Overview of Canine Aggression--Part 2: Fear, Aggression, and Fighting with Other Dogs
World Small Animal Veterinary Association World Congress Proceedings, 2006
Karen L. Overall, MA, VMD, PhD, DACVB, ABS Certified Applied Animal Behaviorist
Center for Neurobiology and Behavior, Psychiatry Department - Penn Med Translation Research Laboratory, Philadelphia, PA, USA

There are at least 2 separate populations of clients whose dogs are aggressive to other dogs: those who think that this is "normal" behavior and that the dogs will "sort it out" and those who are genuinely scared and concerned.

Those in the former group are sometimes right, but often wrong. If they are correct, the activities involved in the aggression should cease in a short period of time. If they are wrong, the activities will intensify and may even become more subtle. If the client believes that all dog-dog aggression is normal they will miss this. If their worries come up against their instilled belief, they will may excuses. Either way, veterinarians are unlikely to know about these clients because the clients will not volunteer the information. It is absolutely essential that every single time a veterinarian sees a client that they screen for behavioral concerns just as they would medical ones. If the veterinarian is not doing this they are missing serious problems (1).

Clients in the latter group often bring dogs to either their personal veterinarian or a specialist in behavioral medicine because their dog has "growled or snarled" at or "fought with" another dog. They are often concerned for reasons of image, liability, and the well-being of their pet. Although these clients seek help, they may be just as disadvantaged as the clients in the first group because what they are labeling aggression may not be aggression. The first step in any work up involving any fighting or aggression between dogs is to learn what the dog did. Just because clients tell you that the dog "growled" does not mean (a) that they really growled or (b) that the behavior was abnormal or a concern. Canine vocalizations are poorly understood although some progress is finally being made (2,3). Many clients mistake normal vocalization that is part of play--especially if the play is exuberant or the breed is deep chested and heavily muscled--as an aggressive "growl". Many clients are concerned about any rough play, tussling, or vocalization that may erupt from their dogs, and if they have had a problem with serious interdog aggression in the past, the clients' concern moves to terror quickly. The first step is to ascertain whether the dog growled. Clients should provide videotapes of interactions about which they are concerned. Second, if the dog actually growled, was the context appropriate. For a species without verbal speech and opposable thumbs, there must be a mechanism for telling another individual the 'it's enough'; hence, growls, barks, grumbles, and a series of other sounds that we humans, no doubt, ignore to our own detriment.

At the outset, whether the dogs are exhibiting rough play, pushy behavior, or true pathological aggression we must always ask the question, "Are both of these dogs behaving appropriately, given the context?". If the answer is no, then some intervention is necessary.

There are many pushy dogs who must go through doors before other dogs, must be able to take toys or rawhides from other dogs, and who are always able to blackmail some other poor canine house-mate from his or her morning biscuit. These same dogs control all dog play and all access to people. The conventional wisdom views such dogs as normal. In fact, these dogs are often hailed as the "alpha", "top", "boss", "head", or "leader" dog and clients who express concern are instructed to encourage the other dogs to take this dog's lead, and then they are encouraged to reinforce the lower "ranks" of the other dogs by a reward and reinforcement schedule. This is truly appalling logic, and with genuinely pathological dogs can be profoundly dangerous advice.

Unfortunately, we know so little about dog behavior that we have no idea how many "pushy" dogs go on to develop truly pathological behavior. If they stay "pushy" without truly threatening the lives of their down-trodden house-mates, they're socially obnoxious, but survivable. In fact, most "normal" dogs will learn to avoid the circumstances that cause the "pushy" dog to exert his or her forcefulness, which has the effect of lowering the overall level of reactivity, and socially ostracizing the bratty dog. The bratty dog likely barely notices this because his or her attention from humans comes when he or she is alone. The close, supportive, playful relationships that the other dogs will share is not truly enjoyed by this dog, but we have no idea if they know this.

