Predatory Behaviour in Dogs: Sequence, Selection and Learning

1.1 What the Term Carries

Classical drive theories described behaviour as the discharge of accumulated internal pressure. Energy specific to an activity builds, and either finds its outlet or forces one — in the strongest versions, appearing without any external trigger once the pressure is high enough.

The vocabulary survived the theory. "Prey drive", "drive level", "needs an outlet" all presuppose the hydraulic model, and they are used by people who would reject it if it were stated explicitly.

1.2 What Follows From It

The model generates specific and consequential predictions. That predatory behaviour must be discharged, so a dog denied opportunity will find one. That an outlet is a physiological necessity rather than a training choice. And that the behaviour is involuntary in a way that makes intervention futile.

None of these follows from the current account, and the third is the one that ends training before it starts.

1.3 What Replaced It

Modern accounts describe motivational states — modulated by internal condition, external stimuli and learning history, without an accumulating reservoir. Reward-related processing is organised around anticipation and prediction error, not around pressure release (with the mechanism set out separately).

1.4 Why the Word Survived

Three reasons, and none of them is that the model held up.

It is intuitive. Something that builds and releases matches how frustration feels from the inside, and the metaphor transfers effortlessly to an animal.

It explains without requiring analysis. "High drive" accounts for the behaviour, its intensity and its recurrence in two words, and it does so without anyone examining what the dog actually did.

And it removes responsibility in a way that is sometimes welcome. A dog that could not help it is a dog whose owner could not have prevented it (with a structurally identical function in other practitioner models).

1.5 What "High Drive" Usually Describes

Used carefully, the phrase points at something real: a dog that orients readily, fixes intensely, pursues persistently and is difficult to interrupt. Those are observable properties of a sequence.

The problem is not that people notice this. It is that a description of behaviour gets converted into an internal quantity, and the quantity then explains the behaviour it was derived from.

1.6 Why This Is Not Semantics

If nothing accumulates, then "he needs an outlet" is a claim requiring evidence rather than a physiological fact. Substitute activities may be valuable for other reasons — exercise, engagement, a trained alternative to running off — and the justification is different, which changes what should be measured to know whether they worked.

2. The Sequence

2.1 The Classical Description

The framework that replaced drive language is Coppinger's predatory motor sequence, usually given as orient → eye → stalk → chase → grab-bite → kill-bite → dissect → consume (Coppinger & Coppinger, 2001).

It describes the complete functional chain in a wild canid: locating prey, fixing on it, approaching, pursuing, seizing, killing, and processing the carcass.

2.2 What Selection Did to It

The central claim is that breeding exaggerated some components and inhibited others, producing breed-typical fragments of the ancestral whole.

Herding breeds show exaggerated eye, stalk and chase with grab-bite substantially inhibited. Retrievers show exaggerated chase and grab-bite with the killing bite suppressed — the reason a soft mouth is a virtue and a hard one a fault. Terriers retain grab-bite and kill-bite intact. Livestock guardians show inhibition across the whole sequence (Udell, Ewald, Dorey & Wynne, 2014).

2.3 Why It Is a Raster and Not a Reflex Chain

The sequence is a description of components and their typical order, not a chain that must run to completion once started. A Border Collie that fixes and stalks and then circles rather than seizing has not failed to complete anything; it is doing what its version of the sequence does.

Coppinger and colleagues themselves noted that a wolf can begin at any pattern and that components can be substituted depending on prey type (with the framing problem this creates addressed below).

2.4 What Each Component Looks Like

The value of the framework is that it makes observation specific. Instead of "he has high prey drive", the question becomes which components this dog performs and how readily.

Orient — the head turn and freeze at first detection, usually to movement. Eye — sustained visual fixation, exaggerated to the point of stillness in some herding breeds. Stalk — the slow, lowered approach. Chase — pursuit at speed. Grab-bite — the seizure. Kill-bite — the crushing or shaking bite. Dissect and consume — processing the carcass.

Most pet dogs show the first four readily and the later ones rarely or never, which is a selection outcome rather than a moral one.

