Safety behaviours

A safety behaviour is something you do inside a feared situation to stop a disaster that was not going to happen. Sitting where you can leave. Carrying the tablets without taking them. Having the first two sentences ready. A bottle of water in the bag. Asking someone to come.

They work. That is not in dispute, and it is the whole difficulty. Anxiety goes down in the moment and stays where it is over the years, because one mechanism produces both: the situation was survived with the action in place, so the action keeps the credit, and what would have had to be learned — that it was survivable anyway — never registers.

The logic was set out by Salkovskis in 1991 and has held up since.

Every anxious prediction resolves the same way: the feared thing does not occur. That is a lot of evidence, and in principle it should wear the fear down. It does not, because the evidence has somewhere else to go. If a safety behaviour was in place, the non-disaster gets attributed to it. I did not faint — because I sat down. They did not think I was odd — because I had the lines ready. The plane landed — because I held the armrest.

Which produces something quite specific: the more often the situation is entered with the behaviour running, the stronger the belief gets. Each repetition is one more occasion on which the precaution appeared to work. Two hundred flights with the armrest gripped is two hundred pieces of evidence for the armrest and none for the aircraft.

And the explanation survives contact with reality. If anxiety was low that day, the behaviour gets the credit; if it was high and nothing happened anyway, it gets the credit for that too. A belief whose evidence is a non-event cannot be disconfirmed while the non-event has an alternative author.

Which explains what people report with real frustration: they have done the frightening thing many times and are no less afraid. Not a failure of courage or of repetition — the predictable result of attending with the insurance on.

Avoidance announces itself. Someone stops flying, stops the presentations, stops going out, and everyone can see what has happened — including the person, who usually does not like it.

Safety behaviours hide inside participation. From outside, the person went. There is nothing to confront and nothing to admit, so the pattern can run for a decade behind a record of attendance. "But I do go to things" is true, and is not the measure.

They also read, from inside, as competence. Preparing, bringing what might be needed, thinking a step ahead — the same actions a well-organised person performs, and what separates them is only what happens when they are unavailable. The test is not whether the action is sensible. It is whether the day becomes impossible without it.

The common forms, by domain:

• Panic. Sitting near the exit, knowing where the toilets are, never going out without water, carrying medication unopened, choosing the stopping train. The attack itself is in panic attack; the route-narrowing version is agoraphobia.
• Social. Rehearsing the opening, holding a glass to occupy the hands, arriving with someone, steering towards questions, sitting at the edge. See social anxiety.
• Health. Body-checking, pulse-taking, reading only about that one illness or refusing to read at all, taking a companion to appointments. See health anxiety.
• Asking. Checking whether it is fine, whether they are annoyed, whether this looks like anything. This one is large enough to have its own entry — reassurance seeking — which is where the shortening half-life and the cost to the person being asked are set out. Asking is the most social of the safety behaviours, and the only one that wears someone else out.

And the quiet one under none of them: rehearsing the event for days beforehand, which feels like preparation and is dealt with in anticipatory anxiety.

This has been tested directly, which is unusual for a claim of this kind.

Social anxiety. Wells and colleagues, in 1995, had people enter a feared social situation twice: once as usual, and once told to drop their safety behaviours and look for evidence instead. One session. Belief in the feared outcome fell substantially more without the behaviours, and anxiety fell with it — the opposite of what people expect.

Panic. Salkovskis and colleagues ran the equivalent experiment in 1999 with panic disorder and agoraphobia, comparing exposure with safety-seeking behaviours dropped against exposure with them continuing. The group that dropped them showed greater reduction in both catastrophic belief and anxiety. Same finding, different fear.

Across conditions. Helbig-Lang and Petermann, reviewing the anxiety disorders, concluded that safety behaviour is best understood as maintaining the problem, while noting that the picture is not uniform and that the practical question — eliminate or tolerate — is not settled by the maintenance finding alone.

Which is the honest complication. Milosevic and Radomsky found that keeping a safety behaviour available during exposure did not interfere with fear reduction, and Blakey and Abramowitz, reviewing the question, concluded that what matters is not presence or absence but what the behaviour does to the belief being tested — whether the person can still find out something they did not already think. A prop that gets someone into the room on day one may be buying the only exposure available; the same prop on day forty is paying for the explanation. That judgement is a clinician's, not a rule.

• Find out what yours are. What do you bring, where do you sit, what do you check, and what would you be unwilling to do the thing without. The last question is the one that works.
• Drop one at a time, inside the situation. Not endurance: going without the insurance and noticing what happened is the only way the credit transfers from the prop to the situation. Half-measures count as the behaviour — bringing it and not using it is bringing it, because the reassurance is in the having.
• Say the prediction out loud first. Without the water I will choke. Then find out. An untested prediction cannot be lost; a stated one can.
• Expect anxiety to rise before it falls. The first time without is worse. That is the price of the information, paid once per behaviour rather than every time.
• The goal is changing the explanation, not getting through it. Afterwards the sentence that matters is not I managed but it was alright and I did not do the thing I thought was holding it together.

The panic attack self-check, the social anxiety test and the health anxiety test each read the pattern these behaviours attach to.

The distinction this page cannot leave out. A safety behaviour guards against a danger that is not there. An inhaler, an adrenaline pen, insulin, a seatbelt, a helmet, telling someone where you are going in a place where that matters — those guard against dangers that are there, and nothing above applies to them. Prescribed medication needs care rather than a rule: whether it is being taken as treatment or carried as a charm is a real question, and one for whoever prescribed it.

The scope note. Most people use some of these and most of the time it costs nothing — there is no threshold count, and no virtue in doing hard things bare. The useful question is whether the list is growing, and whether the situations still feel as dangerous as they did five years ago. If yes to both, the behaviours are not holding the line but holding the fear in place.