1996, V. L. Voith and P. L. Borchelt, Eds., Readings in Companion Animal Behavior, Veterinary Learning Systems, Trenton, NJ. pp 9-18.


Biologic Bases of

Behavior of Domestic

Dog Breeds

Hampshire College

Amherst, Massachusetts

Raymond Coppinger, PhD, Lorna Coppinger, AB, MS





Behavior is the functional component of evolutionary change.  How well an animal runs is the selective force, not its legs.  Paleontologists study the evolution of hard parts because those are what fossilize.  Studying changes in femur lengths, however, leads to the misconception that it is legs that evolved, rather than running or jumping.  For biologists, the evolution of dog behavior is found in the mechanisms of evolutionary change from the antecedent wolf behavior.

Studying the evolution of dog behavior implies some assumptions.  First, one assumes that something called dog behavior can be delineated and that it varies somewhat among dogs.  People often discuss dog behavior or dog training as if it were the same for all dogs.  Second, the origin of some of these variations is assumed to be genetic.  What evolves is the genotype, the specific arrangements of genes.  The only material that can be passed from generation to generation is genes. 



Professional dog breeders and trainers believe that inherent differences in behavior exist among dog breeds.  Breeders select animals that display the breed-typical behaviors, whereas trainers direct the expression of those in­nate behaviors.  For example, trainers do not train border collies to eye sheep, or setters to point birds, or retrievers to retrieve.  These behaviors emerge spontaneously during ontogeny (i.e., the course of the life of an indi­vidual).  Training commences after the emergence of the innate behavior; the trainer teaches the dog when and where to display the behavior.

Breeds are often classified by the job they were selected to perform, such as hunting, herding, or guarding.  They are further subclassified as coon-hounds, pointers, or upland retrievers; heelers, headers, or huntaways; and people-guardians or flock-guardians.  It is assumed that each breed has a unique set of behaviors that endow the members such that no other animal could be taught to perform the task as well.  Not everybody agrees.  Black and Green hypothesized that social attachment and reduced interspecific aggression between dogs and sheep could be achieved using dogs of any lineage if they were raised properly.  These conclusions were based on observations of Navajo dogs that were used to guard sheep but sup­posedly lacked long-term selection preparing them to perform this task.



Selection in dog breeding differs from Darwinian or natural selection.  Darwinian change results from dif­ferential survival and reproductive success among in­dividuals in a geographically isolated population; In contrast, dog breeds are created by humans selecting dogs of a particular phenotype (i.e., with a particular set of observable characteristics and that perform a particular task) and separating them from the rest of the canine population for breeding purposes.  Breeders identify a few single-gene characteristics, such as coat color or achondroplasia (short limbs), as a breed marker, which all animals of the breed must have– re­gardless of the adaptiveness of the characteristic.

Breeds are hybrid saltations, often perfected in a few generations.  Darwinian selection takes longer, perhaps millions of years.  Crossbreeding creates not averages but new phenotypes that can be maintained as new breeds.



Dogs of all breeds have the same karyotype (num­ber and shape of chromosomes) and produce repro­ductively viable hybrid offspring.  The differences among breeds are not gene differences (i.e., different arrangement, ordering, or number of genes).  Pheno­typic (including behavior) differences among dog breeds are produced by slight allelic differences in gene products in three categories:  ontogenetic onset (tim­ing of the activation of the gene), quantity of gene product, and ontogenetic offset (timing of deactiva­tion of the gene).

No gene arrangement has yet been identified that would permit identification of any breed or any par­ticular behavior.  To our knowledge, attempts to detect "fingerprint" genes for pit bull terriers, or for aggres­sive behaviors in pit bull terriers, have so far failed.  Mitochondrial DNA investigations give no clue to re­latedness between breeds, nor do they group breeds in any of the well-known categories, such as sporting, working, or terrier.  Rather, they reveal what has al­ways been obvious but is usually ignored:  dog breeds are local and temporal phenomena recently derived by crossing other breeds or local variants.  Any handbook of dog breeds tells what breeds were used to create a new breed at the turn of the twentieth century or what breeds were used to improve an old breed.

In addition to chromosomal and mitochondrial similarities, most members of the genus Canis have conservative phenotypic similarities (e.g., palate to skull length2) and also fundamental similarities in so­cial and reproductive behaviors.35  This information suggests that despite the claim that Canis familiaris is one of the most varied species on earth, the evolu­tionary distance between dog breeds or within the genus is short– perhaps even nonexistent.



Differences in behavior among dog breeds are often given as differences in temperament.  The com­bined results of these studies indicate some variation in emotionality, vocalization, activity, problem-solv­ing, reaction to human handling, and trainability.  Mahut, Borchelt,2 and Hart and Hart13 reported breed differences in temperament among pet dogs raised in private homes.  They assumed that these dif­ferences were largely innate because pet dogs of all breeds are raised similarly.



