- Psychological safety is a shared belief that it is safe to take interpersonal risks in the team: to ask, admit, challenge and suggest. It is not comfort or lower standards.
- The research base is substantial. A meta-analysis of 136 samples links it with task performance, and the original study was carried out in a manufacturing company.
- Google's Project Aristotle made the idea famous, but it was internal research reported through Google's re:Work site and the press, not a peer-reviewed study.
- The brain treats the risk of being judged as a real threat, and under threat the thinking that improvement depends on becomes harder to reach.
- Leaders build psychological safety through what they do repeatedly, not what they announce. It can be measured, and it can be changed.
What is psychological safety?
Psychological safety is a shared belief, held by the members of a team, that the team is safe for interpersonal risk-taking. In a psychologically safe team, people can ask a question, admit a mistake, raise a concern or challenge a plan without expecting to be embarrassed, rejected or punished for it. It is a property of the team's climate, built by experience, not a personality trait.
That definition comes from Amy Edmondson's 1999 study, which introduced team psychological safety as a research construct 1. The idea is older. William Kahn used the term in 1990 to describe whether people feel able to bring their whole selves to work without fearing damage to their image, status or career 2.
Two words in the definition matter most. Shared means it belongs to the team, not to one confident individual. Belief means it is a prediction, built from what people have seen happen before. In BTFA™ terms, it lives at the Believe stage. That is why it cannot be switched on by a memo.
What psychological safety is, and what it isn't
Edmondson and Lei are clear that psychological safety is about interpersonal risk, not about feeling comfortable 3. Leaders who think it means being soft either reject it or apply it badly.
| IT IS | IT IS NOT |
|---|---|
| Confidence that you can raise a problem and be heard | A guarantee that everyone will agree with you |
| Freedom to admit mistakes so the team can learn from them | Freedom from accountability for results |
| Challenge of ideas, plans and assumptions | Permission to be rude or to attack people |
| Compatible with demanding targets | Lower standards or a quieter life |
Think of two separate dials. Standards set how much is asked of people. Psychological safety sets whether they can tell you the truth about how it is going. High standards with low safety produce anxiety and filtered reports. Improvement needs both dials high.
What the research shows
Edmondson's original study covered 51 work teams in a manufacturing company. Teams with higher psychological safety showed more learning behaviour, such as seeking feedback and discussing errors, and that learning behaviour was linked to team performance 1. The study also pointed to team leader coaching and organisational support as part of what shapes those outcomes.
Since then the evidence has grown considerably:
- Performance. Frazier and colleagues combined 136 independent samples, covering more than 22,000 individuals and nearly 5,000 groups. Psychological safety was linked with task performance and with the helpful, cooperative behaviour that keeps teams working 4.
- Improvement programmes. In a study of 47 mid-sized German companies, Baer and Frese found that climates of psychological safety and initiative were linked to firm performance, and strengthened the relationship between process innovations and results 5.
- Google's Project Aristotle. Google studied 180 of its own teams and reported psychological safety as the most important of five dynamics, alongside dependability, structure and clarity, meaning, and impact 6. It became famous through Google's re:Work site and a 2016 New York Times Magazine article 7. It was internal company research, not a peer-reviewed study, and should be weighed accordingly.
Reviews reach a consistent picture: psychological safety supports learning, speaking up and performance, and leadership is one of the main influences on it 8 9. For what happens when it is missing, our guide to why good teams stop raising problems looks at the research on employee silence.
“Psychological safety is not declared. It is what people predict will happen to them, based on what they have seen happen to others.”
Why the brain treats interpersonal risk as threat
Why should a raised eyebrow in a meeting carry so much weight? Neuroscience offers a useful explanation, read with care.
A leading account of brain function describes the brain as a prediction system. It constantly forecasts what is about to happen, based on past experience, and updates when it is wrong 10. Applied to a team, every previous reaction to bad news becomes data. The person deciding whether to speak is not weighing your values statement. Their brain is forecasting your response.
Being judged by others is a potent stressor. A meta-analysis of 208 laboratory studies found that tasks where performance could be negatively evaluated by others, especially when people had little control, produced the largest cortisol responses and the longest recovery times 11.
Stress also changes what the brain can do. Research in animals and humans shows that even mild, uncontrollable stress can quickly impair the prefrontal cortex, the region that supports flexible thinking, working memory and self-control 12. These are exactly the capabilities that problem-solving and improvement depend on.
Some brain regions involved in physical pain also respond to social rejection 13. How far that overlap goes is still debated, so we treat it as supporting evidence rather than proof.
None of this shows a single "fear centre" that leaders switch on or off. The careful claim is that social threat is processed as real threat, has measurable physiological effects, and narrows the thinking a team needs most.
Psychological safety through the BTFA chain
In BTFA™ we describe behaviour as the visible end of a sequence: Believe, Think, Feel, Act. It is our model, not an established scientific law, but it shows where psychological safety is built and lost. You can read more on the BTFA model page.
- Believe. From experience, the brain predicts what speaking up will cost here. "Problems get you noticed for the wrong reasons."
