Workload Measurement: Peeking into the Brains of System OperatorsSeptember 2012
Topics: Human Factors Engineering, Human Resources Management, Neurosciences
The Workload Conundrum
Imagine you are a researcher evaluating the design for a new aircraft control system. Your evaluation method is to run pilots through realistic simulations and collect quantitative and qualitative data. One of the variables you want to assess is the pilots' "workload" while operating the new system. So you do what is common practice today—you ask the pilots to rate their workload on a numeric scale while they perform their tasks. However, the act of interrupting the pilots to ask about their workload increases their workload, thus confounding the experiment.
So to minimize interruptions, you only ask for workload ratings periodically during the simulation. Let's say you ask the pilots to rate their workload every three minutes. What you are really doing is asking them to generate an average over that time interval. But what if, as is commonly the case, the first minute was exceptionally hard but the last two were easy?
Added to these methodological issues is the questionable validity of workload self-assessment. Operators notoriously have a difficult time accurately assessing and expressing the mental impact of changing task demands. Sometimes operators rate their workload higher in instances when their observable behavior does not appear to change. Are they actually working harder?
This example illustrates the imprecision of current workload measurements. We need a measurement system that captures moment-to-moment changes in workload resulting from changes in the task environment. With such a system, we can identify the precise system and environmental variables that push the operator into a high workload state.
Looking Under the Hood
The need for workload measures that are more objective, time-sensitive, and accurate has spurred interest in physiologically based measures. Various physiological measures correlate with workload, including heart rate, pupil diameter, eye blink frequency, and cortisol levels in saliva. Brain-based measures include changes in blood flow as measured by functional magnetic resonance imaging (MRI) and positron emission tomography (PET) scans. But these measures, while reliable, are also very expensive and intrusive because of the need to place test subjects inside the magnetic scanner or inject them with radioactive isotopes.
Researchers are also studying how electrical activity measured along the scalp using electroencephalography (EEG) relates to cognitive state. EEG measures brainwave activity across the cerebral cortex, the outer layer of the brain that performs higher order cognitive processing. Relationships between EEG brainwaves and workload levels have been widely studied and thus could provide a basis for measuring cognitive workload, vigilance, and attention. However, many of these studies use averaged EEG data over time intervals of stable workload. Developing more sensitive measures of dynamic environmental variables and their effect on operator workload will require the ability to track real-time changes in the EEG as the task environment evolves.
To build this real-time capability, MITRE is using brain-computer interface (BCI) technology. BCIs, which are used for the physically disabled to operate computer-based devices via brain signals, require the ability to track and analyze EEG data in real time. MITRE engineers and neuroscientists are building a workload gauge (called BrainGage) with BCI data processing technology that analyzes streaming EEG data, provides EEG data visualizations,and measures a test subject's workload in real time.
Taking a Flight on BrainGage
The Federal Aviation Administration's Next Generation Air Transportation System (NextGen) program promises to transform the air traffic control system from an aging ground-based system to a satellite-based system, which will require many changes in controller and pilot responsibilities. In evaluating new concepts for automation, policies, and procedures, we need to know when workload is too high, too low, or just right, and how that changes with the operator's tasks, air traffic, and automation factors involved.
Using BrainGage, MITRE human factors researchers will be able to run simulation studies to test how controllers and pilots handle the changes brought on by NextGen. Once we understand how these changes influence the cognitive state of the operator, we can use this knowledge to build policies, procedures, and automation that can adapt to the operator's workload.
The complexity of the EEG signal combined with BrainGage's real-time analysis capability offers the ability to study brainwave-based metrics that go beyond workload and attention. Can BrainGage show when someone is lying? Or when a trainee has achieved a level of expertise?Or what type of cognitive strategy the operator is using for a given task? CanBrainGage show when an operator is bored, fatigued, or anxious? These are all potentially important questions for the variety of sponsors that MITRE supports, making BrainGage valuable in supporting cutting-edge research across the company for applications in healthcare, defense, and homeland security.
—by Monica Weiland