Motor pattern differences among dog breeds have also been reported.14  Studies of motor patterns con­centrate on the kind of behavior, its frequency, and the sequence in which the motor patterns appear.

The kind or quality of the behavior refers to the ac­tion the animal performs in reaction to a specific in­ternal or external signal.  Ideally, this would be done by sequencing anatomical events (e.g., muscle move­ment); but usually, the composite behavior is simply described:  "eye" in herding dogs, "point" in pointers, heeling in heelers.  Male dogs raise a hindlimb during urine marking.  Because their relatives, male wolves, also display this behavior but members of other simi­larly shaped species (e.g., bears) do not, the behavior is assumed to be not only an inherited motor pattern but a taxonomic characteristic.

Two species, or two breeds, may differ not in the kind of motor patterns but in their frequency of ex­pression.  Wolves rarely bark, whereas barking is ubiq­uitous among dogs.  Nevertheless, greyhounds in Ar­gentinean villages rarely bark at strangers, whereas maremmas in Italian villages rarely refrain from bark­ing at strangers.  Behaviors also change frequency on­togenetically.  Wolves bark more frequently as young adolescents and rarely as adults.  Dog pups bark first at 7 days of age and frequently thereafter throughout ontogeny.

The behavior sequence in which motor patterns appear differs between species and between breeds.  Dogs that work with sheep provide an excellent ex­ample.4  Among border collies (a sheep-herding breed) a predatory eye-stalk motor pattern follows a


Figure 1:  Border collie showing "eye" to sheep.  This be­havior, with its characteristic "stalk," causes sheep to bunch and move away from the dog.


predatory orientation 85% of the time, whereas an investigative or a social sequence commonly followed such an orientation among livestock-guarding dogs.  Border collies displayed eye-stalk-chase behaviors both inter- and intraspecifically.  Some authors believe that the eye-stalk-chase motor patterns are homolo­gous to those of wolves.  Most European sheep-guard­ing breeds (e.g., Great Pyrenees, maremma, and Ana­tolian shepherd) do not display these predatory sequences (Figure 1).



Environmental factors cause behavioral variation among dogs.  Scott5 and Fox6 found that puppies undergo a critical period of socialization when they are approximately between 3 and 16 weeks of age.  According to these authors, dogs are susceptible to permanent alterations of behavior as a result of envi­ronmental stimuli during this stage of development.  Cairns and Johnson18 and Fox5 found that social con­ditioning with another species often results in unusu­ally low levels of aggression toward the other species and development of a social attachment, even be­tween species that might normally have a predator-prey relationship (e.g., dogs and sheep).

The critical period hypothesis is evidence for genetic predispositions for development of specific behav­ioral repertoires (see the article by Estep).  For exam­ple, we raised livestock-guarding dogs with sheep and minimized pup-human contact during the critical period when social motor patterns emerge.  Some of these pups became good flock guardians but were permanently shy around people.  These dogs were dif­ficult to manage, but they were less likely to harass strangers in or near the pasture.  How much handling by humans dogs need in order not to become people-shy is also a genetic variable.  We have had strains of dogs predisposed to this shyness; and within those strains, siblings raised in the same pen can be shy or not shy, with the same amount of human contact.



To understand any behavior, whether motor pat­terns, critical periods, or temperament differences, one must discern the underlying biologic determi­nants.  Differences in innate behavior between breeds can ultimately be traced to biologic differences.  These may be gross differences in size or shape or minute differences in the chemical structure of a neu­rotransmitter or a hormone.

A breed is defined by a structural standard (e.g., a Siberian husky should not weigh more than 60 lbs [27 kg]).  This standard was derived through a process of selecting breeding stock from among dogs that performed the task of pulling sleds.  Thus, innate be­havior implies the structural capacity to perform.

Behavior is shape and size moving through space and time.  This definition may seem simplistic.  The evolution of breeds, however, is the selection among allelic variations of genes; but genes simply produce enzymatic reactions for the production of proteins, which give the organism shape and size.  The genetic basis of behavior is a gene-encoded shape that allows the animal to perform in a particular way.  The shape allows, but also limits, the performance.  In the fol­lowing example, sled dogs provide a simple illustra­tion of how size (and consequently shape) determines a particular kind of performance.

Racing sled dogs are the world's fastest quadrupeds for distances over 17 km.  Dogs on championship teams weigh less than 25 kg.  Sled dog drivers would like to use bigger dogs, because big dogs have longer gaits (more reach) and cover more ground with each stride, but bigger dogs do not seem to do well.  Exper­iments by Phillips and coworkers9 showed that big­ger dogs suffered greater heat stresses because surface-to-volume ratio is geometric and not linear with size change.  Big dogs have more heat-storing capacity be­cause they have a larger volume compared to radia­tive surface area, which prevents the dissipation of heat during fast long-distance running.