2.5 What This Buys in Practice

A dog that orients and fixes but breaks off at stalk has a different problem from a dog that reliably reaches chase, and both differ from a dog that has completed the sequence on live prey.

The intervention point, the risk assessment and the prognosis all differ, and none of that is available from a drive score (with the same argument for operational description throughout).

2.6 The Evidence Behind the Model

This is where honesty is required. The sequence is quoted in the training literature as established fact, and the empirical position is weaker.

Coppinger's account is a hypothesis built on observation of working breeds, and for most of it the specific claims about which behaviours comprise the sequences of ancestral and modern breeds remain unverified (Broseghini, Lõoke, Guérineau, Marinelli & Mongillo, 2024). Research on canine predatory behaviour has been, in those authors' words, surprisingly limited.

3. What Triggers It

3.1 Movement, Not Species

The releasing stimulus is characteristically movement of a particular kind — a specific speed, direction and pattern — rather than the identity of the moving thing.

That is why the same dog pursues rabbits, cyclists, joggers and a plastic bag in wind. From the dog's perspective these are not different categories; they share the property that starts the sequence.

3.2 Why Fleeing Makes It Worse

Movement away is the strongest form of the trigger, which produces a pattern owners recognise: the animal that stands its ground is often left alone, and the one that runs is pursued.

It also explains why interrupting a chase by running after the dog is counterproductive, and why a child running is a different stimulus from a child standing still (with the bite-risk implications documented separately).

3.3 Sensory Entry Points

The sequence is usually described as visually triggered, and the other senses enter it too. Sound — a rustle, a squeak, a high-frequency call — can initiate orientation before anything is seen (in a frequency band where dogs hear well), and scent supports tracking once the visual target is lost (with olfaction the dominant modality overall).

The relative weighting between these has not been systematically studied in dogs.

3.4 Internal State

Hunger is the obvious modulator and it is not the main one in well-fed pet dogs — the sequence runs readily in animals that have just eaten, which is itself an argument against a purely appetitive account.

Arousal from other sources lowers the threshold, which is the mechanism behind chases that begin during play or excitement rather than at the sight of prey.

4. What the Ethogram Found

4.1 The First Systematic Description

Sixty videos of dogs of different breeds performing predatory acts — some from public repositories, some recorded for the purpose — were described in detail by two independent observers, and a general sequence was extracted from the descriptions (Broseghini et al., 2024).

That this was necessary in 2024 is itself informative. The sequence had been cited for four decades without a published ethogram behind it.

4.2 Four Phases, Not Eight

The sequence identified consists of four functional phases: search, approach, chase and bite (Broseghini et al., 2024).

This is not a refutation of Coppinger — the phases map onto the classical components at a coarser grain — and it is a meaningful revision. What emerged from systematic description was a simpler structure than the eight-step version taught in training courses.

4.3 What the Ethogram Adds

Behaviours within each phase were organised into three levels of increasing detail, and the descriptions captured differences between dogs of different working groups. The sequence showed consistency across dogs, with breed-specific differences located in the phases (Broseghini et al., 2024).

So the general shape holds and the breed-specific variation is real. What the study provides that the model lacked is an operational description that another researcher can apply (the measurement standard this field usually lacks).

4.4 What It Does Not Establish

Sixty videos, partly from internet repositories, described by two observers. It is an ethogram — a descriptive instrument — not a test of whether breeding exaggerated or inhibited specific components.

The question Coppinger raised is still open. What exists now is a tool for addressing it.

5. Breed and Its Limits

5.1 The Finding

Three breeds with distinct predatory sequences still under selection were compared on a human-guided pointing task: Airedale Terriers with a fully intact sequence, Border Collies with exaggerated eye-stalk-chase, and Anatolian Shepherds with the sequence inhibited (Udell et al., 2014).

Breed differences in the predatory sequence influenced point-following ability. Border Collies performed as predicted from their heightened sensitivity to moving stimuli.

5.2 Why That Result Is Interesting Beyond Predation

A breeding history aimed at how a dog pursues prey turned out to predict how it follows a human gesture. The motor pattern is not confined to the context it was selected in (which the gesture literature approaches from the other direction).

