The Special Case of Our Canine Companions
Besides love and sympathy, animals exhibit other qualities connected with the social instincts, which in us would be called moral; and I agree with [Louis] Agassiz . . . that dogs possess something very like a conscience.
Dogs possess some power of self-command, and this does not appear to be wholly the result of fear.
—Charles Darwin, The Descent of Man
THIS CHAPTER SHIFTS from our closest evolutionary relative to man’s best friend, animals that have been our companions for at least 18,000 years, predating the emergence of agriculture (Thalmann et al., 2013). It is possible that our symbiotic relationship with wolves was the gateway to ultimate domestication, which led to further selection of prosocial traits (Coppinger & Coppinger, 2001; Feuerbacher & Wynne, 2012). From a Darwinian continuity perspective, a key question is how the basic temperaments of dogs may be close to traits we value in humans, while perhaps being less complex than those of chimpanzees.
Methodologies in dog personality research involve more laboratory-type measurements of dogs’ responses to standard challenges, in contrast to what must necessarily be more naturalistic observational research with chimpanzees. Might using these behavioral test batteries, as opposed the observer rating approach, lead to different conclusions? The extent to which temperament findings in dogs and chimpanzees remain similar, despite shifting from more flexible observer rating studies to more laboratory-type challenges, will strengthen the evolutionary case for similar, even homologous (evolutionarily related) emotional foundations of mammalian personalities. We spend a bit more time on dogs than on monkeys and rats because dogs figure so heavily in issues of human-animal companionship. Dogs are, after all, “man’s best friend,” although they may not see it quite that way. Thus, we divide this chapter into two parts: first we describe Scott and Fuller’s behavior genetic studies and related work that followed, and then we discuss work on quantifying socialization, including examples of puppy testing and a long-term project domesticating another canine, the fox.
With dogs, the personality discussion may move into more familiar territory for most readers, because most folks have lived with a dog or even several dogs at some point in their lives. Certainly, those who have had canine companions would agree that dogs have personalities that reflect stable behavioral and emotional characteristics that differentiate their dogs from other dogs. Of course, due to selective breeding started many centuries ago, we now have an enormous variety of dogs available for comparisons. Indeed, the American Kennel Club registers 175 different breeds, with many manifesting distinct behavioral-temperament differences. While we recognize that there are many passionate cat lovers as well, there has been much less research on cat personalities than on dogs (Gartner & Weiss, 2013). Hence, here we have selected domestic dogs, as a unique mammalian species that probably exhibits diverse genetically based personality traits both across and within subspecies, many of them arising from selective breeding for certain behavioral/psychological characteristics across the last few centuries. More important, our long historical relationship with dogs was based on the fact that we developed a symbiotic relationship long before the agricultural revolution, going back to a time when humans were still hunter-gatherers, starting perhaps some 40,000 years ago, during the peak of the last glacial period, with gradually increasing breeding of canine companions for specific utilitarian traits. As noted by Leslie Irvine (2004), “Selective cross-breeding has been done since antiquity, but it really accelerated during the 19th century.” Still, it was another century before humans became interested in the scientific analysis of canine personalities, which had been pragmatically crafted to serve human needs across the past few centuries (Serpell, 1995).
ASSESSING CANINE BEHAVIOR AND PERSONALITY
Scott and Fuller: Social Behavior of the Dog
The classic work of John Paul Scott and John L. Fuller (1965), at the Jackson Memorial Laboratory in Bar Harbor, Maine, set a high quality standard for any future temperament research in dogs. J. P.Scott was recruited in 1946 to initiate a large-scale project in dog behavior genetics at the Jackson Lab. In collaboration with Fuller, Scott set out to fill a void in our understanding of the influence of heredity upon behavior, a subject that was not popular in the zeitgeist following World War II, with the residue of German genetic fantasies, such as Übermensch, fresh in people’s minds. Overall, Scott and Fuller’s work still highlights both the difficulties and promises of large-scale temperament studies in complex domestic animals, a body of work that remains to be equaled.
At least thirteen years were needed to design, set up, and carry out their extensive behavior genetic research plans. Rather than employing the selective breeding frequently used with rats, Scott’s group took advantage of existing breed differences and worked on the assumption that breed differences reflected genetic differences. So, they based their work on five breeds of small dogs that spanned the temperament spectrum: Basenjis, beagles, cocker spaniels, Shetland sheepdogs, and wirehaired terriers.
Although both Scott and Fuller had backgrounds in biology, the neuroscientific and other physiological measures and manipulations that were available in the 1950s were severely limited compared to the research options that are available today. While they did collect some physiological data, they relied largely on diverse objective behavioral tests they designed to expose the dogs to environmental and behavioral challenges that might reveal individual and breed differences. These objective tests mostly allowed for timing or counting specific behavioral responses (rather than relying on subjective ratings by handlers or observers) to over twenty behavioral tests collected throughout the first year of life on large samples of all five breeds. Many of these tests were repeated throughout the dog’s first year of life to quantify developmental changes; this resulted in the administration of over fifty formal tests, plus numerous incidental observations such as whether puppies could learn to hold still when being weighed during weekly health examinations. We refer to many, but not all, of these unique objective tests and behavioral observations in this brief review of their overall study (for a comprehensive summary, see Scott & Fuller, 1965). In a sense, Scott and Fuller devised the first systematic canine emotional-personality tests, which is why we cover it here in some detail.
In addition, two of these breeds, Basenjis and cocker spaniels, were selected for a more complex genetic analysis involving simple crosses of the two breeds, labeled F1s, and crosses of those F1 progeny called F2s (F stands for “filial” or breeding family stock). Back-crosses from F1s to each of these two parent breeds were also completed to check and provide initial verification of any genetic hypotheses. While Scott and Fuller showed that many breed differences could be accounted for by one and two gene models, they surprisingly found that “the vast majority of genetic effects on behavior are highly specific and restricted to one or two situations” (1965, p. 375) rather than cutting across many behavioral traits. This program of research highlighted that trans-situational temperamental variables may be rare and hence difficult to genetically analyze. It is possible that such background traits were emotional ones that are harder to select for than specific behavioral abilities, but Scott and Fuller did not neglect to focus on the emotions of their animals.
