Why Avoidance Can Keep Anxiety Going: What Behavioural Science Tells Us About Fear

Avoidance can bring immediate relief while leaving a feared outcome untested. Behavioural science helps explain this cycle, why it sometimes persists, and why effective treatment involves more than simply telling someone to face their fears.

A meeting invitation arrives. Someone imagines stumbling over their words, being judged and feeling unable to cope. They decline, and the tension eases. Nothing embarrassing happens. Yet the next invitation may feel just as threatening.

The relief was real. What remains uncertain is why the feared event did not occur: was the meeting manageable, or was declining essential for safety? Avoidance can make that question difficult to answer. It is one possible mechanism maintaining anxiety, not a complete explanation of anxiety disorders and certainly not evidence of weakness.

Why avoidance feels so effective

Avoidance means acting to prevent an anticipated unpleasant encounter or outcome. Escape ends an encounter that has already begun. Declining the meeting is avoidance; leaving after arriving is escape. In everyday life the distinction can blur, and both can reduce distress.

Learning theory connects two processes. Through classical conditioning, a previously ordinary cue can acquire threatening meaning after being associated with an aversive experience. Through operant learning, the consequences of an action influence whether it happens again. An uncomfortable presentation might make meetings feel threatening; the relief of cancelling can encourage future cancellation.

This is an illustrative pathway, not a necessary history for every fear. People also learn through observation, information and interpretation. Krypotos and colleagues’ review of avoidance learning describes several competing and complementary accounts. Avoidance can persist even when reported fear is low, so its maintenance cannot always be explained by conscious anxiety alone.

Negative reinforcement: when relief strengthens behaviour

Negative reinforcement does not mean punishment. Reinforcement means that a behaviour becomes more likely. “Negative” refers to something being removed or reduced: an unpleasant feeling, event or anticipated consequence. Punishment, by contrast, makes a behaviour less likely.

The anxiety avoidance cycle can therefore look like this: a situation feels threatening, the person avoids it, anxiety quickly falls, and avoidance becomes more likely next time. Relief rewards the action in a behavioural sense, even if the person dislikes the restrictions it creates.

Timing helps explain its appeal. The relief is immediate; the possible costs, such as fewer opportunities or an unchanged threat belief, emerge later. A strategy that works well over the next five minutes can be less helpful over the next five months.

This account should remain a hypothesis about a particular pattern. Expectancies, habits, perceived control and the effort required to avoid also matter. Pittig and colleagues’ review places avoidance within a broader system of fear acquisition, generalisation, extinction and return. Short-term relief is powerful, but it is not the only process involved.

When safety behaviours keep threat beliefs alive

Someone may enter a situation while trying to prevent catastrophe through smaller protective actions. These are often called safety behaviours. Examples might include repeatedly seeking reassurance, sitting near an exit, rehearsing every sentence, checking bodily sensations, carrying unnecessary “just in case” items or avoiding eye contact to reduce perceived evaluation.

None is a diagnostic indicator. Purpose and context determine the function. Choosing an exit seat for accessibility, rehearsing an important talk or carrying prescribed medical equipment is not inherently problematic. The relevant question is whether the action is treated as indispensable protection against an exaggerated threat.

If a conversation goes well, someone may conclude, “That only happened because I rehearsed everything.” The encounter then provides less evidence that an unrehearsed exchange could be manageable. This interpretation can preserve the original prediction.

The evidence resists a blanket rule. A 2016 meta-analysis by Meulders and colleagues found inconclusive evidence about allowing or removing safety-seeking behaviour during exposure. A systematic review by Goetz and colleagues distinguished behaviours that prevent engagement with the feared experience from those that permit engagement and provide support afterwards; their effects may differ.

A 2026 systematic review and meta-analysis by Goodson and colleagues, examining standalone manipulations outside exposure, found that reducing safety behaviours improved symptoms relative to control conditions. It also identified publication bias and variation between studies. These findings support investigating function, not instructing everyone to abandon every source of reassurance or support.

Avoidance can leave important predictions untested

Avoiding the meeting prevents embarrassment there, but also prevents discovering whether embarrassment would have happened. It can block learning about coping: perhaps words would falter, yet the conversation would continue and colleagues would respond reasonably.

