Domesticated dogs (Canis familiaris) display remarkable sociability toward both humans and conspecifics, yet it remains unclear how they prioritize distinct social and nonsocial motivations when these drives compete. Prior work has focused largely on the human–dog bond, but less is known about how social preferences compare across competing motivational contexts, an important feature of temperament. To address this gap, we developed a Canine Sociability and Social Preference Test, adapted from rodent paradigms and paired with automated tracking to yield reproducible readouts, with high spatial and temporal resolution. Adult pet dogs (n = 32) completed a sequence of two-choice trials designed to capture both Sociability (social vs. toy object) and Social Preference (human vs. dog, as well as comparisons with food and toys). In Sociability trials, dogs preferred both humans and conspecifics over a nonsocial object, confirming independent attraction to each. In Social Preference trials, dogs preferred humans over conspecifics, whereas food and social stimuli elicited comparable levels of preference at the group level, though marked with individual variation. The human-over-dog preference was robust and consistent across repeated trials. Tail wagging served as a complementary socio-emotional measure, with dogs wagging significantly more near humans than conspecifics. Together, these results establish a versatile new paradigm for assessing canine temperament. Findings highlight robust group-level preferences—dogs preferred humans over conspecifics, and both social partners over objects—while also revealing variability in how individual dogs responded to social versus nonsocial stimuli. Such individual differences in motivational priorities are a key feature of temperament and have important implications for welfare, enrichment, and optimizing human–dog relationships.
Research Article
Pet Dogs Prefer Humans to Other Dogs Under Competing Social Contexts
https://doi.org/10.21203/rs.3.rs-7991898/v1
This work is licensed under a CC BY 4.0 License
You are reading this latest preprint version
social behavior
human-dog interaction
attachment
social preference
dog
Domesticated dogs (Canis familiaris) form strong affiliative bonds not only with conspecifics but also with humans. Over the course of domestication, dogs have developed behavioral and neurobiological features that support interspecies attachment, drawing parallels with human infant–caregiver relationships. For example, mutual gaze between dogs and their owners can elicit the release of peripheral oxytocin in both species, a neuropeptide that governs mother-infant social bonding in humans (Nagasawa et al., 2015, but see also Kekecs et al., 2016). Evolutionary accounts, such as the self-domestication hypothesis (SDH), propose that selection against aggression and for heightened attraction to humans during domestication fostered the cooperative and highly social nature of modern dogs (Hare & Ferrans, 2021; Hare & Woods, 2013). A key prediction of this hypothesis is that domestication shaped canine temperament to favor increased attraction to humans, a behavioral trait expected to be expressed in dogs’ social preferences. Supporting this view, recent work has shown that, compared to human-hand-raised wolf puppies, dog puppies are five times more likely to approach a familiar human and thirty times more likely to approach an unfamiliar human (Salomons et al., 2021, see also Gasci et al, 2005). Furthermore, in problem-solving tasks, human-directed behaviors in dogs have been linked to structural variants in genes homologous to those implicated in Williams-Beuren Syndrome (WBS), a condition in humans characterized by hypersociability (vonHoldt et al., 2017). Together, these findings highlight the interplay of behavioral and genetic factors that shape dogs’ social capacities and their remarkable ability to coexist with humans.
Temperament provides a complementary framework for understanding human-directed behaviors in dogs that help form and maintain interspecies bonds. Temperament theories emphasize that individual variation in stable traits, such as sociability, arousal regulation, and approach tendencies, shape behavior across diverse contexts and within and across breeds (Jones & Gosling, 2005; Morrill et al., 2022). Sociability has been proposed as a core temperament dimension in dogs, defined as the willingness to engage in friendly interactions with both humans and conspecifics (Gosling & Jones, 2005; Morrill et al., 2022). In dogs, sociability has typically been assessed using problem-solving paradigms, such as gaze toward a human during an unsolvable task (Hare et al., 1998; Miklósi et al., 2003), rope-pulling tasks (Quervel-Chaumette et al., 2015), or owner-reported survey measures. Another way to measure sociability is to give choices with whom to interact. When given a choice between a dog, human or food, dog puppies tended to communicate with humans more often than age-matched, highly socialized wolf puppies (Gácsi et al., 2005). Subsequent research has extended these paradigms to adult dogs. Free-ranging dogs only develop a preference for taking food from a human if the human also provides them a social reward with petting (Bhattacharjee et al., 2017). Horn et al. (2013) found that dogs preferentially followed cues from caregivers who interacted more frequently with them, compared to equally familiar but less involved caregivers. Tan et al. (2018) found that adult pet dogs prefer information from their owner but did not prefer to approach their owner over a stranger or food in a variety of contexts. Together, these studies demonstrates that dogs’ preferences toward unfamiliar humans can shift dynamically with experience, and that social interaction itself can be a potent reward, particularly in ecologically relevant contexts.
