The books in the Quantum Mind of Safety™ series are built on fifty-two fatality investigations. I was in the room for each of them. This briefing is what I would tell you about them if we had one hour and you had to make decisions by Friday.
I have spent more than thirty years in high-hazard work: oil and gas, mining, nuclear, aviation, chemical manufacturing, transportation. For most of that time I believed what the industry believes. Write the procedure. Train to it. Audit against it. Discipline the gap. I was good at it, and I watched it fail people I had shaken hands with the week before.
Somewhere past the twentieth investigation I noticed that I was writing the same report. The names changed. The equipment changed. The finding did not. The permit was signed. The training was current. The hazard was known. And in the days before the event, the people around the worker had seen something and had no language for it. He seemed quieter. He was rushing. He agreed to everything in the toolbox talk and asked nothing. Nobody wrote that down, because nothing on the form asked for it.
That is the whole thesis, and I will say it plainly so you can decide whether to keep reading. Every fatality I have investigated had warning signs in the psychology that the procedures never captured. The signs were observable. They were early. They were in the nervous system, the language and the social behavior of ordinary people doing a familiar job. Our safety systems were not built to see them.
The frameworks in this briefing are my answer to that. They came out of the investigations first and the science second. I went to psychology and neuroscience to understand what I had already seen, and I have tried to be careful about the science ever since: where a model is debated I say so, and where a physics idea is an analogy I call it one.
What you will get in the next ten pages: the thesis, the patterns the investigations kept producing, the Cognitive Reliability Model™ as a map you can hold in your head, the ten frameworks on one spread, the decisions that follow for a leader, a one-page self-check, and the references behind every outside claim.
Read it with one site in mind. The one that worries you a little, even though its numbers look fine. That is usually the right one.
A procedure describes the work as someone imagined it. A human being performs it with whatever nervous system, attention and social pressure they walked in with that morning. The gap between those two things is where people die.
Erik Hollnagel gave the field a useful pair of terms: work-as-imagined and work-as-done. The procedure is work-as-imagined. The shift is work-as-done. Every site I have walked has a gap between them, and most of the time the gap is harmless adaptation. The problem is that the official system only measures the imagined version, so it cannot see the real one drifting.4
Diane Vaughan called the slow version of that drift the normalization of deviance in her study of the Challenger launch decision. A deviation is tolerated once, nothing happens, and it quietly becomes the standard.2 James Reason described the organizational version: latent conditions that sit inside a system for years until an active failure lines up with them.3 Those books describe my case files.
The second idea is less familiar in safety, and it is the center of the Quantum Mind of Safety™ work. Decision quality is a physiological state before it is a competency. Research on stress and the prefrontal cortex suggests that even mild, uncontrollable stress can rapidly degrade the functions we rely on for judgment, planning and self-monitoring.10 That is a rushed worker at hour eleven of a turnaround.
The widely used dual-process picture of thinking makes the same point from another angle: a fast, automatic mode runs most of the shift, and a slow, deliberate mode is expensive and easily crowded out.9 A procedure assumes the slow mode is always available. It is not.
I use polyvagal theory as a practical language for the states I see on the floor: connected and learning, revved and reactive, or shut down and absent. It is widely used in applied work and still debated among researchers, and I teach it that way.7, 8 The point for a leader is simpler than the debate. The worker who signs the permit in a shutdown state has not read it.
The third idea is social. Amy Edmondson's research on psychological safety showed that teams where people feel safe to speak up learn more and surface more error.6 In my files the inverse shows up with terrible regularity: someone saw it, and the cost of saying so felt higher than the cost of staying quiet.
I do not have a statistic for what follows, and I will not invent one. What I have is a pattern that repeated across industries, countries and decades, strongly enough that I built a model on it. Five sentences came out of the interviews again and again. I now treat each one as the warning signal for a pillar of the Cognitive Reliability Model™.
Said by a crew that had worked beside the person for years. When I pressed, "fine" turned out to mean quieter than usual, short on the radio, skipping the shared break, or flat in a way they had put down to a bad week. Chronic tension and chronic flatness both read as normal once they have lasted a month.
The hazard was on the risk assessment. It had been there for years. The team had walked past it so often that it had stopped being a hazard and become scenery. Known risk underestimated, minor risk over-reacted to, and the near-miss log telling you nothing because reporting had become inconsistent.
Task compression. A job that took forty minutes being done in twenty because the crane was waiting or the shift was ending. A verification step dropped, then dropped again, until "good enough" decisions were accumulating faster than anyone could notice them.
Someone saw it. Often more than one person. They deferred to the most senior person present, or they had been embarrassed the last time they raised a concern, or they assumed someone else would speak. One person's stress had set the tone for the group, and the group had agreed, without a word, that the risk was acceptable.