Greyhounds (30 kg), which are among the fastest of dogs, are limited to sprints.  Their racetracks are ei­ther 3/8 or 5/16 of a mile.  Like human sprinters, they fin­ish their races in a matter of seconds (31 seconds is a common finishing time); so heat load does not limit performance.

Sled dogs have walking gaits (pace, trot, or lope), in which at least one paw is always on the ground. 

Greyhounds have running gaits, a series of leaps into the air with all four feet off the ground.  Running (leaping) is a fault in sled dogs.  Dogs that run (called floaters) are at a severe disadvantage because the back-strap pulls them off balance when all four feet are off the ground (Figure 2).

Racing sled dogs have gait and size characteristics that allow them to run long distances efficiently.  These characteristics are innate because they are ge­netically (allelically) determined.  The shape of their shoulders and pelvic girdles and the length of their back and their size are the products of their genes.  All those differences in shape between the greyhound and the sled dog allow each of them to perform ex­ceptionally well in their own environment but limit their performance in the other breed's environment.

One cannot teach the wrong breed to excel at the wrong task.  A greyhound cannot pull a sled at racing speeds for long distances.  Not only would the grey­hound overheat, but it would be awkward and ineffi­cient in harness because of its gait.  It would be diffi­cult (and probably uncomfortable for or even harmful to the dog) to train a greyhound to run with at least one foot on the


ground.  Similarly, no amount of training or conditioning would enable a basset hound to achieve the speeds of a greyhound.  Struc­tural variation is the biologic basis of behavioral variation.



Breeds differ not only in gross conformation, size, muscle, bone, and hair, but also neurologically and hormonally.  Arons and Shoemaker10 searched for the structural differences (quantity and anatomical distri­bution of neurotransmitters) underlying differences in motor patterns among breeds.  They found differ­ences in motor pattern and motor sequence among sheep-herding dogs, sheep-guarding dogs, and Siberi­an huskies.  They then attempted, with some success, to correlate these breed differences with neurotrans­mitter patterns in the brain.

Their data suggest that breed-typical motor pat­terns reflect differences in the distribution and quan­tity of neurochemicals.  Their findings not only sup­port the assumption of breed-typical anatomical differences but are parsimonious with the direction of those differences.  Low levels of dopamine correlated with the more lethargic temperament of livestock-guarding dogs.  This contrasted with higher dopa­mine levels in the hyperactive border collies and huskies.  Additionally, a lack of this neurotransmitter in a particular region of the brain resulted in a lack of excitability in organs stimulated by that portion of the brain.  The animal therefore cannot be expected to increase the rate of motor function through train­ing or conditioning.  Anatomy, whether structural or chemical, not only allows breeds to perform in a par­ticular way but also limits behavior.



If a dog's anatomy changes over its lifetime, then the “innate” behavior must also change.  Dog behav­ior, then, must also be defined ontogenetically.

As a dog changes anatomically (its size, shape, hor­mones, and neurons) from neonate to puppy to juve­nile to adult, it goes through stages of behavioral changes.  The synonyms grow up and develop mislead­ingly imply that ontogenetically a dog goes from a primitive to a more advanced state.  In fact, neonates are behaviorally complex– just as or even more com­plex than adults.  Many people think of ontogenetic changes as growth or maturation, but the changes might better be thought of as metamorphosis from one complex stage to another.

Each new structural organization of the animal produces a new set of innate behaviors.  Thus, one must specify the dog's current ontogenetic stage when discussing its behaviors.  Too often, people think of dog behavior as adult dog behavior, assuming that the puppy "develops" into the adult, that puppies are less than perfect, and that the purpose of life is to mature into an adult.



Alarm calls by neonates and puppy retrieval behav­iors by dams are innate, reflexive, hard-wired, complex dog behaviors that are ontogenetically limited (i.e., limited to a particular time in the life cycle).  Neonates of all breeds can give species-specific alarm calls from the moment they are born.  One is a care-soliciting call, signaling "I am lost." This call is an innate behavior, a specific motor pattern.  It is stereotyped and can be sonographically measured and described.  It has a single “meaning” and is only given in one environmental context.  Its onset is from the moment of birth.  There is no learning period.  We observed a puppy still at­tached to its placenta giving this call, having been abandoned by its mother in a pasture.  She had depart­ed to the barn to have the rest of the litter.  Puppies continue to emit the call until they are retrieved or un­til they weaken and perhaps die.  The call is unchang­ing from the first to the last vocalization.

Offset (cessation) of the "lost call" occurs before 5 weeks of age.  Neither practice nor reward prolongs the behavior into adolescence or adulthood.

