Fe-BARQ.

Cat owners could voluntarily complete the Feline Behavioral Assessment and Research Questionnaire (Fe-BARQ; https://vetapps.vet.upenn.edu/febarq/). Data were collected from September 2015 to September 2023.

The Fe-BARQ [18] is a validated, frequently used tool to assess cat behavior. The questionnaire asks owners to describe the frequency of 100 different cat behaviors. The responses are five-point Likert scales ranging from never (0), seldom (1), sometimes (2), usually (3) or always (4). The Fe-BARQ includes three questions relevant to fetching behavior: Item 3 (carries small objects/toys in the mouth to interact with), Item 13 (initiates interactive play with people in the home i.e. brings toys, strings, or small objects to play with), and item 91 (plays ‘fetch’; likes to retrieve thrown objects or toys).

The Fe-BARQ has 23 subscales, including Activity/Playfulness, Sociability with People, and Predatory Behavior. Demographic information about the cat was also collected at the time of the survey including age, sex, neuter status, whether the cat lived indoors only, and whether the cat was declawed. The owner was asked about their experience with cat ownership, and also asked to assess whether their cat had behavior problems (no problems, minor problems, moderate problems, serious problems). The Fe-BARQ subscale scores are calculated by averaging all of the responses to that subscale’s items.

C-BARQ.

Dog owners could voluntarily complete the Canine Behavioral Assessment and Research Questionnaire (C-BARQ; https://vetapps.vet.upenn.edu/cbarq/). Data were collected from March 2008 to July 2022.

The C-BARQ [19] is a validated, frequently used tool to assess dog behavior. The questionnaire asks owners to describe either the frequency or the severity of 100 different behaviors. The responses are five-point Likert scales ranging from never (0), seldom (1), sometimes (2), usually (3), to always (4) for frequency, and on a spectrum for severity that differed based on the behavior (e.g., absent (0), mild to moderate (1, 2, 3) and severe (4)). The C-BARQ contains one item relevant to fetching behavior, Item 8: “will "fetch" or attempt to fetch sticks, balls, or objects”.

The C-BARQ comprises 14 subscales, including Stranger- and Owner-directed aggression, Excitability, Trainability, and Stranger-directed fear. Demographic information about the dog was also collected at the time of the survey including age, sex, neuter status and where the dog was acquired. The owner was asked about their experience with dog ownership and whether they were concerned about their dog’s health. The C-BARQ subscale scores are calculated by averaging all of the responses to that subscale’s items.

Hypothesis: Fetching is a form of object play.

In cats, fetching had the highest correlation with the Activity/Playfulness subscale (of which it is an item, r = 0.47); in dogs, there was a moderate correlation with the Energy subscale (which may be analogous to the Activity/Playfulness subscale in cats; r = 0.32). In cats, object play is believed to represent a rehearsal of innate predatory motor patterns; however, there are still open questions as to whether in captive adult animals (including companion animals), object play is influenced by domestication and human behavior [26] (Hall, 1998).

For cats, the correlations with prey interest (r = 0.23) and predatory behavior (r = 0.17) were relatively low. In the C-BARQ, the closest measure of predatory behavior is the Chasing subscale, which had a low correlation with fetching (r = 0.06). In both species, these findings suggest that fetching is more closely related to playful behavior than it is to predation. This suggestion is also borne out by the observed decline in fetching associated with aging in cats and dogs. The sequence of behavior also fulfills the key criteria for "play" in that it does not appear fully functional and appears to be pleasurable or rewarding [27]. The finding that females were less likely to fetch, and that spay/neuter status was not a significant factor in expression of the behavior further suggests that it may be less associated with predation given that presentation of prey to offspring is typical of maternal behavior and related to social facilitation of offspring hunting/killing behavior.

Cats naturally, without any training, will display discrete behaviors used when hunting prey including “…watching, pouncing, hitting, grasping, pulling down, kicking with hindlegs, tossing and biting” [28]. In the predatory sequence these behaviors occur in an established sequence but in the absence of opportunities to hunt, the behavioral sequence may decouple. Predatory behavior in cats is so highly motivated that even in the absence of hunger it will be expressed, albeit with variable frequency across individuals, and it may occur out of context such as during play with objects [28].

