Desensitization and Counterconditioning in Dogs

1. Habituation and Sensitization

1.1 The Two Outcomes of Repetition

Habituation is the simplest form of learning there is: response to a repeated stimulus decreases. It is non-associative — nothing is being paired with anything — and it is not sensory adaptation or muscle fatigue but a genuine change in the nervous system's response.

Sensitization is its mirror image: response to a repeated stimulus increases. Same procedure, opposite result.

Both are ubiquitous across animals, and the practical point is that repeated exposure by itself does not specify which you get. "He'll get used to it" is a prediction with no guarantee behind it, and it is wrong in the cases that matter most (which is how much reactive behavior gets built).

1.2 What Determines Which

Rankin and colleagues (2009) produced the field's consensus description, revising the nine characteristics originally set out by Thompson and Spencer in 1966 and adding a tenth concerning long-term habituation. Several of those characteristics translate directly into practice.

Stimulus intensity is the pivot. Weaker stimuli produce more and faster habituation. Strong or aversive stimuli tend to produce sensitization instead. This single relationship is the reason sub-threshold work is not a refinement of exposure but its precondition.

Frequency matters, and cuts both ways. More frequent stimulus presentation produces faster habituation — but that habituation is also more transient. Spaced presentations habituate more slowly and last longer.

Habituation is stimulus-specific. Habituating to one stimulus does not transfer automatically to a different one, which is why a dog habituated to recorded thunder may not be habituated to actual thunder.

It recovers spontaneously. Withhold the stimulus and the response returns — a point §7 develops, because it applies to the clinical methods built on top of habituation.

Repeated series of habituation training produce progressively faster habituation. This is the empirical basis for doing repeated short sessions rather than one long one.

Dishabituation exists. A strong extraneous stimulus can restore a habituated response. Practically: one unrelated fright can undo accumulated progress.

1.3 Why This Section Comes First

Every clinical method below is built on this substrate. If exposure is too intense, the underlying process is sensitization, and no amount of good food, careful timing, or patient repetition converts it into habituation. The dog gets worse, and the protocol looks like it failed when in fact it was never running.

1.4 Why the Same Exposure Produces Opposite Outcomes

Habituation and sensitization are both products of repetition, and which one occurs depends on intensity relative to the individual animal rather than on the stimulus itself. That is the single most important fact in this article.

It also explains why a protocol that helped one dog made the next one worse: the same arrangement sat below threshold for one and above it for the other.

2. What Desensitization and Counterconditioning Actually Are

2.1 Desensitization

Systematic, gradual exposure to a fear-eliciting stimulus at an intensity below the level that elicits a fear response, increasing intensity only as the dog remains unreactive. The intensity dimension can be distance, volume, duration, movement, or proximity depending on the trigger.

The term comes from Wolpe's systematic desensitization in human clinical psychology (1958), where relaxation was an explicit component — a point that has largely dropped out of the animal literature and which §5 suggests may have been a loss.

2.2 Counterconditioning

Pairing the feared stimulus with something the dog values, so that what the stimulus predicts changes. Formally, the conditioned stimulus that previously predicted something aversive comes to predict something good.

Desensitization changes intensity; counterconditioning changes meaning (the emotional layer beneath trained behavior). In practice they are almost always combined, which is why the literature and this article generally treat them as one method.

2.3 What Distinguishes Them From Flooding

Flooding exposes the animal to the full-intensity stimulus without escape until the response subsides. It is the same procedure that §1.1 identifies as producing sensitization, and where escape is genuinely unavailable, the relevant framework becomes the loss of behavioral control (Maier & Seligman, 2016) (the revised helplessness account).

A dog that has gone quiet under flooding may look like a success. Reduced behavioral output is not evidence of a changed emotional state (which is the central interpretive problem).