Puppy retrieval by mother dogs is similarly hard-wired.  In our experiments, the onset of puppy re­trieval was after the last puppy was born.  Onset varies in other species.  Experiments with rats showed that they can exhibit pup retrieval several days before par­turition.  This motor pattern, displayed only by a dam (in dogs), results from the specific neonatal vocaliza­tion given by a pup.  One of our border collies re­trieved and returned to the nest a small tape recorder producing this vocalization.  Other environmental signals also initiate the response, for dams sometimes retrieve dead puppies.

Offset of puppy retrieval averages 13 days after par­turition.  The dam no longer responds to these vocal­izations from puppies after that day, even though puppies may emit the signal in contextually appropri­ate circumstances for several more weeks.

In both pup and dam, the motor patterns are in­trinsic.  There is no learning or development period.  The behavior is "perfect" the first time it is displayed, the emergence is spontaneous, and it is contextually and motivationally relevant and specific.  Neither be­havior occurs in other behavioral contexts, such as play or courtship.  Neither other puppies nor males respond to the retrieval call.  Nor do adults, of either sex, emit them.

The fact that the mother dog retrieves a tape recorder during the critical period but not a calling pup after the critical period suggests that no cogni­tive process is occurring.  The call is a reflex in neonates, and the retrieval of the calling pup is a re­flex by the dam.  Both motor patterns have a critical period of display.  They are not elicited before or after the particular ontogenetic stage, although the re­trieval behavior can be induced hormonally in experiments.



The care-soliciting calls of pups are not just species-specific but identical across the genus and similar across family and order.  In other words, wolf cubs and dog puppies have the same set of care-soliciting calls, and mammals in general have similar care-soliciting calls.  The similarity between them suggests that the ancestors of dogs and wolves had this same set of modulating tonal signals.

The care-soliciting calls of puppies and cubs are phylogenetically related and are said to be homolo­gous.  The genes that build the structures that allow this reflexive behavior to be displayed were inherited by dogs from their wolf ancestor and by wolves from their carnivore ancestor.  That particular pattern of genes has probably been around for millions of years.  Since the call induces the retrieval response, the two behaviors probably coevolved.  Surely they are hard-wired; if searched, canine anatomy would reveal neu­rologic and hormonal arrangements that predispose dogs to display these behaviors in the presence of an appropriate environmental stimulus.

Because these behaviors coevolved, one is not more primitive nor is the other more complex.  Mammary glands evolved, and so did specialized neonatal mouths that suck on them.  Care-soliciting behavior evolved with care-giving (maternal) behavior.  The be­havior of the neonate is just as old and just as com­plex as the coevolved behaviors of the adult.  The main point is that the emission of the retrieval signal by the puppy is equally as complex, in evolutionary terms, as the retrieval response of the mother.  In fact, mammalian neonatal behavior is a recent evolution­ary advance in vertebrate evolution.  The pup is not a primitive form that "develops" into a behaviorally ad­vanced adult.

For this discussion, just one behavior each was picked from among many in the neonatal and adult repertoires.  During parturition, females remove pup­pies from the placenta; and they do it perfectly (with rare exceptions) with their first pup.  Pups released from the placenta find a teat and begin to nurse.  Sucking is a complex motor pattern that adult dogs cannot perform.



When describing a dog's behavior, an ethologist uses motor patterns to try to determine the animal's motivation.  To say that a dog killed a sheep or bit someone does not help in understanding why the an­imal did it.  Unless one knows why the dog killed the sheep, one cannot begin to correct the problem.

Dogs have many different innate bites; that is, they have several motor patterns that are described as bites.  To understand what may have motivated a dog to bite, investigators need information on the quality of a bite:  where it was directed and what other motor patterns were displayed in the sequence.

Canid behavioral repertoires include several preda­tory biting motor patterns.  Wolves string these patterns together in feeding or foraging behavior.  A typi­cal stereotyped predatory sequence in a canid is:  Orientation toward prey → eye → stalk → chase → grab bite → crush bite (kill) → dissecting bite → con­suming bite.

A field biologist can often tell which predatory species killed an animal by observing where the grab bite, kill bite, or dissecting bite appears on the car­cass.  Typically, wolves and dogs direct the crush bite at a femoral artery– the victim then bleeds to death.  Coyotes more frequently direct the kill bite toward the carotid arteries.  Pumas suffocate their prey.

The dog family has other characteristic bite motor patterns besides those used in the predatory routine.  Intraspecific (within species) dominance bites are di­rected to the top of the neck of an animal refusing to submit, to the loin in a fight, or to the abdomen of a submitting animal.  When a farmer tells us that the dog he hoped would guard his sheep is biting or even killing them instead, we first try to determine the dog's motivation by noting the motor patterns.  If the sheep are torn on top of their necks or their ab­domens, we suspect that the dog is using intraspecific dominance bites interspecifically.  Confidence is bol­stered if further investigation reveals other playful motor patterns (e.g., either play bow or the dog's head and tail held up).