Potential new learning is therefore broader than “nothing bad happens”. It may involve recognising that discomfort is tolerable, that anxiety can fluctuate without escape, or that some uncertainty can be carried without continual checking. These are possibilities, not guarantees that feared outcomes are imaginary.

Behaviour and fear can also change at different rates. In Vervliet and Indekeu’s laboratory experiments, participants learnt to avoid shocks by pressing a button. After fear extinction, low-cost avoidance returned when the button became available again. This narrow experimental finding suggests that reduced fear does not automatically eliminate avoidance; it does not establish how any individual’s anxiety works.

Dymond’s review of avoidance extinction similarly distinguishes learning about threat from changing the avoidance response itself. Both may be relevant when treatment seeks lasting change.

How fear and avoidance can spread

Learning to recognise related threats is useful. Someone who narrowly avoids a speeding vehicle need not repeat the experience with every vehicle before exercising caution. Generalisation allows learning to extend beyond one precise cue.

Problems can arise when relatively safe similarities are treated as equivalent threats. A difficult presentation may be followed by avoidance of team meetings, then informal conversations, then unfamiliar social settings. This example illustrates how restrictions might expand; it is not an inevitable sequence.

A 2022 systematic review and meta-analysis by Fraunfelter, Gerdes and Alpers found heightened fear generalisation associated with several anxiety and stress-related disorders. Findings differed across conditions, and the evidence rarely settled causality. Fear generalisation and the spread of avoidance are related concepts, but they are not identical: anticipating threat does not always produce avoidance.

Learning can also depend on context. Feeling able to speak in a therapist’s office may not immediately transfer to a busy workplace. Treatment may need to support learning across relevant settings rather than assume that one controlled encounter changes every situation.

What exposure therapy is trying to achieve

Exposure-based treatment involves planned contact with feared situations, sensations or other cues, within an appropriate psychological treatment. Depending on the difficulty, it may help someone update threat predictions, discover coping capacity and reduce reliance on avoidance or safety behaviours that obstruct learning.

Evidence supports treatment, while leaving room for differences in response. Carpenter and colleagues’ meta-analysis of 41 randomised placebo-controlled trials found that cognitive behavioural therapy improved anxiety-related symptoms relative to placebo. This evaluated CBT packages, not exposure alone, and cannot establish a single mechanism responsible for improvement.

NICE guidance for social anxiety disorder recommends individual CBT developed for that condition. Its treatment descriptions include behavioural experiments addressing negative beliefs and safety-seeking behaviours in the Clark and Wells model, and graduated exposure in the Heimberg model. These are structured approaches delivered by competent practitioners, not generic instructions to endure distress.

Learning from a mismatch between prediction and experience

Expectancy violation means that experience differs from what was predicted. Someone expects, “If my voice shakes, everyone will ridicule me.” An encounter in which their voice shakes without ridicule could challenge the predicted link. Anxiety need not disappear for that mismatch to be informative.

Craske and colleagues’ updated exposure framework emphasises expectancy violation and making new learning retrievable across situations. An experience need not be uniformly pleasant: a conversation may be awkward, yet more manageable than expected. Neither every trial nor every feared outcome can be arranged to provide reassuring evidence.

From habituation to inhibitory learning

Historically, exposure often emphasised habituation: fear decreasing during repeated or sustained contact. That decrease can occur and may matter, but feeling calmer by the end of an encounter is not a complete measure of what was learnt.

The inhibitory learning approach described by Craske and colleagues proposes that extinction creates new learning that competes with an earlier fear association, rather than literally erasing it. A cue once associated with danger can acquire another meaning: in these circumstances, the expected outcome does not follow.

In accessible terms, an old alarm can remain available while a new understanding becomes stronger and easier to recall. This helps explain why fear can reappear after time or in another setting without implying that all previous progress has vanished.

It also shifts attention from “Did anxiety reach zero?” to “What changed in the person’s expectations and behaviour?” Inhibitory learning is an influential model, not a final verdict replacing all other accounts. Emotional processing, changes in beliefs, self-efficacy and other processes remain under investigation.