2.4 What Distinguishes Them From Extinction

Extinction presents the trigger without the aversive outcome and without adding anything positive. Counterconditioning adds an appetitive outcome — and it is not distraction. Distraction competes for the dog's attention so that the trigger is processed less; counterconditioning requires that the dog notices the trigger and that the trigger comes first, because the trigger is what has to acquire a new meaning. A dog fed to keep it looking away has been distracted, and the association being built is between the handler's hand and the food rather than between the trigger and anything. The practical significance is that counterconditioning gives the dog something to do and something to expect, while extinction only removes a prediction — and extinction alone carries a meaningfully higher rate of behavioral bursting (Lerman & Iwata, 1995).

2.5 Why the Two Are Almost Always Combined

Desensitization reduces the response and counterconditioning changes what the stimulus predicts. Run together, the second gives the dog something to do while the first is working, which is why protocols that use only desensitization tend to produce tolerance rather than comfort.

The combined version is what the canine trials generally test, so the evidence in this article is about the combination rather than about either alone. Anyone asking which component does the work is asking a question the canine literature has not addressed.

2.6 Flooding Is Not a Harsher Version of the Same Thing

It is a different procedure with a different mechanism and a different risk profile. Presenting it as the impatient version of desensitization understates what separates them: one works below the threshold at which the animal stops coping, the other deliberately exceeds it.

A protocol that has drifted above threshold has become flooding regardless of what it is called (keeping arousal below threshold). That drift is usually gradual and is usually invisible to the person running it.

3. The Threshold

3.1 The Variable Everything Depends On

Threshold is the intensity at which the dog begins to show a fear response. Below it, exposure can produce habituation and the dog is available for new learning. Above it, exposure produces sensitization, and the dog is largely unavailable for learning of any kind.

Everything else in a protocol — the food, the timing, the schedule, the session length — is downstream of getting this right.

3.2 It Is Not a Hard Biological Line

Threshold is a practitioner's term for something the nervous system does not do sharply. There is no switch: what the word points at is a range in which the dog can still take in what is happening and learn from it, shading gradually into one in which it cannot.

The longer formulation — working at an intensity at which the dog can still process information and learn — is what the word is shorthand for, and the shorthand is worth keeping provided nobody starts looking for the line.

3.3 Reading It

The most reliable practical marker is food refusal in a dog that normally takes food. It is unambiguous, easy to observe, and it appears before most of the subtler signs are reliably readable.

Beyond that: orientation that becomes fixed rather than glancing, a stiffening of body outline, closed or tense mouth, respiration rate rising without exertion, and the dog becoming unresponsive to its name. Concordance across several of these matters more than any one.

3.4 It Moves

Threshold is not a fixed property. It shifts with sleep (which also consolidates what was learned), with the dog's stress load over previous days (the chronic stress picture), with pain, with hormonal state, and with what has already happened that morning. A distance that worked last week may not work today, and this is not the dog being inconsistent.

Pain deserves specific mention: it lowers thresholds and is regularly missed (Mills et al., 2020). A protocol that stalls for no apparent reason is a reason to look at the body (including the less visible forms of pain).

3.5 Individual Variation

Where any given dog's threshold sits, and how sharply it rises, varies substantially between individuals — which is why protocol parameters have to be set per dog rather than by rule (as temperament and coping-style research would predict).

3.6 Why the Threshold Is Not a Number

It is not a distance or a volume. It is the point at which a particular dog, on a particular day, stops being able to process what is happening, and it moves with sleep, pain, what has already happened and how much has been asked (how load from earlier events shifts the threshold).

Protocols written as fixed distances fail for exactly this reason, and the households following them conclude the dog is inconsistent. The dog is consistent; the threshold is not.

3.7 The Observable Is Recovery

Whether a session was below threshold is answered afterwards: a dog that returns to baseline quickly was worked below it, one that stays lit was not. That is more reliable than any in-the-moment judgment of body language.

It also means a session can only be evaluated once it is over, which is an uncomfortable but accurate description of the method. Handlers who want a real-time verdict are asking for something the method does not provide.

4. Protocol Elements

Identify the actual trigger. Not "fear of visitors" but which component — the doorbell, the knock, the entering, the direct approach, the reaching hand. Protocols fail routinely because they treated a compound situation as one stimulus.

Choose the right intensity dimension. Distance is the default and is not always right. For noise it is volume; for handling, duration and pressure; for a moving trigger, speed.

