problem solving task dogs

• Open supplemental data
• Reference Manager
• Simple TEXT file

People also looked at

Original research article, assistance and therapy dogs are better problem solvers than both trained and untrained family dogs.

• 1 Department of Ethology, Institute of Biology, ELTE Eötvös Loránd University, Budapest, Hungary
• 2 Instituto de Ciencias Biológicas y Biomédicas del Sur (INBIOSUR), Departamento de Biología Bioquímica y Farmacia, Universidad Nacional del Sur (UNS)- Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET), Bahía Blanca, Argentina
• 3 Grupo de Investigación del Comportamiento en Cánidos (ICOC), Instituto de Investigaciones Médicas (IDIM), Consejo Nacional de Investigaciones Científicas y Técnicas, Universidad de Buenos Aires, Buenos Aires, Argentina
• 4 MTA-ELTE Comparative Ethology Research Group, Budapest, Hungary

When faced with unsolvable or difficult situations dogs use different behavioral strategies. If they are motivated to obtain rewards, they either try to solve the problem on their own or tend to interact with a human partner. Based on the observation that in problem situations less successful and less perseverant dogs look more at the humans' face, some authors claim that the use of social strategies is detrimental to attempting an independent solution in dogs. Training may have an effect on dogs' problem-solving performance. We compared the behavior of (1) untrained, (2) trained for recreational purposes, and (3) working dogs: assistance and therapy dogs living in families ( N = 90). During the task, dogs had to manipulate an apparatus with food pellets hidden inside. We measured the behaviors oriented toward the apparatus and behaviors directed at the owner/experimenter, and ran a principal component analysis. All measures loaded in one factor representing the use of the social strategy over a more problem-oriented strategy. Untrained dogs obtained the highest social strategy scores, followed by dogs trained for recreational purposes, and assistance and therapy dogs had the lowest scores. We conclude that assistance and therapy dogs' specific training and working experience (i.e., to actively help people) favors their independent and more successful problem-solving performance. General training (mainly obedience and agility in this study) also increases problem-oriented behavior.

Introduction

Problem-solving behaviors involve a diverse set of cognitive processes, such as perception, learning, memory and decision making, among others ( 1 , 2 ).

Several studies have focused on dogs' problem-solving abilities using a wide variety of tasks (e.g., puzzle boxes in Frank and Frank ( 3 ) and Marshall-Pescini et al. ( 4 ); unsolvable task in Miklósi et al. ( 5 ); string pulling in Osthaus et al. ( 6 ); interactive dog toy in Shimabukuro et al. ( 7 ). Different kinds of tasks require different skills, thus allowing the thorough study of the diverse strategies that dogs use to solve problems [see e.g., Polgár et al. ( 8 )]. While some studies focus on the manipulation of the physical environment, others analyse social strategies, including communicative interactions. With regard to the latter, dogs' gazing behavior has received the most attention. One frequently used protocol to assess dogs' communicative intents toward people is the so-called unsolvable task. In this situation, dogs try to obtain a reward from an apparatus that cannot be opened. When faced with this problem, most dogs tend to gaze at their owners, which can be interpreted as a referential request for assistance by the human partner [e.g., Miklósi et al. ( 5 ), for a review see Cavalli et al. ( 9 )].

Dogs' selection for socio-cognitive abilities during the domestication process might have had a detrimental effect on their physical cognition ( 3 ). This hypothesis has been supported by several comparative studies in which dogs privileged the use of social strategies such as gazing to the human face, while wolves spent more time manipulating an apparatus and were thus more successful in solving the problem ( 10 , 11 ). However, other authors have highlighted that this discrepancy in the performance of the two species may not be (only) due to differences in their ability to solve physical problems, but other factors, such as motivation and persistence ( 12 – 14 ), and vague definitions ( 15 ). Persistence is a reliable predictor of problem-solving ability, and might be linked to trial and error learning strategies ( 16 ). In this regard, persistence has been operationally defined as the time spent interacting with an apparatus ( 17 ). Accordingly, those individuals that persist longer in their problem-solving attempts are more likely to solve a problem than those that give up earlier [e.g., ( 16 )].

Several other factors appear to influence dogs' problem-solving abilities, including their relationship with humans ( 18 ), their living conditions ( 19 , 20 ), and their breed. For example, compared to Siberian huskies, border collies looked more at the owner in an unsolvable problem situation ( 21 ), and herding dogs tended to look more at the person than working and mastiff like dog breeds when confronted with a puzzle box ( 22 ). However, herding dogs did not interact more with the apparatus than other breed groups in this study, and when taking into consideration both breed and training experience, training had a major influence on dogs' orientation to the apparatus ( 22 ).

In line with this, many studies have focused on the role of training experience. This is of particular relevance, considering the importance of training in working dogs' performance and the increased number of tasks in which dogs participate nowadays. For instance, Marshall-Pescini et al. ( 4 ) tested the performance of untrained family dogs and highly trained family dogs that participated in different activities (i.e., agility, schutzhund, retrieving, search and rescue, freestyle performances). All dogs were exposed to a commercial feeding box which could be opened by pressing a paw pad or nosing the lid. While untrained dogs spent significantly more time looking at either the experimenter or their owner; trained dogs interacted significantly longer with the apparatus and were more successful in opening it. Marshall-Pescini et al. ( 22 ) observed similar results using the same apparatus, as dogs with training experience (i.e., agility, police, search and rescue, and man-trailing) were more successful in the task and looked less to people than untrained dogs. It is important to note that in both of the aforementioned studies trained dogs' groups were heterogeneous given that the subjects differed in the types of training they received and their everyday experiences. While some dogs were trained working dogs, others were trained for recreational or sporting purposes such as agility. Thus, to disentangle the relative effects of training for recreational purposes and for specific work, we aimed to compare the performance of dogs trained for assistance and therapy work with family dogs which had been trained for recreational purposes (see subjects' details). Assistance and therapy dogs differed from trained family dogs in the purpose of their training, their everyday tasks and in the methods of training.

Range et al. ( 23 ) carried out a similar experiment, using a wooden box with a handle which could be opened by pushing it down with the mouth or a paw. In line with previous results, trained dogs (i.e., agility and search and rescue) spent more time interacting with the apparatus and were able to open it significantly more often than untrained ones ( 23 ). On the contrary, Brubaker and Udell ( 24 ) found no significant differences between search and rescue dogs and untrained family dogs in gazing or persistence in a similar task. However, significantly more search and rescue dogs opened the container when they received encouragement ( 24 ). The divergence between these studies may be related to differences in the training the dogs from each sample had received [i.e., agility and rescue dogs in Range et al. ( 23 ); only rescue dogs in Brubaker and Udell ( 24 )]. Furthermore, the encouragement in Brubaker and Udell ( 24 ) may have also influenced the results and this difference in the protocols hinders a straightforward comparison. All in all, results regarding the effects of training on dogs' problem-solving skills and strategies are contradicting. This could be due to differences in the protocols and tasks used, samples, the dogs' breed, and the training received as discussed above.

Professional working dogs represent a special group of dogs which, unlike family dogs, are specifically trained to regularly perform a specific activity such as detection of substances, search and rescue or helping disabled people, among others ( 25 ). Importantly, working dogs face a variety of cognitive challenges during their training and working activities which may influence their behavior and performance during cognitive tests. Even more, as different working roles require different sets of skills, it would be expected that working dogs vary in their performance during such tasks according to the specific activities they carry out ( 26 ). In line with this, it must be taken into account that there are variations in the goals of training, the methods employed for it and the frequency in which those abilities need to be performed, which add to the expected variability among working dogs as a whole. Thus, it is important to assess dogs with different training and working experiences to further understand how these aspects influence dogs' problem-solving skills.

In this study we focused on two types of working dogs: assistance and therapy dogs. Assistance dogs are individually trained to perform tasks for the benefit of their owner with a disability affecting everyday life situations ( 27 ). Therapy dogs participate with their owners in planned, goal-oriented therapeutic interventions directed by providers of health and human service ( 28 ). Both types of working dogs need to be sensitive to their owners' wishes, but at the same time they have to be independent in order to solve problems on their own and flexibly adjust to new scenarios.

Gácsi et al. ( 29 ) studied the interactions between assistance dogs and their owners during a carrying task. They observed joint attention during different parts of the task as well as the use of both verbal and non-verbal communication to guide the dogs' actions. In the case of a task that was impossible to perform, they observed that assistance dogs did not give up easily and were very persistent before they showed communicative signals directed at the owner ( 29 ). The results suggest that assistance dogs are not only persistent, but also able to switch between different strategies, such as communicating with the owner, if they failed in independent problem-solving.

Thus, in this study we aimed to compare the problem-solving performance of dogs with different levels of training and working experience. To this end, we tested three groups of dogs in a problem-solving task; untrained family dogs, family dogs trained for specific tasks (e.g., obedience, agility, herding), and working assistance and therapy dogs. For the sake of simplicity, we will refer to dogs working in assistance and therapy as “working dogs.” We expected working dogs to perform better at independent problem-solving and thus to obtain more food rewards than family pet dogs. Also, we expected untrained family dogs to depend more on their owners and prefer the use of a social strategy such as gazing toward people. In the case of trained family dogs, training experience may increase their independent problem-solving abilities [e.g., ( 4 )]. If this is the case, they should behave similarly to the working dog group. Alternatively, the trainings these dogs had (mainly obedience and agility) may have not prepared them for independent problem-solving, thus their performance may be indistinguishable from that of untrained family dogs.

Materials and Methods

Ethical statement.

The procedures comply with national and EU legislation and institutional guidelines and in accordance with the recommendations in the International Society for Applied Ethology guidelines ( www.applied-ethology.org ). In Hungary, the behavioral observations conducted in this study were not identified as animal experiments by the Hungarian Animal Protection Act (“1998. évi XXVIII. Törvény,” 3. §9.), which identifies animal experiments, as this study was non-invasive. The application number of the ethical commission by the Pest County Government Office is PE/EA/2019-5/2017. Each owner filled in a consent form stating that they have been informed of the tests. Our Consent Form was based on the Ethical Codex of Hungarian Psychologists (2004).

We tested a total of 90 dogs between 1 and 12 years of age, of different breeds and mixed-breeds (see below). Owners volunteered to participate in the test and were recruited through the Family Dog Project database of Eötvös Loránd University, Budapest, Hungary. All dogs had been living with their owners for at least 6 months before the test. Dogs were assigned to three groups according to their work and training experience. Size, sex, and breed were balanced across groups:

1. Untrained family dogs had no certification exams. N = 30, 14 males, 16 females, mean age = 4.05, SD ± 2.74, breeds: 1 beagle, 7 border collies, 3 German shepherd dogs, 4 golden retrievers, 3 Labrador retrievers, 1 Maltese, 10 mixed, 1 English cocker spaniel.

2. Trained family dogs are dogs trained for recreational purposes. They had 1–4 certification exams (27 obedience, 23 agility, 11 herding, 5 guarding, 9 other: rescue dog, frisbee, dog dancing, K99). N = 30, 15 males, 15 females, mean age = 4.66, SD ± 2.67, Breeds: 8 Border Collies, 1 Bouvier, 1 Dobermann, 2 Golden Retrievers, 1 groenendael, 1 kelpie, 1 Labrador retriever, 2 malinois, 8 mixed, 1 mudi, 1 sheltie, 2 Hungarian vizslas, 1 Yorkshire terrier.

3. Working dogs worked as certified assistance or therapy dogs. assistance dogs were trained to aid individuals with disabilities by the dogs for human charity ( http://kea-net.hu/ ). Therapy dogs were all certified trained dogs, and lived with their owners at their homes. N = 30, 15 males, 15 females, mean age = 4.47, SD ± 3.32, 1 Airdale terrier, 3 border collies, 1 Cavalier King Charles spaniel, 4 golden retrievers, 1 groenendael, 1 Irish setter, 2 Labrador retrievers, 1 Malinois, 8 mixed, 2 standard poodles, 1 English cocker spaniel, 4 Tervuerens, 1 Hungarian vizsla.

Experimental Setup

All dogs had at least 1 h of fasting time before the testing. Dogs were tested in a room unfamiliar to them at the Eötvös University, Department of Ethology. Four cameras in each corner of the room videotaped all testing sessions. The room was 3 × 6 m 2 and there was a drawer where the problem box was stored before the start of the test and a chair for the owner to sit on ( Figure 1 ).

Figure 1 . The experimental setup. Written informed consents were obtained from the individuals for the publication of this image.

As a problem box we used a commercial wooden dog toy (Nina Ottosson® Dog Brick) that comprised a rectangular base with eight holes where treats could be hidden. A sliding wooden brick covered eight holes on both longer sides of the toy, so dogs had to slide the covers toward the middle with their paws or nose in order to get the treats. The bricks could not be lifted. Eight pellets of dry food in the eight holes on both longer sides were used as treats. None of the dogs were familiar with the apparatus prior to the task.

At the beginning of the test, the owner sat on a chair holding the dog on leash. The experimenter (female, 22 years old), who was the same for all dogs, took the interactive dog toy out from the drawers, placed it on the ground, and put a pellet of dry food inside each hole. Thus, dogs were able to see the baiting. When she was ready, the experimenter stepped back, the owner released the dog and the testing began. The dog had 2 min to obtain the food pellets from the apparatus. During this period, the owner was allowed to encourage the dog to find the pellets, verbally and by pointing at the apparatus, but we forbade the use of any previously trained or known commands relevant to the task such as “catch” or “nose.” The owner could not touch the apparatus nor the dog ( Figure 1 ). After the 2 min had elapsed, the experimenter put the toy back in the drawer. Dogs were allowed to eat only the food pellets they had recovered.

Behavioral Variables

We measured the duration of the vocalizations using a 0–3 score. We also measured the proportion of time dogs spent wagging their tail and the proportion of time oriented to the apparatus (including manipulating it, as gazing at the apparatus was often immediately followed by manipulation, therefore it would have been difficult to separate the two behaviors). We counted the number of times the dog gazed at the owner/experimenter, because gazing was generally a short event (just a glance) and provided more information than duration. We also counted the number of food pellets eaten after the behavior tests, on the spot. Other behavioral measures were coded from the videos using Solomon Coder (András Péter). See Table 1 for details and descriptive statistics.

Table 1 . Descriptive statistics of the measured raw variables and factor loadings of the standardized variables.

Statistical Analysis

We analyzed the inter-rater reliability of the variables using two-way random intraclass correlation, looking for absolute agreement between average measures. The inter-rater reliabilities were satisfactory (ICC > 0.741, N = 10).

After standardizing the variables, we ran principal component analysis and calculated factor scores. Cronbach alpha (CA) was used for checking the internal consistency of the factor. For investigating differences in the behavioral factor score (“social strategy” score, see below), as a function of group, sex (as fixed factors), and age (as covariate) we applied General Linear Model with Student–Newman–Keuls (SNK) post-hoc test, including all main effects and two way interactions. We used backward elimination to obtain the minimum adequate model. SPSS v25 ( 30 ) was used for the analyses.

