Breed and Behavior: Genetic Predisposition versus Individual Variation in Dogs

1. Introduction: The Appeal and Problem of Breed-Based Thinking

1.1 Why This Question Keeps Coming Up

Few claims in dog ownership are repeated as confidently or challenged as rarely as the idea that breed tells you what a dog will be like. It shapes acquisition decisions, insurance premiums, housing rules, shelter placements and in some jurisdictions whether a dog may legally exist.

The evidence base for that confidence is now unusually good, and it does not support the confidence.

1.2 Breed as Shorthand

Sorting dogs into breeds satisfies a deep human pull toward categories, and category labels feel like explanations. At the population level they are not entirely wrong: a random border collie is, on average, more motion-sensitive than a random basset hound. The trouble begins when a population average is read as an individual guarantee – when "this is a family breed" or "that breed is dangerous" is treated as a fact about a particular animal rather than a loose statistical tendency (a dog's actual temperament is an individual matter).

1.3 How to Read the Evidence

The key to this topic is distinguishing two questions that stereotypes fuse. Do breeds differ in behavior on average? Yes, and heritably so. Can you predict an individual dog's behavior from its breed? Largely no. Both answers come from good data, and they are not in conflict – they describe population-level structure versus individual-level prediction. Throughout, this article keeps them apart, because almost every breed myth is a case of treating the first answer as if it settled the second.

1.4 What a Breed Actually Is

The category doing the explanatory work is younger than most people assume. The majority of modern breeds were established in the nineteenth century through closed studbooks and conformation standards — a system that fixed appearance and, as a side effect, reduced genetic diversity within each group.

Before that, dogs were sorted by function and geography, with far more gene flow between types. A breed is therefore a recent administrative boundary drawn across a much older and more continuous population, which is the structural reason it captures behavior poorly.

1.5 Why the Stereotype Feels Confirmed

Three mechanisms make breed expectations self-supporting regardless of their accuracy.

Confirmatory attention: a Labrador that is friendly confirms the rule, a Labrador that is not becomes an exception. Selection of owners: people who want a particular temperament choose the breed reputed to have it, then raise the dog accordingly. And selection of dogs: breeders place puppies with families in light of the same expectations.

None of these requires the stereotype to be true, and all of them produce evidence that it is.

---

2. Heritability and Genetic Architecture of Behavior

2.1 Behavior Is Partly Heritable – but Heritability Is Not Destiny

Canine behavioral traits are partly heritable: genetic variation contributes to differences in fearfulness, sociability, aggression, trainability, and more. Integrating behavioral data from over 14,000 dogs across 101 breeds with genomic data, researchers found high among-breed heritability for fourteen behavioral traits and identified variants in genes expressed in the brain – solid evidence that breed differences in behavior are real and biologically grounded, not merely cultural projection (MacLean et al., 2019). But heritability is a population statistic describing the share of variance attributable to genes; it says nothing about whether an individual's behavior is fixed. Even moderately heritable traits remain highly responsive to environment (experience continues to shape the stress and emotional systems throughout life).

2.2 Polygenic and Probabilistic

Behavioral traits are overwhelmingly polygenic – built from many genetic variants each of small effect, rather than a single "aggression gene" or "trainability gene." This architecture has direct consequences: broad variability within any breed, heavily overlapping trait distributions between breeds, and limited predictive value from genetic background alone. Behavior is best modeled as a probabilistic outcome of many interacting systems, not a trait stamped in by breed identity.

2.3 Which Traits Are Heritable and Which Are Not

Heritability is not uniform across behavior, and the pattern is informative. Surveying owners of 18,385 purebred and mixed-breed dogs and genotyping 2,155 of them, researchers found most behavioral traits heritable above 25 percent — with retrieving the most heritable single trait at 52.5 percent and human sociability the most heritable factor at 67.3 percent (Morrill et al., 2022).

Behaviors related to intrinsic motor patterns and physical traits were more heritable than other behaviors. Emotional and social dispositions can be selected on too — guide-dog and service-dog programs do exactly that. The difference is one of degree and of tractability: a motor sequence is directly visible, easy to judge in a working test and therefore easy to select on, while a disposition has to be inferred from behavior across situations. What breeding shaped most effectively is accordingly the shape of a motor sequence rather than the disposition that stereotypes are built on (as the predatory sequence illustrates directly).

