It’s very hard to learn to swim on land. At some point, sooner than later, ya gotta get in the water.
I recently participated in a simulation with emergency managers, while they learned by engaging in a deliberately chaotic scenario. Fun, practical, and very, very useful. As an observer, it was a fascinating to watch
Typically, we teach skills in calm conditions, assess them in calm conditions, sign the person off as competent, and then hope or expect the skill to survive intact when the working environment adds urgency, uncertainty, noise, interruption, interpersonal difficulties, scrutiny and consequence. This feels especially hopeful when we know what the working environment will be like.
Sometimes it does transfer. Sometimes we discover that what we certified was really classroom competence.
Here’s what happens
The mechanism isn’t mysterious.
Stress changes the state in which knowledge has to be retrieved and used. Working memory shrinks, attention narrows, and people reach for the familiar response even when it’s no longer the right one. That doesn’t mean they don’t know what to do. It means the knowledge now has to be reached through a different physiological and cognitive state from the one it was learned in. The question for transfer is whether anyone ever practised that retrieval.
This is old magic.
Saunders, Driskell, Johnston and Salas meta-analysed 37 studies of stress inoculation training in 1996, covering 1,837 people. Exposing people to manageable stress while teaching coping and performance strategies reduced performance anxiety, reduced state anxiety and improved performance under stress.
That work is thirty years old, so here’s an update. Low and colleagues meta-analysed pressure training in 2021, reviewing 14 studies across sport and law enforcement, and 394 people. Training with deliberately added pressure improved later performance under pressure. Most outcomes were measured straight after training and only one study ran a retention test, so there’s a further study opportunity there.
It’s consistent, still. A 2026 position statement from Exercise and Sports Science Australia endorses pressure training, citing the 2021 meta-analysis, albeit with an extremely useful and valid warning attached about imposing pressure without support. Note that for debriefs.
Some more detail
The individual studies are worth knowing because they’re so concrete. Oudejans and Pijpers trained expert basketballers and expert dart players, 17 in each experiment, with or without induced anxiety, with the dart players throwing from high or low on a climbing wall. Fear of heights and/or falling is a fair way to induce stress in many people! Only the groups that practised with anxiety maintained their performance in the anxiety post-test, despite higher anxiety, heart rate and perceived effort.
Nieuwenhuys and Oudejans then did it with 27 police officers, using an instructor who occasionally fired back with coloured soap cartridges. Before training, anxiety degraded both groups’ shooting accuracy. Afterwards, the trained group’s accuracy no longer deteriorated under threat while the control group’s still did, and when both groups were retested four months later, the trained group’s gains were still present.
Now the warning, and the fix
Lawrence and colleagues had novices learn a golf putting task with anxiety present for all of practice, none of it, or part of it. The group that practised only with anxiety performed worst in the test without anxiety, and performance under pressure was best when the anxiety exposure came in the second half of learning.
That’s critical. It isn’t simply “add stress and stir”. First, build the skill, then bring the pressure in later. That specificity of practice hypothesis is the pattern the study was testing.
Design it in
The design principles run like this.
Build the skill first, in enough psychological space to understand, fail and improve. Then, and only then, add friction, in stages. You could remove some certainty, add time pressure, add interruption, add a difficult stakeholder or an authority dynamic, then combine them in simulation. This becomes graded exposure with learning, rather than surprise punishment.
Importantly, make the pressure representative. Shouting at people, humiliating them or engineering chaos for its own sake teaches one lesson only, which that the instructor is a hazard. The stressors worth training are the ones the job actually contains. Time compression. Incomplete information. Conflicting priorities. A second incident arriving before the first is resolved. A senior person interrupting. A team member challenging the plan. You get the point.
A missing ingredient
Then though, which is a bit we can often miss, debrief the cognition, not just the outcome.
The plan was late and the wrong person was called. Fine, but also, what happened to attention? What disappeared? What changed in how people were communicating? What assumption locked in early? Who held useful information and didn’t speak? What did the leader do that raised or lowered the temperature or pressure in the situation? That turns an exercise into a map of how this team performs under load, and it gives people the vocabulary to name narrowing and overload as predictable patterns rather than personal failure.
Also, train the recovery manoeuvre. People will get overloaded. Attention will narrow. Something will be missed. Teams that can pause, restate the objective, rebuild the shared picture, move key information onto a visible aid and reset the next decision don’t need flawless individuals. That’s a far stronger safety proposition than hoping nobody ever degrades under load.
A key inclusion
And put speaking up in the simulation.
If the exercise tests technical skills but nobody is expected to challenge the incident controller, question the senior clinician, or say “I think we’re solving the wrong problem,” then the most important high-pressure team behaviour was never exercised. Speaking up in a calm workshop is generally easier. Speaking up when the room is fast, the leader is confident, and we’re on the clock, is a mightily different proposition.
But it’s one to rehearse. Watching how a leader responds to speaking up, and what they do next, is educative all on its own.
To be clear, I’m absolutely not saying that all training should be stressful. Neither does the literature say that. Most learning needs calm. Therefore, don’t assume competence demonstrated in one state transfers to another. If the work is done under pressure, pressure itself is part of the task environment, and must be accounted for in learning.
Train the knowledge. Train the skill. Then, carefully and late, train access to both under the time pressure, interruption and scrutiny, or whatever pressure the job actually contains.
Photo by hidde schalm on Unsplash