- Think. That prediction shapes how a new situation is read. A drifting tolerance becomes "not my place to say" rather than "the line needs to know".
- Feel. The interpretation changes state. A rise in threat narrows attention and makes caution feel like good judgement.
- Act. The person stays quiet, softens the message, or works around the problem. The result confirms the belief.
The sequence also runs backwards. A new action, tried where it is safe and supported, produces a different experience. Reflected on and repeated, that experience can revise the belief. This is why leaders change psychological safety by changing what people experience, not by asking them to feel safer. Our article on the neuroscience of change covers how beliefs update.
At scale, we describe the result as the MRD cycle: misalignment, resistance and disengagement, each feeding the next. Low psychological safety sits underneath all three.
The four stages of psychological safety
A popular practitioner model comes from Timothy Clark, who describes four stages: inclusion safety, learner safety, contributor safety and challenger safety 14. The idea is that people first need to feel they belong, then that they can learn and make mistakes, then that they can contribute, and finally that they can challenge the status quo.
It is a helpful way to talk with teams, particularly the point that challenge is the hardest and most valuable stage. But it is a practitioner framework, not a model tested in the way Edmondson's construct has been. Use it as a conversation tool, not a measure.
How to measure psychological safety
The most widely used measure is Edmondson's seven-item team scale 1. Team members rate how far they agree with statements covering whether mistakes are held against people, whether problems and tough issues can be raised, whether people are rejected for being different, whether it is safe to take a risk, whether it is hard to ask for help, whether anyone would deliberately undermine others' efforts, and whether people's skills and talents are valued and used. Several items are worded negatively and reverse-scored.
To make the results useful:
- Measure the team, not just the person. Psychological safety is a shared belief, so look at team averages and at how much members disagree.
- Protect anonymity. A survey about safety that does not feel safe will tell you very little.
- Track the trend. One score is a snapshot. Repeating the measure shows whether what leaders are doing is working.
- Pair it with behaviour. Watch what the climate produces: how many problems are raised at tier meetings, how honest the Check stage of PDCA is, and how early risks appear on projects.
Illustrative examples from manufacturing and Lean
These are illustrations of common patterns, not specific clients.
A plant runs PDCA on a scrap reduction trial. The trial underperforms. Where it is safe to report a disappointing result, the Check stage tells the truth and the team learns. Where it is not, the data is quietly adjusted or the trial is declared a success, and the cycle continues on paper while the learning stops. PDCA improves the process. BTFA asks what the people running it expect to happen if the test fails.
A newly qualified engineer notices that a torque setting in the work instruction does not match the drawing. Two experienced operators have run it that way for months. Whether the engineer says anything depends on whether newcomers' questions have been welcomed or ridiculed before.
A kaizen event produces twenty ideas, and the area manager picks the three he had already planned. Nobody objects. Next time, there are six ideas.
WORKSHOPBuild a team that tells you the truth earlyLeadership workshopHow to build psychological safety: what leaders can do
- 01Frame the work as a learning problem. In Edmondson's operating theatre research, the most effective leaders gave a motivating reason for the change and played down differences of power and status so that people would speak up 15. Say out loud that the work is uncertain and that you need people's eyes on it.
- 02Invite input explicitly. Leader inclusiveness, meaning words and actions that invite and appreciate contributions, predicted psychological safety and helped overcome status differences in improvement teams 16. Ask specific questions, ask the most junior person first, and wait.
- 03Respond productively to bad news. Thank the person, then ask what they saw. Treat the problem as information about the system before asking about people.
- 04Close the loop. Tell people what happened to what they raised. In a randomised field study, a management walk-around programme hurt performance on average, and work areas did better where problems were actually acted on 17.
- 05Regulate your own state. Stress narrows a leader's thinking as much as anyone's 12, and your reaction becomes everyone else's working conditions. Catching it before it reaches the room is a skill that can be practised.
- 06Keep the standards high. Safety without challenge drifts into comfort. Be clear about what good looks like, and make it safe to say when you are not there yet.
Mistakes to avoid
- Treating it as niceness. Avoiding hard conversations erodes safety, because people stop trusting what they hear.
- Running a survey and doing nothing. Asking for honesty and then not acting teaches people that honesty is pointless.
- Launching a speak-up campaign. Posters aim at the Act stage and leave the Believe stage untouched. People trust what they see you do.
- Handing it to HR. HR can design the measure. Only line leaders, in daily operations, create the experience that shapes the belief.
- Blaming the team for a low score. A low score is feedback on the climate leaders have created, not on the people reporting it.
For more, see our Lean leadership guide, resistance to change and, for organisation-wide work, our culture change consultancy.
Questions
What is psychological safety in simple terms?
Psychological safety is a shared belief in a team that you can speak up, ask questions, admit mistakes and challenge ideas without being embarrassed, rejected or punished. Amy Edmondson defined it in her 1999 research as a belief that the team is safe for interpersonal risk-taking. It is built by experience, especially by how leaders respond.
