RESEARCH DOSSIER
Domestication and Human–Animal Social Cognition
Abstract. Domestication and human–animal social cognition examine how evolutionary selection, socialization, developmental experience, and interactional context jointly influence animals' responses to human signals and the formation of interspecies relationships. This critical integrative narrative monograph reviews ancient canine genomics, historical dog–wolf experiments, selectively bred foxes, early puppy development and behavior genetics, pointing and ostensive communication paradigms, cooperative action, attachment-like patterns, oxytocin-associated feedback, and comparisons involving goats, horses, pigs, and cats. It evaluates competing explanations for successful cross-species cue use, differentiates heritable variation from direct historical selection, and examines construct validity, rearing confounds, replication and ecological generalization. The 2023 ManyDogs multisite results establish a reproducible but condition-sensitive pointing response, while contemporary comparative and developmental studies show that selection, experience, and task demands contribute through distinct mechanisms. A synthesis of the evidence supports an integrated multilevel account of interspecies social communication and identifies unresolved questions about communicative intention, population generality, and historically specific causal pathways. The method is a critical narrative synthesis of identifiable scientific research, not a systematic review, meta-analysis or new experiment.
Keywords: domestication; human–animal social cognition; dog cognition; wolves; canine cooperation; human pointing; ostensive communication; developmental socialization; tameness; silver fox selection; heritability; human–animal attachment; oxytocin; comparative cognition; horse social cognition; goat social cognition; pig cognition; cat attachment; experimental control; ManyDogs; ancient dog genomics; interspecies interaction
Scientific Overview and the Research Object
Domestication and human–animal social cognition concern the evolutionary and developmental conditions under which animals perceive, interpret, learn from, and coordinate behavior with human partners. The research object is the interaction of selective breeding, population history, rearing environment, individual social experience, species-typical communication and task-specific opportunity. Human–animal social cognition can be measured through responses to pointing and gaze, attention to human faces, solicitation of assistance, affective regulation in proximity to caregivers, and cooperative action. These measures represent partially separable processes. Performance on one object-choice task cannot establish a general species ranking for social intelligence, nor can an adult pet dog's attachment behavior prove that the entire form of relating was selected during domestication Hare et al. (2002) Miklósi et al. (2003) Topál et al. (1998).
The central questions are how domestic animals acquire sensitivity to human signals, which components vary genetically, what intensive exposure and formal training can achieve in nondomesticated relatives, and whether different taxa converge on common functional solutions. A further question concerns the explanatory scope of cognitive attribution: successful use of a gesture may reflect learned association, attention to motion, sensitivity to communicative intent, or a combination. Correspondingly, reduced response may reflect fear, low motivation, perceptual constraints, or mismatched testing conditions. A rigorous monograph follows the inferential chain from observable behavior through experimental design to proposed mechanism, examining counterexamples as evidence that strengthens causal discrimination rather than as rhetorical exceptions Virányi et al. (2008) Udell et al. (2008) ManyDogs Project et al. (2023).
This work uses a critical integrative narrative synthesis of original controlled comparisons, longitudinal development, behavior-genetic studies, domestication experiments, multisite replication, phylogenetic and ancient-genomic work, and selected conceptual accounts. It does not report new animal experiments, a preregistered literature search, a formal risk-of-bias meta-analysis, or pooled effect estimates. The empirical nucleus concerns canids, because dogs, wolves and selectively bred foxes provide complementary contrasts in domestication history, human socialization and selection. Goats, horses, pigs and cats establish whether human-directed attention is a uniquely canine outcome or a wider consequence of ecological opportunity and social experience Hare et al. (2005) Kaminski et al. (2005) Proops & McComb (2010) Vitale et al. (2019).
The main scientific result of the synthesis is an explanatory architecture: evolved dispositions shape attraction and tolerance, developmental experience calibrates what people predict and reward, and situational interaction converts available capacities into observable cooperative behavior. These are interacting causes with different units of measurement. The empirical record supports early-emerging and heritable variation in canine human-directed communication, alongside the powerful effects of rearing and the procedural sensitivity of gesture tasks. Historical narratives of a singular cognitive transformation are therefore best tested through explicit causal models, multiple species, and more than one behavioral paradigm Bray et al. (2021) Salomons et al. (2021) ManyDogs Project et al. (2023).
Definitions, Units of Analysis, and Terminological Boundaries
Domestication denotes a multigenerational evolutionary process in which populations undergo heritable changes through sustained interactions with human-modified environments, often including human-controlled selection. Taming is a change in the behavior of an individual, whereas socialization is developmentally structured exposure and learning that make other individuals or species familiar and behaviorally significant. Training adds contingencies, repetition and reinforcement to specific actions. These processes can coincide without being equivalent. Human-reared wolves may be extensively socialized yet remain wolves; domestic dogs can experience little human contact and may respond differently from companion animals. This distinction is indispensable when comparing historical selection hypotheses with contemporary behavior Gácsi et al. (2005) Virányi et al. (2008) Udell et al. (2008).