2.4 The Gap Between Heritability and Explained Variation

Two numbers from the same study sit awkwardly together and both are correct. Most traits are heritable above 25 percent — and breed ancestry explained about 9 percent of variation in the measured behavioral traits in that study (Morrill et al., 2022). The figure refers to the owner-reported traits surveyed, not to canine behavior in general.

Heritability describes how much of the variation is genetic. Breed captures only part of that genetic variation, because modern breeds are recent, overlapping subsets of a much older gene pool. A trait can be substantially heritable while breed membership tells you very little about it.

2.5 What the Genome-Wide Analysis Found

Eleven loci were significantly associated with behavior, and characteristic breed behaviors exhibited genetic complexity rather than resolving into identifiable variants (Morrill et al., 2022).

The scale of the shortfall is worth stating. Six associated loci accounted for 42.7 percent of the genetic component of dog sociability — a heritable trait for which most of the genetic contribution remains unlocated, distributed across many variants of small effect.

---

2.6 Why Polygenic Architecture Matters Practically

The genetic structure has a direct consequence that is easy to state and often missed. When a trait depends on many variants of small effect, siblings receive different combinations — and the resulting distributions overlap heavily between any two groups.

Two breeds can differ reliably in their averages while the great majority of individuals from each fall inside the other's range. That is what a small between-group effect looks like at the individual level, and it is why a population-level statement cannot be converted into an individual prediction without losing almost all of its accuracy.

---

3. What Breeds Actually Encode

3.1 Functional Predispositions, Not Fixed Programs

Selective breeding shaped dogs for roles – herding, guarding, hunting, retrieving, companionship – and in doing so tuned sensory sensitivities (motion detection in herding breeds), motivational systems (prey drive in hunters), and responsiveness to human direction. But these are predispositions, not scripts. A border collie's heightened sensitivity to movement can express as elegant herding, as compulsive car-chasing, or as anxious over-arousal, depending on environment and learning history (how arousal is regulated determines which way a predisposition breaks). Breed loads the dice; context rolls them.

3.2 Species-Wide, Not Breed-Specific

Much of what makes dogs remarkable is not breed-specific at all but a species-wide adaptation. The ability to read human gestures, gaze, and expressions (the social cognition dogs share across breeds) evolved in the dog lineage broadly, not in particular breeds. This matters for the stereotype debate: the traits people most value in dogs – their attentiveness to us, their trainability in the broad sense – rest largely on a common inheritance, which is one more reason breed is a coarse instrument for predicting the individual.

3.3 Where the Behaviors Came From

The authors propose that behaviors perceived as characteristic of modern breeds derive from thousands of years of polygenic adaptation predating breed formation, with modern breeds distinguished primarily by aesthetic traits (Morrill et al., 2022).

That reverses the usual assumption, and it needs qualifying. Many modern breeds were strongly shaped by closed studbooks and conformation standards, while behavioral selection varied greatly between breeds and between lines within them — working lines in herding, retrieving and protection breeds continued to be selected on performance. The behavioral tendencies were already distributed through the population — which is why they reappear in mixed-breed dogs according to ancestry rather than being confined to the breeds that claim them.

3.4 What This Means for Breed Health

Where selection did act decisively and recently, it acted on morphology — and there the consequences are measurable. Conformational selection produces documented health outcomes in a way behavioral selection does not (with the flat-faced breeds the clearest case), and pigmentation-linked selection carries its own risks (as congenital deafness illustrates).

The asymmetry is instructive. Breed predicts physical outcomes considerably better than behavioral ones, which is the opposite of how breed information is typically used.

---

4. Gene–Environment Interaction

4.1 Behavior as a Dynamic System

Genes influence sensitivity to stimuli, arousal thresholds, and learning tendencies, but they do not dictate outcomes; behavior emerges from their continuous interaction with environment. Early socialization, maternal care, training methods, stress exposure, and daily human interaction all sculpt neural development and behavioral expression. Early-life stress can raise stress reactivity; enriched environments can build cognitive flexibility; consistent training can strengthen impulse control – all operating on the same genetic starting material to produce very different adults. The sensitive period of early development is a particularly powerful window for this shaping.