One theme running through their work was the similarity of social bonding with conspecifics (members of the same species) and heterospecifics (across other species, including humans), work that highlighted the closely related development of social emotional attachments and place attachments in young puppies. How was this determined? When old enough to maintain body temperature and move about on their own, puppies that were left alone in their home pens emitted loud “distress vocalizations” at a very high rate, signaling their aversion to social isolation. However, they send out even higher rates of distress vocalizations when isolated in a strange place, indicating that distress was amplified when puppies were no longer in a familiar place (Elliot & Scott, 1961). This heightened level of vocalization began tapering off at about eight weeks of age, at the end of what additional research later indicated was the waning of a “critical period” for social attachment (Freedman, King, & Elliot, 1961; Scott, 1958; Scott, Fredericson, & Fuller, 1951). Upon the publication of their book Genetics and the Social Behavior of the Dog (Scott & Fuller, 1965), Scott was invited to be the first Ohio Regent’s Professor of Psychology at Bowling Green State University (BGSU), where the authors of this book worked with Scott in the early 1970s.5
At Scott’s BGSU dog lab, Panksepp’s group discovered that these distress vocalizations were specifically modulated by brain opioids (Panksepp et al., 1978) that had recently been discovered. The key finding was that tiny doses of opioids—namely “feel good” chemistries in the brains of all vertebrates that are naturally very rewarding (indeed addictive, when opioids are concentrated in various drugs)—could reduce separation distress vocalizations in young puppies. Across the late 1970s and 1980s, this led to a research program focused on the identification and description of a separation-distress brain system using direct brain stimulation procedures in guinea pigs and domestic chicks (summarized in Panksepp et al., 1988; Panksepp, 1998a). This research program highlighted a new social emotional system in vertebrates: the PANIC/Sadness system, which contributed substantially to social bonding. In other words, social bonding was partly due to addictive chemistries of the brain, helping explain why the loss of loved ones is so psychologically painful. For the first time a brain social-emotional system had been identified that no one had talked about yet. Parenthetically, at the same time Harry Harlow’s group at University of Wisconsin was demonstrating how emotionally disastrous it was for young rhesus monkeys to live alone (for a poignant summary of that work, see Blum, 2002).
While Panksepp’s (1982, 1998a) remaining primary-process (subcortical) brain emotional systems were worked out in rats and other species, the empirical relationships between the other brain emotion systems and dog behaviors (indirectly evident in Scott and Fuller’s work) still remain to be clarified neuroscientifically. Yet this multigenerational research, spanning the careers of both Scott and Panksepp, highlights the value of how clear animal models for emotions, when combined with direct brain manipulations/understandings, have the power to impact human psychiatric practice (for summaries, see Chapter 18; see also Panksepp, 2015, 2016). The implications of such findings for the understanding of human personality remain to be fully explored, for which we developed the Affective Neuroscience Personality Scales (Davis et al., 2003; Davis & Panksepp, 2011).
Another common social behavior in dogs, enthusiastic tail wagging so common during social interactions, was measured as part of the Jackson Lab project during the research program’s “handling test.” These data showed that Basenjis were less attracted to human handlers and cocker spaniels and beagles were the most attracted. Correspondingly, tail wagging itself appeared two weeks later developmentally in the Basenjis than in cocker spaniels. The difference in appearance of cocker and Basenji tail wagging was highly significant and both differed from the other three breeds, which were intermediate. It was later demonstrated at the BGSU Canine Research Laboratory that tail wagging could also be manipulated in very young and juvenile puppies with opioids and the opioid blocker naloxone (Davis, 1980; Knowles, Conner, & Panksepp, 1989), which like separation distress vocalizations linked canine tail wagging to emotional social bonding/attachments and indirectly to the PANIC/Sadness system.
At Jackson Lab, Scott and Fuller (1965) had also investigated play fighting (also known as rough-and-tumble play) in their dogs. Play fighting with human handlers peaked at about fifteen weeks of age and was also associated with definite breed differences. Among the breeds studied by Scott and Fuller (1965), wirehaired terriers exhibited the most play fighting in response to standard handling tests. Basenjis, Shetland sheepdogs, and beagles played at comparable levels, while cocker spaniels exhibiting considerably less play fighting.
However, the correlation between play fighting and later real fighting between conspecifics (other dogs) was found to be low, suggesting that play fighting was not related to aggressive (RAGE/Anger) sensitivities. Indeed, these distinctions were highlighted in their findings. For example, beagles exhibited a typical high level of play fighting in the handling test (Scott & Fuller, 1965, p. 136) but exhibited the lowest level of serious fighting as observed in dominance tests (p. 155). Cockers showed low play fighting and also low levels of dominance fighting. Wirehaired terriers and Basenjis showed high levels of both play fighting and serious fighting, as manifested in “complete dominance” tests. This segregation of real fighting and play fighting likely reflected the activity of two different brain systems, one linked more to the RAGE/Anger system and the retaining of resources such as food, and the other related more to the PLAY system with its social playfulness and positive socialization, distinctions that were evident in many other species (Panksepp, 1998a). While it was not the subject of a primary research project, Scott and Fuller (1965) had already suggested that play fighting reflected a positive social experience, facilitating positive social integration. The relationship between positive affect and playfulness was later empirically confirmed in rats (as described in Chapter 9).
In the general-purpose handling test, Scott and Fuller also measured the avoidance of human handlers, which was interpreted as timidity or fear, likely reflecting the primary FEAR system. Cocker spaniels showed the lowest avoidance of human handlers. Beagles were also less avoidant in the handling test, with the other three breeds exhibiting higher avoidance levels. In addition to being among the most avoidant breeds on the handling test, Shetland sheepdogs were also the poorest performers on the motor-skill and spatial-orientation tests, both of which required climbing, which Scott and Fuller (1965) attributed to shelties being quite fearful of heights.
Another case of fear affecting test performance was on the trailing test that required dogs to track a scent down a series of branching boards leading to a fish treat. All breeds did well except the wirehaired terriers and Basenjis. However, the poor performance of Basenjis was due a “fear reaction to the apparatus” (Scott & Fuller, 1965 p. 247) and refusal to cross the board trail. Basenjis were not afraid of heights like the shelties and were skilled climbers, but their cautious reaction to the strange apparatus interfered with their scores on the motor-skills test and was another example of the highly specific nature of many instinctive fears.