Establish the starting point empirically. Find the intensity at which the dog notices the stimulus and still eats readily, then work slightly below it.

Use genuinely high-value reinforcement. The reinforcer must outcompete the fear. Kibble in the presence of a real trigger is not counterconditioning; it is a test the dog will fail. Food is the usual choice because it is quick to deliver, easy to repeat and measurable — a dog that stops eating has told you something. It is not the only option: play, movement away from the trigger, access to sniffing or a familiar person can all serve, and for a dog that does not eat under mild pressure they may be the only workable ones.

Get the order right. Trigger first, then food. Trigger predicts food. If food consistently precedes the trigger, the food starts predicting the trigger — the association runs backwards, and the reinforcer itself can become a warning signal.

Keep sessions short and repeat them. Consistent with the habituation characteristics in §1.2: repeated short series produce progressively faster habituation, while long sessions increase the risk of drifting over threshold as the dog fatigues.

Progress on the dog's evidence, not the calendar. Increase intensity when the dog is reliably relaxed and eating at the current level, and be prepared to go back (the same stepwise logic applied to muzzle training).

Manage between sessions. Uncontrolled full-intensity exposure between sessions undoes the work, and given dishabituation (§1.2), a single unmanaged incident can cost weeks. Management is not a failure to train; it is what protects the training.

5. Building a Protocol That Survives Contact With a Household

5.1 Start Below Where You Think You Should

The commonest design error is beginning at an intensity the dog can just about tolerate rather than one it barely notices. Starting too low costs a session; starting too high costs the progress already made and sometimes more than that.

Where a household reports that the plan stopped working, the usual finding is that it never worked at the level it started, and that the early sessions were tolerated rather than processed. Tolerance looks like progress for a while, and then it stops looking like anything at all.

5.2 One Variable at a Time

Distance, duration, intensity, movement and the number of triggers present are separate dimensions, and a session that changes two of them has not tested either. Households change several without noticing, because a real-world trigger moves, makes noise and behaves unpredictably while the plan assumes it will not.

Arranging situations where only one dimension varies is most of the work, and it is why controlled setups beat opportunistic ones. A staged encounter with a known helper dog is worth more than three chance meetings, and it is the part households find hardest to organize (how owners can apply this to leash reactivity).

5.3 Plan the Exit Before the Session

Deciding in advance what would end the session — the dog stops taking food, will not disengage, or does not settle within a set time — prevents the judgment being made under pressure by someone who wants the session to succeed.

An exit criterion written down beforehand is a different thing from one improvised at the moment it is needed, particularly when the session has almost worked and stopping feels like giving up.

5.4 Manage Between Sessions

Progress made at a controlled intensity is undone by uncontrolled exposure between sessions. A dog working through a noise protocol that also meets fireworks unmanaged is being run through two contradictory procedures, and the uncontrolled one is the more intense of the two.

Management is not a supplement to the protocol. On the available evidence it is a condition for it, and where management is impossible the honest answer is that the protocol will not work rather than that the household should try harder.

5.5 Record Something Countable

Recovery time, whether food was taken, and how many repetitions were possible before the dog disengaged are all countable without equipment and all move before the presenting behavior does. Three numbers per session is enough and more than most plans collect.

Without them, the only available measure is the household's impression, which improves over time regardless of what the dog does — a caregiver placebo effect documented in dogs and strengthening the longer the observation continues.

5.6 Session Length and Frequency

Short and frequent beats long and occasional, for reasons the habituation literature makes explicit: the effect depends on repetition at a tolerable intensity rather than on time spent exposed. A ten-minute session done four times a week does more than a forty-minute session done once.

Long sessions also drift upward in intensity as the dog tires and the handler pushes for a result, which is the mechanism by which a careful plan turns into flooding. Ending while the dog is still doing well is counterintuitive and is the correct call.

5.7 Who Runs the Sessions

A protocol run by one household member and undermined by another is a common pattern and rarely discussed. The dog is not learning from the sessions alone; it is learning from everything that happens with the trigger present, including the walks nobody counted as training.