Previous work indicates high sociability toward humans in adult pet dogs, but previous experiments have typically assessed human-directed behavior in isolation, without considering competing needs or drives, and/or under narrowly defined conditions. A critical test remains in characterizing dog sociability. It remains unclear whether adult domesticated dogs are more strongly attracted to humans than to conspecifics when both are presented in direct competition, or how these social stimuli compete with other salient motivators such as food. Addressing this gap requires an assay that evokes dog social preferences within a broader motivational context. To this end, we developed a novel paradigm that measures both human- and conspecific-directed sociability, while also assessing individual preferences across multiple categories of familiar, innately motivating stimuli—including humans, conspecifics, food, and objects—presented in direct competition.
Building on established paradigms in animal behavior research, we applied the standard three-chamber social preference test, widely used for measuring sociability and social novelty in rodents (Yang et al., 2011), and adapted it for use in dogs. In the original assay, rodents freely explore an arena containing a social conspecific in one chamber and a non-social object in the other, with sociability defined as a preference for the social stimulus. This paradigm has also been used to probe more nuanced preferences between competing social stimuli, such as a novel versus a familiar conspecific, in both typical and autism-linked rodent models (Yang et al., 2011). In this present study, we created an area in which dogs could freely explore between two stimulus locations positioned on opposite sides of a room (Fig. 1). In our Canine Social Preference Test, dogs were presented with combinations of familiar stimuli, including a human, dog, food, or a toy. Automated tracking was performed using Noldus EthoVision software, providing continuous, high-resolution measures of the time each dog spent in defined stimulus zones. We also incorporated tail wagging as a measure of social behavior. Tail wagging in dogs is associated with increased arousal and conveys social and emotional information (Leonetti et al., 2024; Quaranta et al., 2007). Wagging patterns have been linked to distinct social or affective states – recent work demonstrates that dogs wagged more frequently in the presence of owners and that spatiotemporal wag patterns shift with increased familiarity to a novel human (Leonetti et al., 2024; Ren et al., 2022). By incorporating comparisons across both social and non-social stimuli, as well as socioemotional tail wag measures, we create an approach to quantify individual variability, assess test-retest reliability, and measure the relative strength of a dog’s preferences for competing stimuli across the full duration of the testing session.
This study was conducted under Duke University IACUC protocol A122-21-06, and all procedures conducted adhered to the approved guidelines. IRB approval was not required, as the data collected involved only dogs was classified as "Non-Human Subject Research."
Subjects
Subjects (n = 32) consisted of all healthy, adult male and female dogs of various breeds, all over one year of age. Breed and size were not restricted to maximize generalizability across companion dogs. Dogs were recruited from the Raleigh-Durham area via flyers and social media advertisements. See Supplementary Table 1 for demographic information. All owners completed a health screening survey and a consent form prior to subjects’ study participation.
Procedure
Testing Arena Setup
All testing was conducted between February 2021 and April 2024 and took place in a quiet testing room with bright lighting at the Duke Canine Cognition Center, at Duke University, in Durham, NC. All testing took place during the daytime between the hours of 9 am and 4 pm. For all tests, dogs were placed in an 8’ x 13.5’ enclosed testing arena (Fig. 1) that contained two 2.75’x 2.75’ “choice boxes” (CBs) in each corner, separated from the main arena by an x-pen gate. In each trial, the subject (S) was allowed to freely explore both sides of the arena, with each containing a distinct stimulus (human, dog, toy, food). All stimuli were kept in a separate room from the subject prior to testing. Subjects were video recorded live, for later offline automated tracking as described below, to measure relative time spent interacting with each stimulus. Subjects were not food or water restricted for the purpose of testing, and water was available ad libitum within the chamber. Brown noise was on during all testing to reduce auditory distractions (i.e. stopwatches, ambient noise, etc.). All tests were filmed with a ceiling-mounted wide-angle camera (Basler aca2500-60uc recorded with Basler Pylon Viewer 6.1.15.7395 at 40 fps) that was positioned ~ 9.5 ft from the floor.
General Methods for Social Preference Assay
Prior to behavioral testing, all subjects were familiarized to the testing room for at least 10 minutes. In each trial type, a handler (H) opened or reached over the arena gate and placed the subject (S) on the ground at the midline of the arena. After an experimenter (E) indicated “Start”, H released the subject and closed the door to the testing room so that the subject was alone in the arena. S was allowed to freely explore the arena for 3 minutes while E watched the subject via a monitor. During the trial, S could approach either stimulus box and interact with the stimulus through the x-pen gate. After each trial, H removed the subject from the arena while the subsequent trial was being set up.