The leash training test measured the tendency of juvenile puppies to adapt to being on the leash or to fight this restraint. Scott and Fuller noted that “Basenjis were outstanding in their vigorous resistance to the restraint of a collar and lead” (1965 p. 209). The heritability for fighting the restraint of the leash was estimated from 0.45 to 0.77 (with 1.0 reflecting perfect heritability). Leash fighting with juvenile puppies may be a good model reflecting the RAGE/Anger system in action, which responds to physical restraint possibly based on an evolutionary adaptation to being captured by a predator (Panksepp, 1998a). Similarly, when placed in the restrictive Pavlov stand for the reactivity test (similar to that used by Ivan Pavlov for his classic studies on canine classical conditioning of salivation), Basenjis also exhibited a strong tendency to bite. While Basenjis showed the highest and cockers the lowest percentage of biting in the Pavlov stand at all ages, the differences became greater with each successive trial. By fifty-one weeks, 83 percent of Basenjis and only 7 percent of cocker spaniels were biting the restraint. Hence, the marked personality differences between aggressiveness and fighting across species—with Basenjis and cocker spaniels being respectively at the high and low extremes—demonstrating hereditary influences from puppyhood through early adulthood, potentially in the responsivity of the RAGE/Anger system.
Scott and Fuller’s (1965) overall emotional reactivity test was administered at seventeen, thirty-four, and fifty-one weeks of age, during which each dog was restrained in a Pavlov stand and subjected to ten stressful episodes with eighteen behavioral and physiological measures. Cocker spaniels showed the lowest overall reactivity scores, and Scott and Fuller interpreted the reactivity test as a measure of inhibitory training. They related the cocker’s relatively calm performance during this test to the breed’s historic capacity to respond to threats, especially hand motions, by ceasing all activity for a moment with no signs of emotional disturbance. This conclusion was further supported by the Jackson Lab’s weekly medical checkup procedures, during which puppies, in order to be weighed, were trained to remain quiet for one minute by the handler holding his or her hands near the animal but attempting not to touch the puppy standing on the scale. By fourteen weeks of age, 80 percent of cocker spaniel puppies would remain quietly on the scale for one minute, which was more than twice the percentage for any other breed. In wirehaired terriers, also the most reactive and least inhibited breed at fifty-one weeks on the reactivity test, only 10 percent of the puppies could quietly remain on the scale for one minute. It is tempting to associate the cocker’s capacity for inhibition as a cross-species form of the human Big Five Conscientiousness trait, which has been thought to represent the cortical regulation of subcortically based primary traits (Davis & Panksepp, 2011) and which in Chapter 7 was identified in chimpanzees and brown capuchin monkeys but has otherwise been difficult to identify in nonhumans.
Shetland sheepdogs showed slightly lower levels of inhibition than cocker spaniels during the reactivity test at all three ages. By contrast, wirehaired terriers, beagles, and Basenjis were consistently much more reactive and less inhibited than cockers and shelties. Like cockers, shelties were less reactive and more easily inhibited in the Pavlov stand, but in contrast to cockers, shelties exhibited high levels of fearfulness in other tests, as noted previously. In fact, among cockers, beagles, shelties, and Basenjis, all four combinations of high and low inhibition (on the reactivity test) and with high and low fear (on the human handler test) were exhibited, suggesting that inhibition training (low reactivity) and fearfulness (high avoidance) segregate independently genetically, as illustrated in Table 8.1.
| Table 8.1. Segregation of Fear and Reactivity Across Four Dog Breeds | ||
| Fear and Avoidance | Reactivity | |
| Low | High | |
| High | Shetland sheepdogs | Basenjis |
| Low | Cocker spaniels | Beagles |
| Adapted from findings reported in Scott & Fuller (1965). | ||
Scott and Fuller (1965) described maternal care of puppies but only indirectly touched on the CARE system by frequently hypothesizing maternal effects, including maternal effects on aggressiveness, which have now been shown in rats (Parent & Meaney, 2008), as discussed in Chapter 9. Also, Scott and Fuller (1965) decided not to study investigative behavior, even though the investigative system was one of Scott’s nine categories of behavior systems (Scott, 1958, 1972). Exploratory/investigative behavior, to monitor the status of what would relate to the primary SEEKING system, was judged to require too much space, given its relationship to hunting in the dog, and would have been too difficult to study experimentally in the field.
With a behavior genetics approach to behavior, these pioneers broadly sampled what we would view as the affective neuroscience temperament dimensions, which, for example, included demonstrations of the importance of distress vocalizations that would later be linked to Panksepp’s PANIC/Sadness system. Their research also included carefully measured individual differences in the PLAYfulness, FEAR, and RAGE/Anger systems. One might also consider cocker spaniel (and to a lesser extent sheltie) inhibition as a nonhuman candidate for the Big Five dimension of Conscientiousness and the regulation of emotion. Scott and Fuller’s discussion of what we would interpret as the SEEKING and CARE systems rounded out a rather complete treatment of the mammalian personality domain. In a highly objective study that predated modern neuroscience, Scott and Fuller impressively documented what Panksepp was identifying as primary mammalian emotion systems. Once again, a comprehensive analysis of mammalian behavior provided a set of traits that have been reported with remarkable consistency starting with Darwin and leading through so many other contributors to neuroscientifically documented blue ribbon emotions of affective neuroscience (Panksepp, 1981, 1982, 1998a; Panksepp & Biven, 2012). Scott and Fuller’s work invites further research exploring the empirical relationships between their many observations and these brain emotion systems underlying dog behavior. In addition, their work set the stage for later dog testing, which we now selectively review.
More Canine Assessments
Scott and Fuller’s work inspired subsequent behavioral dog testing. We begin with studies on dogs tested using canine behavior questionnaires and then briefly review behavioral tests for dogs in military work and one study on pets in a rescue shelter. We reserve for the second part of this chapter research on the quality of social attachments in dogs, followed by a brief review of puppy testing research on selecting guide dogs for the blind (Pfaffenberger, Scott, Fuller, Ginsburg, & Bielfelt, 1976) and police dogs, all with the goal of comparing the breadth of canine personality dimensions to Panksepp’s six primary emotions related to human emotional personality issues.
Swedish Working Dog Association Studies
Svartberg and Forkman (2002) studied personalities of pet dogs using data from a dog behavior rating questionnaire developed by the Swedish Working Dog Association called the Dog Mentality Assessment (DMA). The DMA was developed mainly for dog breeders, but it became a popular general assessment with owners as well that has resulted in a massive database containing over 15,000 pet dogs from 164 breeds. The DMA evaluates dogs interacting with a familiar handler in response to ten separate subtests measuring thirty-three behavioral variables, which measure a dog’s reactions to strangers, opportunities to play with a human, reactions to prey-like objects, and several potentially fearful and aggression-inducing situations. From the forty-seven most popular breeds, Svartberg and Forkman randomly selected 1,175 adult dogs that averaged about two years of age.