Directing these behaviors toward sheep may seem inappropriate until one realizes that the dog spent its critical period of socialization with sheep and "learned" to display normal intraspecific behaviors toward sheep.  As dogs mature sexually, they some­times show reproductive dominance routines toward sheep, including sexual mountings.  Because these are not the motor patterns of a predatory routine, one can assume that the dog is not trying to kill the sheep but has successfully bonded with them and might still be trained to be a good sheep guardian.  If the bite is directed at the hind limb, however, then one suspects that the dog is displaying a grab or kill bite, thus indicating that the motivation is predatory.  One could conclude that the dog is acting interspecifically with sheep and probably will never be reliable (with sheep) or retrainable, since the critical period for so­cial development has passed.

Grab bites, crush bites, and dissect bites are specific motor patterns in dogs and are thought to be homol­ogous with wolf predatory behaviors.  Among dogs, these predatory behaviors differ with age and breed.  Neonates do not display them, and the onsets vary between and within breeds.  Pointers and retrievers have grab bite, but it is a fault for them to have crush bite ("hard mouth").  Grab bite is necessary for cattle-heeling breeds, and not having it is a fault.  Grab bite is a fault in sheep-herding breeds.  Livestock-guarding breeds must not have grab, crush, or dissecting bites.  In fact, they should not even have chase as a motor pattern.  Good livestock-guarding dogs will not chase a ball, a behavior one should try to elicit when judg­ing a dog's fitness for protecting sheep.  Good live­stock-guarding dogs will not (because they do not have the dissect motor pattern) open a stillborn calf to feed.  It is necessary to slice open dead calves for a pen full of livestock-guarding dogs to feed, but not for border collies.



As stated, livestock-guarding dogs develop their so­cial behavior during the critical period from about 3 to 16 weeks of age.  Evidently, the offset of this criti­cal period is delayed in some animals.  Emergence of predatory routines (if they appear) in livestock-guard­ing dogs occurs after six months and occasionally af­ter a year.  Seldom do any of the predatory motor pat­terns occur, although chase and grab bite are the most common.  If, however, the dog has been proper­ly socialized with sheep, the later-emerging motor patterns do not get directed toward sheep but may be directed to other species.  This can lead to problems when, for example, dogs raised with sheep are trans­ferred to goats or when they disrupt wildlife on the sheep range.

Border collies raised with sheep during the critical period of socialization can be ruined as herding dogs.  Although such dogs might eye sheep, they seem un­able to follow the eye with stalk.  In a double-blind experiment in which the trainer did not know the background of the dogs, he reported that the collie raised with sheep "couldn't hold the eye."

An individual may be capable of a motor pattern that it never displays simply because it never gets the environmental signal that elicits it.  Many puppies never use the lost call, for example, because they are never lost.  Because of their early social experience, many livestock-guarding dogs never display grab bite.

Some motor patterns must be temporally rein­forced to become functional.  In many mammalian species, maternal behaviors must be reinforced within a few minutes of parturition or the dam will abandon the offspring.  Similarly, sucking by neonates must be reinforced within a critical period or it is lost from the repertoire.  Animals born with mouth infections or correctable abnormalities often lose the nursing motor pattern in the few days it takes to recover.

Other motor patterns are fixed in behavioral se­quences and cannot be displayed until the preceding behavior is performed.  Some species of cats cannot eat carrion because the motor patterns for consumption are connected sequentially to the whole predatory repertoire.  In some situations, livestock-guarding dogs might be effective simply because they disrupt coyote predatory routines by placing the coyote in conflicting motivational states that do not allow it to complete the


sequence.  Border collies have strong connections between eye, stalk, and chase but weaker and modifi­able connections with the grab bite and kill bite.  Trainers can get a border collie to switch back and forth from chase to eye; but when it is allowed to pro­ceed from chase to grab bite, it becomes more difficult to control the dog.  Trying to get a wolf not to go from chase to grab bite is even more difficult.



In the evolution of behavior, motor patterns can become detached from their ancestral sequence and reincorporated into a new sequence.  Studies of dog breed differences give insights into how these evolu­tionary feats are accomplished.  The reader must re­member that breeds are allelically different and that allelic differences are differences in the onset, rate, and offset of gene products.  Similarly, behavior should be described in terms of onset, frequency, and offset of motor patterns.

As mentioned, the care-soliciting behaviors occur for a few weeks after birth, whereas onsets of care-giv­ing behaviors occur immediately after parturition and are displayed for the next few weeks.  The frequency of display is a response to several internal and exter­nal variables.

