Significant debate surrounds the question of how to organize and classify the dog’s aggressive behavior into functionally discrete and logically coherent categories. Most trainers and counselors have adopted some variation of Moyer’s classification system () — a system that has resulted in a great deal of confusion and misunderstanding (). Other authorities have argued with varying degrees of cogency for a more simple classification system. O’Farrell (1986), for instance, has proposed a bipartite system, suggesting that canine aggression can be divided into two broad functional categories: dominance aggression and predatory behavior. This scheme places fear-elicited aggression under the same heading with dominance aggression: “‘Fear-biting’ is commonly distinguished from dominance aggression, possibly because it is felt to be understandable and excusable in a way that dominance aggression is not. It is, however, a variant of dominance aggression” (). Although simplicity is often desirable, this arrangement is not very edifying or useful when one considers the numerous motivational assumptions it takes for granted and the equally numerous distinctions that it blurs for the sake of Ockham’s razor. Further, the scheme stretches the concept of dominance in a way that further obscures its meaning and usefulness.
Since it is not clear how fear-related aggression might be used to enhance social status, O’Farrell’s position would be made more appealing and defensible if she explicitly replaced the term dominance with the term control related. However, although such a revision would help to reduce some potential confusion in her scheme, the change would only open up another criticism. Reducing aggressive behavior to a control-related motivation still begs the question with respect to the special attributes of aggression that distinguish it from other control-related activities. Presumably, all voluntary behavior is control-related behavior, but not all voluntary behavior is aggressive, except, perhaps, in a philosophical sense. Furthermore, although many forms of aggression appear to be purposive and control related, some forms of aggression appear to occur as reflexive actions in response to specific triggers. Also, defining aggression as a control-related activity tends to obscure the unique motivational and situational factors differentiating aggression into varied species-typical forms — even predatory aggression logically collapses into a control-related category. Broadly speaking, both affective aggression and predatory behavior are control-related activities, but they are significantly different in terms of functional purpose, evolutionary history, and neurobiological origins [see Neurobiology of Aggression (Hypothalamus)]. The most persuasive reason for adopting the concept of control-related aggression is that it conceptualizes offensive and defensive aggression in terms that are functionally compatible with the instrumental learning paradigm — learning concerned with establishing control over the environment. On the other hand, the underlying motivational factors (e.g., irritability, frustration, or fear) differentiating aggression into different forms represent the preparatory establishing operations facilitating the expression and potential reinforcement of control-related aggression. These emotional unitary reactions (conditioned and unconditioned reflexes) are under the influence of classical conditioning. In combination, instrumental control efforts and emotional unitary reactions converge on situations having species-typical significance for the dog, whereupon the specific intention of aggression is revealed (e.g., territory related, possession related, dominance related, or fear related).
Avoidance Learning and Aggression
Tortora (1984) has also suggested an alternative scheme for categorizing aggressive behavior. He interprets the development of aggression in terms of avoidance learning, arguing that what many dog behavior consultants refer to as dominance aggression is better understood as avoidance-motivated aggression (AMA). Social aggressors may not necessarily be dominant; instead, they may merely be incompetent and unable to respond appropriately under social pressure. Tortora argues that aggressive dogs appear to lack a repertoire of confident skills with which to cope and manage everyday challenges and stressors:
The data suggest that the initial source of the aggressive avoidance response was one or more forms of elicited aggression such as species-typical aggressive reactions to pain, frustration, discomfort, territorial intrusion, or threats to dominance. Furthermore, it appears that these aggressive responses were exacerbated by trauma or punishment. Finally, the universal lack of behavioral control over these dogs implies that they had few operant alternatives to gain reinforcement by compliance. From the case histories, it seems that these dogs were channeled down a path that allowed their initial innate aggressiveness to come under the control of the negatively reinforcing contingencies in the environment.
The dogs in this study initially behaved as if they “expected” aversive events and that the only way to prevent these events was through aggression. The consequent reaction of the victim and the family, that is, withdrawal, turmoil, and belated punishment, confirmed the dog’s “expectations” and reinforced the aggression. This positive feedback loop produced progressive escalation of the aggressive response, and the avoidance nature of the aggression presumably retarded or prevented its extinction. ()
The treatment program developed by Tortora is essentially a course of obedience training using various procedures, including remote shock, to enhance confidence and social competence. The operative assumption is that dogs exhibiting avoidance-motivated aggression need to learn systematically that they can safely control threatening or aversive events without resorting to aggression.