An exploratory factor analysis on their large dataset yielded five latent dimensions: Playfulness, Curiosity/Fearlessness, Chase Proneness (including following and grabbing behavior), Sociability, and Aggressiveness (including threat displays and attacks). The Chase Proneness factor included ratings from only a single test, and the researchers tentatively interpreted it as predatory behavior, which based on rat research (discussed in Chapter 9) would align it with Panksepp’s SEEKING system. Although unpublished, some work was done in Scott’s BGSU dog lab examining predatory responses of puppies to small papier-mâché animal models with rudimentary limbs and heads, which also suggested the possibility of studying canine predatory behavior as a distinct temperamental dimension.
Svartberg and Forkman were also able to document Panksepp’s PLAY system with their Playfulness factor and the RAGE/Anger system with their Aggressiveness factor. They also described their Curiosity/Fearlessness factor as an “analogue to fearfulness, but inverted” (Svartberg & Forkman, 2002, p. 152) and compared it to the general fearfulness observed in other studies, which suggests their Curiosity/Fearlessness factor may be related to Panksepp’s FEAR system.
Each of the behavior ratings on the Sociability factor involved initiating physical contact with a stranger along a continuum that included “intense greeting with whining and jumping” (Svartberg & Forkman, 2002, p. 137) at the high end. These social eagerness behaviors were precisely what Davis observed when opioid blockers were administered to dogs, likely activating their PANIC/Sadness systems. Morphine, an opioid, administered to dogs had the opposite effects and basically stopped dogs from initiating social contacts with a human handler or with other dogs (Davis, 1980). Thus, Svartberg and Forkman’s Sociability factor seems to fit with Scott and Fuller’s (1965) distress vocalization and attachment observations and later demonstrations in Scott’s BGSU dog lab that the brain opioids regulate (alleviate) separation distress (Panksepp et al., 1978), which further aligns Sociability with Panksepp’s PANIC/Sadness system (Panksepp, 1982, 1998a).
Altogether, Svartberg and Forkman found all of the primary personality emotions in their analysis of the DMA except for the CARE system, which like in Scott and Fuller’s work was not empirically evaluated as a general temperament dimension. Using rather different methods, adult dogs from forty-seven breeds tested individually in single sessions, and relying on factor analysis, Svartberg and Forkman (2002) added further confirmation of the set of affective neuroscience dimensions that account for a substantial part of the foundation of the mammalian personality factors.
Taking a behavior genetics approach, Saetre et al. (2006), who graciously acknowledged Scott and Fuller as the pioneers of behavior genetic research on personality in dogs, used the large Swedish database of DMA behavior ratings to analyze the pedigrees of 5,964 German shepherds and 4,589 Rottweilers by integrating the DMA data with pedigrees on untested relatives back to grandparents. However, Saetre and colleagues incorporated data from an older, shorter version of the DMA and had complete data on only sixteen behavior observations that only included a single measure of Sociability, which they removed from their analysis. Using the same first-order factors previously identified by Svartberg and Forkman (excepting Sociability), the pedigree relationships and the first-order traits explained 46 percent and 39 percent of the observed genetic correlations for German shepherds and Rottweilers respectively, which supported a genetic basis for canine temperament.
C-BARQ: A Questionnaire for Dog Owners to Rate Their Pets
In dog studies, the behavioral test battery seems to have been historically more popular to identify breed as well as individual personality differences. However, trait ratings by informants—people that know the target individual well—is another method for measuring personality traits, and there is substantial evidence that in humans this approach to collecting personality data can be at least as accurate as when individuals complete the personality questionnaire themselves (Connelly & Ones, 2011).
Accordingly, Hsu and Serpell (2003) have offered a dog-rating questionnaire with sixty-eight items that dog owners can complete on their pets. The instrument was originally named the Pennsylvania Behavioral Assessment and Research Questionnaire, because the authors were at the University of Pennsylvania. However, it was later renamed the Canine Behavioral Assessment and Research Questionnaire (C-BARQ). The C-BARQ measures eleven factor-analytically derived dog temperament scales, which generally overlapped with the scales Svartberg and Forkman derived from the Swedish DMA. However, the C-BARQ scales included multiple measures of RAGE/Anger ( “stranger-directed aggression” and “owner-directed aggression”) and FEAR (“stranger-directed fear” and “nonsocial fear”), and a scale that combined elements of aggression and fear (“dog-directed fear or aggression”). The C-BARQ also provided three measures (“separation-directed behavior,” “attachment or attention-seeking behavior,” and a “trainability” scale that included items like “returns immediately when called while off leash” and “will attempt to fetch sticks, balls, and other objects”), all of which link to social contact and the primary PANIC/Sadness system that is so prominent in the study of mammalian temperaments. Further, a C-BARQ “chasing” factor was described as measuring “a tendency to engage in predatory pursuit,” supporting the Svartman and Fortman chase factor and predatory behavior as a specific expression of the SEEKING dimension in canine personalities. The C-BARQ also has an “excitability” scale, which measures excited overreaction prior to events like “just before being taken for a walk,” which seems like an anticipatory SEEKING response. The C-BARQ also has a scale for “pain sensitivity,” which would have little to do with personality but is probably more relevant for overall veterinary care. Other than C-BARQ not having a scale for the PLAY system, the C-BARQ and Svartberg and Forkman’s DMA analysis are remarkably comparable despite using very different methods.
Taking a page directly from McDougall’s work, Hsu and Serpell (2003) also validated their questionnaire scales by comparing clinical diagnoses from the Veterinary Hospital of the University of Pennsylvania on two hundred dogs having clear behavior problems with the dog’s C-BARQ questionnaire scores. They grouped the diagnoses into seven clusters corresponding to their first seven factor scales. Correlations between the diagnoses and questionnaire factor scores showed that particular behavior problems had the highest correlation with the corresponding questionnaire factor, strongly substantiating all three of the negative primary emotions in canine clinical work. Thus, the C-BARQ and its clinical validation confirms McDougall’s principle that, in their extreme expressions, emotional personality traits can become pathological (see Chapter 4) and strongly supports the need for separate distinct scales for each of the primary negative emotions—RAGE/Anger, FEAR, and PANIC/Sadness—in mammalian personality. A future suggestion could be to add scales targeting more canine PLAYful and CAREing behaviors to broaden the C-BARQ’s coverage of the primary-process positive affective temperamental domains.