Predatory behavior and prey interest may be influenced by whether the cat has outdoor access and opportunities to hunt. In the current study, being housed exclusively indoors increased the likelihood that a cat would fetch, and in previous studies has increased the likelihood of play behavior [29]. This may be because owners of indoor cats may be more likely to play with them [9]. Alternatively, cats allowed outdoor access may have less motivation to fetch as they have more opportunity to engage in physically active and predatory behavior while outside. If this is the case, fetching in indoor cats might be regarded as a form of environmental auto-enrichment.

While there is little research on dogs’ innate tendencies to fetch, there are several studies that utilize fetching as a cognitive task to measure dogs’ cooperative behaviors [e.g., 30, 31]. The current study identified several breeds that were more likely to fetch, including dogs from the Retriever, UK Rural, Poodle, Pointer and Spaniel clades [22]. Predatory motor patterns in these breeds have hypertrophied chase and/or grab-bite behavior that begins to appear during the critical period of socialization. The association between the onset of this motor pattern and socialization may result in the dog displaying the behavior in the presence of their human companion as a form of social play resulting in fetching [24].

Hypothesis: Fetching behavior has been selected for during domestication or breed development.

Breed differences in the number and intensity of behaviors in the predatory sequence exhibited have been identified [24], which may explain the breed differences we found in fetching behaviors in dogs. However, almost all breeds of both cats and dogs had some animals who fetched, including wolves and wolf hybrids (S1 File). A comparative study found that mixed-breed and purebred dogs were more likely to chase and retrieve objects than hand-reared wolves [32] and a study of social play and cooperation found that three out of thirteen wolf puppies offered retrieval behavior spontaneously [33]. Together these results suggest a possible influence of domestication, but also that variation in fetching or retrieval-type behavior may have existed prior to domestication.

Differences in breed expression of behaviors within the predatory sequence that are likely due to human selection are outlined in Table 5 [24]. Early in the domestication process, before intensive goal-directed breed development, dogs were selected for specific traits and behaviors that may have aided humans, most notably with herding and hunting [34]. Several types or landraces of working dogs have been identified, including dogs who move livestock, those who guard livestock, sporting dogs that point and/or retrieve killed prey, and hounds that track prey by scent or chase running prey. In the current study, breeds that were identified as high fetchers included herding, pointing, and retrieving breeds while those identified as low fetchers included hound, livestock guarding (LGD), and wildtype breeds.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Selected behaviors of the predatory sequence in several types of dogs.

(Adapted from Coppinger & Coppinger, 2001).

https://doi.org/10.1371/journal.pone.0309068.t005

Breeds developed to move livestock including headers (e.g., Border Collie), and heelers (e.g., Australian Shepherd, Australian Cattle Dog) and breeds developed as hunting companions including those in the Pointer group (e.g., German Shorthaired Pointer, Vizsla, Weimaraner) and Retriever group (e.g., Labrador Retriever, Golden Retriever), were more likely to fetch. Conversely, breeds in the Hound group including both scent hounds (e.g., Dachshund, Black and Tan Coonhound, Blue Tick Coonhound, Beagle) and sighthounds (e.g., Afghan Hound, Irish Wolfhound, Greyhound) were found to be low fetchers.

A possible explanation for these differences is that dogs identified as high fetchers may get ‘stuck’ in a repetitive pattern of chase >> grab-bite, and fetch, because the predatory motor pattern is truncated [24]. In comparison, low fetching breeds such as hounds have also been selected for hypertrophied chase, grab-bite, and kill-bite and maintain the link from grab-bite to kill-bite. Thus, they can complete the behavior sequence and do not get ‘stuck.’ In the case of LGD, they lack the predatory motor pattern entirely and fetching would be expected to be rare.

Because they are the most common, non-purebred cats (e.g., domestic shorthair, domestic longhair) represented the majority of fetching cats. However, they are not more likely than expected to fetch, and in agreement with Forman et al. [5] and Voith and Borchelt [16], we found breed differences in the prevalence of fetching behavior in cats. The three purebred breeds more likely than expected to fetch (Siamese, Burmese, and their crossbreed, the Tonkinese) are part of a shared origin in the Far East [35], a cluster of cats that are genetically different from cats from other regions of the world (e.g., the Mediterranean Basin, Western Europe, Africa) [36]. Geneticists believe that domestic cats were brought to the Far East early in the process of cat domestication, and then were genetically isolated from other populations for an extended period [36]. The greater prevalence of fetching in this population could be due to a founder effect, genetic drift, or be the result of increased local selection pressures for the tendency to engage in fetching behavior in this region of the world.