Testing was paused if a subject barked nonstop for 1–3 minutes or if they showed signs of extreme stress such as growling, hackles raised, or excessive panting. For each individual dog, all testing took place on the same day with all trial types in the same order (Table 1).
|
Trials/Stimuli |
Trial Length |
|---|---|
|
1) Baseline/Habituation |
3 min |
|
2a) Human vs. dog (1) |
3 min |
|
3) Dog vs. object |
3 min |
|
4) Human vs. object |
3 min |
|
5) Human vs. food |
3 min |
|
6) Dog vs. food |
3 min |
|
2b) Human vs. dog (Repeat) |
3 min |
|
Total |
~ 30 min |
Trial Types
In Habituation trials (Trial 1), no stimulus was present in either stimulus box (SB) so that the dog could acclimate to the arena.
In Human vs. Dog trials (Trials 2a and 2b), a familiar human stimulus (HS) sat on the floor cross-legged in one SB, and a familiar conspecific/dog stimulus (DS) in the other SB. HS was instructed to place both hands on the gate for S to sniff and that they should make eye contact with the subject during the task. HS was instructed not to make vocalizations or call the puppy over to them in any other way during the task.
In Dog vs. Object trials (Trial 3), an age-matched dog stimulus (DS) was placed in one SB, and an object stimulus (OS), a familiar brightly colored stuffed toy, was placed in the other SB for the duration of the trial.
In Human vs. Object trials (Trial 4), a familiar human was placed in one SB, and an object (a familiar brightly colored stuffed toy) was placed in the other SB for the duration of the trial.
In Dog vs. Food trials (Trial 5), a familiar human was placed in one SB, and an object (a familiar brightly colored stuffed toy) was placed in the other SB for the duration of the trial.
In Human vs. Food trials (Trial 6), a familiar human was placed in one SB, and a bowl of familiar food is placed in the other SB for the duration of the trial.
Design
The study was run as a battery of tests, in which subjects participated in all the trials in the same order and completed all tests on the same day. Each dog completed seven distinct trial types presented in a fixed order across all subjects. Trial types varied in structure (stimulus identity), but the sequence was held constant to allow for standardized comparisons. This approach was chosen intentionally to support our primary goal of examining individual differences in behavior across a consistent set of social contexts. By presenting trials in the same order, we could compare across individuals under identical testing conditions. Although this limits the ability to assess order effects, we prioritized interpretability of between-subject differences.
Scoring and Analysis
All analyses were conducted using an automated detection animal tracking software called Noldus, EthoVision (Version 15). For each trial, we defined a “zone” within the software, as a 2-foot radius surrounding the stimulus area in each chamber (e.g., conspecific, human, food bowl, or object) (Fig. 1; Supplementary Video 1). Each dog was tracked using standard automated detection methods, from the dogs’ centroid, across trials, and each trial was validated in its entirety for accuracy by a human observer. From the positional data, the amount of time spent within each zone was calculated in seconds. The amount of time spent in each zone was measured from the moment the subject’s body crossed into the 2-foot radius, until it exited the zone. This metric provided a reliable indicator of subject interaction with and proximity to the stimulus. For each trial type, the percentage of time spent in each zone was calculated for each subject as the proportion of time spent within the stimulus zone relative to the total trial duration.
To quantify each subject’s preference for a specific stimulus (e.g., conspecific vs. object), we calculated a Preference Index. This index was defined as the difference between the time spent interacting with the preferred stimulus (A) and the time spent interacting with the alternative stimulus (B), normalized by the total time spent within both stimulus zones, as follows:
Stimulus A refers to the primary or target stimulus of interest (e.g., conspecific), while Stimulus B refers to the alternative stimulus (e.g., object). ‘Time in Zone’ was defined as the cumulative time the subject’s centroid was located within 2 feet of the stimulus across the trial session. The preference index ranges from − 1 to 1, where a positive value indicates a greater preference for Stimulus A, and a negative value indicates a preference for Stimulus B. A value of 0 indicates no preference between the two stimuli.
Tail Wag
Some behaviors of interest, beyond zone time, could not be captured through positional tracking alone. We coded “tail wagging bouts,” as an additional measure of affiliative engagement. Specifically, tail wag was coded as the duration that each dog wagged its tail while located on the side of the room associated with either the familiar human or the familiar conspecific. Cumulative tail wagging duration (in seconds) was calculated from each annotated video recording across the human versus dog trial. A paired-samples t-test was conducted to compare the total time spent tail wagging on the human side versus the conspecific side.