Descriptive statistics of the variables and factor loadings are presented in Table 1 . Standardized variables loaded on a single factor. The total explained variance of the factor was 51.5%, CA = 0.8. The factor included looking at the owner, looking at the experimenter, tail wagging, and vocalization with positive loadings while orientation toward the apparatus and number of food pellets eaten had negative loadings. We labeled this factor as “social strategy,” because high score indicated that the dog uses communicative signals toward the human partners, including gazing, vocalization, tail wagging.

Only group affected the social strategy score [ F (2, 85) = 16.477, p < 0.001, partial eta squared = 0.275, Figure 2 ), age or sex had no effect and there were no interactions (all p > 0.05). According to the SNK post-hoc tests, all groups differed from each other (alpha = 0.05). Untrained dogs obtained the highest social strategy scores, trained dogs had lower scores, followed by working dogs.

Figure 2 . Social strategy factor scores of the three dog groups.

We set out to investigate the problem-solving abilities and related behaviors of dogs with different levels of training and working experience (trained and untrained family dogs as well as working assistance and therapy dogs) in a problem-solving task. Working assistance and therapy dogs displayed a less social and more problem-oriented strategy with a higher success rate than both untrained and trained family dogs. The frequent use of social strategies (i.e., gazing) is correlated with less persistence on the task (i.e., independent manipulation of the apparatus) and consequently with lower success ( 17 ). The results are also consistent with prior literature stating that animals persisting more on their problem-solving attempts are more successful in actually solving the task ( 16 ).

As it was mentioned in the introduction, the literature is mixed regarding the effects of training on dogs' persistence and gazing behavior during problem-solving tasks. For instance, Marshall-Pescini et al. ( 4 , 22 ) found differences in trained dogs' gazing and persistence patterns, but other authors did not find these differences ( 18 , 24 , 31 ). Results regarding working dogs' abilities should be taken with caution, as dogs from different studies vary in the type and amount of training they have received. For example, dogs in Marshall-Pescini et al. ( 22 ) were trained for different purposes (agility, police, search and rescue, and man-trailing), while Brubaker and Udell ( 24 ) tested search and rescue dogs, D'Aniello et al. ( 31 ) focused on water rescue dogs, and in Topál et al. ( 18 ) dogs were trained for basic obedience. A possible explanation is that specific training and working experience confounded the results. We have tried to independently assess ( 1 ) the effect of training for recreational purposes as dogs in our trained family group were trained for different hobby activities, mainly obedience and agility, and ( 2 ) the effect of specific training, as working dogs were trained as assistance and therapy dogs. Therefore, the type and methods of training could be an important aspect to take into account in future studies. Most probably the broad category “trained vs. untrained” is not precise enough to unravel the effect of training on problem-solving behaviors. Furthermore, working dogs may vary in their independence levels according to the context in which they work. For instance, water rescue dogs did not differ from pet dogs in their interaction with the apparatus during an unsolvable task, but they directed their first gaze significantly more often toward the owner and spent more time gazing at people than untrained pet dogs ( 31 ). Water rescue dogs are rewarded for looking at the handler during their training, and during their service they have to remain inactive for a long time in the vicinity of their owners in order not to cause any disturbance, and they take initiatives only upon command. These specific requirements probably affect their performance during problem-solving tasks.

A direct antecedent in the literature is the study of Mongillo et al. ( 32 ) who measured dogs' attention toward the owner in untrained family dogs, agility trained dogs, and assisted intervention animals. They assessed the number of gazes and the amount of time dogs spent watching their owner in a baseline condition where the owner walked alone in a room, and in a selective attention test where the owner's movements were mirrored by an experimenter. During the baseline phase, agility dogs shifted their gaze frequently toward the owner and were also the ones who spent the lesser amount of time looking at their owners, while assistance dogs gazed longer. In addition, assistance dogs gazed longer at their owners during the selective attention test. These results support the idea that different training and everyday activities may modify dogs' attentional patterns. Contrary to our results, Mongillo et al. ( 32 ) found that dogs participating in animal assisted interventions were the most attentive to their owners. This apparent contradiction could be due to the differences in the task. Unlike Mongillo et al. ( 32 ), we presented dogs with a problem-solving situation, in which dogs had to manipulate an apparatus to access a reward. In this latter scenario we observed that working dogs (which include dogs participating in animal assisted interventions) displayed less social strategies than the other group of dogs. Assistant and therapy dogs have to be attentive to their owners' needs but once they understand them or receive a specific command, they should be independent to succeed in their tasks. This interpretation is also supported by the fact that agility dogs in Mongillo et al. ( 32 ) shifted their gaze toward the owner more frequently than family dogs which is an important feature in the agility sport, but they do not need to solve novel problems independently during it. In our study trained family dogs (which include agility dogs) differed in the use of social strategies from untrained pet dogs. Thus, training for specific purposes may yield different patterns of social behavior depending on the context, emphasizing the plasticity and adaptability of dogs' behavior.

Importantly, according to the SNK post-hoc tests, trained family dogs had lower social strategy scores than untrained family dogs. Possibly, trained dogs were more used to facing novel situations and they could have generalized their training experience to this situation as well. It is possible that during training sessions dogs have to persevere and try different behaviors before getting the reward and that the contextual cues of the testing scenario trigger some of those responses. Indirect evidence supporting this idea comes from studies indicating that dogs are able to generalize and learn to follow novel and complex communicative signals faster when they have previously received a brief training phase with a simpler communicative cue ( 33 , 34 ).

Nevertheless, training for recreational purposes did not seem to be enough for dogs to reach the effectiveness of working assistance and therapy dogs, as the latter were more successful problem solvers and had lower scores in the use of the social strategies component. This result suggests that dogs' everyday experience is an uttermost important aspect to take into account when assessing their skills in a problem-solving situation. There are at least two possible, non-exclusive, explanations for this difference. First, it is possible that working assistance and therapy dogs were more comfortable in the presence of strangers and in novel situations given that they usually accompany their owners to a variety of places. Second, it is possible that dogs that have successfully accomplished the training as assistance or therapy dogs had pre-existing characteristics that distinguished them from other dogs. For instance, it has been shown that personality traits such as boldness are related with the successful training of working dogs ( 35 ). We propose that these two explanations are complementary, because it is possible that those dogs that became working dogs were encouraged during their everyday activities to behave in a more independent manner. Owners were allowed to encourage their dogs during the task, verbally or pointing to the apparatus, but without using commands or touch. Interestingly, Udell ( 11 ) reported that dogs, who were encouraged, spent more time in contact and looking at the puzzle box, but they were not significantly more successful in solving the task. Similarly, in Brubaker and Udell ( 24 ) encouraged family dogs interacted more with the apparatus but their performance was not significantly better. Conversely, encouragement did improve the performance of dogs trained for search and rescue ( 24 ). Given that in the present study we did not systematically manipulate the quantity and quality of the encouragement, we cannot derive unambiguous conclusions regarding this aspect. Udell's ( 11 ) results suggest that the use of encouragement and verbal instructions modulates problem-solving behavior, but their particular effects could depend on the context as well as working and training experience ( 11 ). In this regard, it is also possible that dogs react differently to verbal commands. Working dogs are trained to respond to a command by performing a specific action. For example, if the owner points to a particular object and asks the dog to do something with it, trained working dogs will manipulate the object instead of looking at the owner, while untrained pet dogs may be uncertain about what to do and will gaze at the owner in search for further clues [similarly to young dogs in Miklósi et al. ( 5 )]. Furthermore, not only the type of commands given by the owner affects dogs' performance, but also the bond between them. Topál et al. ( 18 ) compared the performance of dogs categorized according to their relationship with the owners. “Companion dogs” were defined as dogs living indoors as a member of the family and “working dogs” were kept outside the house as a guard or for some other purpose. In a simple manipulative task dogs had to manipulate an apparatus in order to get the reward while the owner could encourage them to retrieve the food. Companion dogs gazed more at the owner, started to manipulate the apparatus later and also retrieved less food than working dogs. The authors also found that obedience training did not affect dogs' performance or gazing patterns to their owners. These results are in line with our findings about the similar gazing patterns between trained and untrained family dogs.

One limitation of the study is that the dogs' characteristics before training were unknown. As it occurs in many studies assessing the effect of training on dogs' cognitive skills, the lack of a baseline measurement before training makes it impossible to guarantee that dogs were not selected for such work based on their pre-existing characteristics such as an increased persistence. Another limitation of these kind of studies is that training methods may differ between specific trainers and yield different results on dogs' problem-solving strategy. Thus, in future research, specific types and methods of previous training should also be taken into account when assessing dogs' problem-solving skills.

Summing up, we have shown that working assistance and therapy dogs were more independent problem solvers compared to both trained and untrained family dogs, who privileged a more social strategy. Thus, although assistance and therapy dogs need to show highly developed social understanding in their interactions with the owner, their special training and work may have increased their persistence and independent problem-solving skills. However, obtaining training certificates (mainly obedience and agility in this study) also increased the independent problem-solving tendency in our task, suggesting that trained family dogs generalize their training experience of facing novel situations and perseverance for obtaining rewards.

Data Availability Statement

All datasets generated for this study are included in the article/ Supplementary Material .

Ethics Statement

The animal study was reviewed and approved by Pest County Government Office, PE/EA/2019-5/2017. Written informed consent was obtained from the owners for the participation of their animals in this study.

Author Contributions

EK, ÁM, and MG designed the experiments and collected the data. EK analyzed the data. FC, CC, and EK wrote the first draft. All authors finalized the manuscript. ÁM and EK provided funding.

This project has received funding from the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation program (Grant Agreement No. 680040), the ERASMUS+ mobility program, the Hungarian Academy of Sciences (Grant F01/031), the National Research, Development and Innovation Office (Grant No. 132372K), and the National Brain Research Program (2017-1.2.1-NKP-2017-00002).

Conflict of Interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Acknowledgments

The authors are grateful for Barbara Gáspár, Borbála Turcsán, and Vera Konok in collecting behavioral data and Ivaylo B. Iotchev for language editing.

Supplementary Material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fvets.2020.00164/full#supplementary-material

1. Miklósi Á, Szabó D. Modelling behavioural evolution and cognition in canines: some problematic issues. Jpn J Anim Psychol . (2012) 62:69–89. doi: 10.2502/janip.62.1.11

CrossRef Full Text | Google Scholar

2. Shettleworth SJ. Clever animals and killjoy explanations in comparative psychology. Trends Cogn Sci. (2010) 14:477–81. doi: 10.1016/j.tics.2010.07.002

PubMed Abstract | CrossRef Full Text | Google Scholar

3. Frank H, Frank MG. Comparative manipulation-test performance in ten-week-old wolves ( Canis lupus ) and Alaskan malamutes ( Canis familiaris ): a Piagetian interpretation. J Comp Psychol. (1985) 99:266–74. doi: 10.1037/0735-7036.99.3.266

4. Marshall-Pescini S, Valsecchi P, Petak I, Accorsi PA, Previde EP. Does training make you smarter? The effects of training on dogs' performance ( Canis familiaris ) in a problem-solving task Behav. Proc. (2008) 78:449–54. doi: 10.1016/j.beproc.2008.02.022

5. Miklósi Á, Kubinyi E, Topál J, Gácsi M, Virányi Z, Csányi V. A simple reason for a big difference: wolves do not look back at humans, but dogs do. Curr. Biol. (2003) 13:763–6. doi: 10.1016/s0960-9822(03)00263-x

6. Osthaus B, Lea SEG, Slater AM. Dogs ( Canis lupus familiaris ) fail to show understanding of means-end connections in a string-pulling task. Anim Cogn. (2005) 8:37–47. doi: 10.1007/s10071-004-0230-2

7. Shimabukuro C, Putrino N, Helbling J, Tognetti S, Bentosela M. Individual differences in social and non-social behaviors in domestic dogs (Canis familiaris) during the acquisition, extinction and reacquisition of a problem-solving task. Behav Proc. (2015) 113:179–86. doi: 10.1016/j.beproc.2015.02.002

8. Polgár Z, Miklósi Á, Gácsi M. Strategies used by pet dogs for solving olfaction-based problems at various distances. PLoS ONE . (2015) 10:e0131610. doi: 10.1371/journal.pone.0131610

9. Cavalli CM, Carballo F, Bentosela M. Gazing behavior during problem-solving tasks in domestic dogs. A critical review. Dog Behav. (2019) 4:23–44. doi: 10.4454/db.v4i3.68

10. Brubaker L, Dasgupta S, Bhattacharjee D, Bhadra A, Udell MAR. Differences in problem-solving between canid populations: Do domestication and lifetime experience affect persistence? Anim Cogn. (2017) 20:717–23. doi: 10.1007/s10071-017-1093-7

11. Udell MAR. When dogs look back: Inhibition of independent problem-solving behaviour in domestic dogs ( Canis lupus familiaris ) compared with wolves ( Canis lupus ). Biol Lett. (2015) 11:20150489. doi: 10.1098/rsbl.2015.0489

12. Marshall-Pescini S, Virányi Z, Kubinyi E, Range F. Motivational factors underlying problem-solving: comparing wolf and dog puppies' explorative and neophobic behaviors at 5, 6, and 8 weeks of age. Front Psychol. (2017) 8:180. doi: 10.3389/fpsyg.2017.00180

13. Range F, Virányi Z. Development of gaze following abilities in wolves ( Canis lupus ). PLoS ONE . (2011) 6:e16888. doi: 10.1371/journal.pone.0016888

14. Rao A, Bernasconi L, Lazzaroni M, Marshall-Pescini S, Range F. Differences in persistence between dogs and wolves in an unsolvable task in the absence of humans. PeerJ. (2018) 6:e5944. doi: 10.7287/peerj.preprints.26913v1

15. Smith BP, Litchfield CA. Looking back at “looking back”: operationalising referential gaze for dingoes in an unsolvable task. Anim Cogn. (2013) 16:961–71. doi: 10.1007/s10071-013-0629-8

16. Chow PKY, Lea SEG, Leaver LA. How practice makes perfect: The role of persistence, flexibility and learning in problem-solving efficiency. Anim. Behav. (2016) 112:273–83. doi: 10.1016/j.anbehav.2015.11.014

17. Marshall-Pescini S, Rao A, Virányi Z, Range F. The role of domestication and experience in “looking back” towards humans in an unsolvable task. Sci. Rep. (2017) 7:46636. doi: 10.1038/srep46636

18. Topál J, Miklósi Á, Csányi V. Dog-human relationship affects problem-solving behavior in the dog. Anthrozoös. (1997) 10:214–24. doi: 10.2752/089279397787000987

19. Barrera G, Fagnani J, Carballo F, Giamal Y, Bentosela M. Effects of learning on social and nonsocial behaviors during a problem-solving task in shelter and pet dogs. J Vet Behav. (2015) 10:307–14. doi: 10.1016/j.jveb.2015.03.005

20. Scandurra A, Prato-Previde E, Valsecchi P, Aria M, D'Aniello B. Guide dogs as a model for investigating the effect of life experience and training on gazing behaviour. Anim Cogn. (2015) 18:937–44. doi: 10.1007/s10071-015-0864-2

21. Kovács K, Kis A, Pogány Á, Koller D, Topál J. Differential effects of oxytocin on social sensitivity in two distinct breeds of dogs ( Canis familiaris ). Psychoneuroendocrinology. (2016) 74:212–20. doi: 10.1016/j.psyneuen.2016.09.010

22. Marshall-Pescini S, Frazzi C, Valsecchi P. The effect of training and breed group on problem-solving behaviours in dogs. Anim Cogn. (2016) 19:571–9. doi: 10.1007/s10071-016-0960-y

23. Range F, Heucke SL, Gruber C, Konz A, Huber L, Virányi Z. The effect of ostensive cues on dogs' performance in a manipulative social learning task. Appl Anim Behav Sci. (2009) 120:170–8. doi: 10.1016/j.applanim.2009.05.012

24. Brubaker L, Udell MAR. The effects of past training, experience, and human behaviour on a dog's persistence at an independent task. Appl Anim Behav Sci. (2018) 204:101–7. doi: 10.1016/j.applanim.2018.04.003

25. Rooney NJ, Clark CCA, Casey RA. Minimizing fear and anxiety in working dogs: a review. J Vet Behav. (2016) 16:53–64. doi: 10.1016/j.jveb.2016.11.001

26. MacLean EL, Hare B. Enhanced selection of assistance and explosive detection dogs using cognitive measures. Front Vet Sci. (2018) 5:236. doi: 10.3389/fvets.2018.00236

27. Brennan J, Nguyen V. Support Animals and Emotional Support Animals: Where Are They Allowed and Under What Conditions? Southwest ADA Center. (2014). Retrieved from: https://adata.org/guide/service-animals-and-emotional-support-animals (accessed December 16, 2019).