The information existed. It was in a toolbox talk nobody absorbed, a permit signed but not read, an observation card that went into a system and never came back out, or a handover that lost the one detail that mattered between nights and days.
The honest reading of those files is uncomfortable for a compliance program. The system worked as designed. It just was not designed to see what mattered. That is why the thesis puts psychology first in time. The procedures keep their importance. You need to look at the person before you trust the paper.
CRM™ is the meta-framework. I describe it as the operating system; the other frameworks are the apps that run on it. Each pillar is a capacity a reliable human system needs, and each one has a sentence I learned to dread in interviews. When you hear the sentence, you know which pillar has degraded and which tools to send in.
| # | Pillar | The warning signal | What it looks like on site | Deploy first |
|---|---|---|---|---|
| 1 | Neurological Baseline | "He seemed fine. Then it happened." | Chronically tense, reactive or flat workers. High absence. Short fuses whenever safety comes up. The crew has normalized a state that is not safe to work in. | Vagal Safety Ladder™, PERSONA™ |
| 2 | Threat Appraisal Architecture | "They knew the hazard was there." | Known risks underestimated, minor risks over-reacted to. Near-miss reporting inconsistent. Precursors visible in hindsight and unrecognized in real time. | PERSONA SIF™, The Backpack™ |
| 3 | Decision-Making Under Pressure | "They rushed it." | Task compression. Time pressure treated as normal. Verification steps skipped. "Good enough" decisions accumulating. ThresholdIQ™ is the assessment tool for this pillar. | Quantum Safety™, PERSONA™ |
| 4 | Social Reliability Systems | "Nobody said anything." | Deference to authority. Silence in safety meetings. One person's stress spreading through the crew. Groupthink on what counts as acceptable risk. | PRISM™, BEACON™ |
| 5 | Communication Fidelity | "I thought they knew." | Toolbox talks that do not land. Permits signed but not read. Observations reported and never acted on. Information lost across shifts. | VOICE™, NISOS™ |
Each framework was built to act on something I kept seeing in the files. I have kept each entry to what a leader needs: what it is, what it sees, and what it does. All frameworks are proprietary to Safety Sense Inc.
Twelve behavioral archetypes: Guardian, Analyst, Achiever, Connector, Adventurer, Mentor, Visionary, Enforcer, Harmonizer, Skeptic, Pragmatist, Stabilizer. Each has a strength (the Skeptic catches flaws; the Guardian is the most reliable person on the crew) and a characteristic way of getting hurt (the Achiever rationalizes shortcuts as experience; the Connector will not contradict the group).
For leaders: a lens for understanding and coaching. Never a diagnosis, never a label, never used to rank, hire or discipline anyone.
Observe your own state before you enter. Broad Scan the whole scene before you narrow. Specific Focus, zone by zone. Evaluate what is driving the behavior. Record, including what you could not see. Validate with other sources. Engage with a coaching conversation.
For leaders: replaces the clipboard walk with an observation discipline that reads people as well as conditions. The rule everyone remembers: scan wide before you narrow.
Maps each archetype to its SIF risk pattern and its observable precursor: the Harmonizer's flat affect replacing warmth, the Adventurer's near-misses framed as achievements, the Enforcer's crew going quiet. Describes a five-stage precursor chain from baseline behavior to the event, and three intervention windows.
For leaders: the early-warning layer. Psychology Before Procedures™ means starting at Window 1, weeks out, while the signal is still cheap to act on.
Five elements: Perception, Response Pattern, Influence Pathway, Safety Integration, Motivation Anchor. Treats resistance as a threat response (acknowledge, validate, then redirect), puts peer voices before authority voices, and asks for meaning over slogans. Tracks a worker response spectrum from Shutdown through Resistance, Compliance and Engagement to Commitment.
For leaders: how a safety message is actually constructed by the person hearing it. Ethical influence only. No manipulation.
The leadership framework from The Plastic Organization. Its field tool is the BEACON™ Conversation in six steps: Build rapport first, Explore with curiosity, Acknowledge before correcting, Clarify at low threat, Offer co-created solutions, Navigate to a forward commitment.
For leaders: the conversation your supervisors have in the sixty seconds after someone speaks up decides whether anyone speaks up again.
Volume Calibration, Operational Tone Mapping, Intentional Pace and Pause, Cadence and Prosody Patterns, Emotional Resonance Alignment. Underneath: rate, volume, pitch, tonality, pause. Three signatures to listen for: the Flat Voice, the VOICE™ Signal, the Spike Voice.
For leaders: a worker's voice changes before their behavior does. So does a supervisor's.
