"Psychological safety" gets used loosely enough in workplace conversation that it can sound interchangeable with "being nice" or "low pressure." The research behind the term describes something narrower and more testable, and a growing number of leaders now treat it as a lever on performance rather than a mood.
What does psychological safety actually mean?
Psychological safety is a shared belief within a team that it is safe to take interpersonal risks: asking a question, admitting a mistake, or challenging a decision without expecting embarrassment or punishment. The term comes from a 1999 study by Harvard Business School professor Amy Edmondson published in Administrative Science Quarterly, which found that psychological safety predicted how much teams engaged in learning behaviors such as seeking feedback and discussing errors openly.
That study also found something leaders still misread: psychological safety was not simply a byproduct of a friendly team. It mediated the effect of structural supports, like coaching from a manager, on whether learning actually happened. A well-resourced team without it still under-performed.
Edmondson's later work sharpened the point further: psychological safety is specifically about interpersonal risk, not risk in general. A team can be safe to admit an error while still operating under real deadline pressure or tight budgets. Those are separate variables, which is part of why the concept has survived years of skepticism from executives who assumed it simply meant lowering the stakes.
Why isn't this just about treating employees kindly?
Because the risk isn't always visible as open conflict or complaints. Reporting in Fortune in August 2026 described a related but distinct failure mode: capable employees often don't voice problems at all. They quietly build workarounds and absorb friction out of loyalty to the team, not fear of it, which means the absence of complaints can mask exactly the operational weak points a leader needs to see.
That distinction matters for how leaders read silence. A quiet team is not necessarily a psychologically safe one. It may simply be a team where the people most aware of the problems have decided raising them isn't worth it.
What actually happens when a team doesn't have it?
Employees disengage, absorb more stress alone, and are more likely to leave once conditions get harder, according to long-run workplace data. A study of psychological safety's protective effect held up even when other resources, like adequate staffing, did not.
Edmondson, along with Michaela Kerrissey of the Harvard T.H. Chan School of Public Health and HBS researcher Hassina Bahadurzada, tracked more than 27,000 workers at a large U.S. hospital system before and during the pandemic, in a study described by Harvard Business School's Working Knowledge. A one-standard-deviation increase in psychological safety measured in 2019 was associated with meaningfully lower burnout and higher intent to stay when researchers checked back in 2021, during acute pandemic strain. The protective effect was strongest among physicians, women, and people of color.
That timing matters. Because the 2019 baseline predicted 2021 outcomes, the researchers could rule out the possibility that psychological safety was simply a symptom of a team already doing well. It was measured before the crisis hit, then tested against how people fared once conditions got harder.
Does building it mean lowering performance standards?
No. The research treats psychological safety and accountability as separate dimensions, not opposites. A team can hold high standards while still being safe to challenge, and a team that is comfortable but has low standards isn't psychologically safe in the research sense — it's just conflict-avoidant.
The distinction is the difference between a team where people stay quiet to avoid friction, and one where people speak up because raising a problem early is treated as useful, not disloyal.
How do leaders build it in practice?
The research points to a small number of repeatable moves rather than a personality trait:
- Frame work as uncertain rather than fully solved, so questions read as appropriate rather than as gaps in competence.
- Acknowledge your own fallibility openly, including mistakes you've made, before asking others to surface theirs.
- Respond to bad news or dissent by asking a follow-up question rather than assigning blame, which is what teaches people whether speaking up was actually safe.
- Ask direct, specific questions of reliable performers about what quietly takes extra effort, rather than waiting for complaints to surface problems on their own.
- Separate the person from the mistake when something goes wrong publicly, so the team learns the failure was addressed rather than that the person who reported it was punished.
None of these are one-time gestures. Edmondson's research frames psychological safety as something a team rebuilds through repeated small interactions, which is also why it tends to erode quickly under a single manager who punishes dissent, even after years of a healthier pattern.
That fragility is also why leaders who inherit a team, rather than build one from scratch, tend to underestimate how long trust-building takes. The hospital-system data suggests the effects compound over years, not weeks, which is a longer runway than most performance-review cycles are built to measure.
For a related careers perspective, read Employers dropped the degree requirement — hiring barely moved.