Agreement within the household about what happens between sessions matters more than technical precision within them, and it is the part a professional cannot supply.

6. What the Canine Evidence Shows

6.1 The Direct Trials Are Small

Butler, Sargisson and Elliffe (2011) tested a systematic desensitization protocol for separation-related behavior in eight dogs, with owner-implemented graduated departures (the neurobiology of that presentation). It is one of the few controlled canine trials of the method and its sample size is what it is.

Shnookal and colleagues (2024) conducted a systematic review of counterconditioning-based interventions for companion dog behavior modification and identified fourteen eligible studies. Their conclusion is the one practitioners should carry: the approach has support, and the evidence base is small, methodologically heterogeneous, inconsistent in how the intervention is defined, and dependent on implementation quality.

Riemer (2023) reviewed the evidence for treating and preventing noise fears specifically and found relaxation training and counterconditioning to real-life stimuli among the better-supported interventions available.

6.2 The Honest Summary

Desensitization and counterconditioning is the most recommended behavioral treatment for canine fear, and it rests on a smaller and softer evidence base than that status implies. It is well grounded in learning theory, it is welfare-compatible, it has genuine canine support, and the number of controlled trials is in the low double figures.

That is not an argument against using it. It is an argument against the confidence with which success rates are sometimes quoted, and for treating implementation quality as a real variable rather than an afterthought.

6.3 What Small Trials Can and Cannot Establish

A study with a handful of dogs can demonstrate that a protocol produces improvement under controlled conditions. It cannot establish how often it works, for which dogs, or how the effect compares with doing nothing.

Those are the questions a household is asking, and the canine literature does not answer them. The outcome studies on noise fears come closest, which is why they carry more of the practical weight here than the protocol trials do.

7. The Finding That Complicates the Protocol

7.1 What Owners Reported

Riemer (2020) surveyed 1,225 owners of dogs with firework fears about which treatments they had used and how well each had worked. The reported effectiveness figures are worth stating precisely:

Counterconditioning delivered ad hoc — providing something valued after loud noises as they occur in everyday life — was rated effective by 70.8% of the 694 owners who used it. Relaxation training came in at 69.3% (n = 433). Noise recordings, that is the formal desensitization protocol that is the standard clinical recommendation, was rated effective by 55% (n = 377). Anxiety wraps came in at 44%.

The standard recommended protocol was rated lower than the informal, opportunistic version.

7.2 How To Read That

Carefully. This is owner-reported perceived effectiveness from a self-selected online sample, not a controlled comparison. Owners who chose different treatments may have had systematically different dogs, and severity is a plausible confound in either direction — the most severely affected dogs are the most likely to end up on a formal protocol.

What it does establish is that ad hoc counterconditioning, which barely features in the clinical literature, is both widely used and widely perceived as helpful. Riemer's own conclusion was not that formal protocols should be abandoned but that ad hoc counterconditioning and relaxation training should complement them.

7.3 Why It Might Be True

Several mechanisms would predict this. Recorded noise is a different stimulus from real noise, and habituation is stimulus-specific (§1.2), so generalization from recordings to the real event cannot be assumed. Ad hoc work happens in the actual context, avoiding that gap entirely. And formal protocols are demanding to implement correctly, so their real-world performance may reflect implementation failure rather than method failure.

The practical implication is modest and useful: pair systematic work with opportunistic counterconditioning to real-life noises, and treat the everyday startle as a training opportunity rather than a setback (practical steps for owners in our fireworks guide).

7.4 The Relaxation Component

Relaxation training performing at nearly 70% is a notable result given that it was integral to Wolpe's original formulation and has since become peripheral in the animal behavior literature. The canine data suggests that omission may have been a mistake.

7.5 Why the Owner-Report Finding Deserves the Space

A result suggesting that the standard protocol performs less well than expected, reported by owners rather than researchers, is the kind of finding a field tends to set aside. Including it and examining it carefully is what separates a review from an advocacy piece.

