Safe enough to try. Not safe enough to coast.
What neuroscience, play theory, and two experiments gone sideways taught me about psychological safety — and where I respectfully part ways with how it's usually understood.
I have a manager who does something I did not fully understand until I tried to do it myself.
When something I have built doesn’t land - a program that flops, a bet that misses, an initiative I was certain about that turns out to be wrong for the context — she does not make me feel stupid. She asks what I learned. And then she moves on. No long debrief designed to feel like accountability but functioning more like a forensics exercise. No subtle shift in how she looks at me the next time I walk into the room. The verdict on the experiment is not the verdict on the person.
I only understood how rare this was when I started noticing what happens in my own body in that moment. The absence of dread. The fact that I can say ‘this didn’t work’ without spending the three days before the conversation rehearsing how to frame it. That is not just management style. That is physiology.
Stephen Porges, the neuroscientist who developed Polyvagal Theory, would call what my manager creates a ‘neuroception of safety’ — the body’s automatic, pre-conscious detection that this environment will not punish vulnerability. Your nervous system makes that assessment before your brain has formed a single coherent thought. And from that state — not from an inspiring culture deck, not from a values poster — creativity becomes possible. Experimentation becomes possible. The willingness to try something that might fail becomes possible.
This essay is about that. About what psychological safety actually is, where the neuroscience sits underneath it, what play theory and parenting research illuminate about it, and — here is the part I feel more strongly about than almost anything else — why it is not, and has never been, a licence to bring your B game.
The nervous system is in the room before you are
In 1994, Porges published what would become one of the more consequential frameworks in applied neuroscience: Polyvagal Theory. The short version is this: the human autonomic nervous system operates in a hierarchy. At the top is the newest evolutionary circuit — the ventral vagal system, associated with social engagement, creativity, collaboration, and learning. Below it are two older, defensive circuits: the sympathetic ‘fight or flight’ system and the dorsal vagal ‘shutdown’ state.
Here is what matters for anyone leading a team: which circuit is active is not a conscious choice. It is determined by what Porges calls neuroception — the nervous system’s ongoing, automatic scan of the environment for cues of safety or threat. Before you have opened your mouth in a meeting, your team’s nervous systems have already read the room. They have clocked the tension in your jaw, the speed of your sentences, whether you made eye contact when you walked in. And they have already begun routing toward openness or toward defense.
When the social engagement system is active, people access higher brain structures — creativity, learning, problem-solving. When it is not, none of that is available. No amount of declared values overrides what the body has already decided.
This is why psychological safety is not a soft concept. It is a prerequisite for any of the things organisations say they want: innovation, candour, experimentation, learning from failure. You cannot mandate those from a stage. You can only create the conditions ,physiologically ,in which they become possible. And those conditions are built or broken in the thousands of micro-interactions between managers and their people: how feedback lands, whether questions are welcomed or managed, whether admitting uncertainty is treated as weakness or wisdom.
My manager, whether or not she has read a word of Polyvagal Theory, has built those conditions consistently. And my journey as a manager is trying to do the same for my team.
What the science of play tells us about safe environments
Stuart Brown is a psychiatrist who spent decades studying play ,not in children, but across the entire human lifespan. He founded the National Institute of Play and conducted more than six thousand ‘play histories’ of humans ranging from Nobel laureates to incarcerated individuals. His finding, stated plainly: play is not a reward for productivity. It is a biological drive, as necessary to human functioning as sleep or nutrition.
What Brown calls ‘improvisational potential’ is one of play’s defining properties: in a true play state, there is no fixed script, no right answer you are converging toward. You are open to serendipity. You follow the thread wherever it leads. You discover things you were not looking for. This, he argues, is how creative work actually happens , not through forced ideation sessions, but through environments that allow people to drop the need to be correct and become genuinely curious.
'The opposite of play is not work. It is depression.' — Stuart Brown
That line is worth sitting with for a moment. Brown is not being provocative. He is making a neurological claim: the absence of a play state , the chronic inability to be curious, exploratory, unguarded , is not just joyless. It is associated with genuine psychological diminishment. When organisations strip out the safety to experiment, they do not get more focus and discipline. They get people who go through the motions with part of themselves switched off.
And this connects to a book that, at first glance, has nothing to do with leadership at all.
