Hypervigilance

Hypervigilance is threat detection running at a level the situation does not require. Not fear exactly — fear is what happens once something is found. This is the scanning that happens before.

It is the least visible of the stress states, because from outside it looks like competence. The person who knows the mood of a room within four seconds of entering it, who hears the shift in a voice on the phone, who registers that something is wrong at the dinner table while everyone else is still eating — that person is not imagining things. They are usually right. The question this entry is about is not accuracy. It is price.

The door opens and before the coat is off you have read four faces and formed a working theory about which one is annoyed.

You notice what changed: the slightly shorter message, the chair angled differently, the fact that they said "fine" with a half-second delay. You track where people are in a room. You sit where you can see the door — a cliché, and also something people do without ever deciding to.

You pre-run conversations that have not happened, checking for where they could go wrong. You are the first to know a relationship is in trouble and the last to be able to relax once it is not, because reassurance settles the question and not the scanner.

And there is the physical layer, which people usually do not connect: startling easily, a jaw that is tight by mid-morning, difficulty getting to sleep even when nothing is pending, because sleep requires standing down and standing down is exactly the thing the system is not doing. The entry on being tired but wired covers the night end of it.

Threat gets attentional priority in everybody. Öhman and Mineka's work established that fear-relevant stimuli are detected fast, learned readily and unlearned slowly — a system built to be wrong in the cheap direction. A false alarm costs a second; a missed alarm costs more. Nothing about that is pathological. It is the default human setting.

Anxious people run it hotter. Bar-Haim and colleagues' meta-analysis, covering 172 studies and more than two thousand participants, found a robust threat-related attentional bias in anxious individuals and not in non-anxious ones. The effect is real, it replicates, and it is not large — which is worth saying, because the popular version of this finding treats anxious attention as a different faculty rather than the same faculty a few notches up.

It is learned where it paid. Pollak and Tolley-Schell's study is the most direct evidence: physically abused children showed altered attention specifically to angry faces — quicker to engage, slower to disengage — compared with non-abused children. In a house where an angry face predicts what happens next, early detection is a genuine skill. The skill does not know when the house has been left.

The bill is cumulative. McEwen's work on allostatic load is the frame for what running the system continuously costs: stress mediators that are protective in bursts do measurable damage when they never switch off — cardiovascular, metabolic, and to sleep first. Hypervigilance is not dangerous because it is inaccurate. It is expensive because it does not stop.

It is not paranoia. Paranoia involves beliefs that are false. Hypervigilance is usually accurate about the small things — the mood really had shifted — and wrong only about the implication, which is nearly always that it concerns you and requires action.

It is not sensitivity. Sensory-processing sensitivity is about depth of processing across all input, pleasant and unpleasant. Hypervigilance is selective and it is about danger. A sensitive person is moved by a piece of music; a hypervigilant one is tracking whether the person next to them is enjoying it.

It is not intuition, although it is sold as such. Reading people well and never being able to stop reading people are different capacities that happen to share a mechanism.

It is not a personality trait. It varies by room. The same person is relaxed with one friend and fully instrumented at a family dinner, which is the most useful diagnostic question available: where does it switch off? If the answer is nowhere, that is a different conversation, and it is the one at the end of this entry.

Three specific bills, in the order people usually notice them.

Rest. Standing down is the one thing the system will not do, so recovery does not happen even when the day is empty.

Accuracy about intent. Constant monitoring generates data faster than it can be checked, and unchecked data gets interpreted in the direction the system expects. A short reply becomes a verdict. This is where hypervigilance feeds rumination and, in a relationship, where it becomes indistinguishable from trust issues even when nobody has done anything.

Other people's comfort. Being read continuously is not a neutral experience for the person being read, and "you're analysing me" is a complaint with something behind it.

What actually helps is unspectacular and mostly about giving the system its job back in a smaller form. Check one reading out loud instead of ten silently — asking "are you annoyed with me?" converts a monitoring task into an answered question, and the scanner has less to work with. Notice the rooms where it switches off, and spend time in them deliberately, because those are evidence that the setting is contextual rather than who you are. And treat the physical layer as physical: this state is maintained in the body, and the things that lower it are boring ones — sleep, food, and the co-regulating presence of a person you are not tracking.

Now the line. Being watchful, even very watchful, is a normal setting with real advantages, and nothing here says otherwise. But hypervigilance is also one of the core symptom clusters of post-traumatic stress, and there is a version of this that a self-knowledge page has no business addressing: constant bracing that does not let up anywhere, startle responses that are out of proportion, intrusive re-experiencing of something that happened. That is treatable, specifically and well, and it belongs with a professional. The Stress Response test reads what you do once a threat is detected, and the Trust Patterns test reads the relationship end of it — but neither is a screen for trauma, and neither is offered as one.