Sucking (feeding behavior) is an innate behavior that appears shortly after birth (Figure 3).  It is ex­pressed for long periods throughout the day for the first few days, declines in frequency over the next few weeks, and disappears at about 8 weeks, never to reappear in the behavioral repertoire.  Consumatory bites (feeding behavior) appear in the repertoire at 4 weeks of age and increase in frequency until an adult plateau is reached.  They are maintained for the rest of ontogeny.

Figure 3 shows that sucking does not grow or ma­ture into feeding.  Sucking offsets at about 8 weeks.  Feeding onsets at about 4 weeks.  Hall and Williams21 showed that sucking behavior has little to do with feeding in either anatomical or neurologic control.  The display of sucking motor patterns is initiated in one section of the brain and that of consumatory bite motor patterns in another.  They are under separate motor control.  Each has separate motivations for per­formance.  Satiation, for example, suppresses feeding but not sucking.

One way to change innate behaviors is to shift on­set and offset times genetically.  As one might expect, this is one of the differences between breeds.  Border collies have early onsets of eye-stalk behaviors, whereas livestock-guarding dogs have late or no onset of these behaviors.

Another way to change behavior genetically is to disconnect motor patterns from other motor patterns.  Some species of wild cats have grab bite, kill bite, dis­sect, and consume bites sequentially connected in the adult but not in the adolescent.  These different adult motor patterns have separate onsets during adoles­cence and cannot be connected until they are all in place.  The timing of the onset of these motor patterns may be due to developmental constraints or they maj be the result of selection.  These interconnected feed­ing motor patterns may be the most efficient way for the adult cat to feed; but at the same time, the con­nectedness prevents them from eating carrion.

If young cats that cannot yet kill their own food had feeding motor patterns sequentially connected, this sequencing would prevent them from eating dead prey provided by the parent.  In this case, the adolescent may not be as efficient in killing prey but is more adaptable than the adult because feeding be­havior is not rigidly innate.



Neonatal and many adult species-typical behaviors have been discussed as innately rigid and reflexive.  The adolescent, however, is quite a different story.  Adolescence is so different that in many ways etholo­gists and psychologists have difficulty understanding exactly what effect it has on adult behavior.  Never­theless, this stage of ontogeny deserves attention be­cause it is the most difficult stage of development to describe, it is the stage in which most mammals play and learn, and it is the stage in which dogs, as a species, are trapped.

The most simplistic definition of adolescence is the period of ontogeny between the neonate and the adult.  It is the period when the organism is experi­encing the offsets of neonatal motor patterns and concurrently the onsets of adult motor patterns.  Juve­nile mammalian behavior is a complex assemblage of waning neonatal and waxing adult motor patterns– the period of metamorphic remodeling.22

In the neonatal stage, behaviors are prepro­grammed (hard-wired).  Neonatal feeding behaviors are all accompanied by specialized anatomy that al­lows sucking to be performed.  The neonate is not primitive or underdeveloped.  It is a highly specialized organism in the sense that it is so perfectly adapted to its environment it does not have to learn anything.  There is no selective advantage to learning, nor is there enough time or energy in the neonatal system to be able to experiment (learn) behaviorally.

The adult mammal is also well adapted to its envi­ronment.  Feeding, reproduction, and hazard-avoid­ance behaviors have large innate components and of­ten vary little or not at all within a species.  In general, most adult mammals have difficulty rearranging se­quences of motor patterns within a specific behav­ioral repertoire; it is therefore difficult to modify their behavior.  They cannot afford the time or energy to experiment.

Adolescence is a period of metamorphosis– anatomical remodeling.  The neonatal organism is taken apart and reconstructed into an adult.  Behav­iorally, the individual is remodeled from innate neonatal feeding and hazard-avoidance behaviors to the adult feeding, hazard-avoidance, and reproduc­tion systems.  Sucking feeding behaviors do not grow, or develop, into predatory feeding behaviors any more than the 18 feet of a caterpillar grow into the six legs of a butterfly.  Instead, the animal is de-differ­entiated, to borrow the words of the embryologist.  New organs are created de novo while old ones are discarded, just as the highly complex placenta and its associated behaviors are discarded at birth.  Skulls do not grow from the neonatal skull (the sucking skull) into an adult predatory skull.  The neonatal skull is resorbed while the adult skull is being laid down.23

Adolescents therefore have a changing access to neonatal and adult motor patterns.  The sequencing of combinations of neonatal and adult motor pat­terns becomes the basis for play and learning.  Se­quences of motor patterns can be adjusted to the en­vironment as the ever-changing adolescent uses them in contextually new and varied ways.