Robyn Gobbel is a therapist who works with children whose behaviour baffles and exhausts the adults around them. Her book, Raising Kids with Big Baffling Behaviours, draws on Polyvagal Theory and attachment research to make a central argument: all behaviour makes sense when you understand the state of the nervous system producing it. The child who shuts down, who acts out, who refuses to engage is not defiant. They are a nervous system in a defensive state, doing the only thing it knows how to do to stay safe.
The translation to organisations is direct and uncomfortable: when employees don’t raise their hand, don’t share the uncomfortable read, don’t admit the experiment failed — that is usually not laziness or indifference. That is a nervous system that has learned, correctly or not, that this environment punishes vulnerability. Your job as a manager is not to demand different behaviour. It is to update the signal.
Gobbel writes about parents needing to regulate their own nervous systems before they can help a child regulate theirs. The same principle applies to anyone leading a team. You cannot build a physiologically safe environment if you yourself are running in chronic threat mode — reactive in reviews, tense when things are uncertain, visibly impatient with imperfect work. The nervous system reads the room. It reads you first.
The experiment that worked sideways — and the one that didn’t
I want to tell you about two experiments, because I think abstractions about psychological safety need to be grounded in what it actually looks like to try something and live with what happens.
THE FRIDAY LINK
A few years ago, I started a Friday learning session for the sales team. The plan was almost embarrassingly simple: same link, every Friday, show up if you want to learn something. No elaborate agenda. No mandatory attendance. Just a consistent door, held open at the same time every week.
What I expected: sales people turning up to get better at selling.
What actually happened: over weeks and months, product teams started joining. Then ops. Then people from functions who had no formal reason to be in a sales learning session but were curious enough to walk through the door. It became without anyone designing it that way one of the more generative cross-functional forums we had. Not because the content was exceptional. Because the consistency of showing up, week after week, with no fanfare and no pressure, created a kind of trust that an expensive offsite rarely manages to manufacture.
The experiment worked. But it worked sideways , in a direction I had not planned for, toward an outcome more valuable than the original goal. That is what happens when you start small, stay consistent, and resist the urge to declare a verdict before the data has had time to arrive.
THE EXPORT THAT DIDN’T TRAVEL
Then there was the experiment that genuinely did not work.
We had a sales enablement program running well in India. Strong uptake, good feedback, the kind of results that make you think: this is working, let’s take it wider. So we tried to export it to Malaysia. Different markets, different cultural contexts, same program — same framing, same examples, same structure. It did not land. What resonated in India felt off elsewhere. The context that made the program make sense was specific to the environment it had been built in. We had lifted the content without lifting the conditions.
We saw the early signs within the first few weeks. And we called it , not with a long post-mortem, not with a round of blame, but with a clear-eyed: this is not working here, let’s stop and understand why before we continue.
Both experiments live in me as equally valid attempts. One succeeded in the way I hoped, and then in a way I didn’t expect. The other failed and told me something I needed to know. Neither outcome changes whether we should have tried.
Start small. Stay consistent. And the moment you see the early signs that something isn’t working, name it and move. The biggest waste is not a failed experiment. It is a failed experiment that nobody was willing to call.
Morgan Housel, the 5-year window, and the decoupling of effort from outcome
Morgan Housel, in The Psychology of Money, makes a point that I keep returning to in the context of organisational experiments. His argument is about investing, but the structure of the insight travels far beyond finance: you can make a good decision with an 80% chance of working out and still land on the wrong side of the 20%. You can make a poor decision and have it work out fine. Outcomes and decisions are not the same thing, and conflating them is one of the most expensive cognitive errors we make.
He writes: “The world is too complex for 100% of your actions to dictate 100% of your outcomes.” Luck and risk are, as he puts it, doppelgangers , inseparable, both real, both easy to underestimate when the outcome confirms the story we wanted to tell.
The managerial implication is sharp: if you evaluate experiments primarily by their outcomes, you will eventually train people to only attempt things they are already certain will succeed. That is not innovation. That is elaborate confirmation bias, dressed up in the language of execution.
Adam Grant sharpens this from a different angle. In his research on originals — people who generate and champion new ideas , he finds that the most creative people fail more often than their peers. Not because they are reckless, but because they take more shots. His principle, stated simply: success doesn’t measure a human being, effort does.
