Why teams that should feel safe still stay silent
The leadership team is around the table. The CEO presents a plan. She asks if anyone has objections. No one does. The decision is made. Three weeks later, in a hallway, she hears a department head say: "I knew from the start this wouldn't work." By then the damage is done.
The pattern shows up across the organizational literature, and often shows up in teams that look polite, harmonious, and well-functioning. The research has a name for the mechanism and several decades of data pointing in a consistent direction.
The study no one expected
In the mid-1990s Amy Edmondson, at the time a doctoral student at Harvard, set out to study medical errors in hospitals. The hypothesis was predictable: the best medical teams make the fewest errors. She measured team quality rigorously and collected error reports. Her surprising finding appeared in a study published in 1996: the opposite of what she had expected.
The best teams reported more errors, not fewer. At first Edmondson read the result as a failed hypothesis. Then she realized she had not been measuring how many errors happened; she had been measuring how many were reported. The good teams were probably not making more mistakes. They were just talking about them.
A few years later, in a study of 51 work teams in a manufacturing company (1999), Edmondson formalized the observation as psychological safety: the belief that it is safe to take interpersonal risk. To admit a mistake. To ask a question that might sound basic. To disagree with the boss. In teams without psychological safety, errors still happen, but nobody talks about them, and so nobody learns from them.
Google reported psychological safety as its leading factor
In 2012 Google launched what it called Project Aristotle: a two-year study of what actually separates good teams from bad ones. They looked at 180 teams, collecting data on composition (age, education, personality types), norms, and results.
The expectation was that it would be a blend of factors. Maybe background, maybe team size, maybe which tools they used. What they actually found, summarized in a New York Times Magazine article in 2016, was that the factor they reported as most important was psychological safety, above factors like team composition, tenure, or shared background. What mattered most, on their reading, was whether people dared to speak up.
The finding has been criticized on methodological grounds, and that is worth keeping in mind; one study from one company is not the last word. But it matches Edmondson's earlier data and Nembhard and Edmondson's study of neonatal intensive care teams (2006), where leaders' inclusive behavior was linked to both safety and improvement work. Several independent lines point the same way.
Three boundaries that keep the concept useful
Edmondson's work separates psychological safety from three related ideas that are often confused with it:
Candor can disrupt harmony
A polite team can have low psychological safety when preserving harmony discourages people from raising concerns. Agreement and silence can look identical in a meeting.
Safety and high standards can coexist
Edmondson's 2x2 model places psychological safety on one axis and performance standards on the other. High safety with low standards yields a comfort zone: pleasant but unproductive. High standards without safety gives you an anxiety zone. Only high safety plus high standards produces a learning zone. The learning zone allows people to surface problems while they work toward demanding results.
Safety belongs to the group
Psychological safety is a group property. The same person can feel highly safe in one team and highly unsafe in another in the same week. Team norms and leadership behavior shape that difference, so hiring for a particular personality cannot substitute for building safer ways of working.
Organizational silence: the structural level
Elizabeth Morrison and Frances Milliken's article Organizational Silence from 2000, published in Academy of Management Review, describes the same phenomenon at the organizational level. They show that silence is often systemic. Entire organizations can develop a culture where employees withhold information, not because they lack it, but because the cost of voicing it feels too high.
Morrison and Milliken argue that organizational silence grows out of two reinforcing factors: managers’ discomfort with negative feedback, and implicit beliefs about employees and disagreement that make silence the default response. They also describe a collective sensemaking dynamic: when people around you are silent, you interpret it as there being good reasons to be silent too.
That last point is close to an older social-psychological concept: Miller and McFarland's (1987) research on pluralistic ignorance. You attribute your own silence to fear or uncertainty. You interpret others' silence differently, reading it as an expression of something else, often agreement. Everyone makes the same attribution error at the same time, and a collective impression of consensus is produced that no one actually shares.
The flat-hierarchy paradox
There is a common intuition that flat hierarchies fix the problem automatically: remove the formal power asymmetry and people will feel free to speak. In our experience the pattern is often different. When there is no formal channel for challenging a decision, the challenge gets attached to a person rather than a question. In an organization with formal procedures you can send a memo that raises concerns without it necessarily being taken personally; it is part of the system. Without those procedures, your relationship with the decision-maker often carries most of the cost of speaking up.
If that reading holds, the practical response is to create formal routes for dissent. A review step, written challenge, or anonymous channel reduces how much the process depends on individual nerve.
Where the concept gets stretched
The term has also become useful cover. When a team wants to avoid an uncomfortable performance conversation, someone can claim it violates psychological safety. Edmondson explicitly includes the ability to hear difficult feedback in the concept. Popular usage often reduces "psychologically safe" to "comfortable," the confusion that her 2×2 matrix was designed to prevent.
The Aristotle finding also deserves more caveats than it usually gets. It was one study, one company, correlational rather than causal, and the effect sizes were smaller than the clean headline suggests. Its popularity owes partly to timing and partly to producing a single-variable story journalists could run with. Edmondson's underlying research, measured across dozens of teams in multiple industries, provides the stronger evidence base. Project Aristotle helped popularize the concept beyond that research.
