II · Protection
Awareness and the Legal Frame
analysistraining
Seeing, avoiding, de-escalating, and the law.
Last reviewed 2026-10-07
Folklore
The Color Code Is Doctrine, Not Science
White, yellow, orange, red — a readiness scale taught in almost every personal-safety course in the country.
It came from Jeff Cooper’s firearms instruction in the 1970s, adapted from a military alert scheme, and was offered as a statement about willingness to act rather than a model of attention. No controlled research validates its stages or shows that self-classifying into them improves any outcome. Even writers inside that training culture note it is routinely misapplied as an attention monitor it was never built to be.
“Stay alert at all times” Teaches a Symptom
Constant vigilance as the core discipline of personal safety.
Sustained vigilance degrades measurably within tens of minutes — this is one of the oldest findings in human factors. And persistent threat scanning has a clinical name: hypervigilance is a diagnostic feature of PTSD, studied as exhausting and self-perpetuating. Eye-tracking work with trauma survivors found hypervigilant attention made appraisal of ambiguous situations worse, not better.
You Cannot Read a Stranger’s Intent in Seconds
Popular training, most famously in the intuition literature, holds that a person can detect violent intent from brief exposure to a stranger.
That claim is not peer-reviewed research; it is practitioner assertion. What the research supports is narrower and pointed the other way — it is about how offenders perceive potential victims over an observation window, not how bystanders perceive offenders in a glance.
Attention Is Not Coverage
With enough discipline you can maintain full awareness of your surroundings.
About half of observers concentrating on a counting task failed to notice a person in a gorilla suit walk through the scene and beat their chest. Expertise does not fix it — radiologists reading scans missed a gorilla image inserted into the lung field. Focused attention reliably costs you salient, unexpected events in plain sight.
Three Levels
Level 1
Perception
What is actually present — people, exits, vehicles, lighting, distance, the ground under your feet. Not interpretation yet. Most failures at the later levels are really failures here: the element was never taken in.
Level 2
Comprehension
What the elements mean together. One person on the corner is an element. One person on the corner who has been in three of the four places you have been in 20 minutes is a situation.
Level 3
Projection
What happens next if nothing changes. This is the level that buys time, because it produces the decision to change route, cross the street or leave before anything has happened.
Mica Endsley’s three-level model, published in Human Factors in 1995, is the situational-awareness framework with an actual research base — used in aviation, air traffic control and human factors work, and measurable with established probe techniques. It is descriptive of how understanding is built, not a readiness scale.
What Is Visible About You
What the Gait Research Found
In 1981 Grayson and Stein filmed 60 pedestrians on New York streets and showed the footage to incarcerated violent offenders, who rated targets for perceived ease of attack. Movement analysis showed the ratings tracked stride length, movement fluidity, and whether limb movement was coordinated across the body — independently of size, sex, or apparent strength.
The original design could not separate movement from clothing, attractiveness and other visible confounds.
The Replication That Fixed the Confound
In 2002 Gunns, Johnston and Hudson repeated the question using point-light displays, which strip away everything except motion. Kinematic features alone — short stride, an up-and-down foot lift, little arm swing, low energy — still accounted for a substantial share of variance in vulnerability ratings.
It measures perceived vulnerability as rated by observers, not subsequent real-world victimization. The inference to actual selection is reasonable, but it is an inference.
What This Does and Does Not License
It supports a modest, actionable claim: how you move is visible, it is read, and it is partly under your control. Purposeful pace, head up, arms swinging, eyes taking in the environment rather than a screen.
It does not license claims about reading strangers, profiling by appearance, or identifying dangerous people on sight. The literature is small and its lineage narrow.
Pathway to Violence
1. Grievance
A perceived wrong, usually specific and personal, that the subject believes justifies action.
2. Ideation
Violence appears as an acceptable answer. Often expressed — to others, online, in writing — which is what makes intervention possible at all.
3. Research and Planning
Target selection, surveillance, learning schedules and security, studying previous attacks.
4. Preparation
Acquisition, rehearsal, logistics, and often final-act behaviors — giving things away, farewell messages, putting affairs in order.
5. Breach
Probing or circumventing whatever security exists.
6. Attack
The point at which prevention is no longer available.
Space, Then Words
Space Before Words
Distance is the only de-escalation tool that works regardless of whether the other person is reasonable, intoxicated, in crisis or committed. Everything else below assumes you have bought room first.
Establish Verbal Contact, One Person Only
A single voice. Multiple people talking at an agitated person compounds the arousal that is already degrading their processing.
Short Sentences, Repeated
Agitation reduces the ability to process complex speech. Brief, concrete, repeated — not reasoned argument.