This is exactly the distinction I want to make and the one I think gets blurred most dangerously in how psychological safety is sometimes applied.
And there is one more reframe worth holding, especially for leaders evaluating experiments that are in motion. Ask yourself: will this matter in five years? Not as a way of dismissing short-term consequences , some short-term things matter enormously. But as a recalibration of what counts as failure. Most experiments, viewed from a five-year window, are not verdicts. They are data. They are the thing that tells you which direction to lean. The problem is that most organisations compress the time horizon until a small experiment that didn’t work feels like an indictment of the person who proposed it.
Psychological safety, at its best, restores the five-year view inside a one-week conversation.
But here is where I part ways — clearly and without apology
Psychological safety is not a comfort zone. And I have watched it become one in the hands of people who have borrowed its language without absorbing its meaning.
Amy Edmondson who coined the term and has spent 25 years researching it is explicit on this point. She defines psychological safety as the confidence that the team will not embarrass, reject, or punish someone for speaking up. She is equally explicit about what it is not: it is not permissiveness. It is not the absence of standards. It is not an excuse for poor performance.
Her framework maps four zones created by the intersection of psychological safety and accountability. High safety with low accountability produces a comfort zone — people feel good, but there is no pull toward genuine performance. High accountability with low safety produces an anxiety zone — people care about results but stay silent about the problems preventing those results. The zone Edmondson calls the Learning Zone — what she also terms the high-performance zone — requires both. High safety and high accountability, operating simultaneously.
Psychological safety is the foundation for healthy accountability — not an alternative to it. You need both. And without both, you have neither.
Here is my version of this, stated as plainly as I know how:
It is okay to fail despite your best efforts. It is not okay to give less than your best efforts and call it psychological safety.
The distinction matters because it is easy to conflate the two. A team that feels psychologically safe should be willing to try harder things, not easier ones. Should be willing to name problems earlier, not later. Should be willing to hold themselves to a higher standard, precisely because they are not spending energy managing how failure will be received.
When I think of the Asterix stories which I wrote about in a different context , Getafix does not give the magic potion to whoever asks for it. He gives it to people who are going to fight. The potion is not a consolation prize for showing up. It is fuel for someone who is going to give everything they have, in the direction of something that matters.
Psychological safety is the same. It is not given to reduce the demands of the work. It is given to make it possible to meet them.
Why this matters even more in the age of AI
I want to stay with this a little longer, because I think the intersection of psychological safety and AI is one of the most underexamined conversations in people strategy right now.
MIT Technology Review Insights, in a 2025 survey of 500 global executives, found that the vast majority of organisations that successfully moved AI projects from pilot to production had cultures that strongly encouraged experimentation. Eighty-three percent of executives surveyed believed that psychological safety measurably improves the success of AI initiatives. And 22% admitted they had hesitated to lead an AI project because they were afraid of being blamed if it went wrong.
That last number is worth pausing on. Nearly one in four leaders , people who understand the stakes of AI adoption , stayed on the sidelines not because the technology wasn’t ready, but because the culture wasn’t safe. The barrier to AI adoption in many organisations is not technical. It is psychological.
I wrote about my own AI experiments in my last piece , how I moved from anxiety to curiosity, how I began letting go and experimenting as an individual practitioner, as a manager, and as an HR leader. What I didn’t say explicitly, but felt throughout, was this: that shift was possible because of the environment I operate in. Because I had a manager who had made it safe to try. Because I had a team I had tried to extend the same safety to. Without that, I might have stayed cautious. Might have waited until AI felt more certain before I moved.
But uncertainty is the condition AI operates in. There is no certain version of this transition to wait for. If you are going to navigate it well individually, as a team, as an organisation , you need people who can try things that might not work, call it early when they don’t, and try something else without that experience becoming a weight they carry.
Research from Harvard Business School found that AI assistance particularly benefits employees with lower initial capability — suggesting that AI has the potential to democratise performance, to give people access to outputs that were previously out of reach. But only when people feel safe enough to experiment without judgment.
The organisations that will navigate the AI transition well are not the ones with the best models. They are the ones where people feel safe enough to be genuinely curious about what the models can do and honest enough to say when something isn’t working.
