Resource-guarding: Are veterinarians lost in interspecies translation?
This catchall phrase used to describe aggressive behavior in dogs can actually do more harm than good.
Evolutionary biologists will tell you that animals make decisions based on four operant environments: demographic, biophysical, social and resource-related. Behaviors change depending on how these environments interact. For example, anyone watching birds at feeders in the spring and early summer will notice that females eat and males forgo eating to attempt courtship. As a result, males of many species lose weight in the breeding season.
GETTY IMAGES/LI KIM GOH
So, the way we view food is not absolute, and an animal's behavior must be both defined and interpreted in the context of other events. Unfortunately, most people rarely use this approach when discussing aggressive reactions in dogs, and we know little about the range of any canine responses, whether "normal" or "abnormal."1-3
Can we define "resource-guarding"?
In the past few years, I have increasingly heard a term that could benefit from this approach: resource-guarding. Few terms are as blurry and inconsistent as this one. I have heard trainers and clients say, "The dog is fine; he just resource-guards," meaning no other dog in the house can approach him when he has a toy. (The implication here is that resource-guarding is "normal," but this dog is not fine.)
I have heard people label the dog that growls and trembles, spilling food from its mouth because someone has instructed its owners to constantly move their hands around in the food as the dog eats (no one should subject any dog to such threats), as one that's "dangerous resource-guarding." In these examples and many more, the term resource-guarding provides no helpful information, and the assumption that it does can be dangerous.
How did we get to the place where everything and everyone is a resource and resource-guarding can apparently be applied to any situation?
I spend endless hours asking clients, trainers, dog breeders, veterinarians, students—even my resident—to elaborate on and clarify the terms they use. In veterinary behavioral medicine we have no protected terminology, so the process of sifting through layers of meaning and possible misdirection is an essential skill.
Diagnostic terminology in this field generally takes one of three approaches:
1. It's named for the recipient of the inappropriate, undesirable or problematic behaviors (e.g. owner-directed aggression, stranger-directed aggression, dog-directed aggression),
2. It's named for the putative driver of the behavior (e.g. offensive aggression, defensive aggression, affective aggression), or
3. It's named phenotypically using criteria that unite the observed behaviors with a pattern or context of concern (e.g. food-related aggression, fear aggression, panic disorder, separation anxiety).
The first approach is largely driven by risk to the recipients of the behaviors and is a human-centric approach. The other two approaches primarily focus on the dog. I personally favor the third, the phenotypic approach, for simple, logical reasons:
> If you can understand a behavior by identifying the circumstances under which it occurs, you can avoid those circumstances (and a good diagnostic description, definition or set of criteria allows you to do this).
> By understanding the overarching rule (i.e. the diagnosis) for the suite of behaviors exhibited by the dog in a certain context, you can recognize patterns of response that might help with anticipating other concerns. You can also ask if those patterns share mechanisms or respond to the same interventions and medications.
> The phenotypic approach allows the clinician to interpret rare nonspecific signs within a context in which the dog can be helped. For example, a dog whose video reveals that, when left alone, he spends all day frantically rearranging fruit bowls is more likely to have an odd form of separation anxiety than he is to have a fruit fetish.
Although it sounds like one, the term resource-guarding not a good phenotypic diagnosis, and a bit of thought reveals why.
> The resources are not defined. In common usage, resources have become everything from toys to dinner to treats to a specific lap or a certain part of the yard. A definition that includes every subject is no definition at all.
> No distinction is made between normal and abnormal behaviors. Clients can find normal behaviors problematic and may like abnormal behaviors; the client's opinion is not what defines "normal."
> In some patients, aberrant behaviors may have been adaptive in another context, a point almost never considered when abnormal behaviors are discussed.
Resource-guarding or human interference?
A brief review of the literature will show additional reasons why the term resource-guarding is so problematic. Resource-guarding has been used as a description that encompasses both food-related and possessive aggressions but not owner-directed aggressions (e.g. fear, impulse control, territory).4 It has been used as an umbrella category for aggression in the presence of wet food, dry food, treats, toys and furniture, such as the couch.5
It's been asserted that aggression toward humans is most frequently driven by some aspect of resource-guarding, but even the dog bite data regarding children—a most selective sample—do not bear this out. Instead, dog bite data are complex and suggest that bites to children occur when the child interferes with some activity in which the dog was engaged and where—at least from the dog's perspective—there is no role for the child,6 one example of which is eating from the food bowl.6,7
In one study that examined the responses of 103 dogs that had bitten children when their resources were removed, the resources were defined as dog food, special food and toys. The data for these three categories were not presented individually, which is a problem because they are very, very different items contextually—at least to the dog.