Military Working Dogs
Studies of working dogs are less concerned with personality as a whole and more focused on trying to solve a very practical problem: identifying dogs that are likely to succeed in their specialty roles. That is, these behavioral tests are designed to predict narrow aptitudes for specific canine work rather than a broad set of personality dimensions. For example, Sam Gosling and colleagues (Sinn, Gosling, & Hilliard, 2010) studied military working dogs at Lackland Air Force base in San Antonio, Texas. They analyzed standardized behavioral tests collected prior to training in which observers rated a dog’s reactions to exercises the military had designed to measure a dog’s aptitude for patrol and detection work.
A factor analysis of twelve observer ratings yielded four factors. The authors labeled the first and largest factor “Object focus,” which measured interest in chasing and playing with a rubber toy that included excited barking, pouncing, throwing, and chewing; general interest in the environment and avoiding potentially fearful distractions; and generally exhibiting no fear. This factor could be measuring predatory behavior with a playful tone in that the dog is excited and vocal rather than quiet. Lack of fear is relevant because it is very clear that fear inhibits play, an effect that has been well studied formally in juvenile rat models.
By contrast to the “vigorous vocalizations” (Sinn et al., 2010, p. 54) observed in the first factor, the second factor, Sharpness, had the highest test loadings on the dog’s tendency to bite the tester’s harm-protected sleeve “strongly, calmly, and quietly” (p. 55), which is strikingly similar to the quiet predatory biting elicited using electric brain stimulation in cats (Flynn, 1976; Siegel, 2004) and rats (Panksepp, 1971), a behavioral tendency that has been shown to require activation of the SEEKING system.
The third factor, Human focus, primarily measured the dog’s tendency to direct aggressiveness toward the human tester with barking, snarling, tooth exposure, and piloerection, which would seem to be a direct measure of agonistic attack and Panksepp’s (1998a) RAGE/Anger system. The fourth and smallest factor, Search focus, was isolated to search activity tests that may reflect Panksepp’s (1998a) dopaminergic SEEKING system.
As the evolutionarily most ancient of the primary emotions, the SEEKING system has widespread influence on the widest variety of behaviors, including exploration and general activity, as well as LUST, CARE and PLAY. It is tempting to speculate that in addition to SEEKING-oriented predatory behavior, these dogs exhibited a curiosity toward the environment that military, police, and hunters take advantage of in search work. It is also relevant that these dogs seem to find the search itself rewarding, because they never get to keep the objects of their search, including the quarry they help the hunters take home. In all cases studied, the activation of the SEEKING system is a strongly positive experience and a potent reward, which relates to all primary emotions having either a subjectively experienced reward or an escape from punishing qualities.
Altogether, the Sinn et al. (2010) factor analysis included a narrower range of temperament traits focusing on what we could call the RAGE/Anger system, as well as multiple elaborations of the SEEKING system. They may also have thus distinguished playful as well the as the serious predatory behavior systems. The common link may be that the quiet predatory attack evoked by brain stimulation requires electrodes to be placed along the dopamine investigatory SEEKING circuits in the brain but with a recognition that the PLAY system (which is heavily represented in the thalamus, as opposed to just the hypothalamus; see Siviy & Panksepp, 1985) also requires substantial brain dopamine activation along the medial forebrain bundle (Panksepp, 1998a), as has been most clearly demonstrated in more recent studies with rat “laughter” (Burgdorf, Wood, Kroes, Moskal, & Panksepp, 2007; Burgdorf et al., 2008). In any case, the predatory behavior in the Sharpness scale, along with the previously described chasing factors reported in the DMA and C-BARQ studies, strongly reinforces predatory behavior as a general temperament trait in dogs.
Pet Adoption: The Need for Low RAGE/Anger
In a very practical example of a test battery, Kelly Bollen and Joseph Horowitz (2008) explored testing the suitability of shelter dogs for future adoption. They used Sue Sternberg’s “Assess-a-Pet” test (Sternberg, 2002, cited in Bollen & Horowitz, 2008), which measured dogs’ responses to nine common household situations, such as a human reaching into the food bowl while a dog is eating, a human taking a valued possession from a dog, and a dog meeting a human stranger. The goal was to increase the percentage of successful home placements by not placing dogs for adoption that exhibited an aggressive personality, such as lunging at the evaluator while growling and snarling, or making any attempt to bite during the tests.
Using this screening test, the shelter increased its rate of successful dog adoptions, with 3.5 percent of dogs being returned for exhibiting aggression compared to 5.1 percent the previous year, which validated Bollen and Horowitz’s work. Importantly, the return rate could have been improved further by not placing “borderline” cases that during the shelter test displayed only mild aggression, such as stiffening or slight growling, because there was a significant trend for these dogs to be returned for displaying aggressive behavior in the home as well.
However, there is another basic personality question: Were there stable differences in aggressive personality tendencies in the dogs they studied? Indeed, 90 percent of the 217 dogs initially brought to the shelter because they were aggressive in their previous homes also displayed aggression during at least one of the test components evaluated in the shelter, which was almost twelve times the failure rate of dogs brought to the shelter for other reasons and highly significantly different. So, a stable aggressive personality trait was confirmed in these dogs across the two completely different settings, in the dog’s previous home and at the shelter. Furthermore, the “borderline” dogs placed for adoption were more likely to be returned to the shelter than dogs exhibiting no aggression during the shelter test, demonstrating consistent aggressiveness across a third different setting. In short, Bollen and Horowitz provided evidence that a behavioral assessment could identify dogs exhibiting easily provoked RAGE/Anger that made it difficult for families to keep them as pets.
SOCIALIZATION AND DOMESTICATION
Human Social Attachment Measures Adapted to Dogs
Social bonding has long been a subject of interest in psychology (Bowlby, 1960; Scott, 1962). In a study specifically focused on canine social attachment, Topal, Miklosi, Csanyi, and Doka (1998), working in Hungary, adapted the Ainsworth Strange Situation Test, originally used to examine attachment behavior in children (Ainsworth, 1969). The original test revealed “secure” and “insecure” social attachment styles in children by observing their reactions in a strange setting to being separated from and then reunited with their mothers, as well as interacting with a stranger. Topal and colleagues were interested in studying these attachment styles between dogs (as the child) and their human owners (as the parent). Like Mary Ainsworth, they set up a series of tests that placed each dog in a strange place with its owner, with a stranger, or alone.
In their attempt to replicate Ainsworth’s attachment types in dogs, Topal et al. (1998) identified three factors: anxiety (separation distress), acceptance of stranger, and attachment to owner. They reported that some of the pet dogs clearly explored the unfamiliar room and exhibited increased play and did so significantly more with their owner present than with the stranger. This pattern was consistent with the Ainsworth (1969) secure attachment phenotype, which similarly found young children showing signs of missing the parent, but then greeting the parent upon reunion, and the child then resuming his or her activities without further interruption. Further, Topal and colleagues reported that the dogs in their study could be placed on a continuum of how securely they seemed to be attached to their owners.