5.3 The Sentence That Matters Most

The same authors state it directly: explicit experience can be used to overcome breed-specific predispositions in some cases (Udell et al., 2014).

Both halves are load-bearing. Selection shapes the starting point, and it does not fix the endpoint — which is exactly the position the breed literature reaches from population data (where breed explains a small share of behavioural variation).

5.4 Why the Assignments Are Softer Than They Look

The breed descriptions in section 2 come from breed function and observation rather than from measurement across individuals. "Herding dogs exaggerate eye-stalk-chase" is a statement about what the breed was selected to do, not a distribution derived from testing.

Udell's study is the exception in having actually compared three breeds experimentally — and it tested nineteen to twenty dogs per group on a pointing task, not the predatory sequence directly.

5.5 The Show-Working Split

A further complication rarely stated. Many breeds have diverged into show and working lines, and the behavioural selection that produced the sequence descriptions applies mainly to the working population.

A Border Collie from show lines and one from working lines share a breed name and may not share the selection history that the framework assumes (with breed as a category doing less work than expected).

5.6 What Breed Does Not Tell You

Breed predicts the shape of the sequence at group level and not its intensity in an individual. A Border Collie that never chases and a Labrador that hunts to the kill are both entirely possible, and the variation within any breed exceeds the difference between breed means for most behavioural traits.

6. Why It Escalates

6.1 The Sequence Reinforces Itself

This is the mechanism that the drive model obscures. Components of the sequence can become reinforcing through repetition and successful completion — the chase can be rewarding whether or not anything is caught, and the seizure whether or not anything is killed.

Which components acquire that property, and through what — the movement itself, the pursuit, the contact, the interaction with the object, the outcome — is not established for dogs. What can be said is that a dog which chases and loses the rabbit has not necessarily had an unsuccessful experience (with the reinforcement mechanics set out separately).

6.2 Why That Explains More Than Pressure Does

The hydraulic model predicts that discharge reduces the tendency: a dog that has hunted should need it less. The reinforcement account predicts the opposite — and the opposite is what owners commonly report, with the behaviour becoming faster, more readily triggered and harder to interrupt over time.

That is owner observation rather than measurement, and it is the observation both models have to account for. One of them can.

6.3 Intermittent Success Is the Worst Case

Most chases fail, which means the behaviour is maintained intermittently — one of the conditions known to produce highly persistent responding and considerable resistance to extinction (with the schedule evidence set out separately).

The transfer from laboratory schedules to a chase in a field is an extrapolation rather than a measured equivalence, and it is worth marking as one. What holds is the direction: rare success sustains behaviour better than no success or constant success.

6.4 Why Interruption Is Hard

Two properties of the sequence make it unusually resistant to recall.

It accelerates. Orientation is a still moment; chase is a dog at full speed with reduced responsiveness to anything behind it. The window in which a cue can compete narrows as the sequence proceeds (with arousal constraining what can be retrieved).

And the competing reinforcer is immediate and large. A recall cue competes against a behaviour that is rewarding while it is being performed, which is a different problem from competing against an anticipated reward.

6.5 The Trigger Set Widens

Sequences generalise. A dog that has run deer will run cyclists, and the common element is movement of a particular speed and direction rather than the species involved.

That is ordinary stimulus generalisation, and it is why the problem rarely stays where it started (with the same process documented in fear learning).

7. Predation Is Not Aggression

7.1 Different Motivation

Aggression, in most of its forms, functions to increase distance or retain access to something (as the aggression evidence sets out). Predation functions to acquire food. The behaviours look superficially similar and the underlying states are not the same.

7.2 Different Signalling

The distinction is visible before contact. Aggression is typically preceded by threat display — stiffening, staring, growling, lifted lip — which exists precisely because it can prevent escalation.

Predatory behaviour has no equivalent. There is no warning phase because there is no negotiation: signalling would defeat the purpose (which is why the absence of warning means something different here).