Google Scholar

28. International Association of Human–Animal Interaction Organizations (IAHAIO). The IAHAIO Definitions for Animal Assisted Intervention and Guidelines for Wellness of Animals Involved in AAI . (2018). Retrieved from: http://iahaio.org/wp/wp-content/uploads/2018/04/iahaio_wp_updated-2018-final.pdf

29. Gácsi M, Szakadát S, Miklósi Á. Assistance dogs provide a useful behavioral model to enrich communicative skills of assistance robots. Front Psychol. (2013) 4:971. doi: 10.3389/fpsyg.2013.00971

30. IBM Corp. IBM SPSS Statistics for Windows, Version 25.0 . Armonk, NY: IBM Corp. (2017).

31. D'Aniello B, Scandurra A, Prato-Previde E, Valsecchi P. Gazing toward humans: A study on water rescue dogs using the impossible task paradigm. Behav Processes . (2015) 110:68–73. doi: 10.1016/j.beproc.2014.09.022

32. Mongillo P, Pitteri E, Marinelli L. Sustained attention to the owner is enhanced in dogs trained for animal assisted interventions. Behav Proc. (2017) 140:69–73. doi: 10.1016/j.beproc.2017.03.024

33. Agnetta B, Hare B, Tomasello M. Cues to food location that domestic dogs ( Canis familiaris ) of different ages do and do not use. Anim Cogn. (2000) 3:107–12. doi: 10.1007/s100710000070

34. Elgier AM, Jakovcevic A, Mustaca AE, Bentosela M. Pointing following in dogs: are simple or complex cognitive mechanisms involved? Anim Cogn. (2012) 15:1111–9. doi: 10.1007/s10071-012-0534-6

35. Svartberg K. Shyness – boldness predicts performance in working dogs. Appl Anim Behav Sci. (2002) 79:157–74. doi: 10.1016/S0168-1591(02)00120-X

Keywords: human-animal interaction, canine-cognition, persistence, gazing, unsolvable task, working dogs

Citation: Carballo F, Cavalli CM, Gácsi M, Miklósi Á and Kubinyi E (2020) Assistance and Therapy Dogs Are Better Problem Solvers Than Both Trained and Untrained Family Dogs. Front. Vet. Sci. 7:164. doi: 10.3389/fvets.2020.00164

Received: 03 January 2020; Accepted: 06 March 2020; Published: 31 March 2020.

Reviewed by:

Copyright © 2020 Carballo, Cavalli, Gácsi, Miklósi and Kubinyi. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY) . The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

*Correspondence: Enikő Kubinyi, eniko.kubinyi@ttk.elte.hu

† These authors have contributed equally to this work

This article is part of the Research Topic

Humans in an Animal’s World – How Non-Human Animals Perceive and Interact with Humans

clock This article was published more than  2 years ago

Thinking about how dogs think

Back in 2002, when Alexandra Horowitz was working toward her PhD at the University of California at San Diego, she believed that dogs were a worthy thing to study. But her dissertation committee, which favored apes and monkeys, needed convincing.

“They were primate people,” she said. “They all studied nonhuman primates or human primates, and that’s where it was thought that the interesting cognitive work was going to happen. Trying to show them that there would be something interesting with dogs — that was a challenge.”

Oh, how things can change in just two decades, especially in a nation that includes about 90 million dogs among its residents — everything from beloved pets to working dogs doing all kinds of tasks, from sniffing out drugs in airports to assisting blind people with crossing a street. Today, Horowitz is a senior research fellow at Barnard College in New York City, where her specialty is dog cognition: understanding how dogs think, including the mental processes that go into tasks such as learning, problem-solving and communication. Dog cognition is now a widely respected field, a growing specialty branch of the more general animal-cognition research that has existed since the early 20th century.

“This field, and animal cognition, really, is all within our lifetimes,” Horowitz said. “It’s not as if nobody ever looked at dogs, but they weren’t looking at their minds.”

Looking at dogs’ minds, so far, has revealed quite a few insights. The Canine Cognition Center at Yale University, using a game where humans offer dogs pointing and looking cues to spot where treats are hidden, showed that dogs can follow our thinking even without verbal commands. The Max Planck Institute for Evolutionary Anthropology in Germany figured out that dogs are smart about getting what they want — they will eat forbidden food more frequently if humans can’t see them. Researchers from Austria, Israel and Britain determined that seeing a caregiver, versus a stranger, activated dogs’ brain regions of emotion and attachment much as it does in the human mother-child bond. Other European researchers showed that negative-reinforcement training (like jerking on a leash) causes lingering emotional changes and makes the dog less optimistic overall.

Read more stories of dogs, humans and the relationship they share

Some dog owners hear about this type of research and think: “They did a whole study to figure out that my dog looks where I point? I could have told you that.” But the studies aren’t just about what a dog is doing. They’re indicating areas to research so that we can better understand why and how the dog is doing it — in other words, what’s happening inside the dog’s mind.

“Maybe they’re not looking at your finger at all. Maybe they’re paying attention to your face and not to your hand,” said Federico Rossano, whose team at the University of California at San Diego is trying to determine whether dogs can translate their thoughts into words that humans can understand through a language device. “A lot of this becomes interesting in terms of how you can train them better.”

An evolving area of research

Right now, with no organizing body in the field, it’s hard to say exactly how many people are doing dog-cognition research. You can count on two hands the number of dedicated university spaces led by professors with graduate students and funding grants. When the leaders from those places get together once a year, it’s usually at someone’s home.

But researchers at universities doing studies on dogs? There are now many dozens of those, and there’s no lack of students wanting to at least dabble in the work.

“The thing that gets my students all abuzz is that people always want to know whether their dog loves them back,” said Ellen Furlong, associate professor of psychology at Illinois Wesleyan University and leader of its Dog Scientists Group.

Every semester, on the first day, she asks students if their dogs are happy. It’s her way of helping them understand why the study of dog cognition is important.

“They’re always kind of offended — ‘O f course my dog is happy. I love my dog,’ ” she said. “But then you dig a little bit and push them and say: ‘Your dog’s life is different from your life. You get to decide when your dog gets to eat and play and go outside. You decide everything about your dog’s life, but your dog isn’t human. They have different wants and needs than you do.’ They have a semester-long assignment where they have to consider how their work on cognition can help to design some enrichment activities to improve the dogs’ lives.”

The topics that dog-cognition researchers focus on today often are chosen based on personal interests. While Furlong is most curious about ethics, welfare and how humans can meet dogs’ psychological needs, Horowitz is focusing her research on what dogs understand through smell. At the Duke Canine Cognition Center in North Carolina, Brian Hare is trying to determine — when a dog is still a puppy — whether the way a dog thinks might make her a good candidate for different jobs as an adult.

“We’re saying, ‘Here are some cognitive abilities that are critical for training for these jobs,’ ” Hare said. “It’s a little bit like talking about personality, but we’re talking about your cognitive personality, in a way. Maybe you have a really good memory for space, or maybe you’re good at understanding human gestures. The question is whether we can identify some of these dogs really early, in the first two to three months of life, who will do well in these programs.”

How the research is done

One example of dog cognition research with a potential training application is a study that Horowitz did on nose work — an activity that lets dogs use their natural abilities with scents to find everything from a treat hidden under a cone to marijuana in somebody’s suitcase.

Horowitz and her team showed the dogs three buckets and taught them that one of the buckets always had a treat under it, and one did not. Then she measured how quickly the dogs went to the “ambiguous bucket” in the middle.

The dogs then attended nose-work classes. These types of advanced classes are widely available at the same types of schools that teach basic obedience. In the nose-work classes, dogs are encouraged and trained to use their noses to search for and find treats or favorite toys that are hidden under boxes or cones, inside suitcases or in other places.

After a few weeks of nose-work classes, Horowitz repeated the bucket test.

“What we found was the dogs in the nose-work class got faster at approaching ambiguous stimulus,” she said, adding that the results suggest that for some dogs, taking nose-work classes could help them feel more optimistic. “The group that had nose work changed their behavior afterward, so I have to say it’s something about the nose work. I don’t know exactly what it was, but if the effect is profound and we keep seeing it, we would go in and try to see what it was that made it useful for the subjects.”

Hare is widely credited with having jump-started America’s dog-cognition research field. In the late 1990s as an undergraduate, he was doing research with chimpanzees when he realized they couldn’t do something that his dogs could do: follow a human’s pointing gesture to find food. Chimpanzees are the closest animal relatives humans have, and dogs could do something they couldn’t. Researchers suddenly wanted to know why dogs could understand something that chimpanzees could not.

In his most recent study , published in July, Hare and his team looked at the difference between wolf and dog pups. There had been some debate in the dog-cognition field about where dogs’ unusual abilities to cooperate with humans originate — whether those abilities are biological or taught. So the team gave a battery of temperament and cognition tests to dog and wolf puppies that were 5 weeks to 18 weeks old. The pups of both species were given the chance to approach familiar and unfamiliar humans to retrieve food; to follow a human’s pointing gesture to find food; to make eye contact with humans, and more. The team found that even at such a young age, the dog pups were more attracted to humans, read the human gestures more skillfully, and made more eye contact with humans than the wolf pups did.

The conclusion? The way that humans domesticated dogs actually altered the dogs’ developmental pathways, meaning their abilities to cooperate with us today are biological — a research result that is likely to have many practical implications.

“It’s highly inheritable, and it’s potentially manipulatable through breeding,” Hare said, adding that dogs might be bred to specialize in certain types of thinking. The finding opens up the idea of studying dogs in ways that could make deep-pocketed entities like the U.S. government want to fund more dog-cognition research, Hare said.

By way of example, he talked about dogs he has worked with for the U.S. Marine Corps, compared with dogs he has worked with for Canine Companions for Independence in California. The Marines needed dogs in places like Afghanistan to help sniff out incendiary devices, while the companions agency needed dogs that were good at helping people with disabilities.

Just looking at both types of purpose-bred dogs, most people would think they’re the same — to the naked eye, they all look like Labrador retrievers, and on paper, they would all be considered Labrador retrievers. But behaviorally and cognitively, because of their breeding for specific program purposes, Hare said, they were different in many ways.

Hare devised a test that could tell them apart in two or three minutes. It’s a test that’s intentionally impossible for the dog to solve — what Star Trek fans would recognize as the Kobayashi Maru. In Hare’s version, the dog was at first able to get a reward from inside a container whose lid was loosely secured and easy to dislodge; then, the reward was placed inside the same container with the lid locked and unable to be opened. Just as Starfleet was trying to figure out what a captain’s character would lead him to do in a no-win situation, Hare’s team was watching whether the dog kept trying to solve the test indefinitely, or looked to a human for help.

“What we found is that the dogs that ask for help are fantastic at the assistance-dog training, and the dogs that persevere and try to solve the problem no matter what are ideal for the detector training,” Hare said. “It’s not testing to see which dog is smart or dumb. What we’ve been able to show is that some of these measures tell you what jobs these dogs would be good at.”

What comes next in the field of dog-cognition research is probably a bit more of everything. Some researchers are following their interests, while others are following the research grants. Those grants can come from a wide array of sources, including the government trying to help soldiers with post-traumatic stress disorder, shelters trying to rehome animals and neuroscience institutes looking for insights across species.

“It’s a really exciting moment,” Hare said. “I think we can continue on with individual researchers pursuing fun, interesting things — the students and the universities love it — but most successful academic endeavors have two parts. Being intellectual is wonderful, but that kind of research tends to struggle with funding. Academic endeavors with practical application tend to be incredibly well funded, and then the field grows.

“If you can have both of those things, then it will grow, and it will grow phenomenally,” he added. “If it’s just, ‘We’re going to do this because people love dogs,’ that’ll be fun, but it will stay small like it is now.”

• In a pandemic, these pups have made all the difference August 17, 2021 In a pandemic, these pups have made all the difference August 17, 2021
• The voices we make when we pretend our dogs can talk August 12, 2021 The voices we make when we pretend our dogs can talk August 12, 2021
• Thinking about how dogs think August 12, 2021 Thinking about how dogs think August 12, 2021

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here .

Loading metrics

Open Access

Peer-reviewed

Research Article

The Importance of the Secure Base Effect for Domestic Dogs – Evidence from a Manipulative Problem-Solving Task

* E-mail: [email protected]

Affiliations Department of Cognitive Biology, University of Vienna, Vienna, Austria, Clever Dog Lab Society, Vienna, Austria

Affiliations Department of Cognitive Biology, University of Vienna, Vienna, Austria, Clever Dog Lab Society, Vienna, Austria, Messerli Research Institute, University of Veterinary Medicine Vienna, Medical University Vienna, University of Vienna, Vienna, Austria

Affiliations Clever Dog Lab Society, Vienna, Austria, Messerli Research Institute, University of Veterinary Medicine Vienna, Medical University Vienna, University of Vienna, Vienna, Austria

• Lisa Horn, 
• Ludwig Huber, 
• Friederike Range

• Published: May 29, 2013
• https://doi.org/10.1371/journal.pone.0065296
• Reader Comments

It has been suggested that dogs display a secure base effect similar to that found in human children (i.e., using the owner as a secure base for interacting with the environment). In children, this effect influences their daily lives and importantly also their performance in cognitive testing. Here, we investigate the importance of the secure base effect for dogs in a problem-solving task.

Methodology/Principal Findings

Using a manipulative task, we tested dogs in three conditions, in which we varied the owner's presence and behavior ( Experiment 1 : “Absent owner”, “Silent owner”, “Encouraging owner”) and in one additional condition, in which the owner was replaced by an unfamiliar human ( Experiment 2 : “Replaced owner”). We found that the dogs' duration of manipulating the apparatus was longer when their owner was present than absent, irrespective of the owner's behavior. The presence of an unfamiliar human however did not increase their manipulation. Furthermore, the reduced manipulation during the absence of the owner was not correlated with the dog's degree of separation distress scored in a preceding attachment experiment.