Silence can also be appropriate. A team that has already worked through the arguments shouldn't relitigate them at every meeting. A junior team member who raises a concern privately with the team lead after the meeting may be choosing a more effective channel. The failure mode the research describes is systematic silence, the kind that reliably hides problems from the people who should see them. Conflating that with "any instance of not speaking up" turns a specific diagnostic into a vague indictment, and makes the concept less useful, not more.
What actually changes it
Edmondson and other researchers have identified interventions that change how teams ask for, receive, and respond to critical input.
1. Separate the problem from the person
"What in this plan is weak?" is a different question from "what do you think of the plan?" The first makes critique an intellectual task. The second makes it a personal judgment. Edmondson calls this framing the work as learning rather than execution.
2. Structural channels for disagreement
Anonymous input, written discussion before a verbal decision, and a rotating "devil's advocate" role make dissent a routine part of the process. Amazon uses six-page memos read in silence before discussion. Pixar has the "Braintrust" with an explicit norm of candor. Each method gives disagreement a defined route through the discussion.
3. The reaction to the first critical voice
The team learns the norm from your reaction to critical feedback. A leader who sighs or defends when challenged has just set the norm for the rest of the year. A leader who says "thank you, I hadn't seen that" has shown that the request was genuine. The first visible response becomes a template for later meetings.
Why meetings and presentations are where it falls apart
Even teams with high psychological safety in day-to-day work can lose it in formal meetings and presentations. Three reasons:
The number of eyes. More people present increases the bystander effect because each person expects someone else to speak. Pluralistic ignorance intensifies too, because it becomes harder to read everyone's face.
Hierarchy becomes visible. In a 1:1 it's you and me. In a meeting it's you, me, your boss, and your boss's boss, all in the same room. The risk scales.
The format offers little time. A presentation squeezes every reaction into the Q&A window at the end. People who need time to formulate don't get it. And the people who feel least certain are exactly the ones who need the time most.
Psychological safety depends on the relative cost of speaking and staying silent. Meeting formats, hierarchy, leadership practice, and reactions to criticism all affect that calculation. When silence is safer, decisions continue to be made while errors, doubts, and useful disagreement remain hidden.
Build regular routes for input, then respond to the first critical contribution in a way that makes the next one easier. Repeated experience turns an invitation to speak into a credible team norm.
Common questions
How does psychological safety differ from team building?
Team building usually focuses on relationships and morale. Psychological safety describes whether people take interpersonal risks such as admitting mistakes, disagreeing, or saying that they do not understand. A pleasant team can still have low psychological safety if maintaining harmony discourages challenge.
Can a team combine psychological safety with high standards?
Yes. Edmondson's 2x2 matrix treats psychological safety and performance standards as separate dimensions. High safety with low standards creates a comfort zone; high safety with high standards creates a learning zone. Teams need both conditions so people can surface problems while working toward demanding results.
How do I know if my team has it?
Edmondson's seven-item scale asks whether people feel safe admitting mistakes, raising concerns, and taking interpersonal risks with teammates. Meeting behavior provides another signal: track whether questions and objections appear before decisions are final. Unanimous silence alone cannot tell you whether the team agreed or withheld concerns.
Sources
- Edmondson, A. C. (1996). Learning from Mistakes Is Easier Said Than Done: Group and Organizational Influences on the Detection and Correction of Human Error. Journal of Applied Behavioral Science, 32(1), 5–28.
- Edmondson, A. C. (1999). Psychological Safety and Learning Behavior in Work Teams. Administrative Science Quarterly, 44(2), 350–383.
- Edmondson, A. C. (2008). The Competitive Imperative of Learning. Harvard Business Review, 86(7/8), 60–67. (Source of the 2×2 safety × standards model.)
- Edmondson, A. C. (2023). The Intelligent Failure That Led to the Discovery of Psychological Safety. Behavioral Scientist. (Edmondson's own retrospective and the source of the biographical framing of the 1996 study.)
- 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.
- Edmondson, A. C. (2018). The Fearless Organization: Creating Psychological Safety in the Workplace for Learning, Innovation, and Growth. Wiley.
- 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.
- Morrison, E. W. & Milliken, F. J. (2000). Organizational Silence: A Barrier to Change and Development in a Pluralistic World. Academy of Management Review, 25(4), 706-725.
- Duhigg, C. (2016). What Google Learned From Its Quest to Build the Perfect Team. The New York Times Magazine.
- Detert, J. R. & Edmondson, A. C. (2011). Implicit Voice Theories: Taken-for-Granted Rules of Self-Censorship at Work. Academy of Management Journal, 54(3), 461–488.
- Miller, D. T. & McFarland, C. (1987). Pluralistic Ignorance: When Similarity Is Interpreted as Dissimilarity. Journal of Personality and Social Psychology, 53(2), 298-305.
- Catmull, E. & Wallace, A. (2014). Creativity, Inc.: Overcoming the Unseen Forces That Stand in the Way of True Inspiration. Random House. (Primary source for Pixar’s Braintrust practice, discussed in ch. 5.)
- Bezos, J. (2017). 2017 Letter to Shareholders. Amazon. (Describes six-page memos read in silence before meetings.)
Make input part of the format
Plenora lets you collect input, questions, and disagreement throughout meetings and presentations. Anonymous response channels give more people a practical way to contribute before the room moves on.