The article's position is that the finding is real, its interpretation is uncertain, and the protocol remains the best-supported approach available. Best-supported is a comparative claim about a thin field rather than a strong endorsement, and saying so is what distinguishes an honest recommendation from a confident one.

8. Failure and Relapse

8.1 Why Protocols Fail

Six reasons account for most of it. Working over threshold, which converts the whole exercise into sensitization. Progressing too fast, which is the same error arriving later. Reinforcement of insufficient value to compete with the fear. Reversed contingency, food before trigger. Unmanaged exposure between sessions, compounded by dishabituation. And an unaddressed physical driver, most often pain.

A seventh belongs on the list because it is invisible: the trigger was misidentified, and the protocol has been faithfully desensitizing the dog to something it was never afraid of (generalized anxiety complicates this further).

8.2 Relapse Is Built In

This is the part most often left out of client conversations. Counterconditioning and desensitization work substantially through new inhibitory learning rather than erasure of the original fear association. The original memory persists, which means the standard return-of-fear phenomena apply: spontaneous recovery with elapsed time, renewal when the dog encounters the trigger in a context different from where training happened, and reinstatement after a fresh frightening experience (Bouton, 2002) (and the reconsolidation route around this remains unvalidated in dogs).

Relapse is therefore not evidence the protocol failed. It is the expected behavior of the mechanism. For the same reason the accurate claim is not that the fear goes away but that the emotional response can change while the older learning history remains partly intact.

8.3 What Reduces It

The human inhibitory-learning literature (Craske et al., 2014) suggests emphases that transfer plausibly to dogs: train across multiple contexts rather than one, vary the stimulus rather than holding it constant, and treat expectancy violation — the feared outcome not occurring — as the operative ingredient rather than fear reduction within a session (which is prediction error doing the work).

That last point matters practically. A session in which the dog stayed sub-threshold and nothing bad happened has done its job even if the dog did not visibly relax further.

8.4 What Not To Do When It Returns

Do not escalate intensity to "push through," and do not correct the returning fear response. Punishment applied to fear-driven behavior worsens emotional state and can remove the warning signals that made the dog's state legible (Schilder & van der Borg, 2004; Vieira de Castro et al., 2020; Casey et al., 2021) (the aversive fallout picture).

Go back to an intensity that works, and rebuild.

8.5 The Commonest Failure Is Pace

Of the ways a protocol fails, moving faster than the dog accounts for most. It is also the one the handler controls completely, which makes it the first thing to check when progress stalls — before the reinforcer, before the setup, and before anything about the dog.

Going back one step is unglamorous and almost always faster than pushing through.

9. What to Tell a Household

9.1 Be Honest About the Numbers

The direct evidence is a small number of small trials, and the outcome literature for noise fears reports partial improvement more often than resolution. Full resolution does occur for some dogs, particularly where the fear is mild, recent or caught early — the point is that partial improvement is the more common result rather than the only possible one. A household told to expect a cure will conclude the plan failed when it produced exactly what the evidence predicts.

Stating the realistic range at the start costs nothing and prevents the commonest reason plans are abandoned, which is not that they failed but that they succeeded less than promised.

9.2 Explain What Relapse Is

A return after a gap, in a new place, or after a single bad encounter is expected rather than a sign that the work was wasted. Households that know this in advance respond by resuming; households that do not conclude the dog is untrainable. The difference between those two outcomes is one sentence spoken at the first consultation.

9.3 Name the Medical Question

Pain lowers the threshold and makes every session harder, and it is common enough to warrant exclusion before a protocol begins rather than after it stalls (how often pain turns out to be involved). A dog that cannot be worked below threshold may be a dog that hurts, and no amount of protocol refinement will reach it.

Saying so at the first consultation is a single sentence and changes what the household does next.

9.4 Set the Effort Expectation

This work is slow, it requires arranging situations rather than waiting for them, and it demands management in between. A household that understands the cost at the start can decide whether to take it on, and one that decides against it can be offered management instead, which is a legitimate answer rather than a failure.

One that discovers it in week four usually stops, which is worse for the dog than never having begun.