A strength of Tortora’s functional analysis is that it rests on a strong body of supporting experimental research (). In addition to being elicited by a variety of natural or learned triggers acquired through classical conditioning, aggressive behavior functions motivationally and behaviorally in a variety of ways, such as providing instrumental control over the physical and social environment. Animals can learn to avoid aversive stimulation by responding aggressively and may even learn and perform arbitrary instrumental responses to obtain an opportunity to attack a target provided as a reward (). The avoidance paradigm also offers an explanation for the persistence of some forms of aggressive behavior, since avoidance learning is marked by a strong tendency to persist over time and resist extinction (see Fear and Conditioning).
Tortora does not reject the notion of elicited aggression (e.g., irritable, territorial, or dominance related), but stresses that the dog’s repertoire of species-typical aggressive behavior and controlling natural triggers only represents part of the picture. Although aggression may be originally elicited by a natural trigger, it can subsequently come under the control of conditioned stimuli or learned triggers through avoidance learning. In fact, dominant-aggressive dogs frequently do exhibit behavior that appears to be influenced by avoidance learning. A common and often confusing characteristic of dominance aggression is the absence of an adequate trigger to explain the ferocity of the attack. In other words, the magnitude of dominance attacks is frequently far in excess to what one might expect to occur under the operative circumstances present at the time. Such attacks often appear to occur under minimal or no provocation at all. Low-threshold or unprovoked attacks may be explained along the lines of avoidance conditioning, whereby neutral stimuli present at the time of attack may become conditioned or learned triggers via association with unconditioned or natural triggers. As a result, these conditioned triggers may become capable of eliciting aggressive behavior in the absence of natural triggers. In other words, aggressors may learn to anticipate aversive arousal (frus-trative, irritable, painful) by association with other stimuli present at the time when aversive arousal led to aggression. As a result, such stimuli may gradually become discriminative signals controlling avoidance-motivated aggression. According to this general account, most aggressive behavior is learned as a means to anticipate and avoid actual or perceived threats, especially threats occurring under circumstances where other means of control are unavailable or ineffectual. Bottom line, according the AMA hypothesis, dominance aggression toward human targets appears to be more about defensive control than offensive status-seeking efforts.
Many attack situations involving defensive aggression (that is, aggression influenced by a component of fear or avoidance) appear to present characteristics consistent with Tortora’s AMA hypothesis; however, attacks motivated by anger do not appear to involve an underlying component of fear. The offensive aggressor may exhibit threatening postures and gestures (standing tall and stiff, ears up and turned forward, tail held erect, lips forming an agonistic pucker), behavioral signs indicating confident aggressive arousal — not preparatory fear. The offensive aggressor may learn that threats and attacks serve to secure or protect vital interests and resources. For example, a dog that has learned to threaten and displace its owner in the presence of food may learn through positive reinforcement (that is, the continued possession of food) that such behavior works. Determining whether the particular behavior is under the control of positive or negative reinforcement depends on whether the behavior functions to terminate or avoid stimulation or serves to obtain or perpetuate stimulation. When aggression occurs while the dog is sleeping, resting, or eating, the attack may be analyzed in terms of the perpetuation of these activities or resources, that is, understood by appealing to positive reinforcement. On the other hand, such attacks may also be analyzed in terms of negative reinforcement, especially if the behavior is primarily motivated to terminate the presence of a threat, a source of irritation, or frustration. In general, the determination of whether aggression is defensive (avoidance motivated) or offensive depends on the presence of behavioral signs at the time of attack. However, distinguishing between these two forms of aggression on the basis of postural signs of fear may become progressively difficult in the case of the experienced avoidance-motivated aggressors, who may not exhibit any signs of overt fear, especially as they become progressively confident and sure about the likelihood of success. In general, though, whether defensive or offensive, aggressive behavior aims at establishing control over some intruding target. The notions of positive and negative reinforcement may actually obfuscate the vital concern; that is, aggression is reinforced by the control it succeeds to establish, regardless of the motivational substrate operative at the moment of attack (see A Brief Critique of Traditional Learning Theory). According to this perspective, punishment occurs when the aggressive threat or attack fails to avoid or terminate an aversive-thwarting situation or when it fails to obtain or perpetuate some gratifying activity or resource. These considerations are of vital importance for the effective control and management of aggressive behavior.
