Psychological Safety for Learning Cultures
Create environments where people report mistakes freely and rethinking becomes routine
Grant contrasts performance cultures, which prioritize excellence of execution, with learning cultures, which prioritize growth and rethinking. Both can produce impressive results, but performance cultures become vulnerable to overconfidence cycles where standard procedures are never questioned.
The framework centers on Amy Edmondson's discovery that psychologically safe teams report more errors but actually make fewer errors. When people feel safe to admit mistakes, root causes are diagnosed and problems are eliminated. When people hide mistakes to avoid punishment, the same errors repeat indefinitely. Google's research found that psychological safety was the single most important factor distinguishing high-performing teams.
Grant argues that psychological safety alone is insufficient. It must be combined with process accountability, the practice of evaluating decisions based on the quality of the thinking process rather than only the outcome. A bad process with a good outcome is luck. A good process with a bad outcome might be a smart experiment. Leaders who model this combination create organizations where rethinking becomes routine.
- Psychologically safe teams report more errors but make fewer errors because they learn from each one
- Performance cultures reward execution but can trap organizations in overconfidence cycles
- Learning cultures thrive when psychological safety is paired with process accountability
- Leaders must model humility by sharing their own mistakes and uncertainties first
- Evaluate decisions on the quality of the thinking process, not just the outcome
- Model vulnerability as a leaderShare your own mistakes, uncertainties, and rethinking moments publicly. When Ellen Ochoa became director of the Johnson Space Center, she began asking questions like: What are the cases where I might be wrong? and What would I need to see to change my mind? This signaled that questioning was not just tolerated but expected.
- Establish process accountabilityCreate systems that evaluate how decisions are made, not just whether they produced good results. Implement a rethinking scorecard that tracks how thoroughly different options were considered. Reward people for the quality of their analysis, even when the outcome is negative.
- Normalize error reportingMake it easy and safe to report mistakes. When someone surfaces an error, respond with curiosity about root causes rather than blame. Publicly acknowledge and thank people who report problems early. The goal is to make hiding mistakes feel riskier than revealing them.
- Replace best practices with better practicesAbandon the language of 'best practices,' which implies the ideal routine already exists. Instead, adopt the language of 'better practices' and continual improvement. This signals that every process is a hypothesis to be tested and improved, not a sacred procedure to be followed.
Astronaut Luca Parmitano noticed water droplets in his helmet after his first spacewalk. The team assumed it was a drink bag leak and replaced the bag. A week later, during his second spacewalk, water flooded his helmet so rapidly he nearly drowned. The root cause was a fan/pump/separator failure that would have been caught if the team had investigated the initial anomaly rather than normalizing it.
Edmondson surveyed hospital teams on psychological safety and compared self-reported errors with covertly observed actual errors. Initially, psychologically safe teams appeared to make more errors. When a covert observer was sent in, the data flipped: safe teams reported more errors but actually made fewer. They learned from each mistake and eliminated root causes.
Grant tells the story of astronaut Luca Parmitano, who nearly drowned in space when water leaked into his helmet during a spacewalk. A week earlier, he had noticed water droplets in his helmet after his first spacewalk. The crew assumed it was a drink bag leak because that had happened before without consequence. They replaced the bag but stopped investigating. This normalization of minor anomalies was the same pattern behind the Challenger and Columbia shuttle disasters, where NASA's performance culture prevented rethinking of accepted assumptions. Grant traces how NASA's culture failed to question the O-ring risk before Challenger and the foam strike before Columbia, killing fourteen astronauts in total.