Human vs. Dog Trials
To assess whether dogs exhibited a preference for a familiar human over a familiar conspecific, a paired t-test was conducted comparing the percentage of time spent in the zone associated with each stimulus. Results revealed that dogs significantly preferred familiar humans over familiar dogs (t(25) = 8.376, p < 0.0001; Fig. 2). The mean difference in time spent in the human zone versus the dog zone was − 39.63% (95% CI: -49.37% to -29.88%), indicating a robust preference for human interaction. The effect size, measured by partial eta squared (η² = 0.7373), indicates a strong and consistent preference across subjects. Furthermore, we leveraged automated tracking to bin the data into 1-minute increments to examine whether the human preference varied over time. We found no differences across time bins in the human zone, with dogs maintaining a sustained preference for the human across the entire trial duration (RM One-Way ANOVA: F(1.926,26) = 0.72, p = 0.49, Fig. S1a). In contrast, time in the dog zone significantly decreased over time (RM One-Way ANOVA: F(1.563,20.32) = 6.31, p = 0.013, Fig. S1a). Dunnett’s post hoc comparisons confirmed that time in the dog zone was significantly lower in the second minute (p = 0.0396) and third minute (p = 0.0239) compared with the first minute bin. In addition, comparisons were made between male and female dogs, with no significant sex differences detected in time spent in each zone across human versus dog trial (Welch’s t(20.89) = 0.6721, p = 0.5089; Fig. S2a.)
In addition to assessing time spent in each stimulus zone, we measured tail-wagging behavior to provide an additional index of social engagement. Dogs spent significantly more time tail wagging when on the side of the room with the familiar human compared to the familiar conspecific, (t(23) = 2.603, p = .0159, 95% CI [3.41, 29.85]; Fig. S5. Tail wag durations on the two sides were also positively correlated (r = .39, one-tailed, p = 0.0283), indicating some consistency in individual differences across contexts.
Figure 2. Human vs. Dog Trial
Diagram of experimental setup (a) and representative heat map (b) for a human vs. dog trial from an individual subject. (c) Engagement time (% of time in each zone) for adult subjects in human vs. food trials (% of time in each zone; mean + SEM; paired t-test, t(25) = 8.376, p < 0.0001, n = 26 pairs). (d) Comparison of Human vs. Dog (1) preference index (for human versus dog) (mean + SEM; paired t-test, (t(23) = 2.603, p = .0159, 95% CI [3.41, 29.85).****P < 0.0001; *P < 0.05.
Human vs. Food Trials
A paired t-test comparing the percentage of time spent in the human zone versus the food zone revealed no significant difference (t(26) = 0.3794, p = 0.7074; Fig. 3). The mean difference in time spent near the human versus the food was − 4.19% (95% CI: -26.89% to 18.51%), suggesting no clear preference for either stimulus. These findings indicate that dogs did not prioritize familiar humans or food in this comparison. Further, no significant sex differences in zone times were detected across these trials (Welch’s t(24.31) = 0.3905, p = 0.6996; Supplemental Fig. 2d).
Diagram illustrates experimental setup (a) and a representative heat map of zone time (b) for a human vs. food trial from an individual dog. c) Engagement time (% time in zone) for adult subjects in human vs. food trials (% time in zone; mean ≤ SEM; paired t-test, t(26) = 0.38, p = 707, n = 27 pairs). Mean difference: -4.19% (95% CI (-26.89, 18.51). Diagram of experimental setup (d) and a representative heat map (e) for a dog vs. food trial from an individual dog. f) Engagement time (% time in zone) for adult subjects in dog vs. food trials (% time in zone; mean + SEM; paired t-test, +(26) = 0.38, p = 701, n = 27 pairs). Mean difference: -2.729% (95% CI [-17.17, 11.721). ***P < 0.0001
Dog vs. Food Trials
A paired t-test comparing the percentage of time spent in the familiar dog zone versus the food zone revealed no significant difference (t(26) = 0.3883, p = 0.7009; Fig. 3). The mean difference in time spent near the dogs versus the food was − 2.729% (95% CI: -17.17% to 11.72%), suggesting that dogs do not exhibit a preference for familiar dogs or food in this comparison. Further, no significant sex differences were detected in zone time across these trials, Welch’s t(22.98) = 0.2800, p = 0.7819; Supplemental Fig. 2f.