7.3 Different Arousal

Aggression is generally accompanied by high sympathetic arousal with the physiological markers that go with it. Predatory sequences frequently run at lower arousal, with focused attention rather than defensive activation (with the arousal distinction relevant throughout).

A dog stalking is not a dog in conflict.

7.4 Why the Distinction Has Consequences

Three of them. A dog that chases livestock is not "aggressive" and treating it as such misdirects the intervention. A dog that shows predatory behaviour toward small animals is not thereby dangerous to people — the contexts and stimuli differ. And conversely, predatory behaviour directed at something it should not be requires management on its own terms rather than being folded into an aggression protocol.

7.5 Why Dog-Directed Cases Are Harder

One complication deserves stating. Aggressive attacks by dogs on other dogs are sometimes reported to display characteristics strongly resembling predatory behaviour, and the classification is not always clear from the behaviour alone.

That overlap is a real diagnostic problem rather than a technicality: the two require different interventions, and the visible sequence may not distinguish them (with aggression itself being situation-specific).

7.6 Where the Distinction Blurs

Two honest qualifications. Predatory behaviour directed at inappropriate targets — including, rarely, at humans — is a recognised clinical presentation, and it is genuinely dangerous precisely because it comes without warning.

And the motivational categories are cleaner in description than in practice. A dog may begin a chase predatorily and switch to defensive behaviour when cornered, and the sequence in a mixed situation does not read as either category cleanly.

8. What "Untraining the Drive" Gets Wrong

8.1 The Question Is Framed Wrong

There is no measurable internal reservoir called drive to untrain. The question that can be asked is which components of the sequence occur, under what conditions, and what happens next.

That is answerable and the drive question is not, which is a large part of why the framing matters.

8.2 Suppression Is Not Removal

Interrupting the sequence stops the behaviour and does not change the motivation that produced it. A dog prevented from chasing still orients and fixes; the components before the interruption are intact.

Prevention is necessary and it is management, not resolution (a distinction that recurs across behaviour work).

8.3 Why Punishment Is a Poor Fit Here

Aversive interruption of a predatory sequence has to compete with a strongly reinforced behaviour at close range, and it carries the documented costs of aversive methods generally (with the fallout characterised separately).

It also risks a specific confusion: a dog punished during a chase may associate the aversive event with the stimulus rather than with its own behaviour, which converts a predatory problem into a fear problem without resolving the first.

8.4 What the Intervention Literature Actually Contains

A systematic review of interventions aimed at resolving problems caused by predatory behaviour found a small number of articles with four consistent limitations: a lack of direct observation, a lack of clear definition of predatory behaviour, a lack of data showing that training generalised, and a lack of consideration of the welfare impact on the dog (McLennan, 2023).

That is the state of the evidence. Every training protocol on offer rests on general learning principles rather than on demonstrated outcomes for this behaviour.

8.5 The Finding That Complicates Management

The same review reports something that sits awkwardly with the standard advice: most research has focused on preventing predatory behaviour, and prevention may carry negative welfare implications for the dog (McLennan, 2023).

This is worth taking seriously rather than explaining away. A dog managed so that the sequence never runs is a dog prevented from expressing a substantial part of its behavioural repertoire, and the review flags that as a cost rather than a neutral default.

It does not follow that management should be abandoned — the consequences of livestock chasing are severe and immediate. It follows that management is a trade-off with something on both sides, which is not how it is usually presented.

8.6 What the Evidence Supports Doing

Management first, because a dog running deer cannot be trained mid-chase and every repetition strengthens the behaviour.

Then intervention at the earliest component. Orientation and fixing are interruptible in a way that chase is not, and the sequence is easiest to affect before it gains momentum — which is the same sub-threshold principle that governs fear work (with graduated exposure the supported method).

One specific caution follows from the review's generalisation gap: a dog that has learned to leave a rabbit in a field has not thereby learned to leave a cat in a garden, and the studies do not show that such transfer occurs.

8.7 What Not to Do to Test It

Progress should not be tested by placing another animal at risk, and exposure protocols that work by steadily increasing viewing time and decreasing distance are not supported by outcome data.