Identify What They Want and How They Feel
Not agreement with it. Naming it accurately is what demonstrates you are listening, which is the precondition for anything else.
Agree, or Agree to Disagree
Find the true thing you can agree with — the specific fact, the principle, or that they are angry — rather than contesting the frame.
Set Clear Limits, Offer Choices
State what cannot happen, then offer two acceptable options. Choice restores a sense of agency, and its absence is often what is driving the escalation.
Do Not Provoke
No barked orders, no challenges to dignity, no audience. Winning the exchange and ending it safely are usually different objectives.
These are the domains of the verbal de-escalation consensus statement published by the American Association for Emergency Psychiatry’s Project BETA workgroup in 2012, developed for agitated patients in emergency settings and adopted widely in healthcare. Two honesty notes. It is an expert consensus statement, not a controlled trial. And the strongest experimental evidence in this area — a randomized trial of de-escalation training in the Louisville Metro Police Department — measured real but modest effects on officer communication behavior and use of force, in trained professionals inside structured encounters. Extrapolating a guarantee to an ordinary person in a spontaneous confrontation outruns what anyone has measured.
The Structure of the Law
Imminence
Was the threat happening now?
Force is justified against a threat that is immediate, not one that has passed and not one that is feared later. A completed assault becomes a matter for police; a threatened future assault becomes a matter for a protective order. Neither is self-defense at the moment you act.
The Model Penal Code’s formulation is force “immediately necessary … on the present occasion,” which is looser than strict common-law imminence. Which version your state uses is a real difference.
Proportionality
Was the force no greater than the threat?
Force must not exceed what is reasonably necessary. Deadly force generally requires a reasonably perceived threat of death or serious bodily injury — not insult, not trespass, not property.
Property is treated very differently from persons. Deadly force to protect property alone is essentially never justified under the Model Penal Code or in the large majority of states.
Reasonableness
Did you believe it, and would a reasonable person have?
Most American self-defense doctrine applies a hybrid test: what the defendant actually believed, and whether a reasonable person in that situation would have believed the same. Both halves must hold.
People v. Goetz (New York, 1986) is the leading case establishing the subjective-and-objective structure.
Retreat
Were you required to withdraw if you safely could?
This is the structural variable that differs most across states. Duty-to-retreat jurisdictions require you to take a safe avenue of escape before deadly force; stand-your-ground jurisdictions remove that requirement anywhere you are lawfully present. The castle doctrine is the near-universal exception inside your own home.
The research on the effect of these laws is not what most people expect — see the note below.
Who Is Being Defended
Self, another person, or property?
Defense of another generally borrows the standard that would apply to the person defended. Defense of property is the most restricted category by a wide margin.
Getting this wrong is a common route from a defensible act to an indefensible one.
Two Systems, Two Outcomes
Criminal, civil, or both?
An act found justified in a criminal proceeding does not automatically end civil exposure. Some stand-your-ground statutes bundle civil immunity; many do not.
Ask about both when you read your own state’s law. People routinely research only the criminal half.
An Honesty Note
Expanded self-defense laws are widely believed to make people safer. The published research does not support that. A systematic review of 25 studies found national patterns ranging from no change to small increases in violent crime following such expansions, with Florida showing robust increases in firearm and total homicide rates, and documented disparities in case outcomes by victim race. An Oxford-led study in JAMA Network Open in 2022 reported a 7.8% mean rise in monthly homicide rates across adopting states. An earlier economics study of castle-doctrine expansions found increased homicide without any deterrent effect on burglary or robbery. None of this tells you what to do in a moment. It tells you that the legal architecture is a policy question with measured consequences, not a safety feature that comes free.
Protective Orders
What a Protective Order Is
A court order creating an enforceable prohibition, the violation of which is itself an offense. That is its mechanism: it converts otherwise ambiguous contact into a documented, chargeable act.
What the Research Says It Does
Orders are associated with reduced violence and re-victimization for a substantial share of recipients, and victims consistently report improved sense of safety — including many who experienced violations. That subjective benefit is real.
What It Is Not
It is not a physical barrier and it does not reliably stop a determined or high-risk offender. Violation rates are meaningfully high, particularly in stalking cases, where behavior often continues through channels an order cannot reach.
The Documentation That Makes It Work
An incident log — date, time, place, what happened, witnesses, screenshots, preserved messages — is the practitioner-standard practice recommended by the Stalking Prevention, Awareness, and Resource Center, which the Department of Justice’s Office on Violence Against Women funds. A pattern is what makes a case; individually trivial incidents are what a pattern is made of.