Avoiding something frightening works. The relief is immediate and real, and that is precisely why avoidance tends to make fear more durable rather than less.
Fear is a prediction, not a fact about the world
Anxiety arises when a situation is appraised as threatening and the person expects a bad outcome they could not handle.
The expectation may be about the situation itself, or about the anxiety, such as a belief that the sensation will escalate uncontrollably.
Either way it is a prediction, and predictions are revised by evidence rather than by argument.
Relief reinforces the escape
Leaving or avoiding a feared situation produces a sharp drop in distress, and that drop follows immediately.
Behaviour followed closely by relief becomes more likely, which is one of the better-established findings in learning research.
Avoidance therefore strengthens itself with every repetition, independently of whether the original fear was warranted.
The prediction is never tested
Because the person leaves, the feared outcome does not occur, and the absence is attributed to having left.
The belief that the situation was dangerous survives intact and is reinforced, since escape appears to have worked.
No information capable of correcting the expectation is ever collected, which is what makes the pattern self-sealing.
Subtle avoidance does the same job
Avoidance is not only staying away. Safety behaviours count: carrying an object for reassurance, rehearsing exits, avoiding eye contact, keeping a companion present.
These allow the situation to be entered while preserving the belief, because any good outcome is credited to the safety behaviour rather than to the situation being manageable.
The prediction remains untested for the same reason as before, which is why such behaviours receive specific attention in treatment.
Structured exposure is how the expectation gets updated
Treatments that work on this pattern involve approaching the feared situation deliberately and remaining until the prediction is disconfirmed.
Current models emphasise violating the expectation rather than simply waiting for anxiety to subside, and variation across situations makes the new learning generalise more effectively.
The original fear association is not erased by this process. A competing memory is built alongside it, which is why fear can return in a new setting and why practice across contexts matters.
This is demanding work, and it is done with a trained clinician who can grade the steps appropriately. Attempting it unsupported can reinforce the fear if the exposure is overwhelming, so it is not a self-directed exercise.