The absence of a validated protocol is itself the finding, and it is a reason for conservatism rather than for improvisation.

8.8 The Substitution Question

Redirecting a dog onto a permitted version of the sequence — a flirt pole, a thrown toy, a tracking task — is widely recommended, and its rationale needs restating without the hydraulic model.

It is not that pressure is being released. It is that an alternative behaviour is being reinforced in the presence of stimuli that would otherwise trigger the problematic one, and that the alternative is under stimulus control while the original is not.

Whether it reduces the original behaviour has not been established. There is a plausible mechanism by which it could make things worse — rehearsing components of the sequence strengthens them — and no canine data settling which effect dominates.

9. Whose Problem It Is

9.1 The Same Behaviour, Opposite Valuations

Predatory behaviour is undesirable in pet dogs because of the risk to livestock, protected wildlife and other pets — and desirable in dogs used for hunting, for managing livestock, for vermin control and in greyhound racing (McLennan, 2023).

It is the same behaviour. What differs is whether the human wants it.

9.2 Why That Matters Scientifically

A behaviour classified as a problem in one context and a working requirement in another cannot be understood as pathology. It has no equivalent among the conditions this library covers — nobody breeds selectively for separation distress.

That framing also explains the research gap. Predatory behaviour has been studied where it is wanted, in terms of how to elicit and refine it, and comparatively little where it is not.

9.3 The Breeding Consequence

Because of historical and current breeding practices, it is difficult to ensure that dogs living as pets do not show an interest in performing predatory behaviour (McLennan, 2023).

That is the honest version of what an owner is told when the dog arrives. The behaviour was selected for, often recently, and the pet home is a context the selection did not anticipate (with the general problem of breed function meeting pet life documented elsewhere).

9.4 What This Does to the Advice

It moves the frame from correction to compatibility. The question is not how to remove something that was deliberately bred in, but which arrangements let this particular dog live safely with what it has.

That is less satisfying than a protocol, and it is what the evidence supports.

10. Where Predation Meets Welfare

10.1 The Practical Stakes

Predatory behaviour toward livestock has consequences beyond training: in many jurisdictions a dog that chases sheep can be destroyed, and the owner is liable.

10.2 Why This Is Not a Behaviour Problem to the Dog

The dog is doing what the sequence does, in a context where the outcome is catastrophic. That framing does not excuse anything and it does locate the intervention correctly — at management and at the early components, not at the dog's character.

10.3 The Differential Nobody Runs

A sudden increase in chasing, or its appearance in a dog that never showed it, is worth a physical examination before a behavioural one. Sensory change, pain and cognitive decline all alter how a dog responds to movement (with hearing loss presenting in ways that are easily misread).

10.4 What Owners Are Usually Told

Three recommendations circulate widely, and their evidential status differs sharply.

Long line and management — supported, in the sense that it prevents rehearsal, with the welfare caveat from section 8.5 attached.

Impulse-control training — plausible and untested for this behaviour specifically. Whether general inhibitory training transfers to a chase in progress is exactly the generalisation question the review found unanswered (McLennan, 2023).

More exercise — the least supported of the three, and the one owners reach for first.

10.5 The Exercise Argument

The claim that a dog which hunts needs more exercise is common and does not follow from anything above. A fitter dog chases further.

What the evidence supports is engagement and trained alternatives rather than volume — and the distinction matters because increasing exercise is the easier intervention and the one owners reach for first.

11. What to Do With a Chasing Dog

11.1 Describe Before Explaining

The first step is not an intervention. It is establishing which components this dog performs, at what distance, toward what, and how far the sequence gets.

A dog that orients and looks back at the handler is in a different position from one that reaches full chase, and the two are routinely given the same advice.

11.2 Record the Conditions

Distance at which orientation begins, stimulus type, whether the dog was already aroused, what the terrain allowed, and what happened afterwards. Without those, progress cannot be assessed and the generalisation problem cannot even be seen.

11.3 Prevent Rehearsal While Working

Every completed sequence strengthens it, so the management and the training run in parallel rather than in sequence. A long line is not a failure of training; it is the condition under which training is possible.

