Prey Drive in Dogs: Why Motivation Explains More Than Classic Drive Theory

In brief

“It’s just his drive, nothing you can do about it”—that sentence is the real problem. The classic drive theories of the early 20th century, above all the idea of pent-up inner pressure that has to discharge compulsively, are now considered outdated. They did not fail because they went out of fashion, but on three concrete points: no mechanism in the nervous system could be found for the supposed buildup of energy, the predictions derived from it did not come true, and the concept explains nothing, because it is inferred from exactly the behavior it is supposed to explain.

But beware of the opposite oversimplification: this does not mean there are no genetic behavioral tendencies—they are real and measurably heritable. The point is a different one: a single “drive” does not explain complex behavior, it merely renames it. Behavior emerges dynamically from genetics, learning, emotion and environment—and that is exactly why it can be influenced.

The pressure-from-within model has been replaced by a different one: motivation as evaluation. How attractive is this goal for this animal, in this situation, with this learning history? The two models make different predictions, and only those of the evaluation model match what actually happens in training.

Instead of “What drive does my dog have?”, the better question is: “What does my dog need in this situation—and what has he learned?”

👉 The related outdated concept: Why Dominance Fails as a Scientific Concept in Dogs

1. What classic drive theory claims

Your dog runs after a deer and doesn’t come back. Someone says: “Well, that’s prey drive.” The question is whether that explains anything at all.

The idea of explaining behavior through inner drives emerged at the beginning of the 20th century and developed in two parallel traditions.

1.1 The ethological line: the psychohydraulic model

Early ethologists observed that animals sometimes showed certain behaviors even when no suitable trigger was present. From this came the idea of drives as inner sources of energy (Lorenz, 1937).

Lorenz put this into a very concrete image, the psychohydraulic model (Lorenz, 1950): imagine a reservoir for each behavior pattern in which “action-specific energy” continuously accumulates. A valve—the innate releasing mechanism—is opened by a key stimulus, the energy flows out and discharges in the instinctive act. Afterward the reservoir is empty, the behavior stops, and the filling process starts all over again.

Two observations were considered the strongest evidence: the vacuum activity—an animal performs the behavior without any recognizable trigger once it has been dammed up long enough—and the lowering of the threshold, the assumption that as the waiting time increases, ever weaker stimuli are enough.

1.2 The psychological line: drive reduction

In parallel, Freud’s drive theory shaped people’s thinking. Even more systematic was Hull’s drive-reduction theory (Hull, 1943): a physical state of deprivation creates a drive, the drive creates restlessness, the behavior reduces the drive—and it is precisely this reduction that explains why learning takes place.

The two lines are not identical, but they share the decisive assumption: behavior is powered by an inner energy that builds up and has to be released.

That very assumption no longer holds up today.

2. Where drive theory failed

This part is usually skipped in popular accounts—even though it is the genuinely interesting one. Drive theory was not abandoned as a matter of taste; it was refuted on concrete points.

2.1 There is no physiological mechanism

Robert Hinde was one of the first to explicitly declare the hydraulic model exhausted (Hinde, 1956). His central argument: no mechanism is known in the nervous system that would correspond to a gradual accumulation and discharge of “action-specific energy.”

The model was a physical analogy, not a described biological process—and the analogy could not be translated into physiology.

2.2 The predictions did not come true

The model implies that the stimulus threshold should drop as waiting time increases. That is exactly what could not be reliably demonstrated in experiments.

2.3 The circular argument

This is the most serious methodological point, and it still applies today.

The drive is inferred from the behavior it is supposed to explain. The dog hunts, so he has prey drive. How do we recognize prey drive? By the fact that he hunts.

Nothing has been explained; it has only been renamed. The term would have explanatory power only if it could be measured independently of the behavior—and that is exactly what it cannot be. Philosophers of science call this a duplication of the problem: instead of having to explain only the hunting, you now also have to explain the hypothetical cause.

2.4 Innate does not mean impervious to influence

Daniel Lehrman published a fundamental critique as early as 1953. One core point: a deprivation experiment—an animal shows a behavior even without the opportunity to learn it—only shows that the factors that were controlled were not necessary for development. It does not follow that the behavior develops independently of the environment.

Even behaviors with a strong genetic basis need specific environmental conditions in order to develop.