Bengals, a hybrid recently developed by crossing the Asian Leopard cat with the domestic cat, were also more likely to fetch than most other breeds. They are described as a very active breed that displays high levels of hunting behavior [37]. While cats are often considered to be ‘self-domesticated’ [38], the breed differences in fetching raise questions about possible working roles for cats in former times. Cats were commonly depicted as accompanying their owners on bird hunts in Egyptian tomb paintings from circa 1450 BCE [39], raising the intriguing possibility that they may have been used by humans for retrieving or chasing small prey items during their early history.

Hypothesis: Fetching is a type of social play.

The domestication process has resulted in the retention of juvenile behaviors into adulthood, specifically increasing prosocial behaviors such as playfulness and sociability and decreased fearfulness and aggressiveness [40]. Companion animals also use body language to indicate their desire for social interactions. For example, dogs offer a play bow to indicate intention to play with other dogs or humans [41, 42]. Kittens have documented play solicitation signals with each other, such as a “side step” and the “open mouth” play face [43].

Early life experiences during each species’ sensitive period may also have an impact on tendency to play with humans, and what forms that play may take. For example, when observed during play with a human caretaker from 6 weeks to approximately one year of age, puppies in a guide dog program became gentler and more self-controlled over time with their human caretaker, suggesting that the puppies learned to cooperate [44]. In kittens, play behavior toward a moving string toy was reduced by the kittens’ fear of humans [45].

In both species, initiating fetching by bringing a toy to a human caretaker appears to be an invitation to play. It is assumed that play with humans is a social activity for dogs [41] and this has recently been explored in cats [9]. Forman et al. [5] reported that fetch behavior was often initiated spontaneously by cats, and fetching was more frequent and resulted in more object retrievals when cats initiated it.

If fetching is a form of social behavior, it would be expected to correlate with measures such as attention-seeking or sociality. However, this was not the case for either cats or dogs in the current study. Some possible explanations include that sociality may only explain a small component of fetching behavior, or that the specific questions in the surveys examined did not permit assessment of the social aspects of fetching behavior. Alternatively, the animal may simply be using the human to perpetuate an inherently rewarding game of chase and catch. The fact that the human also finds the game rewarding may be irrelevant from the animal’s perspective. The possibility that the level of expression of fetching in cats may relate to the strength of the human-animal bond should be explored. While the surveys examined did not permit examination of such an association, understanding of whether such a relationship exists would give greater insight into the potential motivation for fetching behavior in cats.

Hypothesis: Fetching has been reinforced by training or learning effects.

We anticipated a stronger correlation between fetching and trainability in dogs, because dog owners are more likely to provide training than cat owners and often use play as a reinforcer [e.g., 46, 47]. In the C-BARQ survey, fetching was part of the Trainability subscale derived from Exploratory Factor Analyses and had a moderate correlation with the overall Trainability score. Fetching was not highly correlated with any of the individual items on the subscale (0.04 < r < 0.23). For cats, the Trainability subscale has three items, and fetching had similar correlations with the individual items (“Comes when called” (r = 0.18), “Readily responds to simple commands” (r = 0.19), and “Attends and listens closely” (r = 0.23). Thus, it appears that the relationship between trainability and fetching behavior may be similar (and tenuous) in both species. Our findings further support those of Forman et al. [5] that fetching in cats is not likely due to training.

A recent study of behavioral genetics utilizing data from 18385 dogs enrolled in the Darwin’s Dog project reported on two behaviors of interest to the current study, toy-directed motor-patterns (e.g. grab-bite, chase) and biddability (how readily the dog responds to human direction, especially in the training context) [48]. Breeds found to have the highest levels of toy-directed motor behavior were German Shepherd Dogs, Golden Retrievers, and Labrador Retrievers. Breeds with the lowest levels of toy-directed behavior included Great Pyrenees, Malamute, and Greyhound. The breeds identified are similar to those identified as high and low fetchers; our data indicated that German Shepherds, Labrador Retrievers and Golden Retrievers were among the breeds more likely than expected to fetch and had the highest standardized residuals (>13.82; S1 File). Great Pyrenees, Malamutes, and Greyhounds all had standardized residuals of less than -2, indicating that they were less likely than expected to fetch.