11.4 Work at Orientation

The earliest component is the one that is interruptible, and it is where an alternative behaviour can be established before the sequence has momentum.

That is the same sub-threshold logic as fear work, and it has the same requirement: if the dog is already chasing, the session is over (with the exposure principle set out separately).

11.5 Test Transfer Deliberately

Because generalisation is not demonstrated (McLennan, 2023), it has to be built rather than assumed — new locations, new stimulus types, new distances, each treated as a fresh problem rather than as a test of what was learned elsewhere.

11.6 Accept the Ceiling

Some dogs will need lifelong management, and saying so early is more useful than a protocol that implies resolution. That is not a failure of training; it is what a selectively bred motor sequence in an unsuited environment looks like.

12. Summary at a Glance

"Drive" carries an abandoned model — Classical drive theory described accumulating pressure requiring discharge; the vocabulary survived the framework.

The sequence is the replacement — Orient, eye, stalk, chase, grab-bite, kill-bite, dissect, consume, with breeding exaggerating some components and inhibiting others (Coppinger & Coppinger, 2001; Udell et al., 2014).

Most of it remains unverified — The specific claims about which behaviours comprise ancestral and modern breed sequences have not been tested (Broseghini et al., 2024).

The first ethogram found four phases — Search, approach, chase and bite, extracted from 60 videos described by two independent observers (Broseghini et al., 2024).

Breed shapes the sequence and does not fix it — Predatory sequence differences predicted point-following performance, and explicit experience overcame breed predispositions in some cases (Udell et al., 2014).

Completion can reinforce the sequence — Components can become reinforcing through repetition and successful completion, which explains escalation where a pressure model predicts the opposite.

Intermittent success sustains the behaviour — Most chases fail, which is one of the conditions known to produce highly persistent responding.

Predation is not aggression — Different motivation, no threat display, frequently lower arousal, and different intervention.

The intervention literature is thin and flawed — A systematic review found few studies, with a lack of direct observation, no clear definition, no generalisation data and no consideration of welfare impact (McLennan, 2023).

Prevention may carry welfare costs — Most research focuses on preventing the behaviour, and prevention may have negative welfare implications for the dog (McLennan, 2023).

The same behaviour is bred for and against — Undesirable in pet dogs, desirable for hunting, livestock management, vermin control and racing (McLennan, 2023).

13. Research Gaps and Critical Appraisal

The core model is a hypothesis, not a finding. Coppinger's sequence is an observational framework, and the claims about breed-specific exaggeration and inhibition remain largely unverified (Broseghini et al., 2024).

The literature is thin. Canine predatory behaviour has attracted remarkably little research relative to its practical importance, and the first systematic ethogram appeared in 2024.

The ethogram is descriptive and small. Sixty videos, partly from internet sources, coded by two observers. It provides an instrument rather than an answer.

Breed assignments are traditional rather than measured. Statements that herding dogs exaggerate eye-stalk-chase derive from breed function and observation, and have rarely been quantified across individuals within those breeds.

No controlled intervention studies exist. Every recommendation in section 7 rests on general learning principles rather than on trials of predatory behaviour modification specifically.

Substitution is untested. Whether redirecting the sequence onto permitted outlets reduces the problematic behaviour, leaves it unchanged, or rehearses the components has not been established in dogs.

Generalisation is not demonstrated. The intervention studies that exist do not show that training transfers across targets, environments or stimulus types (McLennan, 2023), which is precisely what an owner needs.

The welfare cost of prevention has not been quantified. It is flagged as a concern rather than measured, and the trade-off between management and behavioural expression cannot currently be assessed.

The aggression boundary is under-characterised. Predatory behaviour toward inappropriate targets is clinically recognised and epidemiologically almost invisible, because bite statistics do not classify by motivation.