Human vs. Object Trials
A paired t-test comparing the percentage of time spent in the familiar human zone versus the object zone revealed that dogs spent significantly more time near familiar humans than toys t(26) = 7.551, p < 0.0001; Fig. 4. The mean difference in time spent near humans versus objects was − 45.74% (95% CI: -58.19% to -33.29%), indicating a strong preference for human interaction over non-social stimuli. No significant sex differences were detected in zone time across these trials (Welch’s t(24.62) = 0.1348, p = 0.8939; Supplemental Fig. 2c).
Diagram of experimental setup (a) and representative heat map (b) from a human vs. object trial for an individual dog. (c) Engagement time (% time in each zone) for adult subjects in human vs. object trials (% time in zone; mean ÷ SEM; paired t-test, (26) = 7.551, p < 0.0001, n = 27 pairs). Mean difference: -45.74% (95% CI [-58.19,-33.291). ***P < 0.001. Diagram of experimental setup (d) and representative heat map (e) for dog a vs. object trial from an individual dog. F) Engagement time (% time in zone) for adult subjects in dog vs. object trials (% time in zone; mean ≤ SEM; paired t-test, +(25) = 3.68, p = .0011, n = 26 pairs). Mean difference: − 19.01% (95% CI [-0.174,0.197]). ****P < 0.0001.
Dog vs. Object Trials
A paired t-test was conducted to compare the percentage of time spent in the familiar dog zone versus the object zone. Results indicate that dogs spent significantly more in the familiar dog zone than the object zone (t(25) = 3.678, p = 0.0011; Fig. 4). The mean difference in time spent in the dog versus object zone is -19.01% (95% CI: -29.66% to -8.366%), demonstrating a preference for familiar dogs over non-social stimuli. No significant sex differences were detected in zone time across these trials (Welch’s t(18.32) = 0.5622, p = 0.5808; Supplemental Fig. 2e).
Using a novel Sociability and Social Preference paradigm, we found that dogs’ social motivation reflects both robust group-level preferences and individual variability, consistent with temperament frameworks. In the Sociability trials, adult pet dogs preferred both familiar humans and conspecifics over nonsocial objects, confirming independent attraction to each. In contrast, in the Social Preference trials, when humans and conspecifics were placed in direct competition, dogs consistently favored humans. This pattern suggests that while both social partners are salient in isolation, human presence carries greater relative social value when alternatives are available. Across conditions, dogs spent more time near social than nonsocial stimuli, highlighting social motivation as a primary driver of behavior. Notably, when humans and conspecifics were tested separately against an object, dogs engaged comparably with both, reinforcing their independent salience. By comparison, when either social stimulus was placed against a survival-related motivator, food, no group-level preference emerged, though individual variation was pronounced. Thus, only when humans and conspecifics were directly contrasted did a clear preference for humans emerge, underscoring the particular strength of human-directed motivation. Tail wagging, which increased significantly in proximity to humans, provided convergent behavioral evidence of positive socio-emotional arousal accompanying human preference. These findings demonstrate that dogs prioritize social partners, especially humans, while still exhibiting meaningful individual differences in motivational hierarchies.
Temperament
Our findings highlight the value of temperament frameworks for understanding canine social behavior. Temperament refers to stable individual traits, such as sociability and approach tendencies, that shape behavior across contexts. Traditional measures of canine sociability, such as attachment or problem-solving tasks, often capture behavior in isolated or constrained settings, and typically focus on a single type of social or nonsocial stimulus. Other approaches rely on owner-reported personality or temperament questionnaires (e.g., CBARQ; Hsu & Serpell, 2003); these survey-based tools have revealed stable, heritable trait structures that align with constructs such as sociability, boldness, and trainability (Jones & Gosling, 2005). In order to gain a richer understanding of dogs’ motivation, our new Canine Sociability and Social Preference Test enabled direct comparisons among familiar humans, conspecifics, and non-social stimuli under identical conditions, offering a richer assessment than single-stimulus tests. By incorporating both social and non-social comparisons, this approach also provided a motivational baseline against which to evaluate the relative strength of dogs’ social preferences, as well as individual variation. Furthermore, automated tracking provided precise, continuous measures of engagement, while supplementary coding of tail wagging captured socio-emotional behaviors not detectable from positional data alone. Dogs wagged their tails more near humans than conspecifics, suggesting potentially rewarding arousal associated with human proximity. This interpretation is supported by prior neuroimaging work showing stable neural responses to social versus food rewards (Cook et al., 2016), and strong caudate activation to familiar human scent (Dilks et al., 2015). Collectively, these testing features establish a reproducible assay of canine temperament that quantifies both group-level tendencies and individual motivational hierarchies. Such stable variation is a hallmark of temperament and underscores the importance of tools that can assess and accommodate these differences.