However, Topal’s group did not report finding dogs that matched Ainsworth’s (1969) insecure attachment types, characterized by either showing extreme signs of distress when the parent left and remaining inconsolable upon the parent’s return or showing little if any sign of missing the parent and avoiding and ignoring the parent when the parent reentered the test room. The Topal group did identify a cluster of five out of their fifty-one dogs exhibiting high separation anxiety, high attachment to the owner, but also with high acceptance of the stranger. That is, these dogs were eager to make contact with any human who was with them in the strange room. This type of social need did not appear in the Ainsworth studies, especially with the insecure children who were not consoled by strangers (Ainsworth & Bell, 1970).
In contrast to the Ainsworth findings, all of the dog’s exhibiting high levels of separation anxiety when left alone in the strange room readily made contact when any human returned to the test room. Even though all of these dogs were raised in families, and Topal and colleagues argued these human-dog relationships were like the parent-child relationship, none of the dogs in the Topal study behaved like the Ainsworth insecure or insecure-avoidant types in that they all exhibited separation distress when left alone but all made contact with any human during the reunion episodes of the experiment. This pattern of data may suggest that the insecure types reflect a secondary-process personality dimension, arising from past learning-socialization experiences, rather than a primary emotional one.
Obviously, specific social attachments reflect learned secondary-process associative-learning mechanisms, with the learning of security most likely based on the primary-process PANIC/Sadness system, as well as the social reward systems of CARE and PLAY. Mammals are not born socially bonded, as was dramatically illustrated in the Scott and Fuller work in which puppies with no human contact for fourteen weeks essentially became wild animals that totally avoided humans (Freedman et al., 1961). However, in their work on the critical periods of socialization, Scott and Fuller (1965) also reported that as little as two weeks of contact with humans between four and eight weeks of age were sufficient to socialize puppies to humans. Thus, perhaps not surprisingly, Topal’s group did not find pet dogs that were socially avoidant.
It is tempting to speculate that domesticated dogs have such sensitive PANIC/Sadness systems that they become socialized with even minimal contact with humans. Even though dogs and humans remain dependent on their families for many years, some human children may have sufficiently insensitive PANIC/Sadness systems that they never become securely bonded to their families, especially when surrounded by parental strife that amounts to chaos in family dynamics. As Ainsworth observed, such a child might not show separation distress when left alone and might by choice remain socially isolated when the mother returned. It is also possible that autistic children are deficient in such basic social-emotional sensitivities (Panksepp & Sahley, 1987), an understanding that may lead to novel therapeutics (Moskal, Burgdorf, Kroes, Brudzynski, & Panksepp, 2011).
Palmer & Custance (2008) added additional experimental controls and still confirmed that dog owners provided a secure base for their dogs, showing that dogs were most willing to engage in play with the owner but also more likely to play with the stranger when their owner was present than when their owner was absent. A variation on these studies by an Italian group (Mariti, Carlone, Ricci, Sighieri, & Gazzano, 2014) used Ainsworth’s secure attachment paradigm to test the attachment between two dogs. They used twenty-two pairs of dogs living in the same household. One member of each dog pair was arbitrarily used as the subject and the other dog was assigned the role of the primary attachment figure, rather than using the dog’s owner. A human unknown to any of the dogs was used as the stranger. Remarkably, after being left alone in the room, the dog being studied was more consoled by and made more contact with the human stranger than with the other dog that shared the same household with them, suggesting that dogs may have been genetically selected to form strong social bonds with humans (Serpell, 1995), or simply have a history of finding active comforting from humans rather than other dogs.
While more research is needed in the PANIC/Sadness-related personality differences of dogs, future attachment style research between humans and dogs might benefit from measuring specific canine appropriate social contact behaviors, such as tail wagging and face licking (e.g., Davis, 1980; Knowles et al., 1989), that could add objectivity as well as add qualitative richness to the descriptions of initial social contact the dogs made with either the owner or stranger. However, it would be especially important to include measures of separation distress vocalizations to assess the activation of the specific PANIC/Sadness system, which is characterized by a form of separation anxiety that is distinct from the anxiety generated by the FEAR system (Davis, Gurski, & Scott, 1977; Panksepp, 1998a).
An interesting new approach to detecting separation distress affect and possibly attachment style is the cognitive bias test model. This test is based on the assumption that in ambiguous situations individuals experiencing negative emotions will make more negative (pessimistic) judgments and individuals experiencing positive emotions will make more positive (optimistic) judgments, which has been confirmed in several species, including humans (Mendl, Burman, Parker, & Paul, 2009). Indeed, dogs (9 to 108 months of age) that exhibited higher levels of behaviors related to social separation distress when left alone made more pessimistic-like responses to ambiguous stimuli than dogs exhibiting less social separation distress (Mendl et al., 2010), which we would interpret as validating an animal model of the distressful PANIC/Sadness emotion biasing behavioral choices in a negative direction, and further suggest that it may be subcortical “affective bias” that is influencing these behavioral choices.
Puppy Testing Predictions
Another interesting question that arises from a discussion of early socialization is whether one can tell from examining a puppy what kind of personality it will have when it matures. This is an important question if one is deciding whether to expend the resources required to train a dog for police work or to become a guide dog for the blind. However, families also invest a great deal of money, time, and effort into raising a puppy and often would like some assurance that their new pet is going to remain a welcome addition rather than a burden to the family. This goes doubly so for selecting service companion dogs that are trained to assist medically impaired individuals, including psychiatrically challenged soldiers returning from difficult tours of duty.
While there are critics of puppy testing, especially at six to seven weeks or earlier (Riemer, Müller, Virányi, Huber, & Range, 2014; Wilsson & Sundgren, 1998), there are three validation studies with consistently positive results. These three studies all involve predicting puppies’ aptitudes for later training either as guide dogs for the blind or as police dogs in the latter two studies. The guide dog study was an outgrowth of Scott’s collaboration with Clarence Pfaffenberger, who was trying to reduce the number of dogs that had to be trained to produce a working Seeing Eye dog, because when the program started it could successfully train only 30 percent of its guide dog candidates. A series of thirteen behavioral tests was introduced for puppies from eight to twelve weeks of age, which included many of the kinds of tests Scott and Fuller had used in the Jackson Lab research program. A total of 239 puppies were tested and trained as guide dogs: those that were judged to have failed the tests (n = 33), those that had performed poorly (n = 18), and those that had passed (n = 188). Including puppies they thought could not be successfully trained was important to have sufficient variation in the sample to clearly show that better test results actually were related to better training results.