Conclusions/Significance

Our study is the first to provide evidence for an owner-specific secure base effect in dogs that extends from attachment tests to other areas of dogs' lives and also manifests itself in cognitive testing – thereby confirming the remarkable similarity between the secure base effect in dogs and in human children. These results also have important implications for behavioral testing in dogs, because the presence or absence of the owner during a test situation might substantially influence dogs' motivation and therefore the outcome of the test.

Citation: Horn L, Huber L, Range F (2013) The Importance of the Secure Base Effect for Domestic Dogs – Evidence from a Manipulative Problem-Solving Task. PLoS ONE 8(5): e65296. https://doi.org/10.1371/journal.pone.0065296

Editor: Anna Dornhaus, University of Arizona, United States of America

Received: March 5, 2012; Accepted: April 26, 2013; Published: May 29, 2013

Copyright: © 2013 Horn et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Funding: This research was supported by the Austrian Academy of Sciences (OEAW: www.oeaw.ac.at ; DOC-fFORTE fellowship) and the University of Vienna ( www.univie.ac.at ; Forschungsstipendium) with grants to L. Horn and by the Austrian Science Fund (FWF: www.fwf.ac.at ; P21418-B17). During this research, F. Range received support from the FWF (P21244-B17). The Clever Dog Lab is financially supported by a private sponsor and by Royal Canin ( www.royal-canin.at ). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests: Altough Royal Canin is supporting the Clever Dog Lab with donations, the authors confirm that this does not alter their adherence to all the PLOS ONE policies on sharing data and materials, as detailed online in the guide for authors.

Introduction

Based on ethological principles, Bowlby [1] , [2] formulated the theory that for the survival of infants in humans as well as in many non-human animal species it is essential that infants develop a strong affectional bond with their primary caregiver – usually the mother. Separation from the attachment figure activates the infant's attachment system, which aims at restoring and maintaining proximity with this specific individual [1] . Four particular behavioral components can be used to discriminate a true attachment bond from other affectional bonds [3] : a) staying near to and resisting separation from the attachment figure (proximity maintenance), b) feeling distress upon involuntary separation from the attachment figure (separation distress), c) using the attachment figure as a base for exploring the environment free of anxiety (secure base), d) seeking out the attachment figure for contact and assurance in times of emotional distress (safe haven). Ainsworth [4] argued that the secure base effect was the most important component of the attachment system, because it is crucial for balancing the maturing infants' exploration of the world with maintaining proximity to the caregiver. Although Bowlby's original attachment theory had been developed in regard to human children, the same behavioral components have been found in infant-caregiver relationships in many bird and mammal species (e.g., chicken, macaques, and dogs; for a review see [5] ).

Domestic dogs have been closely associated with humans for about 15,000 years [6] and are so well adapted to their niche in the human society that in many cases the owner has replaced conspecifics as the main social partner. This unique relationship between adult dogs and their human owners bears a remarkable resemblance to an infant attachment bond: dogs are dependent on human care and their behavior seems specifically geared to engage their owners' care-giving system [7] , [8] . Given the broad comparative framework of Bowlby's original theories, several researchers have used attachment concepts and methodology to investigate whether the dog-human relationship conforms to the characteristics of an attachment bond. For that purpose most researchers used an experimental procedure developed for testing human children by Ainsworth and Wittig [9] – the Ainsworth Strange Situation Test (ASST). In this test children are confronted with an unfamiliar setting, an unfamiliar person entering the room and two brief separations from the attachment figure in a fixed sequence. This mildly stressful setting is geared to activate the child's attachment system in order to measure each of its behavioral components separately (e.g., crying at the exit of the parent (separation distress), exploring and playing more in the presence of the parent than in the absence (secure base)). Using a modified version adapted for testing adult dogs with their human owners, two studies found clear evidence for proximity seeking and separation distress in the dogs [10] , [11] . However, although the authors found more exploration in the presence of the owner than in the subsequent absence of the owner, they could not unequivocally attribute this to a secure base effect due to a strong confound with the sequence effects inherent to the procedure [11] . To control for this factor, Palmer and Custance [12] carried out a counterbalanced version of the ASST and found indications for a secure base effect independently from the sequence. However, according to attachment theory, the secure base effect should not only be evident in the ASST, but should influence most of the individual's interactions with the environment. In line with this, Matas et al. [13] showed that in human children the secure base effect had an influence on their performance in an experimental problem-solving task. Their study revealed that children, who were able to use their mother as a secure base for exploring the environment, were also more persistent and enthusiastic while solving the task than children for whom the mother was no secure base. Since the dog has emerged as a model species for behavioral and cognitive research in recent years, it is vital to understand whether the attachment to their owners – particularly the secure base effect – is also relevant for their performance in cognitive tasks as it has been shown for human children [13] .

The aim of our study was to investigate the importance of the secure base effect for dogs in a behavioral test situation. In children the secure base effect is mostly investigated by comparing their motivation to play in the caregiver's presence and absence [9] . However, while children typically spend long periods with solitary play, most dogs do not. Therefore, we used a problem-solving task, which the dogs were motivated to carry out for a long time (i.e. manipulating an apparatus in order to obtain a food reward). In Experiment 1 we tested whether the presence or absence of the owner would influence dogs' motivation to manipulate an apparatus. We predicted that if the owner acted as a secure base for the dog, the dog's performance should be poorer in the owner's absence (Condition “Absent owner”). To control for the possibility that the difference in performance was brought about by the absence of behavioral cues from the owners, each dog was tested in two different conditions with the owner present: a) the owner was blindfolded and did not interact with the dog (Condition “Silent owner”) and b) the owner was allowed to encourage the dog verbally (Condition “Encouraging owner”). To further examine the effects of dogs' general distress during separation from the owner, all dogs were independently tested in a shortened version of the ASST and their degree of separation distress was compared to their performance in the task. In Experiment 2 of this study, we further investigated whether the secure base effect was specific to the owner, which would be predicted if the relationship resembled an infant-caregiver relationship, or whether it would extend to an unfamiliar human (Condition “Replaced owner”).

Experiment 1

Materials and methods, ethics statement..

Owners gave their written consent for participating in behavioral studies with their dogs when entering the database of volunteer participants of the Family Dog Research Program ( http://kutyaetologia.elte.hu ) at the Department of Ethology of the Eötvös Loránd University in Budapest, Hungary. Prior to the experiments reported in this manuscript, the owners were informed about the details of the procedures of each experiment and were given the possibility to withdraw from participation. No special permission for use of dogs in non-invasive studies is required in Hungary. The relevant committee that allows conducting research without special permissions regarding animals is the University Institutional Animal Care and Use Committee (UIACUC, Eötvös Loránd University, Hungary). Since the owners were only required to interact with their dogs in their usual manner during the experiments and their behavior was not coded or analyzed, no approval for human experimentation was obtained. The participants in the video clips that are part of the supporting information have given written informed consent, as outlined in the PLOS consent form, to publication of themselves.

Participants.

Twenty-two dogs were recruited from a database of volunteer participants of the Family Dog Research Program at the Department of Ethology of the Eötvös Loránd University in Budapest, Hungary. Only dogs living permanently in the owner's household as pets were selected and all dogs had at least basic obedience training. All dogs were highly food-motivated. Two dogs had to be excluded because they failed the pre-test (see procedures section below). Therefore, 20 dogs (12M/8F; mean age ± SD = 2.7±2.36 years) completed the experiment. The sample consisted of 14 purebred dogs from three different FCI (Fédération Cynologique Internationale) breed groups (Sheepdogs: N = 9; Toy dogs: N = 3; Primitive types: N = 2) and 6 mixed-breed dogs.

Experimental Design.

Dogs first had to pass a pre-test to ensure that they were motivated to manipulate an apparatus to get food and that they were willing to consume food even in the absence of the owner. Dogs that passed the pre-test by taking food in the absence of the owner progressed to the test phase. In the test phase, we used three different test conditions, in which we varied the presence of the owner and the owner's behavior:

• Condition “Absent owner” (cAO) : The owner is not present in the experimental room during the trial.
• Condition “Silent owner” (cSO) : The owner is present in the experimental room during the trial, but remains silent.
• Condition “Encouraging owner” (cEO) : The owner is present in the experimental room during the trial and is encouraging the dog verbally.

We used a within-subject design so that each dog received one trial in each of the three test conditions. The sequence of the conditions was counterbalanced across dogs.

Apparatuses.

For the pre-test we used a folded cotton towel (13 cm×29 cm×2 cm) under which the food reward was placed. As apparatuses for the test we used four types of commercial interactive dog toys, which could be filled with food rewards: Nina Ottosson © Dog Pyramid (aDP, 13 cm×13 cm×17 cm), Hunter © Snack Bottle (aSB, 9 cm×20 cm×9 cm), Hunter © Snack Cactus (aSC, 20 cm×20 cm×20 cm), and Hunter © Rolling Snack (aRS, Figure 1 ). The latter toy was available in a smaller (10 cm×10 cm×7 cm) and a larger version (13 cm×13 cm×10 cm) and thus the size was adjusted to the size of the dog. All toys had to be manipulated persistently with either the paw or the muzzle to receive the food rewards placed inside. The food rewards were of high quality (i.e., dog sausage) and consisted of 5 pieces per trial. Before the experiment started we asked owners whether their dogs had already interacted with toys that were the same or similar to any of the four toys presented. For each dog we selected three toys that were unknown to the dog. Otherwise toys were selected randomly.

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

a) Folded cotton towel used in the pre-test, b) Nina Ottosson © Dog Pyramid (aDP), c) Hunter © Snack Cactus (aSC), d) Hunter © Snack Bottle (aSB), e) the smaller version of the Hunter © Rolling Snack (aRS).

https://doi.org/10.1371/journal.pone.0065296.g001

Experimental Set-up.

The experiments were carried out between February and June 2010 in a quiet experimental room (3 m×5 m) at the department. The room was equipped with two doors: door 1 could be used to enter the experimental room from the hallway, door 2 led to an adjacent room. A chair for the owner was placed on the left side of door 1. The experimenter was positioned on the right side of door 1 during the pre-test and test trials. Tape markings on the floor indicated a circular area (r = 1 m) around the owner's chair and the experimenter's position, respectively. A clock on the opposite wall was used by the experimenter to time the trials. A line on the floor marked the dog's release point. The apparatus was placed 2.5 m away from this line. The room was equipped with four cameras linked to monitoring and recording equipment in the adjacent room ( Figure 2 ).

The experimental room was equipped with two doors, one connecting it to the hallway (Door 1) and one connecting it to an adjacent room with monitoring and recording equipment (Door 2). At the beginning of each trial, the dog was released by the experimenter from the release point, which was 2.5 m away from the apparatus. During each trial the experimenter stood on the right side of door 1 – timing the trial with a clock on the opposite wall. The owner either sat on the designated chair on the left side of door 1 or was in the adjacent room – depending on the pre-test trial or the condition of the test trial. The dashed lines indicate the floor markings around the owner's chair and the experimenter's position, which were used for later video coding.

https://doi.org/10.1371/journal.pone.0065296.g002

Procedures.

All dogs participating in this experiment had previously been tested in a modified version of the ASST aimed at characterizing their relationship with their owners. This test consisted of several episodes in which either the owner, an unfamiliar human or both were present in the room with the dog. The test also comprised two separation episodes, which allowed us to assign a score of separation-related behavior (SRB) to every dog (see File S1 for a detailed description).

In the main experiment, the general procedure was the same for each trial of both the pre-test and the test phase. Before each trial, the dog always observed the handling of the food reward (pre-test trials, test trials) and/or the baiting of the apparatus (test trials) in the hallway outside the experimental room. The experimenter then entered the room through door 1 and placed the food reward or the apparatus on the designated position (see Figure 2 ). After that, the experimenter returned to the hallway. Right before each trial the owner entered the experimental room through door 1 while the dog waited in the hallway with the experimenter. The owner then took position depending on the pre-test trial or the condition of the test trial. In each trial, the owner had to wear dark sunglasses (either opaque or normal), which allowed us to manipulate the owner's visual access to the dog's actions depending on the condition. After waiting for 10 s, the experimenter entered the room together with the dog on a leash. She directed the dog to the line indicating the release point and released it with one command (i.e. “You can go, it's yours!”). Then she took her position next to the door. During the trial the experimenter never looked at the actions of the dog but kept looking at the clock on the opposite wall. After a pre-defined time (see detailed descriptions below) the trial ended and the experimenter called the dog back, put it on leash and walked it out of the experimental room. The owner waited for 10 s and then also left the room through door 1. There was a break of 5 to 10 min between trials.

Pre-test phase – The pre-test phase consisted of four trials that were administered in a fixed order:

• Trial 1 : Owner present, food reward on the floor
• Trial 2 : Owner present, food reward under the towel
• Trial 3 : Owner absent, food reward on the floor
• Trial 4 : Owner absent, food reward under the towel

In the first two trials of the pre-test phase the owner sat down on the chair and put on opaque sunglasses. The food had either been placed in the location of the apparatus directly on the floor (Trial 1) or under the towel (Trial 2). Trial 1 ended when the dog had consumed all five pieces of food or after a maximum of one minute. Since it was not possible for the experimenter to assess whether the dog had retrieved all pieces of food from underneath the towel, trial 2 ended after exactly one minute. In trial 3 and trial 4 the owner – after first entering the experimental room though door 1 – left the room through door 2 and remained in the adjacent room during the trial. The rest of the procedure was the same as in trial 1 and trial 2, respectively. In the adjacent room the owner could observe the events in the experimental room via monitoring equipment. The owner therefore knew when the trial was over and he or she had to leave this room and exit the experimental room through door 1 again.

Two dogs were not motivated to consume any piece of food from the floor in the absence of the owner (Trial 3) and could therefore not proceed to the test phase. Two further dogs failed to retrieve any piece of food from under the towel in the absence of the owner (Trial 4). However, these dogs were motivated to retrieve the food and actively manipulated the towel during this trial. Therefore we decided to include them in the test phase.

Test phase – In the test phase, we administered three trials, each of which lasted 5 min.

In a trial of the condition “Absent owner” (cAO) the owner left the room in the same way as in the trials 3 and 4 of the pre-test phase (see Movie S1 ). In a trial of the condition “Silent owner” (cSO) the owner sat down on the chair, put on opaque sunglasses and remained silent and passive throughout the trial (see Movie S2 ). In a trial of the condition “Encouraging owner” (cEO) the owner sat down on the chair, put on normal sunglasses and was allowed to encourage the dog verbally and to point at the apparatus throughout the trial. The owner had to remain seated on the chair but was allowed to pet the dog when it came close. However, the owner was not allowed to touch the apparatus (see Movie S3 ). The sequence of conditions was counterbalanced across dogs.

Data Analysis.

All experimental sessions were videotaped for later behavioral coding with Solomon Coder beta (©2006–2011 András Péter). All statistical analyses were carried out with SPSS Statistics 17.0.0 (©2008 SPSS Inc.).