Social Dominance and Aggression
One of the most well-studied areas in ethology is aggression, especially aggression exhibited with the apparent purpose of establishing or defending social status or rank (). The most widely adopted conceptualization of dominance aggression incorporates several interactive components, including an ethological dominance concept, species-typical signalization (gestures and postures signaling agonistic intent and rank — dominance and submission), early socialization, and learning. Dogs and humans both socially organize themselves by rank order (often involving very complex alignments) within a dominance hierarchy. Such organization is accomplished by various means, including the utilization of species-typical gestures and body postures employed to advertise and ritually defend the individual’s status against challenges presented by others belonging to the same group. Sometimes, communication breaks down as the result of a misunderstanding or an outright power struggle, giving rise to conflicts and challenges that may escalate into overt attacks and fighting. The role of social dominance in the expression of aggression is examined in detail in Social Dominance and Aggression.
Fear and Aggression
Normally, fear significantly inhibits aggressive behavior and causes the animal to freeze or flee — if it can. A fearful dog usually makes frantic efforts to escape when it feels threatened or is attacked. It is only under circumstances in which escape or appeasement is thwarted that a fearful dog may resort to aggression. First and foremost, the goal of fearful behavior is to escape or control threatening stimulation, with counterthreats and aggression emitted as a last resort. Fear aggression is always a defensive strategy and is most likely to occur when other means of escape or avoidance are thwarted. However, in cases in which fear aggression succeeds, the defensive threat or attack may undergo reinforcement and, under similar circumstances in the future, the behavior may be triggered by conditioned stimuli associated with the original eliciting situation. As already discussed, the result is the development avoidance-motivated aggression — behavior that may closely parallel dominance aggression but remains essentially defensive rather than offensive. Fear aggressors can be distinguished from dominance aggressors by the exhibition of defensive postures indicative of fear (e.g., ears back, tail tucked under the body, nervous snarling, and showing of teeth) and approach-avoidance conflict. In addition, the fear aggressor may engage in barking (a possible repetitive conflict behavior) and other signs of fearful arousal (licking movements) and agitation that occur when it is exposed to eliciting stimuli, such as a doorbell, the approach of a stranger, noisy children, skaters and other similar stimulation, or the approach of other dogs. Typically, the fear aggressor is most likely to threaten or bite when it is suddenly approached by a fear-eliciting person or dog, where escape is preempted (). Once fear aggression has graduated into avoidance aggression, many of the telltale signs of fear may be replaced with increased confidence and reduced latency and occur under minimal provocation.
Fear-related or defensive aggression stands opposite to dominance-related or offensive aggression on the agonistic continuum. Whereas dominance aggression occurs most often in situations involving competitive conflict between conspecifics, stimulated by the coactive influences of frustration, irritability, and anger, defensive aggression is most often directed toward another group member or species, under the influence of acute threat, fear, or anxiety. The tendency to bite out of fear is most commonly seen among shy or nervous dogs that have learned to rely on biting as means of self-defense. Paradoxically, fear-related aggression and dominance aggression sometimes present together in the same dog. The term bipolar aggression is a good descriptive term for this condition, since opposing ends of the agonistic continuum appear to be alternately involved, depending on the situation.
Since fear-related aggression depends on the presence of fear for its expression, an important initial step in the counseling process is to make an exhaustive inventory of the evoking stimuli and situations where aggression has occurred in the past. Detailed information should be gathered concerning the location, magnitude, and type (superficial, puncture, laceration, etc.) of the bites involved. Further, the originating causes of fear and aggression should be fleshed out and clarified. This is not always practical or possible, but an effort should be attempted in every case since the results are often very useful in terms of accurately describing the problem, prognosticating the likely outcome and benefit of training, and helping owners to understand their dog’s problem. Fear-related aggression appears to be strongly influenced by predisposing genetic factors. Thorne (1944), for example, found that a single “fear-biting” Basset hound had a tremendous influence on a large group of descendants in terms of their relative fearfulness and reactivity. Of 59 dogs related to this highly reproductive female, 43 (73%) were shy and unfriendly. In addition to genetic predisposition, most etiological profiles show significant causality in terms of early socialization and exposure deficits or the contribution of learning. It is not uncommon to find cases involving all three factors. Voith and Borchelt (1996) suggest that excessive punitive interaction with puppies during house training may play a significant predisposing role in the development of fear-related aggression problems in adult dogs.