10. Which Findings Come From Which Species

10.1 The Method Is Borrowed and the Application Is Canine

Systematic desensitization was developed for human patients (Wolpe, 1958), the habituation criteria come from comparative neuroscience (Thompson & Spencer, 1966; Rankin et al., 2009), and the inhibitory-learning account of exposure is human clinical work (Craske et al., 2014; Bouton, 2002).

Everything applying it to dogs is canine, which places this topic in the better-served half of the collection.

10.2 What Has Been Tested in Dogs

Systematic desensitization for a specific problem (Butler, Sargisson & Elliffe, 2011), counterconditioning-based interventions in companion dogs (Shnookal et al., 2024), and the firework and noise-fear literature across three reviews and a longitudinal study (Riemer, 2019, 2020, 2023).

Plus the supporting canine work on training method and welfare (Vieira de Castro et al., 2020; Casey et al., 2021; Schilder & van der Borg, 2004), pain as a driver of behavior (Mills et al., 2020) and emotional-state assessment (Csoltova & Mehinagic, 2020; Flint et al., 2024).

10.3 The Extinction Burst Data Are Human

The prevalence finding that corrects the popular assumption about bursts comes from applied behavior analysis with human participants (Lerman & Iwata, 1995). Whether canine bursts occur at a comparable rate is unknown.

10.4 Where the Canine Evidence Is Thinnest

Direct trials of the protocol are few and small, which the article says at length. What is comparatively well covered is the outcome literature for noise fears, across three reviews and a longitudinal study, and that is where the most useful canine numbers come from.

The mechanism — whether desensitization works by habituation, by inhibitory learning, or by both — has not been tested in dogs at all. The two accounts predict different things about relapse, so this is not a purely academic gap.

10.5 Why the Recommendation Survives Anyway

The practical case rests on three things that are canine: the protocol has been tested and works for some dogs, the alternatives are associated with worse welfare outcomes, and pain is a common confound worth excluding first.

None of those depends on which mechanism turns out to be correct, which is the test worth applying to any article that spends chapters on a borrowed mechanism.

11. Summary at a Glance

Habituation and sensitization — The two possible outcomes of repeated exposure. Intensity determines which: weak stimuli habituate, strong ones sensitize. Habituation is stimulus-specific, recovers spontaneously, and can be undone by an unrelated fright (Rankin et al., 2009).

Desensitization — Graduated exposure below threshold. Changes the intensity the dog can tolerate.

Counterconditioning — Pairing the trigger with something valued. Changes what the trigger predicts. In practice the two are combined.

The threshold — The variable everything depends on. Best practical marker: food refusal in a dog that normally eats. It moves with sleep, stress load, pain, and prior events.

The evidence — Real but thin: a systematic review found fourteen eligible studies (Shnookal et al., 2024); the main controlled desensitization trial for separation-related behavior had eight dogs (Butler et al., 2011).

The complication — In 1,225 owners, ad hoc counterconditioning was rated effective by 70.8% and relaxation training by 69.3%, against 55% for the standard noise-recording protocol (Riemer, 2020).

Relapse — Built into the mechanism, because the method works through inhibitory learning rather than erasure. Expect spontaneous recovery, renewal, and reinstatement.

12. Research Gaps and Critical Appraisal

The controlled trial base is small. Fourteen studies in the systematic review, with heterogeneous definitions and small samples. For the most widely recommended behavioral treatment in the field, that is a thin foundation.

"Counterconditioning" is not consistently defined. Shnookal et al. identified inconsistency in how the intervention is described and implemented across studies, which limits what can be pooled and compared.

The Riemer (2020) result is owner-reported. Self-selected sample, perceived effectiveness, and probable severity confounding. It should change what practitioners try alongside the standard protocol, not what they conclude about relative efficacy.

Implementation quality is a large uncontrolled variable. The method depends on threshold reading, timing, and reinforcement selection, all of which vary by practitioner and owner. Trials of the method are partly trials of the people delivering it.

Generalization from recordings is assumed rather than demonstrated. Given stimulus specificity, whether habituation to recorded stimuli transfers to real events is an open empirical question with direct clinical consequences.