4.2 Maternal and Early Effects

The developmental contribution is not vague hand-waving; it is measurable, and sometimes counterintuitive. Following guide-dog puppies from birth, researchers found that higher maternal investment early in life was associated with lower program success, while a mothering style that required more effort from the pups produced more successful, more resilient adults (Bray et al., 2017). Early environment shapes outcomes in ways that no breed label could anticipate – a reminder that the same genes, raised differently, yield different dogs.

4.3 Why the Interaction Is Not a Hedge

"Genes and environment interact" reads as a way of avoiding the question, and in this case it is the finding rather than a diplomatic evasion.

The same genetic predisposition produces different outcomes depending on what happens to the animal, and the effects are large enough to swamp the breed signal. That is why 9 percent is the figure for breed and not 40 — the environmental contribution is not noise around a genetic signal but a comparable source of variation in its own right.

4.4 What Shapes the Outcome

Several routes are documented in their own right. Early acoustic and social exposure during the developmental window (with the mechanism set out separately). Training approach, where aversive methods carry measurable physiological and behavioral costs (as documented separately). Chronic stress exposure, which alters the systems that generate the behavior (with the physiology described elsewhere).

And physical state, which is the one most often left out: pain features heavily in behavior caseloads and changes what a dog does regardless of what its pedigree suggests (with the evidence set out separately).

---

4.5 The Direction of Effects Is Not Always Obvious

The maternal-investment finding is worth dwelling on because it inverts an intuition that feels unassailable. Following guide-dog puppies from birth, higher maternal investment early in life was associated with lower program success, while a mothering style requiring more effort from the pups produced more successful adults (Bray et al., 2017).

One interpretation offered is that mild, manageable challenge during development builds the capacity to handle challenge later, while its absence does not. Whether that generalizes beyond guide-dog work has not been established, and the finding is a useful corrective to the assumption that more care is always better.

4.6 Why This Undercuts Breed Prediction Further

If maternal style measurably affects adult outcomes, then two puppies from the same litter — same breed, same parents, same genetics to within sibling variation — can diverge on the basis of how their dam handled them.

That variation belongs to the large share of individual differences not captured by breed membership, and it occurs before the puppy has met its owner.

---

5. Epigenetics: Where Experience Meets Gene Expression

Epigenetic mechanisms bridge genes and environment. Through processes such as DNA methylation, experience can alter gene expression without changing the DNA sequence, so early experiences can leave long-term biological marks and behavioral tendencies can be shaped across the lifespan (how experience reaches the genome in dogs). This reinforces the article's central theme: gene expression is dynamically regulated rather than static, so behavior is context-dependent and modifiable rather than fixed at conception – held, as always, with the caveat that direct canine epigenetic-behavioral data remain limited.

6. Within-Breed Variation: The Nine-Percent Finding

This is the empirical heart of the matter. In the largest study of its kind, integrating genome data with owner surveys for thousands of purebred and mixed dogs, researchers found that breed accounts for only about 9% of the behavioral variation among individual dogs (Morrill et al., 2022). Some traits show more breed signal than others – biddability (responsiveness to human direction) is among the more breed-associated – but for the trait people most fear, agonistic threshold (behaviors labeled "aggression"), so little of the variation tracks breed that breed is of very limited predictive value for an individual dog. Within-breed distributions are wide, between-breed distributions overlap heavily, and the upshot is blunt: knowing a dog's breed tells you far less about its behavior than directly assessing the dog. Breed averages do not describe individuals, and individual differences routinely outweigh breed differences (which is why measuring the actual dog matters more than its label).

6.1 What Nine Percent Leaves Unexplained

If breed accounts for 9 percent, then 91 percent comes from elsewhere — and the study identifies part of it. For some traits, age was the better predictor: younger dogs scored higher on toy-directed motor patterns than older ones, regardless of breed (Morrill et al., 2022).

Age, sex, individual genetic variation outside breed-defining regions, early environment, learning history and current physical state all sit in that 91 percent. Several have their own literatures in this library, and none of them appears on a pedigree.

6.2 Why Mixed-Breed Data Were the Key

The methodological move that made the finding possible was including mixed-breed dogs. Admixture patterns in mixed-breed dogs revealed breed propensities (Morrill et al., 2022) — meaning that if a Labrador tendency is genuinely a Labrador tendency, it should appear in proportion to Labrador ancestry in dogs that are part Labrador.