Distinguishing the effects of learning from other potential causes of fear is assisted by obtaining a behavioral history and performing a detailed evaluation. Dogs that are affected by a genetic predisposition are distinguished by a chronic, lifelong, and generalized fearfulness. They often suffer heightened or extreme sensitivity to sensory input and overreact in situations involving novel stimuli (neopho-bia), strangers, or unfamiliar animals. Differentiating cases exhibiting a genetic predisposition from those involving a socialization deficit is not always easy, since undersocialized dogs frequently exhibit similar signs and tendencies as genetically affected individuals. Temperament information about a dog’s sire and dam could be helpful in making such determinations. Puppies isolated until week 14, or in cases where they come into the home at an unusually late date from an unknown situation, should be suspected prima facie as suffering a socialization problem. A lack of proper socialization and inadequate or traumatic environmental exposure occurring early in development are commonly associated with adult dogs’ reactive fear toward strangers (xenophobia), fear of children (pedophobia), or fear of outdoors and new places (agoraphobia). In cases where fear-fulness is the result of learning (e.g., startle, trauma, or abuse), a dog’s reactions are usually limited to a more specific range of eliciting stimuli and situations. Of the three aforementioned etiologies, fearfulness stemming from past learning events is usually the most responsive to remedial training, with problems involving a genetic predisposition being the most difficult to work through in my experience. Fear biting suspected of being predominantly under the control of an underlying genetic causation should be carefully assessed and the owner informed of the limited benefits to be expected from behavior modification before proceeding. Although such dogs may respond to behavioral intervention, the goals of training should be discussed in terms of amelioration and management — not cure.
Obviously, reducing fearfulness is central to effective behavioral control and modification of fear-related aggression. Several methods have been employed for this purpose with varying degrees of success. The most beneficial techniques involve some combination of graded interactive exposure, counterconditioning, relaxation training, modeling, and response prevention. The most important consideration recommending the use of such procedures is that they help to facilitate the disconfirmation of a dog’s adverse expectations of social contact while at the same time encouraging a more affirmative set of expectancies and interactive behaviors.
Cognition and Aggression
Dogs appear to form various prediction-control expectancies about future events based on the accumulation of information extracted from past experiences (see Prediction-Control Expectancies and Adaptation in). In general, these expectancies help to promote a more secure existence by coordinating a dog’s behavior relative to the most probable, although not yet actual (certain), circumstances. Prediction-control expectancies are continually undergoing appraisal and modification in order to most accurately fit or adapt a dog’s behavior to the environment. These expectancies are influenced by emotional concomitants of success (elation) or failure (disappointment). In cases in which a high degree of correspondence exists between what a dog expects to occur and what actually occurs, effects of well-being and confidence prevail; whereas, under opposite circumstances in which their is little correspondence between what the dog expects to occur and what actually occurs, effects of depression and helplessness may ensue. Expectancies are adjusted in accordance with the occurrence of satisfying or distressful emotional concomitants resulting from the confirmation or discontinuation of expectant arousal or action. For example, when a prediction expectancy is discontinued or proves inadequate, then anxiety ensues. On the other hand, when a control expectancy is discontinued or proves inadequate, then frustration ensues. These emotional concomitants of expectancy disconfirmation promote adaptive optimization through the activation of increased sensory vigilance and behavioral invigoration. Adaptive change is mediated through learning, and learning is guided by the affects of anxiety and frustration, resulting from the disconfirmation of prediction-control expectancies. Theoretically, when prediction-control expectancies are fully matched and coordinated with the environment, utopic adaptation is achieved, and further learning is unnecessary and does not occur. Under ordinary circumstances, anxiety and frustration promote learning and adaptive optimization of environmental resources. However, under conditions in which the environment is both highly unpredictable and uncontrollable, then pathological disorganization (learned helplessness) and behavioral disorder (impulsive-compulsive behavior) are prone to follow. In other words, a small amount of anxiety and frustration promotes adaptive success, whereas high levels of anxiety and frustration disturb learning and disrupt behavioral adaptation.
Preparatory arousal, attention, intention, and functional behavior are guided by prediction-control expectancies. Under ordinary circumstances, dogs select courses of action based on cognitive expectancies, unless the particular expectancy has been disconfirmed or the environment provides inadequate information with which to form adequate prediction-control expectancies. Under such circumstances, dogs may depend more on direct sensory information, until a more adequate expectancy is formed. This shift from expectancies to reliance on sensory information may be highly disruptive and stressful. In addition to resorting to sensory information, dogs may also be more inclined to rely on instinctive or species-typical impulses to secure the environment. Under highly threatening social situations which violate a dog’s prediction-control expectancies (e.g., trust), increased sensory vigilance and behavioral invigoration may facilitate intense aggressive arousal and significantly lower thresholds for aggressive behavior. Unfortunately, as a result, the dog may modify prediction-control expectancies so that, under similar circumstances in the future, it may learn to preemptively prepare and respond aggressively under minimal stimulation and continue doing so until the operative expectancy is disconfirmed.
