Behavioral Differences Among Breeds of Domestic Dogs

Article Summary: Mehrkam, L. R., & Wynne, C. D. L. (2014). Behavioral differences among breeds of domestic dogs (Canis lupus familiaris): Current state of the science. Applied Animal Behaviour Science, 155, 12-27.

A recent review paper by Lindsay Mehrkam and Clive Wynne summarizes well the state of the science regarding behavioral differences between dog breeds. Animal behaviorists often encounter breed characterizations, some of which seem to hold true in their work with a variety of dog breeds, but others of which seem irrelevant to being predictive of certain types of behavior problems, training tendencies, or temperaments. This review paper can provide pet behavior professionals and pet owners alike with important information regarding what there is to affirm about true breed differences in the scientific literature.

The paper opens with a discussion of the origin of modern dog breeds and how breeds are defined. Already readers are introduced to areas of debate within the field. While some argue that dog breeds differentiated due to natural selection pressures arising from their particular geographic environments, others have emphasized the role of artificial selection by humans in creating distinct breeds. In terms of defining breeds today, some prefer to do so based on microsatellite DNA that identify clusters of breeds by genotype, while others (including the American Kennel Club) prefer to identify breeds based on the working role with which they are associated.

The majority of behavioral research comparing breeds of dogs has focused on differences in temperament, or personality. The authors prefer the term temperament to personality, and define temperament as “either individual or breed differences that emerge early in development, are elicited in a range of situations, and remain relatively stable over time.” They report that the study of temperament has typically focused on one or more of three behavioral traits - aggression, reactivity, and trainability – and they focus their review on these behavioral constellations.

Aggression can be categorized in multiple ways, and Mehrkam and Wynne distinguish between types of aggression based on the target of aggression. In their review of owner-directed aggression, they report that the scientific findings are inconsistent regarding breed differences. Bite incident reports point to certain breeds (Pit Bull-type dogs and Rottweilers) as being responsible for more than 50% of bite-related fatalities (likely due at least in part to their larger size). Yet surveys of owners, veterinarians, and breed judges point to other breeds as being predominant in this category and their conclusions furthermore are inconsistent with each other. No direct experimental assessments of breed differences in this behavior have been published.

Experimental assessments of breed differences in stranger-directed aggression have been conducted, but do not result in a clear picture of which breeds are more likely to exhibit this behavior. In particular, because different methodologies are employed (different contexts, different trigger situations, and different breeds selected for inclusion), drawing wide conclusions is difficult. For example, while one large shelter study found that the typically “high risk” breeds (Chow, Husky, Pit Bull, Rottweiler) were more likely to fail a behavioral assessment of stranger-directed aggression, these same breeds are ranked as average in this tendency on owner surveys. Data are more consistent on breed differences in dog-directed aggression when assessed by owner survey. Most of our current information on breed differences in dog-directed aggression is based on the early research of Scott and Fuller (1965). Mehrkam and Wynne discuss the role that morphology (e.g., ear structure) may play in social signaling (and thus aggression) between dogs.

In the response category termed “reactivity”, Mehrkam and Wynne differentiate between excitability/general activity and investigative/avoidance behavior. Some consistent breed differences have been noted in these categories. For example, Scott and Fuller (1965) found consistent breed differences among their breeds tested (Basenji, Beagle, Cocker Spaniel, Sheltie, and Wire-Haired Fox Terrier) in observable emotional reactions to social interactions with humans along a variety of specific responses. Some owner survey results have consistently ranked the Miniature Schnauzer, Scottish Terrier, and Westie as high for reactive responses (playfulness, destructiveness, excitability, and excessive barking), and others have indicated certain breeds (hounds) score higher on inattention scores while others (herding breeds) consistently score high on responses such as active play, constant motion, and anticipation. 

Some consistent breed differences are observed in exploratory (investigative vs. avoidant) behavior. Much of this research involves measuring dogs’ responses to the presentation of startling stimuli, and several authors have reported significant effects of breed in these tests. For example, one study found that Labs were least fearful in a startle test while German Shepherds were most fearful (Goddard & Beilharz, 1985). In other studies, Labs and Golden Retrievers have also scored higher than German Shepherds on measures of sociability and curiosity. The authors note that this area has received considerable study in part due to the importance of the emotional reactivity measure in the evaluation of dogs for the working and service dog industries.

The authors also review differences across breeds (or the lack thereof) in tests and surveys of trainability, as well as tests of problem-solving that employ physical manipulation tasks or social cue tasks (which examine a dog’s ability to follow a human gesture to locate a food source.) In some cases, breed differences can be attributed to morphology, which might differentially benefit certain breeds on certain problem-solving tasks. In other cases, the individual variation within a breed is significant, drawing attention to the extensive but often overlooked impact of individual learning history on performance in many behavioral tests. In fact, the authors state, “Although the need to consider the interaction between ontogeny and phylogeny when making conclusions about canid social behavior is not a novel point, it appears to need considerable emphasis with respect to studies of breed differences in behavior (p. 22).” 

Merhkam and Wynne close with some summary statements about the current state of the literature on breed differences and some suggestions for areas of focus in future research. I will end by highlighting some of their conclusions here.

1) Many studies fail to clarify their behavioral dependent measures with  operational definitions. For example, “trainability” is an extraordinarily broad term that means different things to different people and that can be measured in a large variety of ways. Not only does the use of overly broad or vague descriptions of behavior make interpretation of these studies difficult, but it also makes it difficult to replicate these studies in other settings or by other researchers.

2) Survey studies based on owner report or on reports by those who work with pets comprise a large portion of the research on breed differences. These studies are vulnerable to the same pitfalls of all survey research – namely, that people may inaccurately report on a dog’s behavior or may be influenced by cultural or media biases in interpreting the behavior of their own dogs or dogs with whom they interact. (In other words, there is no objective measurement of the behavior of these dogs that can be verified by multiple, unbiased observers.)

3) Many of the studies on breed differences reveal large within-breed variation, and this variation can be due to both genetic differences across individuals and also the individual learning history that each dog brings with it to the experimental situation. The contingencies in place in the experiment itself must also be taken into account – for example, certain breeds or certain individual dogs may respond better to the particular reinforcer being used in a particular study, and thus may perform better on that task. If one dog performs better than another, this would not necessarily speak to that individual’s or that breed’s “trainability” or “intelligence” as a broad behavioral trait – instead, it might speak only to that dog’s increased sensitivity to the particular reinforcer being used.

4) Breeds are not equally represented in this research. Specifically, those breeds that are readily accessible (e.g., Labs, German Shepherds, and Golden Retrievers) are vastly overrepresented in research on breed differences, making conclusions about breed influence on behavior difficult to generalize to dogs as a whole.

5) When breed differences have been identified in the scientific literature, they do not typically match behavioral descriptions provided in breed standards. This is of concern because breed standards can influence owners’ choices in dogs that they buy and also in which dogs they will adopt from shelters. Again, caution must be used in interpreting the results of many of these studies because of the lack of variability in breeds tested.

Mehrkam and Wynne’s (2014) paper was published in the journal Applied Animal Behaviour Science, Volume 155. The paper is available online in full-text HTML as well as PDF format at the journal's website:

Click here to access the PDF version of the paper directly: