Psychological Safety Is a Nervous System State, Not a Policy
Read Time 12 Minutes
Early in my career, I took a leadership role at a community mental health center that was, by every conventional measure, struggling. Turnover was high. Morale was low. Staff were carrying the kind of risk that comes with serving people in genuine psychiatric crisis and doing it with the feeling that no one above them in the org chart truly understood the weight of that work. I was brought in to help stabilize things: build trust, develop clinical acumen across the teams, repair the culture that years of disconnected leadership had eroded.
I dove in. I built individual relationships with everyone I supervised. I made myself available. I listened, I advocated, and I pushed hard for what I believed needed to happen. And within my span of control, things shifted, slowly, imperfectly, but genuinely. People started to show up differently. Trust, when it is actually earned through presence and consistency, does something real to a room.
But I kept running into the same wall. The organization's leadership understood psychological safety and trauma-informed care as something the clinical staff needed to embody, not something that needed to start in the C-suite, move through the director level, and permeate the entire system from the top down. They saw themselves as separate from the work, adjacent to it rather than inside it.
When I advocated for executive leadership to attend trauma-informed care trainings and actually practice what they were asking of everyone else, one senior leader looked at me and said, more or less: "Trauma-informed care is for service providers. That doesn't really apply to us."
That sentence has stayed with me for years, not because it was unusual but because it was so perfectly honest about a confusion that runs through most organizations trying to build psychological safety. The belief that safety is something you implement downward. That the culture you want is produced by the people carrying the work, rather than by the conditions the people at the top create or destroy every time they walk into a room.
The gap between what the policies promised and what people actually experienced was not a failure of implementation. It was a failure of category. Psychological safety has been treated, in most leadership literature and nearly all of the corporate training built on top of it, as a property of organizational culture that can be installed through the right inputs: the right language in the values statement, the right policies in the handbook, the right declared commitments from leadership. What the neurobiology suggests, however, is that psychological safety belongs to a different category of phenomenon altogether. It is an autonomic state, established and sustained through ongoing co-regulation between nervous systems, and no policy, however carefully written, can produce that state on its own.
This piece is an argument for taking that distinction seriously, and for what it asks of leaders who want to build the conditions under which their teams can actually do their best thinking.
What the body is doing when it feels safe
Edmondson's original framing of psychological safety, developed in her work on team learning at Harvard Business School, defined it as a shared belief that the team is safe for interpersonal risk-taking. That definition has been enormously useful and has properly become foundational in the organizational behavior literature. What it does not fully capture is the substrate underneath the belief. A person does not generate the belief that a team is safe through cognitive appraisal alone. The belief follows the state, more often than the state follows the belief.
The state in question is what Stephen Porges, in his polyvagal framework, calls ventral vagal engagement. The ventral branch of the vagus nerve is the most evolutionarily recent component of the autonomic nervous system, present only in mammals, and it is responsible for what Porges calls the social engagement system. When the ventral vagal circuit is active, the body is in a physiological state that supports connection: heart rate is moderated, facial muscles are mobile and expressive, the middle ear tunes toward human speech frequencies, vocal prosody softens, and the prefrontal cortex remains accessible for the kind of executive functioning that complex cognitive work requires.
This last point matters more than it sometimes seems. Amy Arnsten's research on prefrontal cortex function has shown that the PFC is exquisitely sensitive to stress signaling. Under conditions the nervous system reads as safe, the PFC supports the higher-order functions that organizations depend on: working memory, abstract reasoning, cognitive flexibility, the capacity to hold multiple perspectives, and the ability to inhibit reactive responses long enough to think. Under conditions the nervous system reads as threatening, even mildly threatening, the locus coeruleus releases norepinephrine in patterns that suppress PFC function and shift cognitive control to more reactive, habit-driven circuits. The shift can happen within seconds. It does not require a conscious appraisal of threat. It does not require the threat to be real in any objective sense. The autonomic nervous system is a pattern-matching organ, and it matches against patterns it learned long before the current meeting began.
What this means for teams is that psychological safety, when it is real, is identifiable through a constellation of embodied markers. People make eye contact. Voices are modulated rather than flat. There is laughter, including laughter at one's own expense. Disagreement happens directly and is not catastrophized. Silence is comfortable enough that people use it. The room has what could fairly be called a nervous system signature, and that signature is not produced by the policies on the wall.