Although prediction-control expectancies may accurately reflect reality, under the influence of adverse learning dogs may form faulty expectations that may not adequately represent actual circumstances. This risk is particularly problematical in the case of escape and avoidance learning, in which case the avoidance response may preemptively interfere with a dog learning that the response is no longer necessary to control the anticipated threat (see A Cognitive Theory of Avoidance Learning). For avoidance to discontinue, the operative prediction-control expectancy guiding the behavior must be first disconfirmed (e.g., via graduated interactive exposure and response prevention) and replaced with an alternative expectancy more adequately fitted to the actual situation. In addition to forming specific expectancies, dogs also appear to appraise and interpret events in very subtle ways that predispose them to preferentially engage in certain behaviors rather than others. Cognitive appraisal assists dogs in modulating their moment-to-moment arousal levels as well as finely regulating appropriate actions to achieve a more subtle behavioral adaptation to the environment. These interpretive cognitive functions are especially evident in the case of complex social circumstances requiring a high degree of communication and cooperation, such as play. In the case of play, aggressive elements and sequences are interpreted in terms of the play partners intention and various play metasignals confirming that the interaction is just play. Interpretive appraisal of the social intention of others provides the basis of communication. The mutual communication of intent determines whether competitive or cooperative behavior will ensue between interactants. Communication of intent may not only predict aggressive or affiliative action, it may also define the most likely outcome of the encounter. A highly motivated dog (e.g., starving) may show a very strong intent to defend a bowl of food, sufficient to cause a less hungry potential competitor to withdraw — even though under other circumstances the competitor may be dominant and the aggressor submissive. The way a dog interprets the intention of interaction strongly influences how it will respond to it. Petting or hugging coming from one person may be welcomed and reciprocated with expressions of shared affection, whereas the same actions coming from another person may be interpreted as a threat and, perhaps, evoke an aggressive response. Such interpretations of intent are strongly influenced by the quality of attachment and communication between the human and the dog. Interpretive appraisal of social intention under the influence of high levels of affection, familiarity, and trust appears to promote strong and durable inhibitory effects over aggression between closely bonded interactants.
Another important cognitive influence over aggressive behavior is cost-benefit assessment and risk taking. Engaging in aggressive conflict brings with it considerable risk. Cost-benefit assessment appears to play a significant role in the case of offensive aggression, where the goal is to achieve some benefit or resource. In the case of a starving dog, the risk of injury that may result from fighting is offset by the benefit of eating. In situations where the potential cost of behaving aggressively (loss or injury) exceeds what might be conceivably gained by the action, a dog is more likely to steer away from initiating an aggressive conflict. Aggression is most likely to occur in motivationally significant situations, where the risks of aggression are minimal (costs) and the potential benefits are substantial. Finally, dominant dogs appear to be more inclined to engage in risk-taking behavior, whereas submissive dogs may be more conservative and careful regarding risky behavior. A predisposition to take risks may be a genetically expressed trait that is more characteristic of dominant individuals than submissive ones. Submissive individuals may be genetically prone to avoid risk taking, unless the perception of risk is motivationally offset by a pressing biological need or threat and the potential benefit of success is sufficiently enticing.
A potential factor altering risk assessment abilities is stress. Quatermain and colleagues (1996) have found that stressed mice more rapidly engage in risk-taking behavior than unstressed controls. In the case of dogs, stress may lower thresholds for aggressive risk taking, causing otherwise submissive and compliant dogs to become periodically more irritable and aggressive. Stress appears to impair normal attention and memory functions (Mendl, 1999) and cortical impulse control over sub-cortical activity (), potentially lowering behavioral thresholds for aggression or liberating species-typical offensive and defensive behavior in response to wrongly interpreted social signals. The systematic reduction of stress is an important aspect of effective behavior therapy. Such treatment efforts may facilitate risk-assessment normalization and improve other cognitive functions involved in the modulation of aggressive arousal and the regulatory control of aggressive behavior. A neural site of particular interest in this regard is the amygdala (see Limbic System), which appears to serve a central role in social communication by mediating direct eye contact, by interpreting socially significant facial expressions, and by assessing the interactants emotional disposition and intent (). Under the dysfunctional influence of excessive stress, the intent of social signals may be distorted and misinterpreted, causing a dog to respond with inappropriate fear or aggression. The amygdala may play a particularly prominent role in the case of dominance-related aggression ().