As pieces of neonatal motor patterns are offset, the remainder are displayed against the fragments of adult motor patterns.  When pieces of neonatal motor patterns are displayed in sequence with adult fragments in nonfunctional behavior patterns, the behav­ior is called play.  Putting fragmented neonatal motor patterns together with adult behaviors in functional ways is one definition of learning.  The distinction be­tween learning and play is simply whether the result­ing motor sequence is functional and/or repeatable.

The two classes of vertebrates that play are mam­mals and birds, both of which have well-defined neonates.  Play and learning are therefore considered artifacts of being able to use motor patterns that are not contextually appropriate to their phylogenetic origin.  The hypothesis is that adolescents play and learn so well because their neonatal behaviors have become disconnected and their adult behaviors have not been sequentially aligned.  Thus, they can use the dissociated motor pattern of two distinct ontogenetic stages in new ways.



Dogs may be thought of as permanently adoles­cent.  Genetically and evolutionarily they are "arrest­ed" in the adolescent period.  They have become re­productive as adolescents.  Retaining juvenile characteristics into the adult period is called neoteny or paedomorphism by embryologists.  Dechambre24 suggested that canine breeds are differentially neo­tenic; whereas Fox,7 Frank and Frank,25 and we26 suggested that not only are dogs behaviorally neo­tenic, but breeds of dogs are differentially neotenic behaviorally.  The difference between the two types of sheepdogs might simply be that livestock-guarding dogs are arrested in a stage of development before the onset of adult predatory motor patterns (e.g., grab, kill, and dissecting bites), whereas border collies are in a stage that includes eye-stalk and chase but the grab and kill bites are weak and imperfect and not se­quentially linked to the rest of the repertoire.14

It is common practice to try to understand dog be­havior by searching for motor pattern homologues with wolves.  But wolf ontogeny should be taken into account.  People often say that dogs are territorial be­cause wolves are and that wolves form packs with a leader, therefore dogs form packs and the human trainer has to become the pack leader.  Wolf cubs, however, do not have reproductive territories or form packs and are not organized socially around a leader.  If dogs are neotenic, then they should show behavior that is homologous to that of wolf cubs, not that of wolf adults (e.g., territory defense or pack formation).  Nor should a person pretending to be the pack leader or dominant dog make any difference in their behav­ior.  Wolf neonates and young adolescents do not hunt, nor do the neonates have any predatory motor pattern.  Wolf adolescents have some of the predatory motor patterns, but they may not have them all, nor are the patterns connected sequentially, depending on the individual's age.

Exploration of the behavioral differences in breeds does support the hypothesis that dogs are arrested in an adolescent stage of wolf ontogeny.  The difference between guarding and herding sheepdogs is illustra­tive.  The guarding breeds never get to the ontogenet­ic stage when onsets of predatory behavior occur.  The herding breeds get some of the predatory motor pat­terns, but not the final, wolf-like ones.  Border collies and livestock guardians raised from puppyhood to­gether in one large pen separated themselves into dis­tinct "tribes" during play routines.  The border collies played eye-stalk-chase games, whereas the guardians played at dominance-submission.  The two types of sheepdogs segregated, like species, and did not play with each other.  They are different temperamentally because of the kinds and frequencies of motor pat­terns they display.

The selection of type and intensity of play behav­iors has differed among breeds.  For example, the per­formance of sled dogs is based on selection for social play.  What the driver thinks regarding his or her role as "pack leader" is irrelevant to putting together a team.  The driver can train them and discipline them, but any attempt to become "dominant dog" is usually disruptive to the racing potential of the team.  Many women with a gentle and supportive approach to training dogs excel with their racing teams.  Many teams have female lead dogs.  Many teams have two lead dogs, which if one were "dominant" over the other would not work.  Many teams have several lead dogs that race back in the team and can be swapped for a tired leader.  None of this would be possible with a team that behaved like adult wolves.

One should, however, treat the neoteny hypothesis with some caution.  It would be silly to think that the ancestral wolf went through a maremma stage of be­havior, then a border collie stage, then a husky stage, to become finally an adult wolf.  It is just as silly to think of a maremma as representing a particular stage of wolf development, for the following reasons.

Puppies change (remodel) from neonates to adults.  Heads grow (remodel) from little short-faced puppy heads to big long-faced adult heads.  Change is allo­metric, not isometric.  To arrest an animal at a particular growth stage is in effect to continue the allometric changes that are occurring at that stage.  Since allometric changes are not prolonged in the ancestral stage, the resulting dogs are anatomically bizarre.  In­deed, dogs are anatomically bizarre (e.g., head shapes or size variation).

If their anatomy is novel in terms of their ancestor, then their behaviors will also be novel and not direct­ly equivalent to that of their ancestor.  For this reason, one discussion of the evolution of working dogs was ready to drop the notion of behavioral neoteny, sim­ply because although it might account for the hetero­chronic processes resulting in dogs, it was a poor pre­dictor of either anatomical or behavioral result.27  In the five years that have elapsed since the paper was written, optimism about the use of neoteny as a pre­dictor of anatomy and behavior is returning.