Positive outcome measurement is weak. No validated indicator of positive emotional state in dogs exists (Csoltova & Mehinagic, 2020), and the standard indicator set differentiates arousal only under negative valence (Flint et al., 2024). Establishing that a dog has become comfortable, rather than merely less overtly stressed, is harder than the protocols assume.

Prevention is understudied relative to treatment. Riemer (2019) framed firework fear as preventable as well as treatable, and the preventive application of counterconditioning during early development has far less evidence than the therapeutic one (despite the sensitive period being well characterized).

The direct canine trials are few and small. Systematic desensitization and counterconditioning have been tested in dogs (Butler et al., 2011; Shnookal et al., 2024), and neither study establishes how often the protocol works or for which dogs.

The mechanism has not been tested in this species. Whether the protocol works by habituation, by inhibitory learning or by both is unresolved for dogs, and the accounts come from human and comparative work.

The extinction burst data are human. The prevalence correction comes from applied behavior analysis with human participants (Lerman & Iwata, 1995), and no canine equivalent exists.

Owner-reported outcomes and trial outcomes do not fully agree. That disagreement is informative and unresolved, and the article treats it as an open question rather than selecting the more convenient result.

13. Conclusion

Desensitization and counterconditioning is the correct default treatment for canine fear, and the reasons are good ones: it is grounded in well-established learning processes, it is welfare-compatible, it addresses emotional state rather than suppressing its expression, and the canine evidence supports it. It also rests on a smaller evidence base than its clinical status suggests, depends heavily on execution, and cannot erase the original fear — which means relapse is a property of the method rather than a verdict on it. The single most useful thing to take from the underlying science is the one most often skipped: repeated exposure has two possible outcomes, and stimulus intensity decides which. Below threshold, the process is habituation and the dog is learning. Above it, the process is sensitization and the dog is getting worse while the protocol appears to be running normally. Everything else — the food, the timing, the schedule — is secondary to that. And the most useful thing to take from the canine data is that the informal version is also widely used and perceived as helpful. The treat delivered after an unexpected bang on an ordinary walk is not a distraction from the real protocol. On the largest dataset available, it may be doing more of the work than the protocol is.

Key Insights (Takeaways)

• Repeated exposure has two opposite possible outcomes, and intensity decides which. Weak stimuli produce habituation; strong ones produce sensitization (Rankin et al., 2009). "He'll get used to it" is a coin flip, and it lands wrong in exactly the cases that matter.

• Threshold is the variable everything else depends on. Below it the dog can learn; above it, no amount of good food or careful timing converts sensitization into habituation. The most reliable practical marker is food refusal in a dog that normally eats — and threshold moves with sleep, stress load, and pain.

• The evidence base is thinner than the method's status implies. A systematic review found fourteen eligible studies with inconsistent definitions and heavy dependence on implementation quality (Shnookal et al., 2024), and the main controlled desensitization trial for separation-related behavior had eight dogs (Butler et al., 2011).

• Ad hoc counterconditioning outperformed the standard protocol in the largest available dataset. Among 1,225 owners, opportunistic counterconditioning to real-life noises was rated effective by 70.8% and relaxation training by 69.3%, against 55% for formal desensitization with recordings (Riemer, 2020) — owner-reported, but a strong argument for doing both.

• Relapse is built into the mechanism, not a sign of failure. The method works through new inhibitory learning while the original fear memory persists, so spontaneous recovery, renewal in new contexts, and reinstatement after a fright are expected (Bouton, 2002). Training across varied contexts and treating expectancy violation as the active ingredient is what reduces it.