Since we were interested in the dogs' motivation to manipulate and not in potential individual differences in their ability to retrieve the food, we coded their duration of manipulation in the test trials and did not analyze their success. “Manipulating” was recorded continuously whenever the dog was interacting with the apparatus with its muzzle or paw. Additionally, “staying close to the owner” and “staying close to the experimenter” was recorded continuously whenever the dog was with at least one paw and the head within the area marked by the circle around each respective person. A second coder blind to the aim of the experiment and to the experimental conditions coded 20% of the videos of the test trials and Cronbach's alpha was calculated as a measure of inter-observer reliability. Cronbach's alpha was greater than α = 0.9 for all behavioral variables.

A linear mixed model (LMM) with main effects and two-way interactions was used to investigate the effect of the factors “sequence of conditions” (1 st , 2 nd , 3 rd ), “type of apparatus” (aDP, aSB, aSC, aRS), and “condition” (cAO, cSO, cEO) on the variable “duration of manipulation”. To investigate whether the duration that dogs spent manipulating the apparatus when the owner was absent (cAO) was correlated with the degree of SRB scored during the ASST, we used Spearman's rank correlation. To control for individual variation in general manipulation durations, we calculated a “relative duration of manipulation in cAO” by dividing the absolute duration of manipulation in cAO by the dog's average duration of manipulation across all three trials and correlated this score with the dog's SRB score.

Additionally we calculated two separate LMMs (main effects, two-way interactions) where we investigated the effects of the same factors as above on the variable “duration spent in the proximity of the owner” in those conditions where the owner was present in the room (cSO, cEO) and on the variable “duration spent in the proximity of the experimenter” in all three conditions.

Analyses of the residuals confirmed normal distribution for all variables in the three LMMs. Post-hoc comparisons of estimated marginal means (EM means) were carried out with LSD confidence interval adjustment.

We found that the condition in which the dogs were tested had a highly significant effect on how long the dogs manipulated the apparatus (LMM, N = 20, F 2,36  = 12.478, P≤0.001, Figure 3 ). The dogs manipulated the apparatus shorter when the owner was absent (EM means, cAO-cSO: P = 0.001, cAO-cEO: P≤0.001), while there was no significant difference between the two conditions when the owner was present (EM means, cSO-cEO: P = 0.540). Neither the sequence nor the type of apparatus, nor any of the interactions had an effect on how long the dogs continued to manipulate the apparatus (full results of the LMM can be seen in File S1 ). When analyzing the correlation between the dogs' relative duration of manipulating when the owner was absent and their SRB score assigned in the ASST, we found no negative correlation (Spearman's rank correlation, N = 20, ρ = −0.140, P = 0.557), indicating that the effect observed in the LMM was not only present in those dogs with strong separation distress but in all dogs of the sample.

Experiment 1 . Mean duration of manipulating the apparatus in the conditions “Absent owner” (cAO), “Silent owner” (cSO), and “Encouraging owner” (cEO). Shown are mean ± s.e.m. *** represents P≤0.001.

https://doi.org/10.1371/journal.pone.0065296.g003

When investigating the time that the dogs spent in proximity of the two humans we found that the dogs spent overall equal amounts of time close to their owner in both conditions when the owner was present. Also, the sequence and the type of apparatus had no effect on this behavior (full results of the LMM can be seen in File S1 ). However, when looking at the duration that the dogs spent close to the experimenter, we found a highly significant effect of the condition. When the owner was absent the dogs spent more time close to the experimenter than in both conditions when the owner was present (LMM, N = 20, F 2,36  = 17.221, P≤0.001; EM means, cAO-cSO: P≤0.001, cAO-cEO: P≤0.001, cSO-cEO: P = 0.593, Figure 4 ). Further, there was a significant main effect of the sequence (LMM, F 2,36  = 4.611, P = 0.016) and a significant interaction term between condition and sequence (LMM, F 4,36  = 2.923, P = 0.034; full results of the LMM can be seen in File S1 ), indicating that the effect of sequence on the time spent close to the experimenter varied according to the test condition. We therefore split the data into the three conditions. We found that in the condition “Absent owner” the dogs spent significantly more time close to the experimenter when they received this condition in their second trial than when receiving it in either their first or their third trial (pairwise comparisons, Mann-Whitney U test, 1 st –2 nd : Z = 2.429, P = 0.015; 2 nd –3 rd : Z = −1.981, P = 0.048; 1 st –3 rd : Z = 1.000, P = 0.317). In the other two conditions there was no effect of sequence ( Figure 4 ).

Experiment 1 . Mean duration spent in the proximity of the experimenter in the conditions “Absent owner” (cAO), “Silent owner” (cSO), and “Encouraging owner” (cEO). White bars represent the 1 st trials, striped bars represent the 2 nd trials, grey bars represent the 3 rd trials. Shown are mean ± s.e.m. * represents P≤0.05.

https://doi.org/10.1371/journal.pone.0065296.g004

In Experiment 1 we found that in a problem-solving task the dogs' duration of manipulating an apparatus for retrieving food was shorter when their owner was absent than when the owner was present, irrespective of the owner's behavior when present in the room. This significant decrease in manipulation in the absence of the owner was not only evident in dogs with strong separation distress, since the duration of manipulation was not negatively correlated with the dogs' degree of separation-related behavior scored in the ASST. The dogs' proximity to their owner during the experiment did not depend on the owner's behavior, whereas the dogs spent most time in the proximity of the experimenter when the owner was absent – especially when they received this condition as their second trial.

The effect of reduced manipulation in the absence of the owner found in this experiment cannot be attributed to a lack of food motivation in the absence of the owner or to the surprise of not finding the owner in the experimental room because with the pre-test we made sure that all dogs were familiarized with the owner's potential absence from the room and that they were ready to consume food also when separated from the owner. Additionally, the decrease of manipulation cannot be attributed to a lack of behavioral cueing or encouragement from the absent owner. In the condition “Encouraging owner” the owner was constantly encouraging the dog to manipulate the apparatus, while in the condition “Silent owner” the owner could not see the actions of the dog due to being blindfolded and did not interact with the dog at all. Despite these substantial differences in the owner's cueing and encouragement, there was no significant effect on how long dogs persisted in manipulating the apparatus in these two conditions. Therefore, the only factor influencing the dogs' duration of manipulation was the presence of the owner.

In our study we also analyzed whether the dog's decreased duration of manipulation in the owner's absence was due to separation distress rather than the lack of security gained by the presence of the owner. In this case, dogs that experienced strong separation distress would have been expected to manipulate shorter than dogs that were not distressed by the owner's absence. However, since the dogs' duration of manipulation was not negatively correlated with their individual SRB score, we showed that the owner's absence did not affect the dogs differently. These results point to the owner's function as a secure base for the dogs, influencing their persistency to manipulate the apparatus in this cognitive task.

However, we also found that the dogs interacted with the experimenter and spent more time in her proximity in the absence of the owner, possibly indicating that the experimenter had the potential to provide social support to the dogs in this stressful situation. It is interesting that the effect of searching the experimenter's proximity was strongest when the dogs received the condition “Absent owner” as their second trial. This cannot be attributed to the stronger insecurity about the absence of the owner after a trial with the owner present in the room because in that case the same effect should have also been found in the third trial. Additionally, in the pre-test all dogs were familiarized with the potential absence of the owner and therefore it should not have been an unexpected event for the dogs. Although the observed effect was very robust with an equal variation across the dogs, our experiment does not allow us to draw final conclusions and therefore further research will be needed to explain this effect.

The fact that the experimenter also seemed able to provide some social support for the dogs raises the question whether the effect of increased manipulation in the presence of the owner was due to a secure base effect or whether any human can provide security to a dog. In the latter case, the difference we observed between the two conditions “Owner present” and “Owner absent” might be due to a greater security provided by two humans (i.e., owner and experimenter) than by one person (i.e., only experimenter). Therefore, in Experiment 2 we added a new condition, in which an unfamiliar human replaced the owner (Condition “Replaced owner”).

Experiment 2

Prior to the experiment, the owners were informed about the details of the procedures of this experiment and gave their written consent to participate with their dogs. The study was discussed and approved by the institutional ethics committee (Ethik- und Tierschutzkommission) of the Veterinary University Vienna in accordance with GSP guidelines and national legislation (Approval date: 21.08.2012). Since the owners were only required to interact with their dogs in their usual manner during the experiments and their behavior was not coded or analyzed, no approval for human experimentation was obtained. The participants in the video clips that are part of the supporting information have given written informed consent, as outlined in the PLOS consent form, to publication of themselves.

Thirty dogs were recruited from a database of volunteer participants of the Clever Dog Lab Society in Vienna, Austria, using the same criteria as in Experiment 1 . Three dogs had to be excluded because they failed the pre-test, which was the same as in Experiment 1 . For one dog the experiment had to be aborted after the first trial due to considerably high stress levels when interacting with the apparatuses. Therefore, 26 dogs (11M/15F; mean age ± SD = 3.6±2.44 years) completed the experiment. The sample consisted of 17 purebred dogs from three different FCI breed groups (Sheepdogs: N = 10; Toy dogs: N = 5; Primitive types: N = 2) and 9 mixed-breed dogs.

Dogs first had to pass the same pre-test as in Experiment 1 . In the test phase, we used the three test conditions from Experiment 1 and added one new condition, in which we replaced the owner with an unfamiliar human to control for the specificity of the secure base effect:

• Condition “Replaced owner” (cRO) : An unfamiliar human of the same gender as the owner is present in the experimental room during the trial and remains silent.

We used a within-subject design so that each dog received one trial in each of the four test conditions. The sequence of the conditions was counterbalanced across dogs.

The same apparatuses as in Experiment 1 were used in this experiment. The apparatuses were unknown to all of the dogs and the sequence of apparatuses and the condition in which they were used were counterbalanced across dogs.

The experiments were carried out between August 2012 and March 2013 in a quiet experimental room (3 m×6 m) at the Clever Dog Lab. The set-up was the same as in Experiment 1 (see Figure 2 ).

Pre-test phase – The procedure of the pre-test trials was the same as in Experiment 1 . The only difference was that in trials 1 and 2 both the owner and the unfamiliar human were present in the room, whereas in trials 3 and 4 both of them left the experimental room and remained in the adjacent room during the trial.

Test phase – The three test conditions “Absent owner” (cAO), “Silent owner” (cSO), and “Encouraging owner” (cEO) were the same as in Experiment 1 , with the only difference that also the unfamiliar human entered the experimental room together with the owner through door 1 and then left through door 2 and remained silent in the adjacent room during the trial. In the new test condition “Replaced owner” (cRO) the owner and the unfamiliar human entered the experimental room together. The owner then left the room through door 2 and remained silent in the adjacent room. The unfamiliar human sat down on the chair, put on opaque sunglasses and remained silent and passive throughout the trial (see Movie S4 ). After the trial ended and the experimenter had left the room together with the dog, also the owner and the unfamiliar human left the room through door 1. The sequence of conditions was counterbalanced across dogs.

All experimental sessions were videotaped, coded with Solomon Coder beta (©2006–2011 András Péter), and analyzed with SPSS Statistics 17.0.0 (©2008 SPSS Inc.).

The same three behaviors as in Experiment 1 were coded continuously (i.e., “Manipulating”, “Staying close to the owner”, “Staying close to the experimenter”). A second coder blind to the aim of the experiment and to the experimental conditions coded 20% of the videos of the test trials and Cronbach's alpha was greater than α = 0.9 for all behavioral variables. Where possible, the variables were lg10-transformed in order to apply parametric statistics.

A linear mixed model (LMM) with main effects and two-way interactions was used to investigate the effect of the factors “sequence of conditions” (1 st , 2 nd , 3 rd , 4 th ), “type of apparatus” (aDP, aSB, aSC, aRS), and “condition” (cAO, cRO, cSO, cEO) on the variable “duration of manipulation”. Additionally we calculated a separate LMM (main effects, two-way interactions) where we investigated the effects of the same factors as above on the variable “duration spent in the proximity of the experimenter”. Analyses of the residuals confirmed normal distribution for all variables in the LMMs. Post-hoc comparisons of estimated marginal means (EM means) were carried out with LSD confidence interval adjustment.

We compared the duration spent close to the silent owner (cSO) to the duration spent close to the unfamiliar human (cRO) with a Wilcoxon signed-rank test.

As in Experiment 1 , the condition in which the dogs were tested had a significant effect on how long the dogs manipulated the apparatus (LMM, N = 26, F 3,67  = 7.700, P≤0.001, Figure 5 ). The dogs manipulated the apparatus shorter when they were alone with the experimenter than in the two conditions when the owner was present (i.e., conditions “Silent owner” and “Encouraging owner”; EM means, cAO-cSO: P = 0.002, cAO-cEO: P≤0.001). The dogs also manipulated the apparatus shorter when the owner was replaced with an unfamiliar human than when the owner was present and encouraging the dog (EM means, cRO-cEO: P = 0.007). There was only a non-significant trend that dogs manipulated less when the owner was replaced than when the owner was present and silent (EM means, cRO-cSO: P = 0.088). There was no significant difference between the two conditions when the owner was present (EM means, cRO-cEO: P = 0.294) and the two conditions when the owner was absent (EM means, cRO-cEO: P = 0.137).

Experiment 2 . Mean duration of manipulating the apparatus in the conditions “Absent owner” (cAO), “Replaced owner” (cRO), “Silent owner” (cSO), and “Encouraging owner” (cEO). Shown are mean ± s.e.m. *** represents P≤0.001, ** represents P≤0.01.

https://doi.org/10.1371/journal.pone.0065296.g005

In this experiment dogs also differentiated between the apparatuses (LMM, N = 26, F 3,67  = 5.509, P = 0.002). They manipulated the Hunter© Rolling Snack longer than all the other apparatuses (EM means, aRS-aDP: P≤0.001, aRS-aSC: P = 0.013, aRS-aSB: P = 0.004). Neither the sequence nor any of the interactions had an effect on how long the dogs continued to manipulate the apparatus (full results of the LMM can be seen in File S1 ). The dogs spent significantly more time close to the owner in the condition “Silent owner” than to the equally silent unfamiliar human in the condition “Replaced owner” (Wilcoxon signed-rank test, N = 26, W = 73.00, P = 0.009; Figure 6 ).