Some tendencies behaved that way and the effects were small. Studying purebreds alone cannot separate a genuine breed propensity from the expectations of people who chose that breed, and adding mixed-breed dogs is what allows the two to come apart.

6.3 The Prediction Problem in Practice

A statistical association of this size supports statements about groups and not about individuals — the same distinction that governs early behavioral assessment (where heritable traits also failed to predict the individual).

Wilsson and Sundgren's service-dog data make the point from the other direction: behavioral traits with medium-high to high heritability at eight weeks nonetheless showed negligible correspondence with adult working suitability. Heritable does not mean predictable, and the two are confused constantly.

---

6.4 What the Nine Percent Is Not

Three misreadings circulate and each inverts something the study says.

It does not mean breeds do not differ. They do, heritably, and the same paper documents it. Nine percent of individual variation is a small share and not zero, and at population level the differences are measurable.

It does not mean genetics contribute little. Most traits were heritable above 25 percent. Breed is a poor proxy for the genetics, not a demonstration that genetics are unimportant.

It does not mean all dogs are equivalent. Individual differences are large — larger than breed differences, which is the point. A dog is a specific animal, not an average of anything.

---

7. Cognitive and Emotional Systems as Modulators

Genes and environment do not act on behavior directly but through cognitive and emotional systems – attention, learning, arousal, and emotional regulation – that continuously shape how a dog responds. A dog with a low arousal threshold may present as "reactive" whatever its breed (reactivity is a nervous-system property, not a breed trait); impaired emotional regulation can drive anxiety-related behavior (with its own neurobiology); and a strong reinforcement history can override an initial predisposition entirely (learning can rewrite what genetics merely suggested). Behavior is ongoing modulation, not the readout of a fixed trait.

7.1 Why Two Dogs of One Breed Differ So Much

Within-breed variation exceeds between-breed differences for most behavioral traits, and the reasons are identifiable rather than mysterious.

Polygenic architecture means siblings inherit different combinations from the same parents. Developmental timing means the same experience lands differently depending on when it occurs. Learning history diverges from the first week. And current state — pain, fatigue, arousal, hunger — modulates expression on any given day (with arousal constraining what is accessible).

7.2 The Owner Expectation Problem

One source of apparent breed differences is worth naming because it operates invisibly. An owner who acquires a breed with a reputation handles the dog in light of that reputation, and the handling shapes the behavior.

The mechanism runs through four steps, each of which produces apparent breed differences without any genetic contribution. Selection: people who want a particular temperament acquire the breed reputed to have it. Rearing: they raise and train the dog in light of that reputation. Interpretation: the same behavior is read as "typical" in one breed and as an aberration in another. And reporting: the questionnaire answer records the interpretation rather than the behavior.

That is not a small effect and it is difficult to separate from genuine breed propensity in owner-reported data — which is precisely why the mixed-breed design in section 6.4 mattered.

---

8. Function Versus Form

8.1 Two Kinds of Breeding

The distinction that organises most of the confusion is between breeding for what a dog does and breeding for what it looks like.

Working selection judges the animal on performance and keeps the behavior that produces it. Conformation selection judges the animal against a written standard describing appearance. Both are called breeding and they act on different traits with different consequences.

8.2 Why the Split Matters Within Breeds

Many breeds now contain both populations, separated for decades. A working-line Border Collie and a show-line Border Collie share a name, a studbook and increasingly little else behaviorally, because the two lines have been selected on different criteria since the split.

Breed-level statements therefore average across populations that were deliberately made different, which adds variance that no genomic analysis treating breed as one category can resolve.

8.3 What Working Selection Actually Captures

Where behavioral selection has been sustained and specific, it does produce results — and the traits it captures are the motor patterns rather than the temperament. A line selected for retrieving reliably produces retrieving, which is consistent with retrieving being the most heritable single trait measured (Morrill et al., 2022).

What it does not produce with the same reliability is a disposition. Selection for working ability in guide dogs, service dogs and detection dogs continues to reject a substantial proportion of animals from lines specifically bred for the purpose, which is the clearest available evidence of how far heritability falls short of individual prediction.

8.4 The Aesthetic Cost

Where conformation selection has been intense, the health consequences are documented and substantial — which is the asymmetry noted in section 3.4, seen from the other side. Selection on appearance produced measurable outcomes quickly; selection on behavior did not produce comparable predictive power over the same period.