What is remarkable about the course of this debate: Tinbergen largely conceded the physiological weakness of the instinct model; Lorenz held on to it longer. In behavioral biology today, the energy-flow metaphor is considered outdated.

2.5 And the second line? Drive reduction failed on two animal experiments

So far this has been about Lorenz. Hull’s drive reduction from section 1.2, on the other hand, was refuted elsewhere—by two findings that were both published in 1950 and were both hard to argue away.

The first concerns sweeteners. According to Hull, learning takes place because a physical deficit is reduced. Sheffield and Roby showed that rats will work for a non-nutritive sweet solution and keep that behavior up (Sheffield & Roby, 1950). No hunger is satisfied, no deficit is corrected, no drive is reduced—and still learning takes place. The taste itself was the reinforcer.

The second concerns curiosity. Harlow and colleagues gave four rhesus monkeys a mechanical puzzle for twelve days whose solution led to nothing at all—no food, no water, no reward. The animals solved it anyway and became better at it than a control group that merely had the puzzle lying unassembled in their cage (Harlow, Harlow & Meyer, 1950).

The real blow, however, came from the second part of the experiment. When food was subsequently brought into play, the practiced animals’ performance got worse—not better. A model in which learning arises from the reduction of a deficit predicts exactly the opposite.

What is remarkable is how Harlow himself responded: he postulated a manipulation drive, a drive to handle things. In doing so he rescued the concept of drive by expanding it—and unintentionally provided a textbook example of the problem from section 2.3. If a new drive can be introduced for every unexplained behavior, the theory is immune to any contrary finding. That is precisely what makes it worthless.

How far this goes: Both are old studies with very small samples of rats and monkeys. They are cited here not as evidence about dog behavior, but as what they were historically: the findings on which a general model of learning broke down. Nothing follows directly for dogs—but the model that is still invoked in their name today failed on them.

3. Why drive theory persisted for so long

Tradition. Many trainers were themselves trained this way.

Simplicity. Drive models seem to provide clear answers and spare you from looking closely.

Marketing. Some approaches use drives to justify highly controlling or aversive methods along the lines of “The dog can’t help it.”

Confusion with needs. Exercise, food, social contact and mental stimulation are real needs, but not drives in the classic sense.

There is also a linguistic effect: “drive” sounds like an explanation. A label that feels like a diagnosis is rarely questioned.

👉 Dog training myths put to the test · Dog–Dog Social Behavior: What Free-Ranging Populations Show

4. The modern view: behavior arises from many factors

Two dogs of the same breed—one chases every rabbit, the other doesn’t. Why?

Modern behavioral research replaces the question of the one cause with several levels. Tinbergen’s four questions separate two things that are constantly mixed up in the language of drives:

Proximate questions: What triggers the behavior right now? How did it develop in this individual?

Ultimate questions: What selective advantage did it have? How did it evolve?

“Because dogs are predators” answers an ultimate question. It gives no answer to the proximate one—why this dog takes off right here. It is exactly this confusion that makes drive concepts useless in training. The four questions have since been developed further and place more emphasis on the interplay of genetics, development, environment and learning (Bateson & Laland, 2013).

In concrete terms, these factors work together:

• Genetics—behavioral tendencies can be inherited. They influence behavior but do not determine it.
• Learning and experience—dogs learn from consequences, observation and social interactions.
• Emotions—fear, joy, frustration or a sense of safety often influence behavior more strongly than any presumed disposition.
• Social attachment—a securely attached dog behaves differently from an insecure one.
• Environment and context—a behavior shown in one situation will not necessarily occur in another.
• Current arousal—the same dog responds differently when highly aroused than when calm.

Put simply: your dog doesn’t act because he has to, but because it pays off in that situation. And what isn’t rigidly fixed can be changed.

👉 Epigenetics in Dogs · Arousal Regulation in Dogs · Behavior Does Not Equal Emotion

4.1 What has taken the place of drive

“Motivation” is not simply a substitute word for “drive.” Modern behavioral research no longer explains behavior through a single inner factor at all, but through the interplay of motivational systems, learning history and context. Where there used to be one term, there are now three levels.

The decisive difference, though, lies in how it is modeled: motivation is not thought of as pressure from within that has to get out, but as evaluation—how attractive is this goal for this animal, in this situation, with this learning history?