References

Bouton, M. E. (2002). Context, ambiguity, and unlearning: Sources of relapse after behavioral extinction. Biological Psychiatry, 52(10), 976–986. https://doi.org/10.1016/S0006-3223(02)01546-9

Butler, R., Sargisson, R. J., & Elliffe, D. (2011). The efficacy of systematic desensitization for treating the separation-related problem behaviour of domestic dogs. Applied Animal Behaviour Science, 129(2–4), 136–145. https://doi.org/10.1016/j.applanim.2010.11.001

Casey, R. A., Naj-Oleari, M., Campbell, S., Mendl, M., & Blackwell, E. J. (2021). Dogs are more pessimistic if their owners use two or more aversive training methods. Scientific Reports, 11(1), 19023. https://doi.org/10.1038/s41598-021-97743-0

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

Csoltova, E., & Mehinagic, E. (2020). Where do we stand in the domestic dog (Canis familiaris) positive-emotion assessment: A state-of-the-art review and future directions. Frontiers in Psychology, 11, 2131. https://doi.org/10.3389/fpsyg.2020.02131

Flint, H. E., Weller, J. E., Parry-Howells, N., Ellerby, Z. W., McKay, S. L., & King, T. (2024). Evaluation of indicators of acute emotional states in dogs. Scientific Reports, 14(1), 6406. https://doi.org/10.1038/s41598-024-56859-9

Lerman, D. C., & Iwata, B. A. (1995). Prevalence of the extinction burst and its attenuation during treatment. Journal of Applied Behavior Analysis, 28(1), 93–94. https://doi.org/10.1901/jaba.1995.28-93

Maier, S. F., & Seligman, M. E. P. (2016). Learned helplessness at fifty: Insights from neuroscience. Psychological Review, 123(4), 349–367. https://doi.org/10.1037/rev0000033

Mills, D. S., Demontigny-Bédard, I., Gruen, M., Klinck, M. P., McPeake, K. J., Barcelos, A. M., Hewison, L., Van Haevermaet, H., Denenberg, S., Hauser, H., Koch, C., Ballantyne, K., Wilson, C., Mathkari, C. V., Pounder, J., Garcia, E., Darder, P., Fatjó, J., & Levine, E. (2020). Pain and problem behavior in cats and dogs. Animals, 10(2), 318. https://doi.org/10.3390/ani10020318

Rankin, C. H., Abrams, T., Barry, R. J., Bhatnagar, S., Clayton, D. F., Colombo, J., Coppola, G., Geyer, M. A., Glanzman, D. L., Marsland, S., McSweeney, F. K., Wilson, D. A., Wu, C.-F., & Thompson, R. F. (2009). Habituation revisited: An updated and revised description of the behavioral characteristics of habituation. Neurobiology of Learning and Memory, 92(2), 135–138. https://doi.org/10.1016/j.nlm.2008.09.012

Riemer, S. (2019). Not a one-way road — Severity, progression and prevention of firework fears in dogs. PLoS ONE, 14(9), e0218150. https://doi.org/10.1371/journal.pone.0218150

Riemer, S. (2020). Effectiveness of treatments for firework fears in dogs. Journal of Veterinary Behavior, 37, 61–70. https://doi.org/10.1016/j.jveb.2020.04.005

Riemer, S. (2023). Therapy and prevention of noise fears in dogs — A review of the current evidence for practitioners. Animals, 13(23), 3664. https://doi.org/10.3390/ani13233664

Schilder, M. B. H., & van der Borg, J. A. M. (2004). Training dogs with help of the shock collar: Short and long term behavioural effects. Applied Animal Behaviour Science, 85(3–4), 319–334. https://doi.org/10.1016/j.applanim.2003.10.004

Shnookal, J., Tepper, D., Howell, T., & Bennett, P. (2024). Counterconditioning-based interventions for companion dog behavioural modification: A systematic review. Applied Animal Behaviour Science, 276, 106305. https://doi.org/10.1016/j.applanim.2024.106305

Thompson, R. F., & Spencer, W. A. (1966). Habituation: A model phenomenon for the study of neuronal substrates of behavior. Psychological Review, 73(1), 16–43.

Vieira de Castro, A. C., Fuchs, D., Morello, G. M., Pastur, S., de Sousa, L., & Olsson, I. A. S. (2020). Does training method matter? Evidence for the negative impact of aversive-based methods on companion dog welfare. PLoS ONE, 15(12), e0225023. https://doi.org/10.1371/journal.pone.0225023

Wolpe, J. (1958). Psychotherapy by reciprocal inhibition. Stanford University Press.