---

9. Research Gaps and Methodological Challenges

The evidence is strong but has real limits worth naming.

Population versus individual. The central results describe variance across populations; they establish that breed predicts the individual poorly, not that genetics is irrelevant to any dog.

Measurement. Much behavioral data comes from owner questionnaires, which are practical but subjective, and different instruments recover different structures (operationalizing behavior is genuinely hard).

Visual breed guessing. People – including professionals – are unreliable at identifying breed by appearance, and morphology maps loosely onto ancestry, so "breed" as applied in daily life is often wrong before behavior even enters the picture.

The heritability–prediction gap. That a trait is heritable and that breed differences exist (MacLean et al., 2019) can coexist with breed being a weak individual predictor (Morrill et al., 2022); conflating the two, in either direction, is the recurring error – and the same caution applies to marketed genetic tests, which predict appearance far better than behavior.

Beyond breed genetics. Coat-color and structural genes carry their own welfare implications independent of behavior (as with the merle gene), a reminder that "breed genetics" is not a single thing.

9.1 What the Owner-Report Basis Limits

The large-scale genomic work rests on owner questionnaires, which are the only way to reach eighteen thousand dogs and carry a known limitation: they measure owner perception of behavior rather than behavior directly (with the measurement problem set out separately).

That cuts in a specific direction here. If owners perceive their dogs in line with breed expectation, owner-reported data would overstate breed differences rather than understate them — which makes the 9 percent figure an upper bound rather than a conservative estimate.

9.2 What Has Not Been Measured

Working lines and show lines within the same breed have diverged behaviorally, and most genomic work treats breed as a single category. Village dogs and free-ranging populations, which represent the majority of the world's dogs, are barely represented. And no study has followed the same dogs across development while genotyping them, which is what would separate genetic propensity from developmental trajectory.

---

9.3 What a Better Study Would Look Like

The design that would settle the remaining questions is describable and has not been run. It would genotype dogs rather than relying on owner-reported breed, follow the same animals from placement into adulthood, record the rearing and training environment systematically, and measure behavior by direct observation rather than by questionnaire.

Each element exists somewhere in the literature. None of the large studies has all four, which is why statements about the relative contribution of breed and environment remain estimates rather than measurements.

9.4 Why It Probably Will Not Be Run Soon

Cost is the obvious barrier and not the only one. Direct behavioral observation across thousands of dogs in their homes is impractical, and the alternative — bringing dogs to a testing facility — introduces the situation-specificity problem that undermines single-occasion assessment generally.

What is achievable is narrower: more genotyping, better environmental records, and outcome measures targeted at specific behaviors rather than at broad personality dimensions.

---

10. The Identification Problem

10.1 A Prior Question

Everything above assumes the breed is known. For a large share of dogs it is not, and the assumption fails before the prediction does.

Roughly half the dogs in the Morrill sample were mixed-breed (Morrill et al., 2022), and for those animals a breed label is an inference from appearance rather than a record.

10.2 How Poorly Visual Identification Performs

The failure has been measured. Comparing adoption agency breed designations against DNA analysis for 20 mixed-breed dogs from 17 agencies, only four of the sixteen dogs assigned a predominant breed were in fact predominantly that breed (Voith, Ingram, Mitsouras & Irizarry, 2009).

A follow-up using video of the same dogs tested 923 canine professionals — shelter staff, veterinarians, animal control officers, dog club members. For 14 of the 20 dogs, fewer than half of the participants gave a visual identification matching the DNA analysis, and inter-observer agreement was low.

10.3 Why Appearance Is a Poor Guide

The reason follows from section 3. Coat color, ear set, muzzle length and body shape are governed by relatively few loci with large visible effects, and they assort independently of the thousands of variants underlying behavior.

A dog that looks like one breed may carry little ancestry from it, and a dog carrying substantial ancestry from a breed may not look like it. That is why genotyping was necessary for the analysis at all — visual assignment would have introduced exactly the error the study was designed to avoid.

10.4 What This Does to the Stereotype

The chain has two weak links rather than one. Breed must first be correctly identified, and it must then predict behavior. The second link accounts for 9 percent; the first is not measured at all in most everyday judgments.