This is more than cosmetic rewording, and the difference can be tested. A pressure model predicts: if the behavior is not performed, it builds up and eventually bursts out. An evaluation model predicts: if what pays off changes, the behavior changes.

Only the second prediction matches what actually happens in training.

4.2 The test anyone can do

There is an everyday observation on which the pressure model fails especially clearly—and you don’t need a lab for it.

A dog rests for an entire day. According to the reservoir model, his play reservoir should be full to the brim and overflow at the first opportunity. But if you offer a toy to a dog who is exhausted, ill or in pain, as a rule he won’t play. The full tank does not discharge.

Conversely, the same dog plays at length the next day after ten minutes of warming up—without anything having “built up” in the meantime.

Play behavior arises when two things come together: an inner readiness—rested, free of pain, in a suitable range of arousal—and an attractive external occasion. That is exactly what the evaluation model predicts. The pressure model predicts the opposite.

👉 The Neurobiology of Play in Dogs · Recognizing pain in dogs

4.3 Wanting and liking are not the same thing

A distinction from neurobiology fits especially well here, because it explains what the drive model can only name: wanting something and liking something are two different processes.

Berridge and Robinson published an extensive review on this, combined with their own experiments in rats (Berridge & Robinson, 1998). Animals whose dopamine levels in the relevant brain structures had been depleted by up to 99 percent still showed the normal pleasure reactions to sweet tastes and the normal aversion to bitter ones. They also continued to learn which stimulus predicts what. What they lacked was the drive to pursue it.

This leads to the distinction that has been standard ever since: the dopaminergic system does not produce the pleasure, but the pull of a goal—the authors speak of incentive salience. That is why wanting and liking can diverge.

How far this goes: This is basic research, mostly in rodents. That the systems involved are similarly organized across mammals is well established; transferring this to specific training situations with dogs is nevertheless an inference, not a finding.

For practice, the consequence is still considerable. A dog can show a behavior with high motivation that does him no good at all while he is doing it—and those are exactly the cases the drive model interprets as a particularly strong drive. The evaluation model describes them more accurately as behavior whose pull has taken on a life of its own. And it says something the concept of drive cannot: high motivation is no evidence that the dog needs something.

👉 Dopamine in Dogs: Reward, Motivation and Learning · Prediction Error in Dogs · The role of motivation in dog training

5. The myth of prey drive

Your dog sees a rabbit, takes off—and you wonder whether you will ever get control.

Myth	What actually applies
“Prey drive controls the dog uncontrollably.”	Hunting behavior is a multi-part, partly learned and motivated behavior pattern.
“There’s nothing you can do about it.”	Impulse control, alternative behaviors and management can change it.
“All dogs of this breed are the same.”	Individual learning history, context and arousal level also play a decisive role.

5.1 Why the behavior chain is the better model

What looks like an inner automatic mechanism is in fact a sequence of distinguishable links: orienting, eyeing, stalking, chasing, grabbing, killing, dissecting, eating.

Ethology has changed its terms for these building blocks twice, and the change itself is revealing. Lorenz and Tinbergen spoke of Erbkoordinationen—in English, fixed action patterns—that is, rigid programs that, once triggered, run through to the end regardless of what happens in the environment.

That is exactly what did not survive scrutiny. Today the more common term is modal action pattern (Barlow, 1977): the pattern is species-specific and genetically pre-shaped, but statistically modal rather than rigid—it shows considerable variability in execution, intensity and completeness and can be influenced or interrupted by arousal, context and learning.

This is not hair-splitting. “Fixed” would mean: if you’re too late, there’s nothing more you can do. “Modal” means: there are points where you can step in.

And this is where the decisive advantage over the drive model shows: breeding lines have emphasized some links and weakened others. In herding dogs, eyeing and stalking are pronounced, while the later links recede strongly. In retrievers, grabbing is preserved and the killing bite reduced. A livestock guardian dog ideally shows hardly any of the chain.

A single prey drive could not explain this—you cannot cut a reservoir off in the middle in a breed-specific way. A modular behavior chain, on the other hand, can.

That is exactly why this model has prevailed. It not only explains more, it explains something that the other model logically rules out.

In practical terms, this means: training does not work through suppression, but through early interruption at the early links of the chain, alternative behaviors, impulse control, management and species-appropriate outlets. If you only step in once the dog is already chasing, you are almost always too late.