The foregoing discussion of the selection process for behavioral differences among canine breeds has focused on breeds that were developed for work.  This discussion underscores the biological bases of behavior, emphasizing details of physical structures that enable or limit behavior.  We believe, however, that this information will also provide new ways of looking at the behavior of companion animals and offer alternative interpretations of problem behaviors.  Such increased understanding of behavior should lead to improved techniques for the prevention, treat­ment, and management of canine behavior problems.




1.   Black ML, Green JS:  Navajo use of mixed-breed dogs for management of predators.  I Range Management 38:11-15, 1985.

2.   Wayne RK:  Cranial morphology of domestic and wild canids:  The influence of development on morphological change.  Evolution 40:243-261.1986.

3.   BekoffM (ed):  Social play in mammals.  Am ZooLogist 14(1):

26c¼427, 1974.

4.   Scott JP:  Evolution and domestication of the dog.  in Dobzhansky T, Hecht MK, Steere WC (eds):  Evolutionary BioLogy.  vol 2.  New York, Appleton-Century-Crofts.  1968, pp 243-275.

5.   Fox MW:  A comparative study of the development of facial expressions in Canids, wolf, coyote and foxes.  Behaviour


6.   Fuller JL:  Hereditary differences in trainability of purebred dogs.JGen Psychol87:229-238, 1955.

7.   Freedman DG:  Constitutional and environmental interac­tions in rearing of four breeds of dogs.  Science 127:585-586, 1958.

8.   Elliot 0, Scott JP:  The development of emotional distress reac­tions to separation, in puppies.! Gen Pqchol99:%22, 1961.

9~  Elliot 0, Scott JP:  The analysis of breed differences in maze performance in dogs.  Anim Behav 13:5-18,1965.

10. Scott JP, Fuller JL:  Genetics and the Social Behavior of the Dog Chicago.  University of Chicago Press, 1965.

ii.   Mahut E:  Breed differences in the dog's emotional be­haviour.  CanjPsycholl2:3544.  1985.

12. Borchelt P:  Aggressive behavior of dogs kept as companion animals:  Classification, influence of sex, reproductive status, and breed.  ApplAnim Ethol i0:45~i, 1983.

13. Hart BL, Hart LA:  Canine and Feline Behavioral Therapy.  Philadelphia.  Lea & Febiger.  1985.

14. Coppinger RP, Glendinning JI, Torop E, et al:  Degree of behavioral neoteny differentiates canid polymorphs.  EthoLogy


15. Scott JP:  The domestic dog:  A case of multiple identities, in

iS Readings in Companion Animal Behavior

Roy MA (ed):  Spedes Identi~ andAttachment A PhyLogenetic

Evaluation.  New York, Garland STPM Press, 1980, pp 129-


16. Fox MW:  Integrative DeveLopment of Brain and Behavior in the Dog.  Chicago, University of Chicago Press, 1971.

17. Fox MW:  The Dog:  Its Domestication and Behavior.  New York, Garland STPM Press, 1978.

18. Cairns RB, Johnson DL:  The development of interspecies social attachments.  PsychonometrkSci2:337-338, 1965.

19. Phillips CJ, Coppinger RP, Schimel DS:  Hypothermia in running sled dbgs.JApplPbysiolSl:135-142, 1981.

20. Arons CD, Shoemaker WJ:  The distribution of catechola­mines and ~-endorphin in the brain of three behaviorally distinct breeds of dogs and their F, hybrids.  Brain Res

594:31-39, 1992.

21. Hall WG, Williams CL:  Suckling isn't feeding, or is it?  A search for developmental continuities.  Advances in the Study ofBehavior 13:219-254, 1983.

22. Coppinger RP, Smith CK:  A model for understanding the evolution of mammalian behavior, in Genoways H (ed):

Current MammaLogy vol 2.  New York, Plenum, 1990, pp


23. Enlow DH:  Handbook of Facial Growth.  Philadelphia, WB Saunders Co, 1975.

24. Dechambre B:  La th6orie de foetalization et Ia formation des races de chiens et de porc.  Mammalia 13:129-137, 1949.

25. Frank H, Frank MG:  On the effects of domestication on ca­nine social development and behavior.  AppI Anim Ethol


26. Coppinger R, Coppinger L:  Livestock-guarding dogs that wear sheep's clothing.  Smithsonian Apnl:64~73, 1982.

27. Coppinger R, Schneider R:  The evolution of working dogs, in Serpell JS (ed):  The Domestic Dog:  Its Evolution, Behaviour and interactions with People.  New York, Cambridge Univer­sity Press, 1995.