution.

Related

• Same group: A9.09.1 Pupil dilation grows with task difficulty, but is also confounded by lighting and emotional arousal · A9.09.2 EEG and heart-rate variability reflect arousal level, not the specific source of load · A9.09.4 Physiological measures suit tracking load's continuous change over time, while subjective scales suit an overall post-hoc evaluation
• Adjacent: A9.08 Performance-Based Measurement and the Secondary-Task Method
• Search terms: physiological artifact · signal quality · unobtrusive measurement