5.2 Why hunting becomes entrenched even though it almost never succeeds

The reason is not an overpowering drive, but something more sober: performing the behavior is itself reinforcing. The technical term for this is intrinsic reinforcement—the behavior rewards itself through its own performance, not through a consequence from outside.

So the chase does not have to end in a catch in order to become entrenched. To keep it clear what we’re talking about, it helps to separate three things that are often lumped together in everyday life.

First, the external reinforcer. This is the actual catch—a consequence that occurs after the behavior and comes from outside. In hunting it almost never happens; that is exactly why it cannot explain why the behavior becomes entrenched anyway.

Second, intrinsic reinforcement. Here the performance itself is the reinforcer: the stalking, the taking off, the pursuit. No consequence is needed afterward, because the reinforcement lies in the doing. This is the mechanism that can explain the persistence.

Third, arousal. It accompanies the sequence and raises the level of arousal, but it is not itself reinforcement—it is a state. It makes the behavior more likely and the dog harder to reach—one reason why stepping in late fails—but it does not explain why the behavior is learned.

So the dog doesn’t need to get the rabbit for hunting to pay off. And the arousal you see along the way is the side effect, not the cause.

And this is exactly where the distinction from earlier comes in: what drives the dog is the pursuit—not the outcome. That is why behaviors of this kind are so persistent, even though objectively they almost never reach their goal: they reward themselves along the way.

👉 The predatory motor sequence · Self-Control in Dogs · Retrieve training with dummies · Mantrailing · Recall training

6. The other “drive” concepts

Guarding drive. Barking at the property line is often full of insecurity, territorial behavior, learned successes and a lack of impulse control. The apparent drive here is often a behavior that can be reinforced again and again—because the consequence occurs reliably: the mail carrier leaves.

Herding drive. Circling children is a natural but trainable pattern made up of the early links of the hunting chain, which can be strengthened through learning and context.

Play drive. Not a drive in the classic sense, but an expression of social behavior, hunting motivation and positive arousal. Play is now regarded as a behavioral system in its own right.

Protection drive. Particularly problematic, because here the explanatory error has direct consequences: what is interpreted as protection is frequently fear, insecurity or resource guarding. If you read it as a drive, you train past the actual problem—and may make it worse.

👉 Territorial behavior in dogs · When your dog barks at visitors · Aggression in Dogs · Warning signals in dogs

7. What remains valid in the idea of drive

Fairness demands this question—because replacing the old oversimplification with a new one would be an oversimplification of its own.

What does not hold up: the idea of pent-up pressure that has to discharge compulsively; that drive is unchangeable and resistant to training; that a single drive fully explains complex behavior.

What research does support, on the other hand:

7.1 Genetic behavioral tendencies are real

Breed differences in behavior are measurable (MacLean et al., 2019; Serpell & Duffy, 2014). MacLean and colleagues combined behavioral data from more than 14,000 dogs of 101 breeds with breed-averaged genotype data from over 100,000 loci and found high between-breed heritability for 14 behavioral traits; 131 individual genetic variants could be linked to breed differences in behavior, mostly in genes that are highly expressed in the brain.

Here, however, a distinction is worth making that regularly gets muddled.

Heritability is not the same as breed membership. Morrill and colleagues (2022) surveyed the owners of 18,385 dogs and genotyped 2,155 of them. The result: most behavioral traits are indeed heritable—heritability was above 25 percent. But breed explains only about 9 percent of the variation in behavior between individual dogs.

Both figures are population values. They describe how differences are distributed within a studied group—not how much is fixed in the dog in front of you. What this means in detail is explained further down in section 7.3; the note belongs here as well, because without it both figures are regularly read as statements about the individual animal.

So genetics matters more than the counter-movement sometimes suggests. Breed is only a weak proxy for it.

7.2 How two studies can both be right

At first glance, the two large studies seem to contradict each other: MacLean et al. (2019) have “highly heritable” in the title, while Morrill et al. (2022) emphasize how weakly breed predicts behavior.

In fact, they measure different things. MacLean and colleagues integrated behavioral data from more than 14,000 dogs of 101 breeds with breed-averaged genotype data and determined heritability between breeds—that is, the proportion of trait differences that goes along with the genetic similarity of breeds. Morrill and colleagues, by contrast, asked how well breed predicts the behavior of an individual dog.