Notably, dogs preferred familiar humans even when humans remained passive (e.g., no speech or gestures), while conspecifics could actively vocalize or display affiliative behaviors (e.g., play bows, tail wagging), suggesting that human presence alone carries greater social value than a potential conspecific interaction. Unlike attachment or communicative tests, our paradigm required no stress, training, or explicit interaction, yet dogs still prioritized humans. These findings align with prior evidence that domesticated dogs are motivated to approach both familiar and unfamiliar humans in ways that wolves are not (Palmer & Custance, 2008; Salomons et al., 2021; Topál et al., 1998), and suggest that dogs’ sensitivity to human intentional communication may, in part, arise from an underlying attraction to humans that supersedes intraspecific motivation (Hare et al., 1998; Miklósi et al., 2003; Miklósi & Topál, 2013; Tan et al., 2018).
Individual variation
Although a strong overall human preference was observed, and group-level patterns favored social over non-social stimuli, meaningful individual variability was observed. In particular, the absence of significant group differences in social versus food comparisons (e.g., human vs. food, dog vs. food), produced individual variation that indicate the strength of social preferences are not uniform across dogs. Such variability likely reflects individual differences in motivational hierarchies shaped by context or prior experience. These findings are consistent with prior studies showing that dogs tend to prefer social over non-social rewards yet vary in the strength of these tendencies. For example, an awake fMRI study revealed stable individual differences in caudate activation, with some dogs exhibiting stronger neural responses to human praise than to food rewards (Cook et al., 2016). Future studies could manipulate contexts to disentangle within-subject flexibility from stable between-subject variation. Along these lines, since we could not control for hunger levels in the present study, we cannot rule out the contribution to meal history on the current variation. Regardless, the consistent presence of individual variation underscores the importance of tools that can capture diverse motivational drivers. Further, the observed individual differences in preference for food versus social stimuli, has implications for individualized training plans. Practically, tailoring enrichment strategies to individual dogs may enhance welfare—for instance, dogs with strong food motivation may benefit from training or activities centered on food rewards, whereas socially motivated dogs may thrive with increased opportunities for human or canine interaction.
Application of findings
Social preference, as a temperament trait, offers valuable insights into dogs’ individual social tendencies and can help guide better dog-owner matches. For example, dogs showing a stronger preference for conspecifics may thrive in multi-dog households or with regular dog-dog interactions, while dogs with greater food motivation may benefit from training and enrichment strategies that incorporate food-based reward. Aligning care and environment with individual motivational profiles also has clear implications for canine welfare. The robust preference for humans emphasizes the key role of human interaction in canine well-being. This is especially relevant in rescue or shelter settings, where consistent familiar human contact may be limited. Our findings highlight the welfare benefits of structured programs that increase positive human interaction, such as volunteer-led playgroups, foster care, or consistent caretaker assignments. Providing familiarity and predictability in human–dog interactions may help reduce stress and support healthy attachment processes in these environments.
Limitations & Future work
While the novel preference assay used in this study yielded quantitative and reproducible measurements, several limitations should be addressed in future research. Replication in larger and more diverse populations – spanning age, breed, and rearing history – will be critical for generalizability. Notably, for logistical safety, all dogs in our sample were raised in multi-dog households, avoiding potential aggression during conspecific trials. This design choice, however, limits conclusions about dogs from single-dog homes, where conspecific companionship may carry greater value, or about dogs with negative histories of dog–dog interactions, for whom the task could be less motivating or even aversive. Incorporating screening for aggression toward humans or conspecifics in future work could further refine individualized interpretations of preference measures.
Future work could extend this assay to compare preferences for familiar versus unfamiliar humans and conspecifics, offering insights into attachment and novelty-seeking processes. Linking behavioral task performance to validated, owner-reported survey constructs such as the CBARQ (Hsu & Serpell, 2003) could help clarify how temperament dimensions such as sociability and approach/avoidance map onto broader personality profiles. Such cross-method validation would strengthen interpretability and support the use of brief owner-report tools for large-scale screening of social motivation. Longitudinal studies are also needed to track how social preferences develop across the lifespan or shift with environmental change. Such approaches could assess the stability of social preferences over time, incorporate novelty conditions, and examine how social preference patterns relate to factors such as attachment security, temperament, or working potential across diverse canine working roles (e.g., emotional support, guide dogs, or search-and-rescue). Extending this paradigm to puppies, service or working dogs, free-ranging dogs, and dingoes or wolf populations would further clarify how developmental and ecological factors shape social motivation across canine domestication.