Three states: connected and learning (asks questions, stops work when unsure); fight or flight (rushed, tunnel vision, skips steps); shutdown (flat, monosyllabic, compliant but absent, signs the form and cannot recall it). Four steps: recognize the state, regulate your own first, activate connection (stand beside, same direction, "What's the biggest risk you see right now?"), then sustain it by fixing the chronic stressors.
Five layers of learned response that every worker carries in, from early attachment ("Am I safe?") through "What happens when I speak up?" to peer-defined acceptable risk. Used to recognize disproportionate reactions and respond to the root without ever asking about history. The Backpack™ is safety intelligence, and it is never therapy.
Observer Effect: presence changes behavior. Superposition: before the event the worker was both compliant and non-compliant; ask what shifted the odds. Entanglement: what happens on nights does not stay on nights. Wavefunction Collapse: decisions at a risk threshold are irreversible, so intervene before them. Uncertainty Principle: the more precisely you measure compliance, the less you know about psychological state.
Measures how close to the edge a person operates before judgment breaks down. Built on the Wavefunction Collapse idea: the threshold state before an irreversible decision. Free, no signup, at safetysenseinc.com/resources.
| SIF prevention | PERSONA™ → PERSONA SIF™ → Vagal Safety Ladder™ → NISOS™ |
| Culture rebuild | PRISM™ → Quantum Safety™ → BEACON™ → VOICE™ |
| Post-incident | PERSONA™ → The Backpack™ → Vagal Safety Ladder™ → PERSONA SIF™ |
| New site baseline | CRM™ audit → PERSONA™ → NISOS™ → PRISM™ |
You will not run these frameworks yourself. You will decide whether your organization is allowed to see what they see, and whether anyone is rewarded for acting on it. Five decisions follow from the thesis.
Lagging indicators tell you about the past. Compliance audits tell you about the paper. Neither will show you a Harmonizer going flat or a crew going quiet. Ask your safety function what they currently record about nervous-system state, language and social behavior. The honest answer is usually nothing, and that is the gap.
The PERSONA SIF™ precursor chain runs from baseline behavior, weeks out, through stress loading and signal emergence, to a precursor cluster hours before the event. Most organizations intervene at the cluster, when a supervisor finally sees something alarming. Window 1 is cheaper, quieter and far more likely to work. Give your supervisors permission to act on a quiet signal without needing a near-miss to justify it.
Your open-door policy is irrelevant. What matters is the face a supervisor makes when a worker says the job does not feel right, and whether the concern comes back with an answer. Train the BEACON™ Conversation and audit the loop closure.
A dysregulated leader produces a dysregulated crew. The Vagal Safety Ladder™ starts with the person holding the clipboard regulating their own state before they try to change anyone else's. That is a trainable skill, and it is the one most senior people have never been taught.
A permit signed in a shutdown state is a legal document and nothing more. Communication Fidelity is a pillar for a reason. Ask for evidence that the message landed, in the worker's own words, before you count it as communicated.
These are the questions I ask in the first hour on a new site. They are organized by pillar so the answers tell you where to look. Ask them of your safety lead, then ask the same questions of a supervisor on the floor, and pay attention to the distance between the two sets of answers.
Mark each statement as it is today, with one site in mind. This is a structured self-reflection prompt built on the five CRM™ pillars. It is a conversation starter for a leadership team, and it is not a validated psychometric instrument or a prediction of anything.
You now have the thesis, the map, and the questions. Most leaders who read this far want one of three things next: to understand the science properly, to train the people who will run the frameworks, or to talk through a specific site. Here is where each of those lives.
The Quantum Mind of Safety™ books hold the fifty-two investigations in full, the science told as history, and each framework shown working. The Plastic Organization is the book that holds BEACON™. Available in paperback and Kindle.
Discover Your Safety PERSONA™ (sixty questions, twelve to fifteen minutes, radar chart results, no signup). ThresholdIQ™ for risk tolerance. The PERSONA SIF™ Precursor Analyzer for an organizational precursor index. The other three free downloads sit beside this one, including the one-page field checklist 7 Warning Signs You're About to Have a Fatality and the Safety Culture Diagnostic Checklist.
HAVEN University™ teaches the frameworks across eight thresholds, from the foundations of behavior and neuroscience through leadership, culture and communication, to professional certification (PCP, NCO, VCP, LPI, QSP and others). There is a live session every Thursday at 5:00 pm Pacific.
If you have a site in mind, the fastest route is a short discovery call. Bring your answers to page 9 and page 10. I will tell you which pillar I would start with and why, and whether I am the right person to help.
Every outside claim in this briefing points to one of the sources below. Where a model is contested I have included the critique alongside it. Field observations and the count of investigations are my own and carry no citation.