Why “just face your fear” is poor advice

The slogan leaves out assessment, consent, planning and support. A clinician needs to understand what is feared, which responses maintain the difficulty, what genuine risks exist, and how severity, physical health, trauma, circumstances and readiness affect treatment.

Our article on psychological formulation beyond diagnosis explains why apparently similar behaviours may have different meanings. Leaving a room could reflect a fear prediction, an overwhelming sensory environment or an unsafe interaction. The same instruction would not suit all three.

Exposure is usually planned collaboratively and individualised within evidence-based care. Appropriate professional support matters when difficulties are severe, complex or substantially restrict daily life. The NHS describes CBT assessment as a way to decide suitable treatment with a therapist.

Access matters too. As discussed in our article on why people delay seeking psychological help, stigma, knowledge and practical barriers can obstruct care. Difficulty obtaining treatment should not be mistaken for unwillingness to change.

Not all avoidance is unhealthy

Leaving an unsafe environment, avoiding genuine danger, setting interpersonal boundaries and reducing exposure to real threats can be sensible and protective. Anxiety is sometimes an appropriate signal that circumstances need to change.

Clinical concern is more likely when repeated avoidance of relatively safe situations preserves disproportionate threat predictions and increasingly restricts a person’s life. Even then, context and the person’s priorities matter. Declining an event is not automatically pathological, and engagement is not automatically beneficial.

Understanding a pattern should expand choice, not demand constant confrontation or turn courage into a treatment requirement. No account of learning justifies deliberately entering dangerous situations or removing necessary safeguards.

Anxiety is more complex than avoidance

Avoidance can be one mechanism maintaining anxiety in some circumstances. It is not a sufficient explanation of why anxiety develops. Craske and colleagues’ broader review of anxiety disorders describes interacting biological, psychological and environmental influences.

Temperament, genetic vulnerability, previous experiences, trauma, cognitive biases, uncertainty, relationships, social disadvantage, environmental stress, physical health and neurobiological processes may all contribute. Different people encounter different combinations. Avoidance can follow anxiety, influence its persistence, or both. An association between the two does not by itself demonstrate causation.

What the evidence still cannot tell us

Laboratory fear conditioning offers experimental control, but a newly learnt association between a cue and a shock is not the same as years of social anxiety, panic or trauma. Student and non-clinical samples may not represent clinical populations. Scheveneels and colleagues’ review argues that the validity of extinction as a model of exposure treatment needs systematic scrutiny.

Avoidance is difficult to measure. A questionnaire, missed appointment and button press capture different actions, while covert strategies may be hard to observe. The same outward behaviour can serve different purposes. Anxiety disorders also involve different patterns of worry, bodily threat, evaluation and learning; results from one condition should not automatically be applied to another.

Exposure outcomes vary, and improvement does not identify its cause. Learning, beliefs, attention and confidence may change together. Establishing a mechanism requires more than finding that one change correlates with another. Benito and colleagues’ research agenda highlights unresolved questions about exposure’s mechanisms and how to test them rigorously.

Safety-behaviour findings illustrate the uncertainty: studies differ in the behaviours manipulated, participants, treatment context and follow-up. Publication bias and limited evidence about lasting effects further constrain conclusions. The evidence supports careful, individualised treatment more confidently than a universal rule about every avoidant action.

Relief and learning are different outcomes

Avoidance makes sense because it can work immediately. The longer-term question is whether it protects someone from a real threat or repeatedly leaves a manageable situation unexplored. When the latter happens, relief can strengthen a response while the original fear remains untested.

Behavioural science helps explain that possibility without blaming the person. Effective care considers learning alongside genuine risk, individual history and present circumstances. The aim is greater flexibility and a less restricted life, rather than the elimination of every anxious feeling.

This article provides general educational information and is not a substitute for individual psychological or medical assessment, diagnosis or treatment.