Both can be true at the same time: breeds differ measurably and in genetically traceable ways on average—and yet breed tells you little about the dog in front of you.

This is the core of the misunderstanding behind many statements about drives and breeds: a group average is read as a characteristic of the individual.

Broken down by area, the picture varies. For traits related to biddability and trainability, a moderate breed effect emerged—for aggression-related behaviors, practically none. The authors suggest that the behavior-relevant adaptations of dog domestication predate the emergence of modern breeds by a long way, and that modern breeds differ mainly in physical traits.

The difference between the two measures becomes most visible precisely where the debate is loudest. MacLean and colleagues report the highest between-breed heritability values for trainability, chasing behavior and stranger-directed aggression, among others. Morrill and colleagues find no meaningful breed effect for aggression-related traits at the level of the individual dog.

This is not a contradiction, but the same facts from two angles: breed averages differ clearly and in genetically traceable ways, and at the same time individual animals within each breed vary so widely that breed membership yields almost nothing for predicting an individual dog. Anyone who turns the first figure into a statement about the dog in front of them makes the same mistake as someone who concludes from the second figure that genetics plays no role.

7.3 What heritability does not mean

Heritability is a population statistic, not a statement about an individual.

A value of 30 percent does not mean that 30 percent of your dog’s behavior is genetic and 70 percent learned. It describes how much of the differences within a studied population goes along with genetic differences. And the variation within a breed is usually greater than the variation between breeds.

7.4 And what else remains

Some dispositions are more stable than others. Hunting motivation in selectively bred working breeds can be managed and toned down—but often not reduced to zero.

Motivational systems are real. What has changed is not just the word, but the model behind it: from pressure that has to get out to an evaluation that depends on the situation and on learning.

Training has limits. Not every behavior can be shaped at will; realistic goals take the individual, genetics and context into account.

The accurate statement, therefore, is not “There are no drives,” but: the classic drive model does not adequately explain behavior. Genetic dispositions and motivational systems are real, but they always act in interaction with learning, emotion and environment—and can therefore be influenced.

👉 Breed and Behavior: Genetic Predisposition versus Individual Variation · The “dangerous breed” myth · Temperament, Personality, and Coping Styles in Dogs

8. What this means for training

Reward-based instead of punishment-based. Reward strengthens desired behavior and the relationship; punishment-based methods can promote stress responses, avoidance behavior or aggressive reactions.

Individual instead of one-size-fits-all. Every dog brings their own experiences, temperament and learning strategies.

A secure relationship. A secure dog learns better.

Work with emotions and needs. Instead of “a drive outburst,” the question is: What need lies underneath, which emotion is in control, and how can the dog learn to deal with it appropriately?

Management is part of the solution. As long as an alternative behavior isn’t yet reliable, management prevents the old pattern from being reinforced further. That isn’t failure, it’s a prerequisite.

👉 Force-free dog training · Modern learning theory in dog training · Aversive Training Methods in Dogs · Building trust with your dog

9. How to tell outdated approaches from current ones

Red flags: Behavior is explained exclusively in terms of drives. Dominance or coercive methods are recommended. Sweeping statements about all dogs of a breed. Outdated terms such as alpha roll, rank order or dominance training.

And one warning sign that is easy to overlook: when a method fundamentally cannot fail, because every failure is blamed on the drive or the owner. An explanation that explains everything and can never be wrong explains nothing.

Current practice means working with testable methods, individual assessment—why this dog, this behavior, in this situation?—positive reinforcement, alternative behaviors and transparency about the limits of one’s own method.

👉 Dog trainer or behavior consultant? · Choosing the right dog trainer

10. What the research does not show

• The exact share is an open question. How much of a specific disposition is due to genetics, early development or learning can rarely be quantified cleanly in an individual case.
• “Motivational system” is a model, not a mapped mechanism. The neurobiology behind motivation and arousal is partly understood, but not completely—and some of the findings come from other species.
• Owner surveys have limits. Much of canine behavioral genetics is based on questionnaire data. These are indispensable for large samples, but susceptible to expectation effects—especially where owners are familiar with breed stereotypes.
• The building blocks of the hunting chain have been described, not measured breed by breed. That breeding lines emphasize individual links is considered an established ethological description; there are no controlled comparative studies by breed on this.
• Case examples are not studies. Successes in practice illustrate principles, but they do not replace controlled evidence of effectiveness.