A prediction made from a visual breed guess is therefore weaker than the 9 percent figure suggests, because that figure was calculated using genetic ancestry rather than appearance.

10.5 The Consequence for Records

Shelter intake records, veterinary histories and bite statistics routinely carry breed designations assigned by eye. Any analysis built on those designations inherits the identification error, which is one reason breed-based incident data are difficult to interpret (with the broader problem of aggression statistics documented separately).

---

11. Practical Implications: Beyond Stereotypes

Overreliance on breed assumptions produces predictable errors: misreading a dog's behavior, choosing inappropriate training strategies, and holding unrealistic expectations. Assuming a "family-friendly breed" needs little training can lead to a badly under-structured dog; labeling a dog "genetically aggressive" can obscure the modifiable environmental and medical factors actually driving the behavior (including that so-called aggression is weakly breed-linked and often has other causes, from fear to pain). The same categorical thinking underlies discredited frameworks like breed-based "dominance" (a concept that failed scientific scrutiny). A better approach centers the individual: direct behavioral assessment, functional analysis of what actually drives a given behavior, and context-specific intervention. This is not a rejection of biology – a high-drive working breed genuinely does tend to need more physical and mental outlet – but a refusal to let a label stand in for the dog.

11.1 Where Breed Information Is Genuinely Useful

Three places, and none of them is temperament prediction.

Physical health screening. Breed-linked conditions are real, well documented and worth testing for — conformational, ocular, cardiac, orthopaedic and, in specific pigmentation patterns, auditory.

Anticipating motor patterns. The most heritable behaviors are the intrinsic motor sequences, which means a herding breed is more likely to show the eye-stalk-chase configuration than a retriever is. That is a statement about which behaviors are available, not about whether this dog will perform them (with the sequence described separately).

Setting expectations about size and physical needs. Adult weight, coat maintenance and exercise capacity are considerably more predictable from breed than anything behavioral.

11.2 What to Ask Instead

Where breed is the question being asked, four others carry more information about the individual dog: what the parents were like as adults, how the dog was reared before placement, what it has been exposed to since, and what it does now in the situations that matter.

The last is the only one that is directly observable, and it is the one most often skipped in favour of the label.

11.3 Where the Stereotype Does Real Harm

Breed-specific legislation rests on a premise this literature bears on directly: that breed identity predicts individual risk well enough to justify categorical restriction. Two findings in this article speak to it — the small share of behavioral variation attributable to breed, and the unreliability of visual breed identification on which such rules are usually enforced (Voith et al., 2009).

Bite risk is shaped by context, learning history, physical state and the specific situation far more than by breed membership (with aggression being situation-specific rather than trait-like), and a substantial share of severe cases involve circumstances that no breed rule addresses (as the bite data on resource guarding show).

---

12. Summary at a Glance

Breed explains 9 percent of behavioral variation in individuals — Across 18,385 surveyed dogs with 2,155 genotyped (Morrill et al., 2022).

Most traits are nonetheless heritable above 25 percent — Both figures are correct; breed captures only part of the genetic contribution (Morrill et al., 2022).

Retrieving is the most heritable single trait at 52.5 percent — And intrinsic motor patterns are more heritable than other behaviors (Morrill et al., 2022).

The behaviors predate the breeds — The behavioral tendencies associated with modern breeds reflect older functional adaptations, while modern breed formation also involved strong selection on appearance and closed breeding populations (Morrill et al., 2022).

Eleven loci reached significance, and most of the genetic contribution is unlocated — Six loci accounted for 42.7 percent of the genetic component of sociability (Morrill et al., 2022).

Mixed-breed data were what made the finding possible — Admixture patterns reveal breed propensities in a way purebred-only samples cannot (Morrill et al., 2022).

Maternal style matters and not in the expected direction — Higher maternal investment was associated with lower guide-dog program success (Bray et al., 2017).

Breed predicts physical outcomes better than behavioral ones — Which is the reverse of how breed information is typically used.

---

12.1 For Anyone Choosing a Dog

The evidence supports a reordering rather than a rejection. Breed is worth knowing for physical health screening, for anticipating which motor patterns are likely to be in the repertoire, and for size and care requirements.

For temperament, the informative sources are the adult behavior of the parents where it can be observed, the rearing environment before placement, and — for an adult dog — what the animal actually does in the situations that matter. Each of those carries more signal about this individual than the label does, and the last one requires no inference at all.

