Your brain runs a threat check all the time, mostly below your awareness. It is fast and rough on purpose, because missing a real danger costs far more than reacting to a false alarm.

Under stress, a network that handles alertness and scanning gets louder and more tightly linked. In a brain-scanning study, the shift grew with how stressed each person felt.

That network is driven by the fast nerve chemicals of the stress response, because blocking them softened the shift, while blocking the slower stress hormone did not.

You rarely notice the switch itself. What you notice is the result, like jumping when a door slams, checking your phone again and again, or scanning a room before you sit down.

Being unsure is the key. When a threat is clear, fear rises and falls quickly, but when it could come at any time, people stay on edge for the whole stretch.

Staying braced has a cost, because scanning uses up attention you would normally use for other things, so focus and short-term memory slip while the alarm is running.

Then the loop closes. Feeling jumpy and worn out becomes its own worry, and that worry keeps the alarm turned up, which keeps you watching for the next problem.

The good news is that this system can be retrained. In a trial of adults with anxiety, eight weeks of practice brought their startle response back into a normal range.

Your threat system is built to check the world around you very fast, because it spots movement, noise and faces that look upset, and it gets your body ready to act before you have had time to think it through.

That speed is useful for a few minutes at a time, when there is something real to deal with. Stress asks the same system to stay switched on for days, through deadlines, money worries or a tense home, where there is nothing to run from.

Lab studies show what the alert state does. When people know a mild shock might come, they get better at spotting threats and faces, and worse at holding numbers in mind. The brain trades focus for watchfulness.

So hypervigilance is not a broken part. It is a normal alarm that has been set too sensitive for the life you are living right now, and that setting can move back down with time and practice.

Two paths carry the alarm. The fast one uses nerve chemicals and works within seconds, sharpening your senses, tightening muscles and priming the startle reflex. The slower path uses hormones and takes longer to build up.

A brain-scanning study tested which path drives the alert state. Under a scary stressor, a network for scanning and alertness grew louder. A drug that blocks the fast chemicals softened it, and blocking the slow hormone did not.

The stressor does not have to be physical. In one lab study, people waited to give a speech to a virtual audience. There was no shock and no danger, only the feeling of being watched, and their bodies still reacted more strongly than when they were waiting to count backward.

Not knowing when something will happen matters most. Researchers measure this with the startle blink. A clear warning brings a quick spike of fear, but a threat that could come at any moment keeps people braced for the whole time.

Many people rule themselves out because they are not jumping at every noise. But the common form is mild and comes and goes. In a US survey of 30,954 adults, most people who felt on edge felt it on several days, not every day.

The amount of stress also matters. After the 2013 Boston Marathon bombings, a national survey of 4,675 adults found that stress symptoms rose steadily with each extra hour of news coverage people watched each day.

People who watched six or more hours a day reported more stress symptoms than some people who were at the scene. One of the questions asked about feeling on edge. That is one event and one survey, so hold the size loosely.

The point is not the headline number but the pattern behind it, which is that low, steady input keeps the alarm warm. A little tension that runs all day can add up to more than one big scare that passes.

Most of the input is ordinary: a phone that buzzes all day, a manager who might message at any time, and a news feed you check on the bus. Each one is small, but each one tells the alarm to keep scanning.

Sleep sits under all of it. One night of lost sleep raised next-day anxiety in a lab study, and across a large online sample, even small cuts in sleep from one night to the next went with more anxiety the next day.

Caffeine can mimic the feeling, and in lab tests, doses close to five cups of coffee raised anxiety even in healthy adults. Some cold medicines with decongestants can cause nervousness and restlessness too.

Hormones can play a part as well, since an overactive thyroid can cause nervousness, shaky hands and trouble sleeping, which is why a simple blood test is worth asking for if the feeling came on without a clear reason.

The more you watch for danger, the more you find. Across 172 studies, anxious people’s attention moved toward threat faster than other people’s. The effect was modest, and it did not show up in people who were not anxious.

It can turn into a loop. A three-year US study of 4,165 adults found that heavy news use after one disaster predicted more worry later, which predicted more news use and more stress when the next one came.

The strongest good news comes from a trial of 276 adults with anxiety. An eight-week mindfulness course worked about as well as a common anxiety medicine. In a linked study, the startle response of both groups fell back into a normal range.

Movement helps too. When researchers pooled 15 trials with 675 patients, aerobic exercise eased raised anxiety compared with waiting for treatment, and harder sessions did more than gentle ones.

One idea deserves retiring: that a fear centre in your brain hijacks you. Leading researchers argue that the threat alarm and the feeling of fear run on separate systems. The alarm can fire with no fear at all.

The evidence also disagrees with itself. In one lab study, people with PTSD were more jumpy while waiting for an unpredictable threat, but people with long-running worry were not, even though both feel on edge.

Training attention away from threat looked promising. But pooling 49 trials, the effects were small and faded once the weakest studies and missing results were taken into account.

The direction of cause is unclear too. Stress may drive the scanning, but feeling on edge also pushes people toward more news and less sleep, so it is best treated as a loop with more than one way in, and more than one way out.