This study investigated the social preferences of a diverse sample of adult pet dogs using a novel two-choice preference assay, adapted from rodent models of competing social and non-social motivation (Rein et al., 2020; Yang et al., 2011). The paradigm combines automated tracking with supplementary behavioral coding to provide highly quantitative, precise, and reproducible measures of social behavior. Unlike prior studies that assess social preference under stress (e.g., separation in attachment tests) or in response to social cues (e.g., pointing tasks), our designs tested dogs’ preferences between familiar social partners in a calm, low-demand setting, thereby minimizing external influences.
Collectively, our findings show that adult companion dogs are spontaneously motivated to approach and maintain proximity to humans over other familiar stimuli when presented in competition, reflecting the deep affiliative bonds that characterize human–dog relationships and suggesting these bonds may be prioritized over intraspecies interactions. Importantly, when presented independently, we observe high sociability for either human or conspecific stimulus, over non-social objects but not survival needs such as food. By capturing both group-level tendencies and individual variability, this paradigm advances our understanding of canine social motivation and establishes a new canine temperament tool with broad applications for welfare, enrichment, and owner–dog compatibility. Further research exploring the interplay of social, environmental, and genetic factors will further illuminate the motivational forces that shape this unique interspecies bond and inform strategies to better meet the needs of individual dogs.
Acknowledgements
We would like to acknowledge the many research assistants and handlers who assisted with data collection for this project, especially Samantha Kefer, Jessica Shoemaker, Shelby Carter, Blake Roper, Candler Cusato, and Mia Goodson. We would also like to thank Olivia Timmermans, Nicholas Lusk, Erika Blaine, and Daniel Needs for their help in experimental and analysis setup, as well as design feedback and technical troubleshooting. This work was funded in part by grants from the Office of Naval Research (N00014-16-12682), the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NIH-1RO1HD097732) and the American Kennel Club Canine Health Foundation (Grant #02700). Additional support was provided by Duke University start-up funds awarded to J.M.
Author contributions
J.M. and B.H. supervised the study. M.F. and J.M. designed the experiments and analyzed the data. M.F. collected the data. M.F. wrote the initial draft of the manuscript, and all authors edited the manuscript. All authors contributed to data interpretation and approved the final version of the manuscript.
Data availability
The datasets generated and analyzed during the current study are available from the corresponding author on reasonable request.
- Bhattacharjee, D., Sau, S., Das, J., & Bhadra, A. (2017). Free-ranging dogs prefer petting over food in repeated interactions with unfamiliar humans. Journal of Experimental Biology, 220(24), 4654–4660.
- Cook, P. F., Prichard, A., Spivak, M., & Berns, G. S. (2016). Awake canine fMRI predicts dogs’ preference for praise vs food. Social Cognitive and Affective Neuroscience, 11(12), 1853–1862. https://doi.org/10.1093/scan/nsw102
- Dilks, D. D., Cook, P., Weiller, S. K., Berns, H. P., Spivak, M., & Berns, G. S. (2015). Awake fMRI reveals a specialized region in dog temporal cortex for face processing. PeerJ, 3, e1115. https://doi.org/10.7717/peerj.1115
- Gácsi, M., Győri, B., Miklósi, Á., Virányi, Z., Kubinyi, E., Topál, J., & Csányi, V. (2005). Species-specific differences and similarities in the behavior of hand-raised dog and wolf pups in social situations with humans. Developmental Psychobiology, 47(2), 111–122. https://doi.org/10.1002/dev.20082
- Hare, B., Call, J., & Tomasello, M. (1998). Communication of food location between human and dog (Canis familiaris). Evolution of Communication, 2(1), 137–159.
- Hare, B., & Ferrans, M. (2021). Is cognition the secret to working dog success? Animal Cognition. https://link-springer-com.proxy.lib.duke.edu/article/10.1007/s10071-021-01491-7
- Hare, B., & Woods, V. (2013). The genius of dogs: How dogs are smarter than you think. Penguin.
- Hsu, Y., & Serpell, J. A. (2003). Development and validation of a questionnaire for measuring behavior and temperament traits in pet dogs. Journal of the American Veterinary Medical Association, 223(9), 1293–1300.
- Jones, A. C., & Gosling, S. D. (2005). Temperament and personality in dogs (Canis familiaris): A review and evaluation of past research. Applied Animal Behaviour Science, 95(1–2), 1–53.