Further, in the Darwin’s Dog study, the behavioral trait with the highest heritability was retrieving (52.5 +/- 9.2%) and the factor with the highest heritability was human sociability (67.3 +/- 12%). Breeds found to have the highest levels of biddability were German Shepherd Dogs, Golden Retrievers, and Labrador Retrievers. Breeds with the lowest levels of biddability were Great Pyrenees, Malamute, and Basset Hound (less likely than expected to fetch, see S1 File). Herding breeds were reported to be more toy-directed and biddable while sporting breeds were more toy-directed. These results suggest that, in dogs, biddability and training may have an effect on frequency of fetching especially in breeds predisposed to exhibit the behavior. It is likely that fetching behavior, similar to other behavioral characteristics, is complex, influenced by several interacting environmental and genetic factors in both species [e.g., 48].

Fetching behavior can certainly be reinforced and shaped by human responses. For example, if a cat or dog presents a toy to a human, and the human responds by tossing the toy and engaging in play with their pet, this is likely to be rewarding to the animal. If the owner does not recognize that the animal is soliciting play, fetching behavior may be ‘extinguished’ due to lack of reinforcement. The human movement of tossing a toy may serve as an indicator (discriminative stimulus) of the human’s intention to participate in the fetching game [49] and chasing the thrown object could trigger a predatory sequence that is likely rewarding [50]. The cat or dog is then further reinforced (via praise, petting, attention, or the continuation of the game) when they retrieve the item [49]. However, we are still left with the question of how the fetching behavior began in the first place.

Factors that decreased fetching.

Our analyses also identified factors that were associated with decreased likelihood of fetching behavior. Older cats and dogs were less likely to be reported as fetching, suggesting a possible association with general play and activity, which also tend to decrease with age [18, 51]. Animals with health issues were also less likely to fetch. Health issues could have a similar effect as aging, and it is also possible that owners interact with their older or unwell cats and dogs differently, generally engaging in less play and activity.

Female animals of both species were also less likely to fetch. This may seem surprising, given the tendency of female cats to carry weakened prey to their kittens to aid them in learning to hunt [11]. All predators must hunt to survive, and sex has not been found to be a significant factor in hunting or related behaviors in cats (e.g., Ceccheti, personal communication, 2021) [52, 53] or dogs [e.g., 54, 55]. However, sex differences in activity or playfulness could also explain this finding.

In dogs, there was an association between being neutered and being less likely to fetch. Neutered cats were less likely to carry objects, but there was no relationship with fetching. However, many more cats in our study were neutered than dogs (95.3% vs 71.5%). It is possible that a larger sample size of unneutered cats would have resulted in a different relationship between desexing and fetching behavior.

For both cats and dogs, living with dogs reduced fetching behavior. This could be due to changes in human-animal interactions in cat-dog households, or cats may be more comfortable in homes without dogs [56]. Cats in households with dogs may be inhibited from developing or expressing fetching behavior due to the behavior of the dog (e.g., if the dog chases or expresses predatory behavior toward cats or objects the cat may want to play with).

The effects of living in a multi-dog household on dog behavior may be highly dependent on the relationship between the dogs; if the relationship is tense, play behavior may be negatively impacted [57] but if the relationship is positive dogs may engage in more conspecific play in multi-dog households. In either case, it is also possible that people are less likely to observe individual behaviors in their pets when there are dogs in the home.

We also posit that brachycephaly may inhibit fetching behavior, as indicated by our breed analyses. Brachycephalic dogs (Pugs, Shih Tzu, Bulldog, Cavalier King Charles Spaniel, Mastiff, Brussels Griffon, Japanese Chin, Lhasa Apso, Pekinese, Chow Chow) and cats (Persians) were less likely to fetch or carry objects (the exception was Boston Terriers). If certain physical traits (e.g., bite and occlusal pattern) aid or hinder predation, it is possible that selection for extreme brachycephaly restricts the ability to perform certain parts of the predatory sequence [e.g., 24, 58] or constrains play behavior in other ways.