For more than three decades, psychology labs have had a dependable recipe for making a healthy adult miserable on command. Someone finally asked the people who cook it.
Forty six researchers who administer the Trier Social Stress Test responded to a call for their experiences with the paradigm, and their accounts appear in a new paper in the open access journal Comprehensive Psychoneuroendocrinology. The collection covers participant reactions, panel member reactions, and the practical headaches of running the protocol. The conclusion is not the one most readers would expect. The test appears to place a substantial burden not only on the volunteers being stressed but, in a different way, on the researchers sitting across the table.
The authors then turn that observation into an open research question. Does imposing stress on another person produce a physiological stress response in the person imposing it? Nobody has measured it.
A Protocol Built to Break Composure
The Trier Social Stress Test was developed by Clemens Kirschbaum and colleagues at the University of Trier in 1993, and it has been the workhorse of human stress research ever since. A participant gets about 10 minutes to prepare a short speech arguing why they are the ideal candidate for a job. They deliver it to a panel that offers no encouragement, no nodding, no reassuring noises. Then, without warning, they are asked to count backward from 1,022 in steps of 13, and corrected whenever they slip.
Two ingredients do the work: social evaluative threat and uncontrollability. Across the six studies described in the original report, the protocol produced two to four fold increases in salivary cortisol above baseline, along with changes in ACTH, growth hormone, prolactin and heart rate. A meta-analysis of 76 sub-studies covering 5,171 participants confirmed the cortisol response holds up, found that swapping the job interview for a different speech topic changed nothing measurable, and found that men respond more strongly than women. A larger pooled analysis of 171 studies and 8,452 healthy adults reported that cardiovascular responses to the test are actually bigger than hormonal ones.
That reliability is why the test spread. It is also why the ethical arithmetic has always been slightly uncomfortable. The design has to be strong enough to move hormones and mild enough to clear an ethics board.
What the Panel Sees from the Other Side of the Table
Volunteers have been asked before what the experience feels like. A qualitative study of participants in PLOS One found that men who went through the procedure described exactly the two dimensions the theory predicts: being judged and having no control over it. What has almost never been documented is the experience of the judges.
Panel members are typically research assistants in white lab coats, trained to hold a neutral expression no matter what happens in front of them. They watch people stammer, go blank, tear up or freeze. They do it repeatedly, across dozens of sessions, for a study whose success depends on their refusal to offer comfort.
The new compilation records that this is not emotionally free. Alongside the accounts of strain, the authors include a section of humorous anecdotes involving unusual participant performances and unexpected mishaps, which reads as a reminder that a standardized protocol keeps colliding with the unpredictability of actual people.
A Paradigm That Was Never as Standardized as It Looked
The reports also cover implementation problems, and this is where the paper has practical teeth. The 1993 protocol left a great deal unspecified. An introductory guide to running the test published in Neuroscience and Biobehavioral Reviews documented how much labs differ: the timing of events, the number and method of cortisol sampling, whether a glucose drink is given, how rooms and cameras are set up, panel composition, and how much the panel interacts with the participant.
Panel composition is not a trivial variable. If the humans enforcing the social threat are themselves tired, uncomfortable or inconsistent, that is a source of variance sitting inside the instrument. Some groups have responded by moving the panel into software. Researchers at Trier have published an open virtual reality version of the test, and online adaptations have multiplied since the pandemic. Virtual panels tend to produce a smaller cortisol response than a live one, which is itself a clue about how much of the effect comes from real people in the room.
What This Paper Can and Cannot Establish
This is a collection of anecdotal reports, not a study. The researchers who responded chose to respond, which means the sample skews toward people with something to say. No cortisol was measured in the panel members. There is no control condition. The authors present the material as observation and invitation rather than as evidence.
That distinction matters, because the interesting claim, that administering stress may stress the administrator, is at this point a hypothesis. It would be testable with the same saliva assays already sitting in these labs, applied to the people in the lab coats. Until someone runs that study, the burden described here is reported experience, not a measured physiological effect.
Anyone who has sat on an interview panel, a disciplinary hearing, or a performance review will recognize the underlying question. It has simply never been put to the people whose job is to make the discomfort happen.
Key Questions Answered
What is the Trier Social Stress Test? A standardized laboratory procedure created in 1993 in which a person delivers an impromptu job interview speech and performs mental arithmetic in front of an unresponsive panel. It reliably raises cortisol and heart rate.
What did the new paper actually do? It gathered written accounts from 46 researchers who have run the test, covering participant behavior, panel member reactions, implementation problems and recommendations for future stress studies. It is a compilation of experiences, not an experiment.
What was the surprising part? The accounts indicate the protocol is demanding for the researchers serving on the panel, not only for the volunteers. The authors flag this as a question worth studying rather than a proven effect.
Does this mean stressing someone else raises your own cortisol? Not established. No hormone measurements were taken from panel members. The paper raises it as a hypothesis for future research.
Is the test harmful to participants? It is designed to produce a short, controlled stress response and is approved by ethics committees. Effects are transient and volunteers give informed consent, though the experience is genuinely unpleasant while it lasts.
Are there alternatives? Yes. Virtual reality and online versions exist and are easier to standardize, but they generally produce a weaker cortisol response than a live panel in the room.