Four tests best predicted future success as guide dogs: the come test, the fetch test, the heel test, and the footing-crossing test. The use of these four tests improved training success rates to 60 percent, with those puppies failing the test succeeding only 20 percent of the time (Scott & Bielfelt, 1976). The first three tests could be easily conducted outdoors by a new puppy buyer. The come test involved walking a good distance away from the puppy, squatting down facing the puppy, and calling out “come puppy” while clapping or using its name (if it has one). The best result is when the puppy comes immediately. Poorer results are when the puppy has to be coaxed repeatedly or does not come at all. In the fetch test, the handler shows the puppy a small rubber ball while trying to get the puppy excited before throwing the ball several feet away and calling “puppy fetch” in a lively voice. The handler continues clapping and encouraging the puppy. The best result is when the puppy retrieves the ball to the handler immediately. For the heel test, if the puppy has never been on a leash, the puppy can be carried a short way from a familiar place, put on the leash, and encouraged to walk back on the left side of a right-handed handler. Otherwise, the puppy can be put on leash and walked away from the familiar place. The puppy can run slightly ahead of the handler, and the session is not turned into an obedience lesson. The issue (i.e., test result) is whether the puppy cooperates with the handler or fights the leash. These first three tests assess whether the puppy is motivated to interact with a human handler and, in the case of heeling on a leash, whether that extends to tolerating some physical restriction. The footing-crossing test involves a puppy being led over a patch of metal and a shallow curb. Observers rated whether the puppies were alert to these differences in their footing, which would be useful later as guide dogs.
The first of the two police dog studies was conducted in South Africa (Slabbert & Odendaal, 1999). They gave 167 young German shepherd puppies eight behavioral tests and later recorded whether the puppies passed a 103-day training course, which had to be completed by two years of age. Most of their objective tests were completed by twelve weeks of age, and three of those were most predictive of later successful training. The first was the obstacle test, a variation of the guide dog come test, in which puppies at eight weeks of age had to go through a large pipe (five feet long and two feet in diameter) and climb over stairs about eighteen inches high to reach a handler who called them. A fetch test, similar to the guide dog test, examined whether a puppy at eight weeks retrieved a small toy to its handler and was a strong predictor of later training success. At twelve weeks a startle test subjected each puppy to a stranger jumping out from behind a wall in front of the puppy. Puppies that made no effort to run away received the highest scores. A fourth test, another stranger test that the puppies received much later at six and nine months, involved a stranger provoking the puppies by striking them firmly but gently with a rag. If the dog bit the rag and held on, it received a high score; if it showed fear it received a low score. These tests were able to identify 91.7 percent of the successful dogs and 81.7 percent of the unsuccessful dogs.
The second of the police dog studies comes from the Czech Republic (Svobodová, Vápeník, Pinc, & Bartoš, 2008). In the Czech study, 206 German shepherd dogs were given ten tests and weighed at seven weeks of age. Interestingly, the strongest predictor of whether a puppy would later pass certification was its weight at seven weeks, which the authors attributed to a general preference of police to use large dogs. However, among the behavioral measures, the strongest predictors were the puppy’s willingness to chase, catch, and fetch a tennis ball and whether a puppy would bite and fight a rag being drawn away from it, which the authors collectively labeled “attitude to predation.” Although a weaker predictor, in three of the ten tests the puppies were exposed to loud distracting noises. Puppies that showed the most fear received the lowest scores; those that exhibited less fear and even showed exploratory interest in the noise receiver higher scores. Lastly, there was evidence that the more the puppy approached the tester in a friendly way, the more likely the puppy would later be successfully certified for police work.
All three of these studies aimed to identify dogs that could be trained to perform complex tasks. One finding all three studies have in common was a puppy fetching at an early age, which was strongly related to later training success. Scott and Bielfelt (1976) found that retrieving at later ages was not as predictive, because eventually almost all dogs in their sample could learn to fetch. However, the willingness at seven or eight weeks to fetch may indicate intense prosocial motivation that can carry into a dog’s adult life and facilitate its trainability to work closely with humans. While there may be a predatory SEEKING element that energizes the chasing component of fetching, there also seems to be a strong social motivation to interact with the handler that could indicate a sensitive PANIC/Sadness or PLAY system, which makes the ensuing contact with the handler very motivating as well. Along these lines, the guide dog study and the South African study found strong evidence that a puppy’s willingness and motivation to come to a human handler at eight weeks was also a strong predictor of later trainability, which could be related to the extent of human interactions in the Czech study, all of which support the idea that, in the dogs that excelled in their training, already as young puppies the PANIC/Sadness system had established a strong motivation to maintain contact with humans.
In the police studies, quiet biting (as opposed to angry noisy biting) likely reflects increased emphasis on SEEKING-based predatory behavior (with angry noisy biting more likely reflecting dispositions toward RAGE/Anger). In puppies, an aptitude for this capacity may be promoted and stimulated by taunting the puppy with a rag. This predatory (SEEKING) feature was also highlighted in the military dog training study cited above. In police or military work, biting hard and holding the bite to subdue a human target is an important tool, which the officer-handler must be able to turn on and off with commands given to the dog.
Having a less sensitive FEAR system seems to be an important affective dimension for such working dogs as well. Being able to tolerate being startled by loud noises or not being afraid of loud noises in general was a consistent predictor of future success as a police dog. Lastly, the puppy’s willingness to tolerate restraint as measured by the guide dog heeling test may be a good proxy for a less sensitive RAGE/Anger system, which along with the puppy’s willingness to please a human handler is an obvious requirement of all guide dog work. However, it would be surprising if this simple test were not equally effective in predicting training success in police and military working dogs.
In summary, both brain emotion network analysis and modern genetic profiling will allow investigators to have better anchored measures of emotional temperament than was available in previous eras. With ready access to modern genetic tools, such work is likely to be a major focus of future inquiries. Such analyses may already be coupled to the most remarkable canine selection study ever conducted, as described next.