14. Conclusion

The most useful thing to say about predatory behaviour is that the word most people use for it carries a model that mainstream behavioural science largely left behind. "Prey drive" carries a hydraulic account — pressure accumulating until it must discharge — and with it three claims that do not follow from anything current: that the behaviour must be released, that an outlet is a physiological necessity, and that the whole thing is involuntary and therefore beyond intervention. What replaced it is a sequence of motor patterns whose components were exaggerated or inhibited by breeding, and which is shaped further by everything the dog has experienced. The evidence for that framework is thinner than its confident use suggests — Coppinger's account remains largely unverified, and the first systematic ethogram of the canine predatory sequence appeared only in 2024, finding four functional phases where the training literature teaches eight. What the reframing buys is a better explanation of the thing owners actually report. Predatory behaviour can escalate because components of the sequence may be reinforced through repetition and completion, while most attempts fail — one of the conditions known to produce highly persistent behaviour. A pressure model predicts that hunting should reduce the tendency; the reinforcement account predicts it should increase it, and the second better matches the persistence owners commonly report. That also relocates the intervention: not at discharging something, but at management, at the earliest components of the sequence, and at the honest acknowledgement that a dog which has run deer has been rewarded for it (much as other self-reinforcing behaviours resist the interventions aimed at them).

Key Insights (Takeaways)

• The word carries a model that mainstream behavioural science largely no longer uses. "Prey drive" presupposes accumulating pressure requiring discharge — the hydraulic account of classical drive theory. What follows from it, that the behaviour must be released and cannot be trained, does not follow from any current framework, and it is the reason owners are told their dog needs an outlet rather than a plan.

• The sequence is the replacement, and it is less established than it sounds. Orient, eye, stalk, chase, grab-bite, kill-bite, dissect, consume (Coppinger & Coppinger, 2001), with breed-specific exaggeration and inhibition (Udell et al., 2014) — but for most of it the claims about which behaviours comprise ancestral and modern breed sequences remain unverified (Broseghini et al., 2024).

• The first empirical ethogram found four phases, not eight. Sixty videos described by two independent observers produced a sequence of search, approach, chase and bite, with breed differences located within the phases (Broseghini et al., 2024). That it took until 2024 says something about how little this area has been studied.

• Escalation is better explained by reinforcement than by pressure. Components of the sequence can become reinforcing through repetition and completion, and most chases fail — intermittent success being one of the conditions known to produce highly persistent responding. The hydraulic model predicts that hunting reduces the tendency; what owners commonly report is the opposite, which the reinforcement account accommodates and the pressure account does not — owner report being the observation to be explained rather than a measurement of escalation.

• The intervention evidence is close to absent, and one review says so plainly. A systematic search found few studies, each limited by a lack of direct observation, no clear definition of the behaviour, no evidence that training generalised, and no consideration of what prevention costs the dog (McLennan, 2023). Every protocol on offer rests on general learning principles rather than on outcomes for this behaviour.

• Predatory behaviour is not aggression, and breed is not destiny. Different motivation, no threat display, frequently lower arousal — and a dog that chases livestock is not thereby dangerous to people. The same researchers who documented breed-specific sequences also found that explicit experience can overcome breed predispositions in some cases (Udell et al., 2014).

References

Broseghini, A., Lõoke, M., Guérineau, C., Marinelli, L., & Mongillo, P. (2024). Ethogram of the predatory sequence of dogs (Canis familiaris). Applied Animal Behaviour Science, 279, 106402. https://doi.org/10.1016/j.applanim.2024.106402

Coppinger, R., & Coppinger, L. (2001). Dogs: A Startling New Understanding of Canine Origin, Behavior and Evolution. Scribner.

McLennan, T. (2023). Review of literature on interventions aimed at resolving problems caused by predatory behaviour in dogs (Canis familiaris). Applied Animal Behaviour Science, 266, 106037. https://doi.org/10.1016/j.applanim.2023.106037

Udell, M. A. R., Ewald, M., Dorey, N. R., & Wynne, C. D. L. (2014). Exploring breed differences in dogs (Canis familiaris): Does exaggeration or inhibition of predatory response predict performance on human-guided tasks? Animal Behaviour, 89, 99–105. https://doi.org/10.1016/j.anbehav.2013.12.012