Experiment 2 . Graph depicts box plots of duration spent in the proximity of the unfamiliar human (cRO) and of the owner (cSO), respectively. For each box plot, median values are indicated by the line within the box. The box represents 50% of the values (25th and 75th percentiles), with the upper bar representing the 90th percentile and the lower bar representing the 10th percentile. Circles indicate outliers. ** represents P≤0.01.

https://doi.org/10.1371/journal.pone.0065296.g006

The dogs did not spend the same amount of time close to the experimenter in the different conditions (LMM, N = 26, F 3,67  = 8.257, P≤0.001). As in Experiment 1 , the dogs spent more time close to the experimenter when they were alone with her than in both conditions when the owner was present (EM means, cAO-cSO: P = 0.009, cAO-cEO: P≤0.001) and there was a non-significant trend that they spent more time close to her when the owner was replaced (EM means, cAO-cRO: P = 0.089). Further, the dogs also spent more time close to the experimenter when the owner was replaced than when the owner was present and encouraging (EM means, cRO-cEO: P = 0.016). There were no significant differences between the other conditions (EM means, cRO-cSO: P = 0.369, cSO-cEO: P = 0.110). Neither the sequence nor the type of apparatus, nor any of the interactions had an effect on how long the dogs continued to manipulate the apparatus (full results of the LMM can be seen in File S1 )

In Experiment 2 we replicated our findings from Experiment 1 and showed that the dogs manipulated the apparatuses for shorter durations when they were alone with the experimenter than when the owner was present. When the owner was replaced with an unfamiliar human then dogs manipulated for shorter periods than when the owner was encouraging them. The difference in the duration of manipulation when the owner was present and silent and when the unfamiliar human was present was not significant. Importantly, the presence of the unfamiliar human did not significantly increase the dogs' manipulation compared when they were alone with the experimenter. Furthermore, the dogs spent significantly more time in the proximity of the owner than of the unfamiliar human. As in Experiment 1 , the dogs spent most time in the proximity of the experimenter when they were alone with her and they spent more time close to her when the owner was replaced by an unfamiliar human than when the owner was present and encouraging.

By including the control condition “Replaced owner” in this experiment we could show that the owner had a specific effect on the dog's behavior in this manipulative problem-solving task. The fact that the presence of an unfamiliar human did not significantly increase the duration of manipulation in the dogs compared to when they were alone with the experimenter provides evidence for a secure base effect in dogs that is specific for the owner and therefore comparable to the one found in infant-caregiver relationships [4] . While the dogs also manipulated the apparatus significantly shorter when the unfamiliar human was present than when their owner was encouraging them, the difference in the duration of manipulation in the presence of the silent owner and the unfamiliar human was not significant. The smaller difference between those two conditions can be attributed to the fact that the dogs manipulated the apparatuses for somewhat – albeit not significantly – shorter durations when the owner was silent than when they were encouraging (see Figure 5 ). In our experiment we asked the owners to wear opaque sunglasses and not to react in any way to their dogs in the condition “Silent owner”. It is possible that the dogs perceived this behavior as unnatural. In human children a comparable situation (i.e., the still-face paradigm: after an initial social interaction, the adult suddenly becomes unresponsive) evokes negative affect, confusion and attempts to re-establish reciprocity with the interaction partner [14] , [15] . In line with that, the dogs spent a considerable amount of time in the proximity of the owner in this situation, often in close physical contact. To avoid this effect it might have been better to allow the owners to interact naturally with their dogs while instructing them not to encourage the dogs. Interestingly, the dogs spent significantly more time in the proximity of their owners when they were silent compared to the equally silent unfamiliar human. This indicates that the lack of interaction on the part of the unfamiliar human did not evoke the same negative affect and attempts to establish proximity in the dogs. As a follow-up to this study, it might also be interesting to investigate how the interaction with and encouragement coming from the unfamiliar human would influence the dogs' motivation to manipulate.

In this experiment the dogs also differentiated between the apparatuses and there was a preference to manipulate one of them longer than the others, which was not the case for the dogs tested in Experiment 1 . Before testing the dogs in both experiments the owners confirmed that the toys used as apparatuses were unknown to their dogs. However, more owners from the second sample of dogs, which was tested in Austria, reported that they occasionally provided their dogs with the type of toy, which had to be manipulated to receive a food reward. Therefore, it is possible that the varying manipulation durations in Experiment 2 were due to the resemblance of this specific toy to toys that the dogs already knew. However, despite of these differences between the toys, the pattern of manipulation in the four conditions was still evident.

As in Experiment 1 we found that the dogs spent most time close to the experimenter when they were alone with her and some more time when the owner was replaced by an unfamiliar human. This again indicates that she acted as a source of social support for the dogs.

General Discussion

Our study provides an important piece of evidence for the similarity between the secure base effect found in dog-owner and infant-caregiver relationships. Further, our study is the first to show that the secure base effect in dogs extends from the ASST [12] to other areas of dogs' lives and that it can also manifest in cognitive testing. A comparable effect has been shown in human children when they were confronted with a problem-solving task: those children that were able to use their mother as a secure base were found to be more motivated and persistent in solving the task [13] . However, while the secure base effect is usually only evident in infanthood, where it balances the infants' exploration of the world with maintaining the crucial proximity to the caregiver [4] , dogs seem to be unique in having retained this behavior into adulthood. Dogs living in animal shelters have even been found to establish preferences for specific humans after short positive interactions in adulthood, which already strikingly resemble attachment bonds [16] .

Although the secure base effect we found in this study was specific for the owner, unfamiliar humans like the experimenter also seem to be able to provide some social support for the dogs. This is also suggested by the fact that the replacement of the owner with an unfamiliar human slightly – albeit not significantly – increased manipulation in the dogs. A similar effect has been observed in human children in the ASST when they seek social support from non-attachment figures with whom they had been familiarized prior to the test [17] . Although in adult dogs it has so far mainly been shown that owners are the ones who provide social support for their dogs [18] , [19] , in dog puppies social support can also be provided by an unfamiliar human [20] .

Finally, our results also have important implications for behavioral testing in dogs. Although in our task the dogs did not need to apply sophisticated problem-solving skills and we were interested in their general motivation to manipulate the toys and not in their success, it is likely that the presence or absence of the owner might also substantially influence dogs' motivation in other more complex test situations. The owner's absence in the generally unfamiliar experimental setting might cause a lack of security, which in turn could influence the outcome of the test.

Supporting Information

Supplementary material. Supplementary methods for Ainsworths's Strange Situation Test (ASST) carried out prior to Experiment 1 and full results of the Linear Mixed Models (LMM) calculated in Experiment 1 and 2 .

https://doi.org/10.1371/journal.pone.0065296.s001

Condition “Absent owner”. This video clip shows the beginning of a test trial in the condition “Absent owner” with the apparatus Hunter© Rolling Snack. Neither the owner nor the unfamiliar person is present. The experimenter releases the dog and then takes her position. The duration of the full trial is 5 minutes.

https://doi.org/10.1371/journal.pone.0065296.s002

Condition “Silent owner”. This video clip shows the beginning of a test trial in the condition “Silent owner” with the apparatus Hunter© Snack Bottle. The owner (sitting on the chair) is present, wears opaque sunglasses and remains silent. The experimenter releases the dog and then takes her position. The duration of the full trial is 5 minutes.

https://doi.org/10.1371/journal.pone.0065296.s003

Condition “Encouraging owner”. This video clip shows the beginning of a test trial in the condition “Encouraging owner” with the apparatus Nina Ottosson© Dog Pyramid. The owner (sitting on the chair) is present, wears normal sunglasses and is allowed to encourage the dog. The experimenter releases the dog and then takes her position. The duration of the full trial is 5 minutes.

https://doi.org/10.1371/journal.pone.0065296.s004

Condition “Replaced owner”. This video clip shows the beginning of a test trial in the condition “Replaced owner” with the apparatus Hunter© Snack Cactus. The unfamiliar person (sitting on the chair) is present, wears opaque sunglasses and remains silent. The experimenter releases the dog and then takes her position. The duration of the full trial is 5 minutes.

https://doi.org/10.1371/journal.pone.0065296.s005

Acknowledgments

We are grateful to Á. Miklósi for the generous opportunity to use the volunteer database for recruiting participants and to conduct Experiment 1 of this study at the Department of Ethology at the Eötvös Loránd University. Additionally, we would like to thank the students of the Department of Ethology for their support in recruiting and communicating with the Hungarian participants. Further, we would like to thank N. Atrissi, C. Müller, E. Pitteri, and S. Schäfer for acting as unfamiliar humans in Experiment 2 and A. Racca for carrying out the double coding. Finally, we thank all the owners and dogs for volunteering to participate in our experiment.

Author Contributions

Conceived and designed the experiments: L. Horn FR. Performed the experiments: L. Horn. Analyzed the data: L. Horn. Wrote the paper: L. Horn L. Huber FR.

• View Article
• Google Scholar
• 2. Bowlby J (1969) Attachment and Loss: Vol. 1. Attachment. New York: Basic Books.
• 3. Cassidy J (1999) The nature of the child's ties. In: Cassidy J, Shaver PR, editors. Handbook of attachment: Theory, research and clinical applications. New York: Guilford Press. pp. 3–20.
• 8. Askew HR (1996) Treatments of behaviour problems in dog and cat. A guide for the small animal veterinarian. Oxford: Blackwell Science. 9 pp.
• 9. Ainsworth MDS, Wittig BA (1969) Attachment and exploratory behavior of one-year olds in a strange situation. In: Foss BM, editor. Determinants of infant behavior (Vol. 4). London: Methuen. pp. 111–136.

Human encouragement might influence how dogs solve problems

Human encouragement might influence how dogs solve problems, according to a new Oregon State University study.

The study, published in the journal Applied Animal Behaviour Science , sheds light on how people influence animal behavior, said study lead author Lauren Brubaker, a doctoral student in OSU's Human-Animal Interaction Lab.

Brubaker evaluated the behavior of search and rescue dogs and pet dogs when presented with the same problem-solving task. Both sets of dogs persisted at the task for about the same proportion of time, but the search and rescue dogs were more successful at solving the task when encouraged by their owners.

However, the search and rescue dogs didn't solve the task when they were alone. Further, pet dogs that solved the task with their owner present -- but not encouraging them -- also solved it when they were alone, Brubaker said.

"We thought that was unusual," Brubaker said. "Because search and rescue dogs are trained to work independently, we expected that they would out-perform pet dogs on this independent task and that wasn't the case. This suggests that the behavior of the owner, including their expectation of their dog and how they engage with their dog on a day-to-day basis, may influence the dog during a problem-solving task.

"This leads us to believe that communication between search and rescue dogs and their owner could be more effective than communication between pet dogs and their owners," she said.

In the study, the dogs were given a solvable task with a person present: open a puzzle box containing a sausage within two minutes. They compared a group of 28 search and rescue dogs and a group of 31 pet dogs.

Search and rescue dogs were used as a comparison to pet dogs because they are traditionally trained to work independently from their owner. The search and rescue dogs were provided by Mountain Wave Search and rescue in Portland, Douglas County Search and Rescue in Roseburg, and Benton County Search and Rescue in Corvallis.

Pet dogs were recruited at random from the community through online advertisement and by way of word of mouth. Data from pet dogs from a 2015 study conducted by Udell were also used in the analysis. The dogs in both groups were from a variety of breeds.

The dogs were given the puzzle box under two conditions: alone in the room, and with their owner in the room standing neutrally. During the neutral phase, owners were instructed to stand in the room with their arms by their side and to avoid communicating with the dog. In the encouragement condition, the owner was instructed to encourage the dog however they saw appropriate, typically by using verbal praise or gestures, but without touching the dog or the container and without making contact with the dog or the container.

Before each condition the owner was instructed to "bait" the container by picking the container up, placing the food inside the container while the dog watched, and showing it to the dog to allow the dog to see that the container had food in it. Then they placed it on the ground in a designated location. In the neutral-human condition, the owner took three steps back and stood neutrally for two minutes. During the alone condition the owner left the room after placing the object on the ground.

In the human-neutral condition, three of the pet dogs and two of search and rescue dogs solved the task. Two pet dogs solved the task in the alone condition. In the encouragement condition, nine of the search and rescue dogs solved the task, while only two pet dogs did.

"When the owner's social cues direct the dog towards the independent problem-solving task, then we see something interesting," said Monique Udell, an animal scientist who directs the Human-Animal Interaction Lab in the College of Agricultural Sciences. "While most dogs increase the amount of time they spend attending to the puzzle when encouraged, pet dogs often end up treating the puzzle like a toy. Instead of engaging in goal directed behavior, they act as if their owner was encouraging them to play."

Udell continued, "It's possible that when directed by their owners, search and rescue dogs instead see opening the box as their job. Their owners may be more effective at communicating about the task at hand. Or maybe there is something inherently different about dogs that are selected for search and rescue that makes them more apt to solve the problem. More research is needed to know for sure."

• Animal Learning and Intelligence
• Veterinary Medicine
• Behavioral Science
• Agriculture and Food
• Spiders and Ticks
• Hunting dog
• List of dog breeds
• Service dog
• Dog intelligence
• Dog skin disorders

Story Source:

Materials provided by Oregon State University . Note: Content may be edited for style and length.

Journal Reference :

• Lauren Brubaker, Monique A.R. Udell. The effects of past training, experience, and human behaviour on a dog’s persistence at an independent task . Applied Animal Behaviour Science , 2018; 204: 101 DOI: 10.1016/j.applanim.2018.04.003

Cite This Page :

Explore More

• Soft, Flexible 'Skeletons' for 'Muscular' Robots
• Toothed Whale Echolocation and Jaw Muscles
• Friendly Pat On the Back: Free Throws
• How the Moon Turned Itself Inside Out
• A Welcome Hug Is Good for Your Health
• Climate Change Threatens Antarctic Meteorites
• Precise Measurement of Our Expanding Universe
• Little Research On 'Polycrisis' Humanity Faces
• Prebiotic Molecular Kitchen
• A Neurodegenerative Disease Triggered by Virus

Trending Topics

Strange & offbeat.

The importance of the secure base effect for domestic dogs - evidence from a manipulative problem-solving task

Affiliation.

• 1 Department of Cognitive Biology, University of Vienna, Vienna, Austria.
• PMID: 23734243
• PMCID: PMC3667003
• DOI: 10.1371/journal.pone.0065296

Background: It has been suggested that dogs display a secure base effect similar to that found in human children (i.e., using the owner as a secure base for interacting with the environment). In children, this effect influences their daily lives and importantly also their performance in cognitive testing. Here, we investigate the importance of the secure base effect for dogs in a problem-solving task.

Methodology/principal findings: Using a manipulative task, we tested dogs in three conditions, in which we varied the owner's presence and behavior (Experiment 1: "Absent owner", "Silent owner", "Encouraging owner") and in one additional condition, in which the owner was replaced by an unfamiliar human (Experiment 2: "Replaced owner"). We found that the dogs' duration of manipulating the apparatus was longer when their owner was present than absent, irrespective of the owner's behavior. The presence of an unfamiliar human however did not increase their manipulation. Furthermore, the reduced manipulation during the absence of the owner was not correlated with the dog's degree of separation distress scored in a preceding attachment experiment.

Conclusions/significance: Our study is the first to provide evidence for an owner-specific secure base effect in dogs that extends from attachment tests to other areas of dogs' lives and also manifests itself in cognitive testing - thereby confirming the remarkable similarity between the secure base effect in dogs and in human children. These results also have important implications for behavioral testing in dogs, because the presence or absence of the owner during a test situation might substantially influence dogs' motivation and therefore the outcome of the test.

Publication types

• Research Support, Non-U.S. Gov't
• Dogs / physiology*
• Dogs / psychology
• Human-Animal Bond*
• Mother-Child Relations
• Motor Activity / physiology
• Pets / physiology
• Pets / psychology
• Problem Solving / physiology*

Grants and funding

• P 21418/FWF_/Austrian Science Fund FWF/Austria

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

• Publications
• Account settings

Preview improvements coming to the PMC website in October 2024. Learn More or Try it out now .