If the dog goes on to be truly pathological, it is not sufficient that the other dogs get out of the aggressor's way and allow him or her first access to everything. Even if these dogs defer to everything the "boss" dog wants, it is not sufficient. The aggressor targets a dog that they wish to banish from the social hierarchy, and they continually attack that dog, looking for an opportunity to kill.

The key to understanding this difference in these 2 types of dogs--the truly dangerous, aggressive dog and the socially undesirable "pushy" dog--lies in understanding why social systems evolved, in the first place. Social systems, which are rule structures designed to minimize counterproductive error, evolved because, by distribution of effort and clear, redundant signaling, everyone in the group does better than they would alone. Is the behavior exhibited by the "pushy" dog above consistent with a social network where tasks are shared, responsibilities are distributed, and play is likely to be about learning to make mistakes successfully? Absolutely not. No one should have to control everything. The thing that makes most social interactions work is the pattern of deference that's found in every true social system: individuals defer to those who may make more appropriate decisions, given the context. If clients are allowed to watch their dogs without pre-imposed labels, they usually cannot identify an "alpha" dog.....instead they note that one dog is more outgoing with new people, another with new dogs, and perhaps, yet another in situations when the dogs aren't on their own turf. Status is always conferred by a group--it cannot be taken and maintained--and when it is conferred it's given to the individual who makes the most contextually appropriate decisions. All members of groups are not all good at the same things.

In such social systems, people often fail to recognize the elegant dance that maintains order because by using terms like "alpha" we have convinced clients that dogs fight for status. Nothing could be further from the truth. Unless a dog is defending itself against an attack, fights are the exception and a signal that there is social breakdown. Normal dogs can accommodate pushy compatriots. They have trouble accommodating truly pathological bullies because--by definition--that's not on the agenda of the aggressor. With a video or a few observant minutes clients can tell the different. If behaviors involving deference (e.g., looking away, turning the neck, disengaging by rolling on the back, trying to leave, et cetera) are offered to the aggressor and the aggressor continues to block access, intensify the threat, or bite, there is a pathological problem here. Dogs who are working out relationships grab around the ears, sides of the neck, and shoulders, and although the noise is terrible, there is almost never any damage. True aggressors go for the front legs, the throat, the belly, and the eyes. The recipient of such aggression may grab the aggressor by the ear and hold on, because the aggressor then cannot get hold of the recipient's throat.

If there is true pathological aggression (4, 5) clients must realize that one of their dogs can be killed, and it will likely be the sweetest one. This situation cannot be ignored. Clients have 2 initial choices: 1) treat the dogs; 2) place one dog. If placement is the option the aggressor can only go to a single dog home where the new clients have a complete grasp of the situation and can and will comply with the degree of protection necessary. If the aggression involves only a 2 dog household, the victim may be very easy to place. If the clients decline placement and refuse treatment--they must either live with the dogs separately. The people--as the guardian of the dogs--must put their needs aside and address those of the dogs. Clients and dogs can live safely and happily with the dogs separated, but it is a lot of work.

The biggest determent to fixing or controlling the problem is the client who wants to have the peaceable kingdom. These clients believe that they can love and reason with the dogs enough and the behavior will stop. This client behavior is generally followed by the most serious attack yet. Clients who are committed to keeping their dogs and keeping them safe, can live well and happily with baby-gates, doors, and rotating car, play, training, and access schedules. The dogs can live happily, as well, with these rules.

These rules are almost impossible to enforce if you have young children or a deliberately non-cooperative housemate (dogs and cats are often victims of hidden agendas), and the smaller the house, the harder it is to succeed. However, before time-share co-habitation can happen, people have to realize in their gut--not intellectually--but at the gut level--that they can be responsible for the needless death of one of their dogs. People don't generally get this at once. But when they do, they often comment on what a "nice" dog the aggressor has become, and how calm the rest of the dogs are. Of course they are calm. They now have a rule structure that guarantees them safety, whereas before they had a totally unpredictable, mobile explosive device.