When the resource was removed from the dog, 48 dogs were found to be "aggressive" or "anxious" (neither of which was defined in this study), and 31 were found not to react to resource removal. For an additional 24 dogs that had bitten, no one reported taking resources from the dogs and no bites were reported for these dogs in this context. This finding strongly suggests that the dogs may not actually guard these items and, instead, that it was the act of removal of the item that triggered the "aggression" or "anxiety." If this is so, the concern from the dog's viewpoint may be less about the item itself than the interference with the dog's activity (e.g. eating dinner) and relative disruptions to that.
In a worst-case scenario, 48 of 79 dogs (61 percent) that had a resource removed were "aggressive" or "anxious"; however, only 48 of 103 dogs (48 percent) that had bitten children showed "aggression" or "anxiety" when the resource (e.g. dog food, special food, toys) was present, whether or not it was removed.8
Because the original data are not recoverable from many publications using the term resource-guarding, we cannot compare actual responses when different definitions of resource-guarding are used. This is unfortunate because the questionnaires used by authors usually ask separately about the individual contexts, which are later lumped into whatever definition of resource-guarding the authors use, and so those data could—and should—be published.
In extreme cases, humans are even defined as resources in special cases where children approach another human with whom the dog is actively or passively interacting.9 The problem with a diagnostic algorithm that lumps contextual responses is that no a priori testing is done to know whether those responses come from the same population. When such testing is done, phenotypic diagnostic strategies usually show that populations of dogs exhibiting the related behaviors within a certain context are distinct, and they can overlap when co-morbid diagnoses occur.10,11
In other words, not all dogs with impulse control aggression have food-related aggression, dogs with food-related aggression need not have any other aggression, including impulse control aggression. But some dogs could have both—and knowing that would allow you to predict situations to avoid.
What do the studies show?
Studies that lump various responses into the catchall term resource-guarding forfeit the ability to clearly understand the triggers (e.g. dog food, human food, bones, treats, toys) that most frequently elicit the behavior of concern. By lumping these very different resources into one group (forgetting for the moment the concern that the response may be more about interference in an activity or interaction than about a resource—and that, regardless, perhaps we should not expect dogs to tolerate the interference), we lose the ability to understand discrete contexts that matter to the dog. We also lose the chance to explore any role the dog's history may play in its response.
Understanding food-related aggression
Simply put, the dog matters here. There is an assumption that shelter dogs, particularly those that have been homeless or neglected, may have a history of food deprivation and, as a result, are protective of food. A parallel assumption is that if these dogs exhibit food-related aggression in the shelter, they'll make poor pets. So if we concentrate only on real, edible food as an example of resource-guarding, what do the data say?
In a questionnaire study of 72 adopted shelter dogs, 7 dogs (11 percent) were reported to be aggressive over their food, 14 dogs (22 percent) were reported to be aggressive over a bone, and none were considered to be problematic by the humans, nor were any dogs returned to the shelter for either of these two reasons.12
Similar results were obtained in an interventional test for shelter dogs designed to evaluate true food-related aggression. In this test, food-related aggression was defined as the dog displaying any of five behaviors (showing teeth, growling, snapping, lunging or biting) when the dog was in one of two contexts (the dog was touched when eating or chewing, or the food item was approached or touched when near the dog).13
Twenty of 97 dogs evaluated (21 percent) were deemed to show food-related aggression in the shelter, and 11 of these 20 dogs (55 percent) also showed food-related aggression at home. Of the 77 dogs that did not exhibit food-related aggression in the shelter, 17 dogs (22 percent) showed food-related aggression in their new home. Here, the positive predictive value (PPV=0.55) was much lower than the negative predictive value (NPV=0.78) of the test—a pattern also reported for shelter dogs in tests of interaction with humans.14
Finally, a study of 66 rehomed shelter dogs found that three of the 66 dogs showed signs of resource-guarding toward humans, and four of the 66 showed signs toward other household animals when in their new home.4
However, in none of these shelter dog studies did adopters consider the food-related aggression to be a problem for keeping the dog. I suspect that this is the case for the same reasons phenotypic diagnoses are so logical and helpful—the trigger for the behavior is easy to recognize, the dog's behavior is easy to recognize, the association is clear and, hence, the problem is easily avoided or controlled.
Take a look at context
When we're talking about reactivity to true food items, the problem is not that much of a problem from the human's viewpoint. But what about the dog's viewpoint?
The concept that resource-guarding is a useful umbrella term for a range of what humans consider bothersome dog behaviors dismisses the dog's role in the interaction and does not consider that the dog's behavior could be a variant of normal.3,15 We know nothing about the true range of canine behaviors across time and varying contexts. Our knowledge of "normal" is assumed, not measured. If we do not know where "normal" behavior ends, it is difficult to know where "abnormal" behavior begins and what its contributing causes might be.