- Kekecs, Z., Szollosi, A., Palfi, B., Szaszi, B., Kovacs, K. J., Dienes, Z., & Aczel, B. (2016). Commentary: Oxytocin-gaze positive loop and the coevolution of human–dog bonds. Frontiers in Neuroscience, 10, 155. https://doi.org/10.3389/fnins.2016.00155
- Leonetti, S., Cimarelli, G., Hersh, T. A., & Ravignani, A. (2024). Why do dogs wag their tails? Biology Letters, 20(1), 20230407. https://doi.org/10.1098/rsbl.2023.0407
- Miklósi, Á., Kubinyi, E., Topál, J., Gácsi, M., Virányi, Z., & Csányi, V. (2003). A Simple Reason for a Big Difference: Wolves Do Not Look Back at Humans, but Dogs Do. Current Biology, 13(9), 763–766. https://doi.org/10.1016/S0960-9822(03)00263-X
- Miklósi, Á., & Topál, J. (2013). What does it take to become ‘best friends’? Evolutionary changes in canine social competence. Trends in Cognitive Sciences, 17(6), 287–294. https://doi.org/10.1016/j.tics.2013.04.005
- Morrill, K., Hekman, J., Li, X., McClure, J., Logan, B., Goodman, L., Gao, M., Dong, Y., Alonso, M., Carmichael, E., Snyder-Mackler, N., Alonso, J., Noh, H. J., Johnson, J., Koltookian, M., Lieu, C., Megquier, K., Swofford, R., Turner-Maier, J., … Karlsson, E. K. (2022). Ancestry-inclusive dog genomics challenges popular breed stereotypes. Science, 376(6592), eabk0639. https://doi.org/10.1126/science.abk0639
- Nagasawa, M., Mitsui, S., En, S., Ohtani, N., Ohta, M., Sakuma, Y., Onaka, T., Mogi, K., & Kikusui, T. (2015). Oxytocin-gaze positive loop and the coevolution of human-dog bonds. Science, 348(6232), 333–336. https://doi.org/10.1126/science.1261022
- Palmer, R., & Custance, D. (2008). A counterbalanced version of Ainsworth’s Strange Situation Procedure reveals secure-base effects in dog–human relationships. Applied Animal Behaviour Science, 109(2), 306–319. https://doi.org/10.1016/j.applanim.2007.04.002
- Quaranta, A., Siniscalchi, M., & Vallortigara, G. (2007). Asymmetric tail-wagging responses by dogs to different emotive stimuli. Current Biology, 17(6), R199–R201.
- Quervel-Chaumette, M., Dale, R., Marshall-Pescini, S., & Range, F. (2015). Familiarity affects other-regarding preferences in pet dogs. Scientific Reports, 5, 18102. https://doi.org/10.1038/srep18102
- Rein, B., Ma, K., & Yan, Z. (2020). A standardized social preference protocol for measuring social deficits in mouse models of autism. Nature Protocols, 15(10), Article 10. https://doi.org/10.1038/s41596-020-0382-9
- Ren, W., Wei, P., Yu, S., & Zhang, Y. Q. (2022). Left-right asymmetry and attractor-like dynamics of dog’s tail wagging during dog-human interactions. iScience, 25(8). https://doi.org/10.1016/j.isci.2022.104747
- Salomons, H., Smith, K. C. M., Callahan-Beckel, M., Callahan, M., Levy, K., Kennedy, B. S., Bray, E. E., Gnanadesikan, G. E., Horschler, D. J., Gruen, M., Tan, J., White, P., vonHoldt, B. M., MacLean, E. L., & Hare, B. (2021). Cooperative Communication with Humans Evolved to Emerge Early in Domestic Dogs. Current Biology, 31(14), 3137-3144.e11. https://doi.org/10.1016/j.cub.2021.06.051
- Tan, J., Walker, K. K., Hoff, K., & Hare, B. (2018). What influences a pet dog’s first impression of a stranger? Learning & Behavior, 46(4), 414–429. https://doi.org/10.3758/s13420-018-0353-y
- Topál, J., Miklósi, Á., Csányi, V., & Dóka, A. (1998). Attachment behavior in dogs (Canis familiaris): A new application of Ainsworth’s (1969) Strange Situation Test. Journal of Comparative Psychology, 112(3), 219.
- vonHoldt, B. M., Shuldiner, E., Koch, I. J., Kartzinel, R. Y., Hogan, A., Brubaker, L., Wanser, S., Stahler, D., Wynne, C. D. L., Ostrander, E. A., Sinsheimer, J. S., & Udell, M. A. R. (2017). Structural variants in genes associated with human Williams-Beuren syndrome underlie stereotypical hypersociability in domestic dogs. Science Advances, 3(7), e1700398. https://doi.org/10.1126/sciadv.1700398
- Yang, M., Silverman, J. L., & Crawley, J. N. (2011). Automated three‐chambered social approach task for mice. Current Protocols in Neuroscience, 56(1), 8.26. 1-8.26. 16.
No competing interests reported.