Belling the aggressor (Bear Bells, Silverfoot, Squamish BC, Canada; silverft@mountain-inter.net) also helps: the other dogs always know exactly where he or she is. It should be noted that in such households some dogs may be able to spend time with the victim and aggressor and never be attacked. When this is true, that dog can go back and forth and all lives will be enriched.

Behavior modification involves 2 passive first steps: 1) teaching each dog to sit and be calm before they can have any human attention (e.g., walks, tick removal, play, food, head collar placement, et cetera) [The Protocol for Deference] (6) , and 2) the clients must reward the dog who is behaving most appropriately. That determination is made as discussed above. The reward structure means that the most appropriate dog is fed, played with, let out, et cetera first, and the least appropriate dog, last. Size, age, and health, believe it or not, are second order concerns unless the differences are extreme.

If the clients start this early enough in the ontogeny of the conflict, the conflict will stop. If serious fights have occurred, this, alone, will not be sufficient. Medication (generally a selective serotonin re-uptake inhibitor [fluoxetine, paroxetine, sertraline] or a tricyclic antidepressant [e.g., amitriptyline, clomipramine]) and active behavior mod to desensitize and counter-condition the dogs to each other, first on lead, and then off lead must be combined. It's important to remember that most is aggression is the result of an underlying anxiety disorder (7): the aggressive reaction, itself, when exhibited by the aggressor, is a non-specific sign of underlying anxiety. Aggression is the outcome of a rule system gone wrong, not a normal, first line response in social interactions. It's equally important to remember that although the victims may have started out 'normal', they are likely not after a few months of unrecognized torture from one of their housemates. Most often, the victim becomes fearfully aggressive, and if they have any confidence at all they become adroit at pre-emptive strikes. Hence, both the victim and aggressor are likely to benefit from medication, albeit possibly not the same one.

Finally, the behavior mod must be intensive and consistent. Clients with serious fights more often opt for some behavior mod, +/- medication to decrease anxiety, and separation of the dogs. These clients know, to their sadness, that the victim often now needs medication, too.

Regardless, if the clients are at all worried about the dogs, or if the veterinarian is worried about the dogs, the dogs should be separated behind locked doors when no one supervises them. Threats can be as debilitating as a blow, and when a dog cannot escape a threat (and they cannot in most kennel situations) they become profoundly debilitated. By putting on "hold" the artificial concept of the peaceable kingdoms, clients can have a different kind of harmony, and develop a deeper understanding of, and relationship with, their dogs.


1.  Overall KL. Early intervention and prevention of behavior problems: Step 2--Routine screening for behavioral problems, Veterinary Forum August 2001: 42-46.

2.  McConnell PB. Acoustic structure and receiver response in domestic dogs, Canis familiaris. Anim Behav 1990; 39: 897-904.

3.  Ohl F. Ontogeny of vocalizations in domestic dogs, breed standard-poodle (Canis lupus f. familiaris). Zool Beitr 1996; 37: 199-215.

4.  Overall KL. Animal Behavior Case of the Month: Use of fluoxetine (Prozac) to treat complicated interdog aggression. J Am Vet Med Assoc 1995; 206: 629-632.

5.  Overall KL. Animal Behavior Case of the Month: Intrasexual interdog aggression in two pugs. J Am Vet Med Assoc 1995; 207: 305-307.

6.  Overall KL. Clinical behavioral medicine for small animals. Mosby, St. Louis, 1997.

7.  Overall KL: Neurobiology and neurochemistry of fear and aggression (1 hour CE). Proceedings--NAVC 1997; 11: 33-39.

Speaker Information
(click the speaker's name to view other papers and abstracts submitted by this speaker)

Karen L. Overall, MA, VMD, PhD DACVB
Center for Neurobiology and Behavior
Penn Med Translation Research Laboratory
Philadelphia, Pennsylvania, USA

MAIN : Animal Welfare : Canine Aggression
Powered By VIN