The most unfortunate outcome of this approach is that it encourages people to think that they can and should teach dogs not to guard resources. As a result, people provoke dogs that otherwise would not—and should not—have been provoked. Careless assignment of the label of resource-guarding can group the mildly and restrictively aggressive dog with the seriously and dangerously aggressive one without any clear plan for mitigating risk, because the context in which the behavior occurs is ignored. In other words, context matters.
Normal vs. abnormal behaviors
So where do the concepts of normal and abnormal behavior fit into this discussion? The idea that dogs are supposed to be protective of their food is as much a myth as the idea that all dogs should allow you to take their food dish away. In fact, dogs, like other social species, are supposed to behave appropriately—in context. Few of us would be sanguine if guests in a restaurant took our dinner because they liked the way it looked. In fact, humans do not eat from each other's plates until they share some level of familiarity.
If a dog has never been teased or threatened about food or starved by illness, neglect or abuse, there is no valid reason for any aggressive behavior when food is present. Given those caveats, if the dog is still aggressive in the presence of food, we conclude that the behavior is not normal. Because the cost of error is so high, and because responses to food—a fundamental resource that's one of the four essential environments governing animals' decisions—may be hardwired, protecting the dog from the concern that his food will be taken can keep everyone safe—and help the dog to become more calm by making clear the lack of risk.
Is the term resource-guarding useful?
Terminology is supposed to enlighten. A term that is used for both normal and abnormal behaviors, which may or may not include certain contexts, does not do this. We would be better served by simple explanations of what happened, followed by thoughtful consideration. If labels affect the way we think, we should take great care with the label.
Simply put, if knowing something matters and if we know less after using a term than we knew before we used it, we need to reconsider our terminology.
Dr. Karen L. Overall is a researcher, editor of The Journal of Veterinary Behavior: Clinical Applications and Research, and author of more than 100 publications, dozens of chapters and a new book, The Manual of Clinical Behavioral Medicine for Dogs and Cats.
1. Mertens PA. The concept of dominance and the treatment of aggression in multidog homes: a comment on van Kerkhove's commentary. J Appl Anim Welf Sci 2004;7(4):287-291.
2. Bradshaw JWS, Blackwell EJ, Casey RA. Dominance in domestic dogs—useful construct or bad habit? J Vet Behav: Clin Appl Res 2009;4:135-144.
3. Overall KL. Manual of veterinary behavioral medicine for dogs and cats. St. Louis, Mo: Elsevier, 2013.
4. Christensen E, Scarlett J, Campagna M, et al. Aggressive behavior in adopted dogs that passed a temperament test. Appl Anim Behav Sci 2007;106:85-95.
5. Mornement KM, Coleman GJ, Toukhsati S, et al. A review of behavioral assessment protocols used by Australian animal shelters to determine the adoption suitability of dogs. J Appl Anim Welf Sci 2010;13:314-329.
6. Cornelissen JM, Hopster H. Dog bites in The Netherlands: a study of victims, injuries, circumstances and aggressors to support evaluation of breed specific legislation. Vet J 2010;186:292-298.
7. Reisner IR, Shofer FS. Effects of gender and parental status on knowledge and attitudes of dog owners regarding dog aggression toward children. J Am Vet Med Assoc 2008;233:1412-1419._
8. Reisner IR, Shofer FS, Nance ML. Behavioral assessment of child-directed canine aggression. Inj Prev 2007;13:348-351.
9. Reisner IR, Nance ML, Zeller JS, et al. Behavioural characteristics associated with dog bites to children presenting to an urban trauma centre. Inj Prev 2011;17:348-353.
10. Bamberger M, Houpt KA. Signalment factors, comorbidity, and trends in behavior diagnoses in dogs: 1,644 cases (1991-2001). J Am Vet Med Assoc 2006;229:1591-1601.
11. Overall KL, Dunham AE, Frank D. Frequency of nonspecific clinical signs in dogs with separation anxiety, thunderstorm phobia, and noise phobia, alone or in combination. J Am Vet Med Assoc 2001;219:467-473.
12. van der Borg JAM, Netto WJ, Planta DJU. Behavioural testing of dogs in animal shelters to predict problem behaviour. Appl Anim Behav Sci 1991;32:237-251.
13. Marder AR, Shabelansky A, Patronek GJ, et al. Food-related aggression in shelter dogs: A comparison of behavior identified by a behavior evaluation in the shelter and owner reports after adoption. Appl Anim Behav Sci 2013;148:150-156.
14. van der Borg JAM, Beerda B, Ooms M, et al. Evaluation of behavior testing for human directed aggression in dogs. Appl Anim Behav Sci 2010;128:78-90.
15. McGreevy PD, Starling M, Branson NJ, et al. An overview of the dog-human dyad and ethograms within it. J Vet Behav: Clin Appl Res 2012;7:103-117.