---

13. Conclusion: Breed as One Factor Among Many

Breed matters – but it is one component in a complex system, not the system itself. Selection left real, heritable behavioral tendencies at the population level (MacLean et al., 2019), yet breed explains under a tenth of the behavioral variation among individual dogs, and next to none of the variation in aggression (Morrill et al., 2022). Behavior emerges from the interaction of many small-effect genes with environment, development, and lifelong experience, channeled through cognitive and emotional systems that keep it modifiable. The scientifically honest stance shifts the focus from categorical assumption to individual analysis – which improves training outcomes, sets fairer expectations, and, not least, protects dogs from being judged by their label rather than their nature.

13.1 What to Take Away

Breed is real, heritable and informative at the population level. It is a weak instrument for predicting the animal in front of you, and it is a considerably better one for predicting that animal's physical health risks.

The practical consequence is not that breed should be ignored, but that it belongs in the same category as any other probabilistic risk factor: useful for framing what to look for, insufficient for concluding what is there.

---

Key Insights (Takeaways)

• Breed explains 9 percent of behavioral variation in individuals. Across 18,385 surveyed dogs with 2,155 genotyped, most behavioral traits were heritable above 25 percent while breed accounted for less than a tenth of individual variation (Morrill et al., 2022). Both figures are correct: breed captures only part of the genetic contribution, because modern breeds are recent subsets of a much older gene pool.

• What breeding shaped best is motor patterns, not temperament. Retrieving was the most heritable single trait at 52.5 percent, and behaviors related to intrinsic motor patterns were more heritable than others (Morrill et al., 2022). The traits stereotypes are built on — friendliness, aggression, trainability in the broad sense — are not the ones selection captured most effectively.

• The behaviors predate the breeds. The authors propose that characteristic breed behaviors derive from polygenic adaptation over thousands of years, while modern breed formation involved strong selection on appearance and closed breeding populations (Morrill et al., 2022). Behavioral selection nonetheless continued in working lines within many breeds.

• Breed predicts physical outcomes far better than behavioral ones. Conformational and pigmentation-linked health risks are well documented and breed-specific, while behavioral prediction from breed is weak. That is the reverse of how breed information is usually applied.

• Heritable does not mean predictable. Traits can be substantially heritable at the population level and still fail to forecast an individual, which is the same result found in early behavioral assessment, where heritable puppy-test traits showed negligible correspondence with adult outcomes.

• Breeds encode predispositions, not fixed programs. A herding breed's motion sensitivity can become herding, chasing or anxiety depending on context, and gene–environment interaction — including counterintuitive maternal effects (Bray et al., 2017) — keeps behavior modifiable.

• Assess the individual, not the label. Breed can inform general needs, and reliance on stereotypes causes misread behavior, wrong training and unfair judgments — with "genetically aggressive" in particular often masking fear, pain or environmental causes that are treatable.

References

Bray, E. E., Sammel, M. D., Cheney, D. L., Serpell, J. A., & Seyfarth, R. M. (2017). Effects of maternal investment, temperament, and cognition on guide dog success. Proceedings of the National Academy of Sciences, 114(34), 9128–9133. https://doi.org/10.1073/pnas.1704303114

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: Biological Sciences, 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., Alonso, J., Noh, H. J., Johnson, J., Koltookian, M., Lieu, C., Buckley, R. M., Ostrander, E. A., Ruple, A., … 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., & Jagoe, J. A. (2017). Becoming a dog: Early experience and the development of behavior. In J. Serpell (Ed.), The Domestic Dog: Its Evolution, Behavior and Interactions with People (2nd ed., pp. 93–117). Cambridge University Press.

Voith, V. L., Ingram, E., Mitsouras, K., & Irizarry, K. (2009). Comparison of adoption agency breed identification and DNA breed identification of dogs. Journal of Applied Animal Welfare Science, 12(3), 253–262. https://doi.org/10.1080/10888700902956151

Wilsson, E., & Sundgren, P.-E. (1998). Behaviour test for eight-week old puppies — heritabilities of tested behaviour traits and its correspondence to later behaviour. Applied Animal Behaviour Science, 58(1–2), 151–162. https://doi.org/10.1016/S0168-1591(97)00093-2