• Advanced Search
• Journal List

Breed group differences in the unsolvable problem task: herding dogs prefer their owner, while solitary hunting dogs seek stranger proximity

Enya van poucke.

Linköping University, IFM Biology, 581 83 Linköping, Sweden

Amanda Höglin

Lina s. v. roth, associated data.

All data generated or analysed during this study are included in this published article, and its supplementary information files.

The communicating skills of dogs are well documented and especially their contact-seeking behaviours towards humans. The aim of this study was to use the unsolvable problem paradigm to investigate differences between breed groups in their contact-seeking behaviours towards their owner and a stranger. Twenty-four dogs of ancient breeds, 58 herding dogs, and 17 solitary hunting dogs were included in the study, and their behaviour when presented with an unsolvable problem task (UPT) was recorded for 3 min. All breed groups interacted with the test apparatus the same amount of time, but the herding dogs showed a longer gaze duration towards their owner compared to the other groups and they also preferred to interact with their owner instead of a stranger. Interestingly, the solitary hunting dogs were more in stranger proximity than the other groups, and they also showed a preference to make contact with a stranger instead of their owner. Hence, we found differences in contact-seeking behaviours, reflecting the dog–human relationship, between breed groups that might not only be related to their genetic similarity to wolves, but also due to the specific breeding history of the dogs.

Supplementary Information

The online version contains supplementary material available at 10.1007/s10071-021-01582-5.

Introduction

The domestic dog is well known for its abilities to communicate with humans (Miklósi et al. 2000 ) and this ability is already present at an early age (Passalacqua et al. 2011 ). One commonly used method to trigger dogs’ communication with humans is to present the dog with an unsolvable problem, and in a pioneering study Miklósi et al. ( 2003 ) compared the contact-seeking behaviours of dogs to those of similarly socialised wolves. They found that dogs made quicker and longer eye contact than their ancestors, the wolves, and since then, various unsolvable problem tasks (UPT) have been used to study the effects of breed, age, sex, etc., on the contact-seeking behaviours of dogs (see reviews by Cavalli et al. 2018 ; Mendes et al. 2021 ).

Today, there are over 400 dog breeds officially recognised by the Fédération Cynologique Internationale, and these are further divided into breed groups depending on their phenotypical traits, such as behavioural skills, appearance, but also according to their genetic origin. Genetically closest to the ancestral wolf are the ancient breeds (Parker et al. 2017 ), such as Shiba Inu and basenji, which are not specifically bred for human cooperation. On the other hand, we have the herding dogs, such as border collies, that are heavily selected for human cooperation. There are also breeds that are primarily selected for their sensory skills and natural behaviour, such as hunting dogs. In Scandinavia there is even a special type of hunting dog that is released into the forest during the hunting season to work on its own, separated from the owner. These solitary hunting dogs are not specifically selected for human cooperation and contact-seeking behaviour, but instead for their skills of searching the terrain and tracking down prey animals using mainly their olfactory sense. These dog breeds can belong to different genetically divided breed groups, and examples of solitary hunting breeds are Swedish and Norwegian elkhounds, dachshunds, and hunting terriers.

Hence, hunting breeds can be diverse and include both solitary hunting breeds and breeds that work in close proximity to humans, such as the retrievers. Passalacqua et al. ( 2011 ) found that their retrieving and herding breeds gazed for a longer duration compared to ancient dog breeds. In addition, Maglieri et al. ( 2019 ) found that the genetic closeness to wolves reduced contact-seeking behaviours, and that their group of retrievers gazed the most towards humans. In this study, we investigated contact-seeking behaviour in actively used solitary hunting breeds, bred for hunting independently of humans, and compared them to herding dogs selected for human cooperation, and also to ancient breeds, genetically closer to the wolf.

Recently, we have shown that herding dogs synchronize with their owners in long-term stress levels (Sundman et al. 2019 ), while solitary hunting dogs and ancient dog breeds do not (Höglin et al. 2021 ). The aim of this paper was to investigate the behavioural differences between these three breed groups, using the same dogs, focusing on the contact-seeking behaviour towards both the owner and a stranger using an unsolvable problem task (UPT). We hypothesized that herding dogs would show more contact-seeking behaviours towards the owner compared to the other breed groups, but also that the preference for the owner could be less obvious in the breeds that are not specifically selected for human cooperation.

Subject information

Dogs and their owners were recruited through social media and personal contacts, and consisted of 24 dogs belonging to ancient dogs (15 females and 9 males) with a mean age of 4.83 years ± 0.60 SE, 17 solitary hunting dogs (14 females and 3 males) with a mean age of 5.06 years ± 0.84 SE, and 58 herding dogs (23 females and 35 males) with a mean age of 4.7 years ± 0.38 SE. The herding dogs could also be divided into 32 competing and 26 companion (non-competing) dogs, where competing dyads reported that they actively trained and competed in either agility, obedience, or both disciplines. All dogs in this study lived indoors as pet dogs, even though the solitary hunting dogs were also actively used for hunting purposes. For more information about the dogs and their specific breeds see Supplementary 1.

The unsolvable problem task

The behavioural experiment took place outdoors at Linköping University, southeast of Sweden, during September–October in 2018 and 2019. The UPT, consisting of both two solvable and one unsolvable task, has previously been described in detail by (Persson et al. 2015 ; Sundman et al. 2018 ). In short, the apparatus consists of three compartments, where the outer two lids are possible to slide to the side, making the treat accessible to the dog (Fig.  1 a). The lid covering the middle compartment is fastened and hence, unsolvable. Before being presented with the UPT the dog’s motivation was tested by the female test leader. This was done with a separate compartment without lid, wherefrom the dog was allowed to eat three treats. If the dog succeeded, the owner and an unfamiliar female experimenter walked into their positions within the marquee that was novel to the dog. The experimenter placed herself in the front left corner, and the owner and the dog in the front right corner (Fig.  1 b). Then the test leader positioned the UPT apparatus in the middle of the back side of the marquee, and tapped with her finger on the apparatus to obtain the dog’s attention. When the dog looked towards the UPT apparatus it was released and the test leader left the marquee and the surrounding test area. During the following 3 min the experimenter and the owner were motionless, ignored the dog, and faced the UPT apparatus. However, if the dog did not manage to open any of the solvable tasks within the first minute, both experimenter and owner would simultaneously walk up and each open one lid halfway, and then return to their initial positions. The dog’s behaviour during 3 min was video recorded (Canon Legria), and later continuously recorded using the software Observer XT (Noldus) with a predetermined ethogram (Table ​ (Table1). 1 ). Note that the hair samples used to assess long-term stress levels in Sundman et al. ( 2019 ) and Höglin et al. ( 2021 ) were obtained after the behavioural part of the study (and could, therefore, not have affected the behavioural test).

a Test apparatus consisted of two outer solvable compartments and one middle unsolvable compartment. The apparatus was placed b in the middle back of the test arena with the owner and experimenter standing in each front corner of the marquee

Ethogram used in the behaviour analysis of the problem-solving test,

adapted from Persson et al. ( 2015 )

Data analysis

The behavioural data were not normally distributed, and therefore, non-parametrical tests were used, using the software IBM SPSS Statistics (version 27). Inter-rater reliability between two observers was tested using Spearman’s correlation for the human-related behaviours in 10% of the dogs, which revealed high reliability ( rs  = 0.89, p  = 0.001).

When comparing breed groups independent Kruskal–Wallis tests were used and pairwise comparisons were adjusted by the Bonferroni correction for multiple tests. To compare the breed groups’ behaviour towards owner and experimenter Wilcoxon Signed-Rank tests were used.

The behavioural interaction with the test apparatus was correlated with eye-contact related behaviours (including latencies for eye contact) using Spearman’s Correlations. Ages of the dog were also correlated with behaviours using Spearman’s Correlations. Mean and SE is reported in the results.

Breed group differences

Since there were no significant differences in recorded behaviour between competing and non-competing herding dogs, these dogs were considered as one single group. In addition, there were no significant sex differences within breed groups, so females and males were pooled together (see Supplementary 2 for these non-significant results).

Eye contact-seeking behaviour towards the owner differed significantly between the three breed groups (χ 2  = 24.80, p  < 0.001; Fig.  2 a), where herding dogs showed longer duration of eye contact than both solitary hunting breeds ( p  < 0.001) and ancient dog breeds ( p  < 0.001). There was also a tendency for eye contact-seeking behaviour towards the unfamiliar experimenter to differ between breed groups (χ 2  = 5.32, p  = 0.070; Fig.  2 a).

Mean duration (s ± 1SE) of a eye contact-seeking behaviour, b proximity, c physical contact, towards the owner and the stranger, and d interaction with the test apparatus, for ancient breed group, solitary hunting dogs and herding dogs *** p  < 0.001, ** p  < 0.01, * p  < 0.05, (*) p  < 0.1

Proximity behaviour towards the owner did not differ between breed groups (χ 2  = 0.39, p  = 0.824; Fig.  2 b). However, proximity towards the experimenter differed significantly (χ 2  = 11.63, p  = 0.003), where the solitary hunting dog breeds revealed longer duration of experimenter proximity than both ancient ( p  = 0.034) and herding breeds ( p  = 0.002; Fig.  2 b).

Similarly, there was no difference between breed groups in duration of physical contact with the owner (χ 2  = 1.71, p  = 0.425; Fig.  2 c), while we found significant differences for experimenter physical contact (χ 2  = 9.55, p  = 0.008). Again, the solitary hunting breed group revealed longer duration of physical contact with the experimenter than herding dogs ( p  = 0.008), but there was no significant difference compared to ancient dogs ( p  = 0.42; Fig.  2 c).

Age did not correlate with any of the behaviours ( p  > 0.1; See supplementary 2).

Owner and stranger preferences

Comparing the behaviour towards owner and the unfamiliar experimenter revealed that herding dogs showed significantly more eye contact-seeking behaviour with the owner compared to the experimenter ( N  = 58, z  = −3.67, p  < 0.001; Fig.  2 a). In addition, herding dogs were also significantly more in owner proximity than experimenter proximity ( N  = 58, z  = −3.02, p  = 0.003; Fig.  2 b). On the contrary, solitary hunting dogs tended to show both more proximity ( N  = 17, z  = −1.71, p  = 0.088) and physical contact ( N  = 17, z  = −1.96, p  = 0.050) behaviours towards the experimenter compared to the owner.

Test apparatus-related behaviours

There was no difference between breed groups in test apparatus proximity (χ 2  = 1.17, p  = 0.56), and similarly, no difference for physical contact with the test apparatus (χ 2  = 0.46, p  = 0.80; Fig.  2 d).

The dog’s proximity to the test apparatus correlated significantly with the latency to make eye contact with both the owner ( N  = 99, rs  = 0.26, p  = 0.009) and experimenter ( N  = 99, rs  = 0.45, p  < 0.001). Similarly physical contact with the test apparatus correlated significantly with latency to make eye contact with the owner ( N  = 99, rs  = 0.29, p  = 0.004) and experimenter ( N  = 99, rs  = 0.46, p  < 0.001).

Age did not correlate with any of the behaviours ( p  > 0.1; See Supplementary 2).

The aim of this study was to investigate three different breed groups with regard to their human contact-seeking behaviour. Using the unsolvable problem task, we found that herding dogs gazed longer towards the owner compared to both solitary hunting breeds and ancient breeds. In addition, while herding dogs preferred their owner, the solitary hunting dogs showed most contact-seeking behaviours towards the unfamiliar experimenter.

In our study, herding dogs showed a much longer eye contact duration during UPT compared to the other breed groups, which is in line with Passalacqua et al. ( 2011 ). However, we found no difference between the ancient breed group and the solitary hunting breeds, indicating that not only relatedness to the wolf affects the eye contact-seeking behaviour as suggested by Konno et al. ( 2016 ), but also selection and function of the dog breed as suggested by Passalacqua et al. ( 2011 ). In Passalacqua et al. ( 2011 ) both herding dogs and retrieving breeds are grouped together as a hunting/herding group, which might give the impression that all hunting breeds are selected for human cooperation and show much human contact. Our solitary hunting dogs gazed only short durations towards humans during the UPT, which is also different from the hounds in Konno et al. ( 2016 ) that found similar gazing durations for all breed groups except for the ancient breeds that gazed the least. In both studies the experimenter was training the dogs in solvable tasks before the task was made unsolvable. This might increase some dogs’ contact-seeking behaviour due to the association between treats and the experimenter. In our UPT, the experimenter was not associated with either treats, the motivation test plate, or the actual test apparatus and was, therefore, a neutral stranger to the dog.

In this study, both the owner and an unfamiliar experimenter were present during the UPT, allowing the dog to choose whom to seek contact with. Since we analysed the contact-seeking behaviour towards the humans separately it was possible to study whether the dog revealed a preference. As hypothesized, we found differences in human preference between the breed groups. While the herding dogs gazed longer towards the owner and were more in proximity to the owner compared to the unfamiliar experimenter, the solitary hunting dogs preferred to be in experimenter proximity. In previous UPT studies, there is usually a preference for the owner or no obvious preference (see review Cavalli et al 2018 , but note Maglieri et al. 2019 ). Note, however, that the preference in the solitary hunting dogs in this study refers to the proximity and physical contact to the experimenter, and not eye contact-seeking behaviour. Still, the results might indicate that solitary hunting dogs are curious towards strangers which could be a result from experience, hunting in large teams, or it could be related to personality or breed traits. Since this study is part of a larger study, the personality of these dogs has been investigated previously (Höglin et al. 2021 ). However, the only personality trait, where the solitary hunting dogs differed significantly from the other groups was Activity/Excitability , where herding dogs revealed the highest scores. Another possible reason for the experimenter preference in the solitary hunting dogs could be differences in the human–animal relationship as suggested by Cavalli et al. ( 2018 ) and Mendes et al. ( 2021 ). Indeed, as reported earlier in Höglin et al. ( 2021 ), the owner-reported relationship scores (assessed by MDORS) for both the subscale Dog – Owner Interaction and Perceived Emotional Closeness were lower for the solitary hunting dogs compared to both ancient dog breeds and herding dogs. In addition, the score for the subscale Perceived Cost was high for solitary hunting dogs compared to the other breed groups. Hence, in line with Topál et al. ( 1997 ), where the type of relationship was linked to the dog’s behaviour, this weaker relationship might be related to why the solitary hunting dogs seek more contact with an unfamiliar experimenter instead of their owner. However, note that ancient dog breeds and herding dogs were similar in their relationship scores but still differed in their contact-seeking behaviour in the UPT in this study. Hence, future studies should investigate this human preference further to disentangle the effect of breed group and the human–dog relationship.

In addition to breed group, the dog’s training experiences is suggested to influence the gazing behaviour, and Marshall-Pescini et al. ( 2016 ) found that dogs that are more trained gaze less towards humans during a problem-solving task compared to non-trained dogs. In our study, we did not find any differences between the two lifestyles within the herding dogs, i.e., actively competing dogs (in agility or obedience) and dogs kept as pet dogs. This could suggest limited effect of training in the herding group, but the dog’s training experience could still be important to consider when investigating contact-seeking behaviour in other breeds. Topál et al. ( 1997 ) found untrained dogs to play more with strangers, which might add to the explanation for the solitary hunting dogs’ behaviour towards the unfamiliar experimenter in our study. However, since training activities were not assessed for the ancient and solitary hunting breed group we will not speculate further on this point.