11. Conclusion

Dogs are not machines controlled by drives, but learning, social and emotional beings.

Hunting, guarding or herding behavior is trainable, not a matter of fate. What matters is learning history, emotions, context and the relationship. Genetic tendencies are real—but they are not a verdict, and breed says less about the individual dog than many people assume.

If you take away one sentence, make it this one: don’t ask “What drive does my dog have?”, ask “What does my dog need in this situation—and what has he learned?”

Key takeaways on drive theories in dogs

The pressure model failed on concrete points; it didn’t just go out of fashion. No mechanism in the nervous system could be found for the supposed accumulation of “action-specific energy” (Hinde, 1956), and the predicted lowering of the stimulus threshold with increasing waiting time could not be reliably demonstrated.

The classic concept of drive explains nothing; it renames. It is inferred from exactly the behavior it is supposed to explain; it would have explanatory power only if it could be measured independently of that behavior. As a descriptive shorthand in everyday life, “prey drive” can still be useful—as a reason for a behavior, it is not.

The psychological line didn’t hold up either. Rats learn for a non-nutritive sweet solution (Sheffield & Roby, 1950), and monkeys solve a puzzle without any reward—with the later introduction of food actually making their performance worse (Harlow et al., 1950). Learning through the correction of a deficit explains neither.

It hasn’t been replaced by a substitute word, but by a different model. Motivation is not thought of as pressure from within that has to get out, but as the evaluation of a goal in a situation—and evaluations can be changed.

Wanting and liking are two processes. The dopaminergic system produces the pull of a goal, not the pleasure in it (Berridge & Robinson, 1998). High motivation is therefore no evidence that a dog needs something.

Hunting behavior is a chain, not a reservoir. Breeding lines have emphasized some links and weakened others—a single drive could not explain this, a modular pattern can (Barlow, 1977).

Genetics matters more than the counter-movement suggests—breed matters less. Most behavioral traits are more than 25 percent heritable, but breed explains only about 9 percent of the behavioral differences between individual dogs (Morrill et al., 2022). Both figures are population values and say little about any particular dog.

Heritability is a population statistic. A value of 30 percent does not mean that 30 percent of your own dog’s behavior is genetically fixed. It describes differences within a studied group.

👉 Behavioral Flexibility in Dogs

References

Barlow, G. W. (1977). Modal action patterns. In T. A. Sebeok (Ed.), How animals communicate (pp. 98–134). Indiana University Press. No DOI has been assigned to this book chapter.

Berridge, K. C., & Robinson, T. E. (1998). What is the role of dopamine in reward: Hedonic impact, reward learning, or incentive salience? Brain Research Reviews, 28(3), 309–369. https://doi.org/10.1016/S0165-0173(98)00019-8

Bateson, P., & Laland, K. N. (2013). Tinbergen's four questions: An appreciation and an update. Trends in Ecology & Evolution, 28(12), 712–718. https://doi.org/10.1016/j.tree.2013.09.013

Harlow, H. F., Harlow, M. K., & Meyer, D. R. (1950). Learning motivated by a manipulation drive. Journal of Experimental Psychology, 40(2), 228–234. https://doi.org/10.1037/h0056906

Hinde, R. A. (1956). Ethological models and the concept of “drive.” The British Journal for the Philosophy of Science, 6(24), 321–331. https://doi.org/10.1093/bjps/VI.24.321

Hull, C. L. (1943). Principles of behavior: An introduction to behavior theory. Appleton-Century-Crofts. No DOI has been assigned to this book.

Lehrman, D. S. (1953). A critique of Konrad Lorenz's theory of instinctive behavior. The Quarterly Review of Biology, 28(4), 337–363. https://doi.org/10.1086/399858

Lorenz, K. (1937). Über die Bildung des Instinktbegriffes. Die Naturwissenschaften, 25(19), 289–300. No DOI has been assigned to this work.

Lorenz, K. (1950). The comparative method in studying innate behaviour patterns. Symposia of the Society for Experimental Biology, 4, 221–268. No DOI has been assigned to this work.

