Subjective scales and secondary-task methods measure different things

What it is

Cognitive load has no single readable physical quantity — different measures capture different facets of the construct. A subjective rating scale asks "how effortful did this feel," while a secondary-task technique asks "how much spare capacity is left for another activity." The former is a consciously accessible experience report; the latter is a behaviorally inferred estimate of residual capacity. They are conceptually different things, and their scores need not move together.

Why it happens

The mismatch comes from load being processed at more than one layer. The consciously reportable effort experience is one layer, directly accessible to subjective scales. But not all processing cost enters awareness — with heavy practice, an action becomes automatized: the operator no longer feels it as effortful, yet the processing resources it consumes are still occupied, and this layer only shows up through behavior, such as whether a secondary task gets crowded out. On top of that, people often invest extra effort to keep performance stable when a task feels demanding — effortful compensation — which holds behavioral performance steady without changing what gets reported as subjective fatigue. These layers vary independently, which is exactly why the two measures sometimes correlate and sometimes come apart entirely.

Studying it

A common design collects both measures from the same participants: a subjective rating taken during or immediately after the task, alongside a secondary-task channel for behavioral data, then compares how strongly the two scores correlate. The trap to avoid is reading a low correlation as one measure being "wrong" — a more defensible reading is that load has separable subjective and behavioral dimensions, so the two should be treated as complementary evidence rather than two measurements of the same variable, and neither should be used to calibrate the other as a gold standard.

Where it stops holding

When a task is highly practiced and automatized, subjective feeling and behavioral resource occupancy diverge the most, and relying on the subjective scale alone will underestimate the true processing cost. Conversely, when a task leaves no spare attention to divert to anything else — a continuous high-load tracking task, say — inserting a secondary task disrupts the very task being measured, leaving the subjective scale as the only workable option.

Applying it

• When assessing the load of a product interaction, don't settle for a single measure: collect at least one subjective report and one behavioral indicator alongside it — even a simple completion time and error rate is better than a single score.
• When the two diverge (users report "it felt fine" while operations slow down and errors climb), report that divergence itself as a finding rather than picking one number to stand for the result.
• Verification: repeat both measures on the same user pool and check their correlation; a high divergence rate signals hidden load that subjective self-report is masking, and warrants further investigation.

Related

• Same group: A9.02.2 Load measures need to be read jointly with task performance
• Adjacent: A9.07 Subjective load measurement · A9.08 Performance-based measurement and the secondary-task method
• Search terms: subjective workload assessment · secondary-task technique · residual capacity · effortful compensation