Is psychological safety the same as trust?
They are related but not the same. Trust is usually about whether you can rely on a particular person. Psychological safety is about the climate of a whole team, and whether people expect interpersonal risk-taking to be accepted there. A team can contain individual relationships of trust and still feel unsafe for challenge in a group setting.
What are the four stages of psychological safety?
The four stages are a practitioner model by Timothy Clark: inclusion safety, learner safety, contributor safety and challenger safety. It describes a progression from belonging to being able to challenge the status quo. It is useful for team discussion, but it has not been tested as rigorously as Edmondson's research measure.
How do you measure psychological safety?
The most common tool is Amy Edmondson's seven-item team survey, which asks how far people agree with statements about mistakes, raising problems, risk-taking, asking for help and having their skills valued. Results are most useful when gathered anonymously, looked at by team, tracked over time and set alongside behavioural signs such as how early problems are raised.
Can a team have psychological safety and high standards?
Yes, and the best teams do. Psychological safety does not lower the bar. It makes it possible to tell the truth about where the team is against that bar, so problems surface early when they are cheaper to fix.
Sources
- Edmondson, A. (1999). Psychological safety and learning behavior in work teams. Administrative Science Quarterly, 44(2), 350–383.https://doi.org/10.2307/2666999
- Kahn, W. A. (1990). Psychological conditions of personal engagement and disengagement at work. Academy of Management Journal, 33(4), 692–724.https://doi.org/10.2307/256287
- Edmondson, A. C. & Lei, Z. (2014). Psychological safety: the history, renaissance, and future of an interpersonal construct. Annual Review of Organizational Psychology and Organizational Behavior, 1, 23–43.https://doi.org/10.1146/annurev-orgpsych-031413-091305
- Frazier, M. L., Fainshmidt, S., Klinger, R. L., Pezeshkan, A. & Vracheva, V. (2017). Psychological safety: a meta-analytic review and extension. Personnel Psychology, 70(1), 113–165.https://doi.org/10.1111/peps.12183
- Baer, M. & Frese, M. (2003). Innovation is not enough: climates for initiative and psychological safety, process innovations, and firm performance. Journal of Organizational Behavior, 24(1), 45–68.https://doi.org/10.1002/job.179
- Google re:Work. Guide: Understand team effectiveness.https://rework.withgoogle.com/intl/en/guides/understand-team-effectiveness
- Duhigg, C. (2016, 25 February). What Google learned from its quest to build the perfect team. The New York Times Magazine.https://www.nytimes.com/2016/02/28/magazine/what-google-learned-from-its-quest-to-build-the-perfect-team.html
- Newman, A., Donohue, R. & Eva, N. (2017). Psychological safety: a systematic review of the literature. Human Resource Management Review, 27(3), 521–535.https://doi.org/10.1016/j.hrmr.2017.01.001
- Edmondson, A. C. & Bransby, D. P. (2023). Psychological safety comes of age: observed themes in an established literature. Annual Review of Organizational Psychology and Organizational Behavior, 10, 55–78.https://doi.org/10.1146/annurev-orgpsych-120920-055217
- Clark, A. (2013). Whatever next? Predictive brains, situated agents, and the future of cognitive science. Behavioral and Brain Sciences, 36(3), 181–204.https://doi.org/10.1017/S0140525X12000477
- Dickerson, S. S. & Kemeny, M. E. (2004). Acute stressors and cortisol responses: a theoretical integration and synthesis of laboratory research. Psychological Bulletin, 130(3), 355–391.https://doi.org/10.1037/0033-2909.130.3.355
- Arnsten, A. F. T. (2009). Stress signalling pathways that impair prefrontal cortex structure and function. Nature Reviews Neuroscience, 10(6), 410–422.https://doi.org/10.1038/nrn2648
- Eisenberger, N. I. (2012). The pain of social disconnection: examining the shared neural underpinnings of physical and social pain. Nature Reviews Neuroscience, 13(6), 421–434.https://doi.org/10.1038/nrn3231
- Clark, T. R. (2020). The 4 Stages of Psychological Safety: Defining the Path to Inclusion and Innovation. Oakland, CA: Berrett-Koehler. ISBN 9781523087686.
- Edmondson, A. C. (2003). Speaking up in the operating room: how team leaders promote learning in interdisciplinary action teams. Journal of Management Studies, 40(6), 1419–1452.https://doi.org/10.1111/1467-6486.00386
- Nembhard, I. M. & Edmondson, A. C. (2006). Making it safe: the effects of leader inclusiveness and professional status on psychological safety and improvement efforts in health care teams. Journal of Organizational Behavior, 27(7), 941–966.https://doi.org/10.1002/job.413
- Tucker, A. L. & Singer, S. J. (2015). The effectiveness of management-by-walking-around: a randomized field study. Production and Operations Management, 24(2), 253–271.https://doi.org/10.1111/poms.12226
Creator of BTFA. Time-served precision engineer with 30+ years in WCM and Lean change across Europe, and a Masters in Applied Neuroscience.
More from David →