MacLean, E. L., Snyder-Mackler, N., vonHoldt, B. M., & Serpell, J. A. (2019). Highly heritable and functionally relevant breed differences in dog behaviour. Proceedings of the Royal Society B, 286(1912), 20190716. https://doi.org/10.1098/rspb.2019.0716

Morrill, K., Hekman, J., Li, X., McClure, J., Logan, B., Goodman, L., Gao, M., Dong, Y., Alonso, M., Carmichael, E., Snyder-Mackler, N., … Karlsson, E. K. (2022). Ancestry-inclusive dog genomics challenges popular breed stereotypes. Science, 376(6592), eabk0639. https://doi.org/10.1126/science.abk0639

Serpell, J. A., & Duffy, D. L. (2014). Dog breeds and their behavior. In A. Horowitz (Ed.), Domestic dog cognition and behavior (pp. 31–57). Springer. No DOI has been assigned to this book chapter.

Sheffield, F. D., & Roby, T. B. (1950). Reward value of a non-nutritive sweet taste. Journal of Comparative and Physiological Psychology, 43(6), 471–481. https://doi.org/10.1037/h0061365

Tinbergen, N. (1951). The study of instinct. Oxford University Press. No DOI has been assigned to this book.

Standard works for context (not cited individually in the text):

Miklósi, Á. (2015). Dog behaviour, evolution, and cognition (2nd ed.). Oxford University Press. No DOI has been assigned to this book.

Overall, K. L. (2013). Manual of clinical behavioral medicine for dogs and cats. Elsevier. No DOI has been assigned to this book.

Notes on the sources

The figures for Morrill et al. (2022) were checked against the publication: owner survey covering 18,385 purebred and mixed-breed dogs, genotyping of 2,155 animals, heritability of most behavioral traits above 25 percent, share of behavioral variation between individuals explained by breed about 9 percent. Eleven loci showed a significant association with behavior. For aggression-related traits, no meaningful breed effect was found.

Resolving the apparent contradiction with MacLean et al. (2019) rests on the fact that the two studies measure different quantities: heritability between breeds versus the predictive power of breed for the individual. The two findings are compatible.

Both studies rely to a considerable extent on owner reports from questionnaire instruments. For samples of this size that is methodologically unavoidable, but it makes the data susceptible to expectation effects—especially for traits about which widespread breed stereotypes exist. This limitation applies to both sides of the debate.

The refutation of drive reduction rests on Sheffield and Roby (1950)—rats work for a non-nutritive sweet solution—and on Harlow, Harlow and Meyer (1950): four rhesus monkeys, a mechanical puzzle with no reward at all, a decline in performance after food was introduced. Both studies come from general learning psychology and do not concern dogs directly; they are cited here as the historical refutation of the model, not as statements about dog behavior.

The historical account follows Lorenz (1937, 1950), Hull (1943), Lehrman (1953) and Hinde (1956). The critique of the psychohydraulic model is considered settled in behavioral biology; it is directed against the energy-flow model, not against the existence of genetically grounded behavioral tendencies.

The description of the hunting behavior chain and its breed-specific expression reflects established ethological description. It is not supported by controlled comparative studies by breed.

The term modal action pattern follows Barlow (1977) and has largely replaced the older term Erbkoordination, or fixed action pattern, in the scientific literature. The difference lies in variability: modal action patterns are species-specific and genetically pre-shaped, but variable in execution, intensity and completeness.

The statement that performing chasing behavior is itself reinforcing is treated here as intrinsic reinforcement: the performance of the behavior is the reinforcer, not an external consequence. This should be distinguished from autoshaping, or sign tracking, in which a conditioned stimulus that signals a goal is itself approached and engaged with—there, the pairing of signal and reinforcer is constitutive. The two mechanisms are often conflated in popular accounts.

The figures for MacLean et al. (2019) were likewise checked against the publication: behavioral data from more than 14,000 dogs of 101 breeds, breed-averaged genotype data from more than 100,000 loci, high between-breed heritability for 14 behavioral traits, 131 associated genetic variants. The traits with the highest values include trainability, chasing behavior and stranger-directed aggression.

The distinction between wanting and liking follows Berridge and Robinson (1998). The evidence comes from basic research, mostly from experiments in rats with targeted depletion of dopamine. Transferring it to training situations with dogs is an inference from the similarity of the systems involved, not a dog-specific finding.

The observation about play behavior serves as an illustrative counterargument against the reservoir model and is not a study reference.

The case example described is a single observation from practice and is explicitly marked as such.

Frequently asked questions about drive theory in dog training

Why the old concept no longer holds up