The Fox Domestication Project
About fifty years ago, a selection project aimed at domesticating the Russian silver fox (Vulpes vulpes) was initiated to explore the domestication process in mammals. Starting with a population of silver foxes that had been farm bred for about fifty years (starting in the early twentieth century), foxes displaying the weakest fear and aggressive responses were selected to begin a line of potentially tame foxes (Trut, Oskina, & Kharlamova, 2009). Subsequent generations were subjected to attempted touching, petting, and hand feeding, and less than 10 percent of the pups exhibiting the most tameness were selected as parents for the next generation. After three generations of such selection, “aggressive and fear avoidance responses were eliminated from the experimental population” (p. 351). In the fourth generation, spontaneous tail wagging in response to humans was observed. In the sixth generation, 4 out of 213 fox pups emerged that eagerly sought human contact and wagged, whined, and licked faces like domestic dogs. The frequency of such pups increased to 17.9 percent by the 10th generation, 35 percent by the 20th generation, and 49 percent in the 30th generation. Reduced reactivity of the FEAR system seemed to have been closely involved in selection for tameness in the silver fox, because the age at which fear was first seen was shifted to about 4 months in the tame line from about 45 days in the unselected line (Trut et al., 2009). Of course, it is likely that sensitivities of PANIC/Sadness, PLAY, RAGE/Anger, SEEKING, and perhaps even CARE were also involved (as discussed further below), but without in depth neuroscience work, such issues will remain empirically unresolved.
A second line of silver foxes was selected for aggressiveness toward humans, and it was discovered that the tame and aggressive lines always exhibited different vocalizations toward humans: Tame foxes produced “cackles” and “pants,” and the aggressive line “coughs” and “snorts” (Gogoleva, Volodin, Volodina, Kharlamova, & Trut, 2010). When foxes from the same lines were observed together in pairs, despite directional testing for generations, tame foxes retained their capacity for agonistic behavior and vocalizations toward other foxes and aggressive foxes retained their capacity for affiliative behavior and vocalization, suggesting the genetic independence and complexity of these emotional systems.
From video tapes of the foxes during standard tests, 311 measures of physical activities such as position in the cage, body postures, ear position, and sounds were reduced to fifty key behaviors. The first principal component (a statistical clustering) of this reduced set of behaviors provided an objective measure that accounted for 48.4 percent of the variance (a statistical measure of overall percent relatedness or how well findings are explained by the chosen measures). This first principal component distinguished the aggressive and unselected lines from the tame line, as well as the F1 backcross to tame foxes, with the F1 line (tame line bred with aggressive line) being intermediate (Kukekova et al., 2008), clearly demonstrating the genetic differences between the tame and aggressive lines. While no specific genetic hypotheses were offered, one must assume that many genes and brain systems must be involved considering that the proportion of foxes that could be characterized as tame almost tripled from the 10th to the 30th generation. In addition to the FEAR and RAGE/Anger systems that were selected against in the tame parent line, it is likely that the PANIC/Sadness social bonding system was selected for as well, because tail wags and face licking in response to humans—both of which are known to increase in response to opioid blockers increasing PANIC/Sadness sensitivity (Davis, 1980; Knowles et al., 1989)—amazingly emerged in the fourth and sixth generations of this remarkable selection project.
CANINE SUMMARY
In canines, many of the descriptive details of social bonding and separation-induced distress have been documented (Scott & Fuller, 1965), the C-BARQ (Hsu & Serpell, 2003) has provided a “separation-related behavior” trait scale for dogs, and Topal et al. (1998) has developed a new test of social attachment. However, fuller understanding of sensitivity to separation distress as a temperament dimension in dogs and of the range of social bonding/dependence in dogs extending to pathological separation-related behavior (Mendl et al., 2010), and their genetic and neural underpinnings, will require much additional research. Clearly, better linking the social separation experience to the mammalian brain’s PANIC/Sadness system (Panksepp, 1998a) and the underlying brain and genetic mechanisms is needed to explore the canine model for separation-induced pathology in humans.
While foxes may naturally be less intensely social than dogs and wolves, the silver fox domestication project has shown that foxes possess the genetic basis for pronounced positive social motivation, including heterospecific bonding with humans. The counter suggestion that domesticated dogs have retained predatory tendencies, which are expressed toward humans, may provide insight into another pathological dimension of canine behavior in relation to humans. Although this has yet to be studied, one might start with breeds that have been found to be extra aggressive to humans historically, such as pit bulls (Sacks et al., 2000) and potentially determine whether predatory SEEKING and/or RAGE/Anger are responsible for their aggressiveness.
Another area that is once again highlighted in dog research is the confounding of fear and exploration. FEAR and SEEKING (Panksepp, 1998a) are two ancient but distinct brain emotion systems, which may often interact as animals negotiate their environments. Indeed, all the primary emotional networks tend to overlap in lower regions of the brain, such as hypothalamus and midbrain periaqueductal gray, providing many opportunities for interactions. However, a temperament profile must allow for FEAR and SEEKING to vary independently; for instance, low exploratory tendencies do not automatically imply high levels of fear. A solid understanding of temperament requires the use of models that can better distinguish the anticipatory dopaminergic SEEKING system from the danger-oriented benzodiazepine-receptor-regulated FEAR system.
Overall, the key dilemma in temperament research is that there are no generally accepted experimental strategies to decode the neural nature of primary-process emotions in human subjects. A study of such brain systems in animals may eventually provide us with a causal neuroscience understanding of what it means for the mammalian/human to experience distinct affective feelings, but the primal neural mechanisms of affect simply cannot be deciphered through human research. There are at least seven emotional command networks as determined by distinct instinctual tendencies aroused by localized deep (subcortical) brain stimulation. Three of the positive affective systems, LUST, CARE, and PLAY, are all heavily dependent on the foundational influence of the general enthusiasm-promoting power of the SEEKING system, which may participate in all types of affective sentience (Panksepp, 1998a; Panksepp & Biven, 2012). However, these four as well as the three affectively negative ones (RAGE/Anger, FEAR, and PANIC/Sadness) all provide the motivation (i.e., unconditioned responses) for various kinds of learning (i.e., secondary processes) and surely higher cognitive (tertiary) processes that are very hard to systematically evaluate in animal models but that, in combination, surely guide higher cognitive-affective decision making and states of mind. Most dog owners (including Darwin) believe that their pets exhibit these higher cognitive capacities, which are so difficult to study rigorously in nonhumans.
Stated differently, a key dilemma in temperament research may be that more complete understandings at the primary-process genetic-neuroscience level are needed to provide solid foundation for all levels of personality development. Such knowledge is essential for understanding the development of higher cognitive capacities. Thus, further work on even lowly domesticated laboratory rats and mice will help provide insights and clearer empirical windows into mammalian minds, as we will discuss from genetic perspectives in the next chapter.