Also, one limitation of this study is that there were relatively more males in the herding group compared to the other breed groups. Even though we did not find any sex differences and, therefore, pooled the data, there might be a skewness that affects the results. However, in studies testing for possible sex differences in UPT, sex has been suggested to have little effect on the gazing behaviour (Konno et al. 2016 ; Passalacqua et al. 2011 ; Persson et al. 2015 ; Sommese et al. 2019 ; Topál et al. 1997 ) but note that female laboratory beagles show higher proximity to humans than male beagles (Persson et al. 2015 ).

Genetics and relatedness to the wolf is, as earlier mentioned, suggested to play a key role in eye contact-seeking behaviour of dogs (Konno et al. 2016 ; Maglieri et al. 2019 ; Sommese et al. 2019 ), where breeds more closely related to wolves gaze the least towards humans during an UPT. In our study, both ancient dog breeds and solitary hunting breeds showed little gazing behaviour towards humans. However, since our solitary hunting dogs belonged to different breed groups it is difficult to fully untangle selective breeding for solitary hunting behaviour and relatedness to the wolf in this study.

The persistence in the UPT has been associated to the latency to seek eye contact with humans and has, therefore, been raised as an issue when comparing animals in their contact-seeking behaviour (Mendes et al. 2021 ). Indeed, we did find correlations between persistence and latency in seeking eye contact, similar to Marshall-Pescini et al. ( 2017 ). However, we found no difference in persistence in the UPT between groups, since all breed groups interacted with the test apparatus for a similar amount of time. Therefore, the differences we found in contact-seeking behaviour between groups in this study cannot be explained by the dogs’ persistence and motivation for the task.

In conclusion, all dogs showed similar interest in the UPT and while the herding dogs gazed longer at the owner, the solitary hunting dogs revealed a preference for the unfamiliar experimenter which might be linked to both breed selection and differences in the dog–human relationship.

Below is the link to the electronic supplementary material.

Acknowledgements

We are grateful to all the owners that voluntarily participated with their dogs in this study, and we also want to thank Ann-Sofie Sundman for additional help during behavioural experiments. The study was partly supported by Agria/Swedish kennel club (N2016-0020) and partly by Sveland’s Foundation in Sweden.

Author contributions

LSVR and PJ designed and organized the study, while EVP and AH carried out the behavioural experiments and analyses. LSVR and EVP performed the statistical analyses in discussion with PJ, LSVR wrote the paper and created Fig.  1 , and EVP created Fig.  2 . All authors critically revised the manuscript, gave final approval for publication, and agree to be held accountable for the work performed therein.

Open access funding provided by Linköping University. The study was partly supported by Agria/Swedish kennel club (N2016-0020) and partly by Svelands Stiftelse.

Availability of data and materials

Declarations.

The authors have no conflicts of interest to declare that are relevant to the content of this article.

The study was conducted in line with ethical regulations and approved by the Linköping regional ethical committee for animal experiments in Sweden (Permits: 51-13, 6065/2019). All dog owners were informed about the study and gave their written consent to participate in the study.

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

• Cavalli C, Carballo F, Bentosela M. Gazing behavior during problem solving tasks in domestic dogs. A critical review. Dog Behav. 2018; 4 (3):23–44. doi: 10.4454/db.v4i3.68. [ CrossRef ] [ Google Scholar ]
• Höglin A, Van Poucke E, Katajamaa R, Jensen P, Theodorsson E, Roth LSV. Long-term stress in dogs is related to the human–dog relationship and personality traits. Sci Rep. 2021; 11 :8612. doi: 10.1038/s41598-021-88201-y. [ PMC free article ] [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Konno A, Romero T, Inoue-Murayama M, Saito A, Hasegawa T. Dog breed differences in visual communication with humans. PLoS ONE. 2016; 11 (10):e0164760. doi: 10.1371/journal.pone.0164760. [ PMC free article ] [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Maglieri V, Prato-Previde E, Tommasi E, Palagi E. Wolf-like or dog-like? A comparison of gazing behaviour across three dog breeds tested in their familiar environments. R Soc Open Sci. 2019 doi: 10.1098/rsos.190946. [ PMC free article ] [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Marshall-Pescini S, Frazzi C, Valsecchi P. The effect of training and breed group on problem-solving behaviours in dogs. Anim Cogn. 2016; 19 :571–579. doi: 10.1007/s10071-016-0960-y. [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Marshall-Pescini S, Rao A, Virányi Z, Range F. The role of domestication and experience in 'looking back' towards humans in an unsolvable task. Sci Rep. 2017; 7 :46636. doi: 10.1038/srep46636. [ PMC free article ] [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Mendes JWW, Resende B, Savalli C. A review of the unsolvable task in dog communication and cognition: comparing different methodologies. Anim Cogn. 2021 doi: 10.1007/s10071-021-01501-8. [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Miklósi A, Polgárdi R, Topál J, Csányi V. Intentional behaviour in dog–human communication: an experimental analysis of “showing” behaviour in the dog. Anim Cogn. 2000; 3 :159–166. doi: 10.1007/s100710000072. [ CrossRef ] [ Google Scholar ]
• Miklósi Á, Kubinyi E, Topál J, Gácsi M, Virányi Z, Csányi V. A simple reason for a big difference: wolves do not look back at humans, but dogs do. Curr Biol. 2003; 13 :763–766. doi: 10.1016/S0960-9822(03)00263-X. [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Parker HG, Dreger DL, Rimbault M, Davis BW, Mullen AB, Carpintero-Ramirez OEA. Genomic analyses reveal the influence of geographic origin, migration, and hybridization on modern dog breed development. Cell Rep. 2017; 19 :697–708. doi: 10.1016/j.celrep.2017.03.079. [ PMC free article ] [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Passalacqua C, Marshall-Pescini S, Barnard S, Lakatos G, Valsecchi P, Previde EP. Human-directed gazing behaviour in puppies and adult dogs, Canis lupus familiaris. Anim Behav. 2011; 82 :1043–1050. doi: 10.1016/j.anbehav.2011.07.039. [ CrossRef ] [ Google Scholar ]
• Pérez Fraga P, Gerencsér L, Lovas M, Újváry D, Andics A. Who turns to the human? Companion pigs' and dogs' behaviour in the unsolvable task paradigm. Anim Cogn. 2021; 24 (1):33–40. doi: 10.1007/s10071-020-01410-2. [ PMC free article ] [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Persson ME, Roth LSV, Johnsson M, Wright D, Jensen P. Human-directed social behaviour in dogs shows significant heritability. Genes Brain Behav. 2015; 14 (4):337–344. doi: 10.1111/gbb.12194. [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Sommese A, Nováková K, Šebková NF, Bartoš L. A wolfdog point of view on the impossible task paradigm. Anim Cogn. 2019; 22 :1073–1083. doi: 10.1007/s10071-019-01298-7. [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Sundman A-S, Persson M, Grozelier A, Halldén L, Jensen P, Roth LSV. Understanding of human referential gestures is not correlated to human-directed social behaviour in Labrador retrievers and German shepherd dogs. Appl Anim Behav Sci. 2018; 201 :46–53. doi: 10.1016/j.applanim.2017.12.017. [ CrossRef ] [ Google Scholar ]
• Sundman A-S, Van Poucke E, Svensson Holm A-C, Olsen Faresjö Å, Theodorsson E, Jensen P, Roth LSV. Long-term stress levels are synchronized in dogs and their owners. Sci Rep. 2019; 9 :7391. doi: 10.1038/s41598-019-43851-x. [ PMC free article ] [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Topál J, Miklósi Á, Csányi V. Dog–human relationship affects problem solving behavior in the dog. Anthrozoös. 1997; 10 :214–224. doi: 10.2752/089279397787000987. [ CrossRef ] [ Google Scholar ]

share this!

September 16, 2015 report

Wolves found to be better at problem-solving task than domesticated dogs

by Bob Yirka , Phys.org

(Phys.org)—Monique Udell, a researcher with Oregon State University, has found via experimentation, that domestic dogs appear to have lost some of their problem solving abilities as a result of their long history with humans. In her paper published in the journal Biology Letters , she describes a study she carried out and offers some theories on why she believe domesticated dogs may have lost some of their natural skills.

Udell notes that dogs have long been known to work with people as they go about their lives, in contrast to animals in the wild—one such striking behavior is their tendency to look back at their human companion when faced with a perplexing situation—seemingly asking for help. To learn more about this behavior, Udell enlisted the assistance of ten dogs that live as pets (and their owners), ten that live in shelters, and ten wolves that have been raised by humans.

Each of the animals was presented with a tasty sausage, which they were allowed to sniff, but not eat. Instead, the sausage was placed inside of a plastic container with a snap-on lid connected to a short length of rope. To open the container , the animals needed to pull on the rope while holding down the container—a task Udell deemed relatively easy for animals as smart as dogs and wolves. Udell conducted the experiments in two ways, one where the animal was left alone with the container, the other where there was a human (their owners) standing close by.

Udell reports that none of the pet dogs was able to open the container and just one of the shelter dogs was able to do so, but eight of the ten wolves succeeded. The presence of a person nearby didn't help much, the same number of wolves succeeded and one pet did so. She notes that all of the dogs from both groups spent a lot more of their time looking at the person, than did the wolves. Next, Udell allowed a human to offer encouragement to the dogs—doing so increased the success rate of the shelter dogs, four of them opened the container, but still just one pet dog was able to do it.

The experiment is intriguing Udell notes, because all of the dogs and wolves were capable of opening the container, but only the wolves were truly motivated to do so, as demonstrated by a much higher level of persistence—the dogs on the other hand appeared much more ready to ask for help.

Abstract Domestic dogs have been recognized for their social sensitivity and aptitude in human-guided tasks. For example, prior studies have demonstrated that dogs look to humans when confronted with an unsolvable task; an action often interpreted as soliciting necessary help. Conversely, wolves persist on such tasks. While dogs' 'looking back' behaviour has been used as an example of socio-cognitive advancement, an alternative explanation is that pet dogs show less persistence on independent tasks more generally. In this study, pet dogs, shelter dogs and wolves were given up to three opportunities to open a solvable puzzle box: when subjects were with a neutral human caretaker, alone and when encouraged by the human. Wolves were more persistent and more successful on this task than dogs, with 80% average success rate for wolves versus a 5% average success rate for dogs in both the human-in and alone conditions. Dogs showed increased contact with the puzzle box during the encouragement condition, but only a moderate increase in problem-solving success. Social sensitivity appears to play an important role in pet and shelter dogs' willingness to engage in problem-solving behaviour, which could suggest generalized dependence on, or deference to, human action.

Journal information: Biology Letters

© 2015 Phys.org

Explore further

Feedback to editors

A microbial plastic factory for high-quality green plastic

2 hours ago

Can the bias in algorithms help us see our own?

5 hours ago

Humans have converted at least 250,000 acres of estuaries to cities and farms in last 35 years, study finds

Mysterious bones may have belonged to gigantic ichthyosaurs

Hurricane risk perception drops after storms hit, study shows

Peter Higgs, who proposed the existence of the 'God particle,' has died at 94

Scientists help link climate change to Madagascar's megadrought

7 hours ago

Heat from El Niño can warm oceans off West Antarctica—and melt floating ice shelves from below

9 hours ago

Peregrine falcons expose lasting harms of flame retardant use

The hidden role of the Milky Way in ancient Egyptian mythology

Relevant physicsforums posts, potentially fatal dog parasite found in the colorado river.

19 hours ago

What do large moles on the body indicate?

Mar 30, 2024

Avian flu - A new study led by a team from the University of Maryland

Mar 27, 2024

Are all biological catabolic reactions exergonic?

Mar 20, 2024

A First of Its Kind: A Calcium-based signal in the Human Brain

Mar 18, 2024

Biological culture and cultural biology

Mar 17, 2024

More from Biology and Medical

Related Stories

Canine telepathy study explores how dogs think and learn about human behavior.

Jun 9, 2011

Myth of tolerant dogs and aggressive wolves refuted

Apr 21, 2015

Teaching young wolves new tricks

Jan 31, 2014

Dog-human cooperation is based on social skills of wolves

Jan 20, 2015

Behavior of man's best friend shaped by breed and hunting instincts

Feb 4, 2014

Domestication of dogs may have elaborated on a pre-existing capacity of wolves to learn from humans

Dec 3, 2013

Recommended for you

Archaeological study suggests cultural diversity increases biodiversity of ecosystems

10 hours ago

How whales struggle to navigate in a sea of noise pollution

Impact of climate change on marine life shown to be much bigger than previously known

11 hours ago

Study finds schools of fish can make less noise than a solitary swimmer

12 hours ago

Mediterranean marine worm has developed eyes 'as big as millstones'

Apr 8, 2024

Toothed whale echolocation organs evolved from jaw muscles, new research suggests

Let us know if there is a problem with our content.

Use this form if you have come across a typo, inaccuracy or would like to send an edit request for the content on this page. For general inquiries, please use our contact form . For general feedback, use the public comments section below (please adhere to guidelines ).

Please select the most appropriate category to facilitate processing of your request

Thank you for taking time to provide your feedback to the editors.

Your feedback is important to us. However, we do not guarantee individual replies due to the high volume of messages.

E-mail the story

Your email address is used only to let the recipient know who sent the email. Neither your address nor the recipient's address will be used for any other purpose. The information you enter will appear in your e-mail message and is not retained by Phys.org in any form.

Newsletter sign up

Get weekly and/or daily updates delivered to your inbox. You can unsubscribe at any time and we'll never share your details to third parties.

More information Privacy policy

Donate and enjoy an ad-free experience

We keep our content available to everyone. Consider supporting Science X's mission by getting a premium account.

E-mail newsletter

Bridging the Gap between Theory and Practice: Solving Intractable Problems in a Multi-Agent Machine Learning World - Elmore Family School of Electrical and Computer Engineering - Purdue University

Bridging the Gap between Theory and Practice: Solving Intractable Problems in a Multi-Agent Machine Learning World

Emmanouil-Vasileios (Manolis) Vlatakis Gkaragkounis is currently a Foundations of Data Science Institute (FODSI) Postdoctoral Fellow at the Simons Institute for the Theory of Computing, UC Berkeley, mentored by Prof. Michael Jordan. He completed his Ph.D. in Computer Science at Columbia University, under Professors Mihalis Yannakakis and Rocco Servedio, and holds B.Sc. and M.Sc. degrees in Electrical and Computer Engineering. Manolis specializes in the theoretical aspects of Data Science, Machine Learning, and Game Theory. His expertise includes beyond worst-case analysis, optimization, and data-driven decision-making in complex environments. Applications of his work span multiple areas from privacy, neural networks, to economics and contract theory, statistical inference, and quantum machine learning. 

David Inouye, [email protected] 

2024-04-10 08:00:00 2024-04-10 17:00:00 America/Indiana/Indianapolis Bridging the Gap between Theory and Practice: Solving Intractable Problems in a Multi-Agent Machine Learning World Manolis Vlatakis UC Berkeley 1:30 pm