What happens when the body reads threat
The contrast is instructive. When the autonomic nervous system shifts away from parasympathetic regulation, it moves in one of two directions, depending on what the body has learned to do under threat. The system mobilizes for fight or flight: heart rate accelerates, muscles tense, attention narrows to the perceived threat, peripheral vision contracts, and the cognitive bandwidth that was previously available for nuanced thinking collapses into the much smaller bandwidth required for survival arithmetic. If that is not an effective strategy, the nervous system defaults to shutting down: energy withdraws, affect flattens, motivation evaporates, and the person checks out while remaining technically present.
Erno Hermans and colleagues, in a 2014 paper in Trends in Cognitive Sciences, mapped the neural shift that accompanies acute stress. Under threat, the brain reallocates resources away from the executive control network, anchored in the prefrontal cortex, and toward what they call the salience network, which prioritizes rapid detection of and response to perceived danger. The shift is adaptive in genuinely dangerous situations. It is catastrophic in meetings. The work that organizations need from their people, the integrative thinking, the considered judgment, the willingness to surface a difficult observation, the capacity to take a creative risk, all of it depends on the executive network being online. When the salience network takes over, those capacities are not available, no matter how many times the leader has assured the team that this is a safe space.
The everyday manifestations are familiar to anyone who has spent time in organizational life. The person who had a sharp point to make but cannot remember it now that they have the floor. The team that produces consensus in the meeting and surfaces real concerns only after, in twos and threes, in the parking lot. The brilliant junior employee whose ideas get steadily less ambitious over months and quarters. The senior leader who used to push back and now does not. These patterns are not explained by the absence of policy. They are explained by nervous systems that have learned, through repeated experience, that this particular room is not safe for the kind of thinking the organization claims to want.
What I have seen consistently in this work, across both clinical practice and consulting engagements, is that the consequences compound. In the absence of psychological safety, isolation increases, productivity drops, motivation and engagement suffer, and team dynamics become unhealthy in ways that do not resolve on their own. A nervous system that has read a particular room as unsafe does not arrive in that room neutral the next week; it arrives primed. The work of rebuilding takes considerably longer than the work of breaking, which is part of why the consequences of unsafe teams are so durable and so expensive to organizations that eventually have to address them.
Why policies alone cannot do this work
The intuition behind policy-based approaches to psychological safety is reasonable, and it is worth taking seriously before explaining why it falls short. If people are not speaking up, make speaking up explicitly welcomed. If people are being punished for surfacing concerns, formalize protections against that punishment. That logic is sound as far as it goes. The problem is that psychological safety is not primarily about speaking up or about punishment. It is about what leadership teams demonstrate through their actions, in the small unremarkable moments that no policy ever captures and that every nervous system in the room is quietly cataloguing.
A person who has worked in an organization for any length of time has learned, somatically, what the actual rules are. Not the written rules. The real ones. They have learned them from how disagreement was received in the meeting last quarter, from whose ideas got credited and whose got absorbed without attribution, from what happened to the colleague who raised the difficult question at the wrong moment, from whether the leader who talks about vulnerability has ever actually been vulnerable in a room where it cost them something. The values statement on the wall is processed cognitively and filed. The way a leader's energy shifts when challenged, the subtle withdrawal of warmth when someone pushes back, the pattern of who gets the benefit of the doubt and who does not: all of that is processed autonomically and stored as somatic data. When the next meeting happens, the somatic data is what determines whether someone speaks. The cognitive memory of the values statement is, by then, beside the point.
This is the gap that most policy-based approaches never close. Policy can clarify expectations, name protections, and provide recourse when something goes wrong. What policy cannot do is regulate the nervous systems of the people in the room, and it cannot manufacture the embodied credibility that comes from leaders who have actually done their own work. Psychological safety is not something that gets communicated downward through memos and training mandates. It gets transmitted, or it doesn't, through the quality of presence a leader brings into the room and whether that presence, over time, gives people's nervous systems a reason to trust that it is safe to think out loud.
The senior leader I mentioned earlier, the one who believed trauma-informed care was for service providers and not for people like him, was not a bad person. He was operating from a genuinely common misunderstanding: that culture is something you shape in others, not something you are responsible for embodying yourself. What that misunderstanding costs, in attrition, in suppressed ideas, in the slow departure of people who once had a lot to give, tends not to show up cleanly on a balance sheet. But it shows up.
