Not all stress is the same: VTT’s study gathers unprecedented data on human responses

When a research volunteer dips a hand into a water bowl filled with ice cubes, an acute stress reaction is triggered. A different kind of stress reaction may be caused by a tricky mathematical task or performing in front of an audience.

The 4-year project, supported by the Research Council of Finland, focuses on creating methods to differentiate between various forms of stress. The research collects measurements from participants, including their backgrounds, various indicators, and physical responses.

“Machine learning can be used to determine a person's state by analysing a set of physiological signals and the latency between these signals and the actual stimuli. For instance, heart rate alone does not provide information whether a person is optimally stressed or overloaded,” says Senior Scientist Kati Pettersson.

When modelling succeeds accurately and yet in real time, it enables development of systems that support human performance and information processing, particularly in cognitively demanding environments. These include defence and security sectors, among others.

Senior Scientist Johanna Närväinen specialises in brain imaging and medical physics. She says that physiological reactions held the main role in this research project:

“The brain is capable of transitioning rapidly between states, such as shifting from one task to another. However, physiological responses, like changes in cortisol levels and reductions in heart rate, occur more gradually. In the project, we studied how different states, from concentrating on a mental arithmetic task to experiencing a painful cold-water test, show up in measurements and how people recover from them. we also studied what happens when these states are changed more rapidly and how the state transition happens.”

The researchers developed new measurement methods as well. These include sweat rate plasters, biochemical sensors and solutions based on printed electronics. An interesting direction of development are technologies that allow data collection without the need for the user to wear any separate measuring devices.

“For instance, sensors integrated within the environment, such as meters embedded in chairs and space-monitoring radars, provide means for continuous monitoring of the cognitive state and stress in real life settings,” Närväinen says.

Observing eye movements is one way to measure stress. Kati Pettersson explores eye movements and other biosignals from the head-area are related to human performance. She describes how eye movements of the test subjects were tracked using a video-based method with two infrared cameras. The eye was also measured using electro-oculography (EOG). This method provides valuable data about eye blinks, their number and duration, which can indicate task difficulty or pain.

“Analysing eye movements can provide insight into the strategies a person is using: when an individual experiences stress, cognitive overload, or fatigue, their gaze may focus on a narrower area, whereas in a relaxed and attentive state, the observation may cover a broader field. Thus, we can distinguish a positive flow state from panic stress” Pettersson describes.

In addition, eye movement analysis can be used to detect fatigue – not sleep deprivation but rather boredom during the task. Even more accurate information of the person’s state can be acquired by combining eye movement data with other parameters. For instance, the pupil size can indicate the level of concentration as well as the electrical conductivity of the skin.

The researchers also gathered saliva samples from participants for cortisol analysis, along with detailed information on participant background and context-dependent appraisals and performance data. The data collected may be used in future collaborative projects, such as those with defence forces, and can be repurposed as technology advances.

“Once we know, which physiological signals to measure, the collected material can be used in the development of new methods and systems. The research material has been designed and collected with such care that it will serve for a long time, up to five years onwards, and enable further development,” Närväinen says.

Promoting individual well-being and preventing harmful stress contributes to societal cost savings and generates opportunities across multiple industries. Device manufacturers, such as Oura, utilise extensive data bases to develop more accurate sensors and algorithms. In the security and defence sectors and the educational sector, monitoring stress and the cognitive state in real time may help, for example, in developing different training simulators. When tasks are optimised to an optimal level of difficulty – neither excessively simple nor unduly challenging – learning processes become more efficient, and trainees are kept motivated.

Cortisol is a classical biomarker of stress but as cortisol levels in saliva and urine samples require laboratory analysis, it is not a very convenient indicator – yet. “In the future, fast measurement of cortisol levels from saliva or sweat may become routine, enabling stress to be tracked as an integrated component of broader health monitoring. Even if the solutions were not initially available to consumers, they can be important in medical care and professional work,” Pettersson reminds.

The most important outcome of the research is the academic and scientific foundation for building future solutions. It is not merely about how to measure physiological states but how they are modelled and comprehensively understood. The funding by the Research Council of Finland is critically important as an enabler of this kind of slow basic research.

The research project is beginning its final year, and Johanna Närväinen says that the team is eagerly looking forward to dive into the research data.

“For the first two years, we focused on developing sensors, but this year we have finally been able to perform experimental research. During the summer, we finalised the data collection and now we have cleaned and organised the data. Now, we are holding a real treasure chest full of information in our hands, which we will soon examine thoroughly.”