References

Benito, K., Pittig, A., Abramowitz, J., et al. (2025). Mechanisms of change in exposure therapy for anxiety and related disorders: A research agenda. Clinical Psychological Science, 13(4), 687–719. First published online 2024. https://doi.org/10.1177/21677026241240727

Carpenter, J. K., Andrews, L. A., Witcraft, S. M., Powers, M. B., Smits, J. A. J., & Hofmann, S. G. (2018). Cognitive behavioral therapy for anxiety and related disorders: A meta-analysis of randomized placebo-controlled trials. Depression and Anxiety, 35(6), 502–514. https://doi.org/10.1002/da.22728

Craske, M. G., Treanor, M., Conway, C. C., Zbozinek, T., & Vervliet, B. (2014). Maximizing exposure therapy: An inhibitory learning approach. Behaviour Research and Therapy, 58, 10–23. https://doi.org/10.1016/j.brat.2014.04.006

Craske, M. G., Stein, M. B., Eley, T. C., Milad, M. R., Holmes, A., Rapee, R. M., & Wittchen, H.-U. (2017). Anxiety disorders. Nature Reviews Disease Primers, 3, 17024. https://doi.org/10.1038/nrdp.2017.24

Craske, M. G., Treanor, M., Zbozinek, T. D., & Vervliet, B. (2022). Optimizing exposure therapy with an inhibitory retrieval approach and the OptEx Nexus. Behaviour Research and Therapy, 152, 104069. https://doi.org/10.1016/j.brat.2022.104069

Dymond, S. (2019). Overcoming avoidance in anxiety disorders: The contributions of Pavlovian and operant avoidance extinction methods. Neuroscience & Biobehavioral Reviews, 98, 61–70. https://doi.org/10.1016/j.neubiorev.2019.01.007

Fraunfelter, L., Gerdes, A. B. M., & Alpers, G. W. (2022). Fear one, fear them all: A systematic review and meta-analysis of fear generalization in pathological anxiety. Neuroscience & Biobehavioral Reviews, 139, 104707. https://doi.org/10.1016/j.neubiorev.2022.104707

Goetz, A. R., Davine, T. P., Siwiec, S. G., & Lee, H.-J. (2016). The functional value of preventive and restorative safety behaviors: A systematic review of the literature. Clinical Psychology Review, 44, 112–124. https://doi.org/10.1016/j.cpr.2015.12.005

Goodson, J. T., Patel, T. A., Zech, J. M., Sala, M. C., & Cougle, J. R. (2026). The standalone effect of safety behavior manipulations: A systematic review and meta-analysis. Clinical Psychology Review, 124, 102703. https://doi.org/10.1016/j.cpr.2026.102703

Krypotos, A.-M., Effting, M., Kindt, M., & Beckers, T. (2015). Avoidance learning: A review of theoretical models and recent developments. Frontiers in Behavioral Neuroscience, 9, 189. https://doi.org/10.3389/fnbeh.2015.00189

Meulders, A., Van Daele, T., Volders, S., & Vlaeyen, J. W. S. (2016). The use of safety-seeking behavior in exposure-based treatments for fear and anxiety: Benefit or burden? A meta-analytic review. Clinical Psychology Review, 45, 144–156. https://doi.org/10.1016/j.cpr.2016.02.002

National Institute for Health and Care Excellence. (2013). Social anxiety disorder: Recognition, assessment and treatment (CG159). Recommendations 1.3.1–1.3.2 and 1.3.13–1.3.14. Accessed 6 October 2026. NICE recommendations

NHS. (n.d.). Cognitive behavioural therapy (CBT). Accessed 6 October 2026. NHS CBT information

Pittig, A., Treanor, M., LeBeau, R. T., & Craske, M. G. (2018). The role of associative fear and avoidance learning in anxiety disorders: Gaps and directions for future research. Neuroscience & Biobehavioral Reviews, 88, 117–140. https://doi.org/10.1016/j.neubiorev.2018.03.015

Scheveneels, S., Boddez, Y., Vervliet, B., & Hermans, D. (2016). The validity of laboratory-based treatment research: Bridging the gap between fear extinction and exposure treatment. Behaviour Research and Therapy, 86, 87–94. https://doi.org/10.1016/j.brat.2016.08.015

Vervliet, B., & Indekeu, E. (2015). Low-cost avoidance behaviors are resistant to fear extinction in humans. Frontiers in Behavioral Neuroscience, 9, 351. https://doi.org/10.3389/fnbeh.2015.00351. Figure correction published 2021: corrigendum; the authors reported unchanged conclusions.