What Sigal Barsade's research on emotional contagion established is that the emotional state of one team member, particularly a leader, propagates through the group and shapes collective performance independent of the cognitive content of what is being discussed. The leader who walks into a meeting carrying unprocessed stress, even stress they believe they are concealing, is broadcasting that state to a roomful of nervous systems trained to read leaders for signals of safety or threat. The room adjusts accordingly, and it adjusts faster than any conscious decision-making could intervene. Trust, in this context, is not a value. It is a nervous system prediction, built up over many small instances, that this person and this room can be relied upon to remain regulated under pressure. Embodied values are values that have shown up in the leader's behavior under conditions where it would have been easier for them not to. Consistency is what allows the prediction to stabilize. None of these can be installed through policy. All of them have to be earned, repeatedly, in the moments where the leader's response is what the room is actually reading.
What this asks of leaders
The implication for leadership is significant and is not what most leadership development programs are organized around. If psychological safety is a co-regulated nervous system state, then the leader's most important contribution to it is not the policies they write but the state they bring into the room. A dysregulated leader cannot produce a regulated team. A leader who has not developed the capacity to notice when they are activated, to recognize the somatic signals of their own sympathetic mobilization, and to return themselves to a regulated baseline before responding, is broadcasting dysregulation into every meeting they enter.
This is a competence, and it can be developed. The capacity to self-regulate under pressure, to remain in parasympathetic engagement when the conversation is hard, to bring affect into the room that the team's nervous systems can settle alongside rather than brace against, is a learnable skill with measurable physiological correlates. It is also a skill that most leadership development programs treat as adjacent to the real work rather than as the work itself. The implicit assumption is that leadership is what happens above the neck, in strategy and decision-making and communication, and that what happens below the neck is a wellness concern. The neurobiology suggests this division of labor is wrong. The body is the medium through which leadership is actually transmitted. Leaders who develop sophisticated cognitive frameworks for psychological safety while neglecting their own regulatory capacity are like physicians who can name every pharmacological intervention while neglecting the basic exam.
There is a related implication for how organizations think about culture. The research on team effectiveness, including Google's Project Aristotle, has consistently identified psychological safety as a central, perhaps the central, determinant of team performance. That finding is now well-circulated. What is less often discussed is what it implies for where investment should go. If the determinant of team performance is a nervous system state, then the organizational practices that affect that state, the meeting structures, the feedback rhythms, the leadership selection criteria, the conditions under which difficult information surfaces, are not soft-skill concerns. They are the operating system.
The teams that learn to operate from this understanding tend to develop several characteristics that distinguish them. Innovation pipelines fill because people are willing to articulate the half-formed idea. Conflict surfaces directly because the cost of surfacing it is no longer prohibitive. Information moves through the organization without the distortion that fear introduces. Problems get named earlier, when they are still solvable, rather than later, when they are not. KPIs become more achievable, not because the team is working harder but because the team is working with the full cognitive bandwidth that regulated nervous systems make available.
What we are actually building
If we take this seriously, the work of building psychologically safe organizations changes shape. It stops being a project of writing better values statements and starts being a project of developing leaders whose nervous systems can hold a room under pressure. It stops being something HR owns and starts being something the leadership team owns, in the way that strategy or capital allocation is owned, with accountability and development paths and real consequences for not doing the work. It asks something of leaders that most leadership development programs are not currently asking them to give: the willingness to attend to their own regulatory capacity, in their own bodies, before asking their teams to do anything that requires regulation those teams may not be physiologically equipped to access.
This does not require abandoning policy, or the careful work that organizations have been doing on inclusion, protections, and the structural conditions that make safety possible to even attempt. The policies still matter. The structural conditions still matter. The argument here is not that they are unnecessary, but that they are insufficient, and that the work they cannot do has to be done somewhere else, in the bodies of the people leading the rooms.
The teams that learn to work from this understanding will keep producing the kind of thinking that organizations need from their best people. They will surface difficult information when it is still useful, hold disagreement without rupture, and recover from the small ruptures that do occur. The teams that continue to treat psychological safety as a deliverable to be produced through policy will keep encountering the same gap, between what their values say and what their people do, and will keep treating that gap as a communication problem rather than as the autonomic signal it actually is.
What we make of that signal is, at this point, a choice we are equipped to make.
If you want to sit with these ideas at greater length, I write longer essays on Substack each week, where there's room to follow the research further than a blog post allows. You can read them, and subscribe, here.