For leaders and managers: what this actually asks of you
I want to end the practical section with something concrete. Not a framework, there are enough of those. But a set of permissions and their paired demands, because I think that is the honest shape of this.
GIVE PEOPLE PERMISSION TO FAIL — AND MEAN IT
Not in the abstract. Not in a townhall statement about learning culture. But in the specific moments when it costs you something: when a project misses and your instinct is to find out whose fault it is, when a team tries something new and it doesn’t land, when someone brings you an honest read you didn’t want to hear. Permission to fail is not a declaration. It is a pattern of behaviour over time, and people will test it before they believe it.
CALL THINGS EARLY, NOT LATE
The experiments that damage organisations are not usually the ones that fail. They are the ones that fail slowly and publicly because nobody felt safe enough to name the problem when it was still small. Psychological safety is not just about trying. It is about the courage to say ‘this isn’t working’ before it becomes ‘this has completely broken down’. Reward early calls. Treat them as a form of operational excellence.
REGULATE YOURSELF FIRST
Gobbel’s insight for parents applies to managers: you cannot help someone else’s nervous system feel safe if yours is in a chronic threat state. If you are visibly reactive in reviews, if uncertainty makes you terse, if you treat imperfect work as evidence of inadequate people — your team’s nervous systems will read that and route accordingly. This is not about performing calm. It is about actually developing the capacity for it.
HOLD THE STANDARD AND THE SAFETY TOGETHER
Do not let high warmth become low standards, and do not mistake pressure for accountability. The learning zone Edmondson describes is not a place where expectations are relaxed. It is a place where high expectations can finally be met because people are not spending half their energy managing how failure will be received. Ask yourself not ‘am I being demanding enough?’ but ‘are my people able to bring everything they have to what I’m asking of them?’
THINK IN LONGER TIME HORIZONS THAN THE REVIEW CYCLE
Most organisations are not well set up to take a five-year view on experiments. The quarterly rhythm, the pressure of performance cycles, the manager who is also under pressure , these compress the window until small failures feel catastrophic. One of the most important things you can do for your team is explicitly name the time horizon for a given bet. ‘We are going to run this for three months and see what we learn’ is a fundamentally different frame than ‘let’s see how this goes’. The first has a shape. The second invites anxiety.
A closing thought — from the page itself
There is something worth noticing about the way Uderzo drew Obelix that has nothing to do with the story and everything to do with what the story is trying to say.
Obelix is always slightly too large for the space he is given. Not in a way that breaks the panel , Uderzo was too precise for accidents, but in a way that makes you aware of the frame. His shoulder grazes the border. His menhir nudges the edge. Dogmatix, small and certain, sits at his feet entirely within the lines. But Obelix himself presses against the boundary of the page as though the panel was drawn first and he arrived second and found it just a little too small for what he actually is.
It was a choice. Uderzo made it repeatedly, across decades, and readers loved it without always knowing why. I think they loved it because it told the truth about aliveness: genuinely alive things do not fit neatly inside the boxes drawn for them. They push. They overflow slightly at the edges. They make the frame aware of itself.
Psychological safety is not the absence of the frame. You still need the frame — the standard, the expectation, the clarity about what the work demands. What psychological safety does is make the frame large enough that people can bring their actual size to it. Not perform a smaller version of themselves in order to stay safely in the centre. Not shrink their ideas, their questions, their honest reads, because the frame has historically punished people who pressed against it.
Large enough to push. Demanding enough that the push means something.
The Friday learning session found its shape because nobody declared it too early. The India programme called its failure honestly, because the environment had made honest failure sayable. My manager gave me a panel large enough to try things that didn’t work and from that space, I have tried to give my team the same.
This is not a complicated idea. But it is one that requires something genuinely difficult from leaders: the willingness to hold a high standard and a large frame at the same time. To want both the container and the life inside it pressing against the edges.
Obelix never apologised for his size. He also never stopped carrying the menhir.
Give people a panel large enough for what they actually are. And then expect them to fill it — menhir and all.
About the author
Raji has spent 20+ years working on people systems — as a consultant helping organisations design their culture, and as a practitioner owning this work from the inside. At Pine Labs, she built and continues to own the culture code. The Operating Layer is where she writes about the architecture that makes organisations either coherent or exhausting to be part of.