Social cognition concerns information processing used to predict or coordinate with other agents. In the present field, an experimentally accessible unit is often the choice between two containers after a human supplies a gesture, but the construct extends to attention, learning, communication, social orientation, cooperation and proximity regulation. An object-choice outcome is a measurement of response to a compound situation, not a transparent view of an animal's representation of a human mind. A dog may use the indicated location as information, follow a salient movement, or employ a familiar reinforcement history. Identifying mechanism requires systematically varying the gesture's spatial position, duration, communicative context and relation to reward Hare et al. (2002) ManyDogs Project et al. (2023).
Species identity, population ancestry, breed, litter, individual, caregiver relationship, testing site and repeated trial are nested or crossed levels of inference. Genetic heritability describes a proportion of observed variation in a specified population and environment; it does not determine the fraction of one animal's behavior that is genetic, and it should not be extrapolated unchanged across populations. Likewise, an early appearing behavior can be biologically prepared while still requiring ordinary developmental input. Findings from an assistance-dog breeding population, for example, invite replication among village dogs, shelter dogs and populations with different human-contact histories before population-wide generalizations are made Bray et al. (2021) Udell et al. (2008).
Attachment-like interaction denotes structured proximity-seeking, distress regulation, discrimination of familiar partners and behavior changing under separation and reunion. It is an empirical comparative term whose diagnostic components must be demonstrated within each species. Oxytocin-associated responses refer to measured endocrine changes correlated with social episodes or experimentally manipulated physiological pathways; they do not themselves supply a complete account of psychological bonding. Cooperation designates coordinated action producing an outcome involving partners, and its mechanism may include learned routines, mutual tolerance, monitoring of partners or flexible shared goals. Clear definitions keep evolutionary explanations, proximate processes and relational descriptions in their appropriate scopes Topál et al. (1998) Nagasawa et al. (2015) Range & Virányi (2015).
Historical Development of Comparative Research on Domestication
Research connecting domestication and animal psychology draws on evolutionary thinking, ethology, experimental learning theory and modern cognitive biology. For much of the twentieth century, scientific debates about intelligence in animals were shaped by the need to distinguish inherited tendencies from conditioned responses and human projection. Later comparative designs made human-directed behavior itself the target of controlled study. The question changed from whether a dog could obey a familiar gesture to whether dogs and closely related canids differed under carefully specified rearing conditions, and whether such differences were visible before prolonged training. This shift was critical because it made evolutionary and ontogenetic mechanisms empirically separable Hare et al. (2002) Miklósi et al. (2003).
The late-1990s application of a modified Strange Situation procedure to dog–owner relationships helped establish the investigation of caregiver-specific interaction as a measurable comparative phenomenon. Topál and colleagues assessed variation in adult dogs during structured separations and reunions, showing patterns that invited comparison with attachment research. The experimental series in the early 2000s then introduced human pointing as a prominent object-choice paradigm. Hare and colleagues' 2002 report linked young dogs' competence and human-reared wolves' poorer performance to a domestication hypothesis; Miklósi and colleagues' 2003 study contrasted cue use and gaze toward humans in insoluble-problem situations Topál et al. (1998) Hare et al. (2002) Miklósi et al. (2003).
Follow-up work progressively complicated an all-or-none picture. Carefully hand-raised dogs and wolves were compared across early development; longitudinal studies showed that wolves could learn to respond to more distant gestures after extensive training; and other researchers found wolves following momentary distal points without such training when rearing and daily human interaction differed from earlier samples. Experimental foxes bred for low fear and aggression supplied a separate selection contrast, while studies of goats and horses opened the analysis beyond carnivores. The methodological transformation was from broad species narratives toward the analysis of experience, temperamental readiness and cues with different perceptual properties Gácsi et al. (2005) Hare et al. (2005) Virányi et al. (2008) Udell et al. (2008) Kaminski et al. (2005).
Recent work has introduced developmental behavior genetics, controlled puppy comparisons and multisite collaborations. Bray and colleagues measured early social-communicative capacities in a large puppy sample, and Salomons and colleagues compared dog and wolf puppies on social and nonsocial tasks. ManyDogs 1 subsequently examined the robustness of pointing results across multiple laboratories and conditions of communicative signaling. Ancient DNA research independently refined the timing and population structure of canine domestication, with major genomic studies in 2020 and 2026 providing a far deeper temporal context for evolutionary claims. These advances allow historical explanations to be tested with complementary behavioral, developmental and genomic evidence without assuming that the same dataset identifies every causal step Bray et al. (2021) Salomons et al. (2021) ManyDogs Project et al. (2023) Bergström et al. (2020) Bergström et al. (2026).
Canine Domestication in Archaeological and Genomic Perspective
Evolutionary explanations of human–dog communication require a historically accurate account of domestication. Dogs originated from ancient wolf-related populations; the demographic structure of those populations and the geographical sequence of domestication remain subjects of active genomic investigation. Ancient DNA demonstrates that domestic dog lineages had already diversified substantially during the late Pleistocene and early Holocene. Bergström and colleagues' 2020 analysis identified at least five major ancestry lineages by approximately 11,000 years ago. This observation sets a minimum temporal depth for diversification without identifying one behavioral mutation or a single event when social-cognitive capacities were acquired Bergström et al. (2020).
Earlier proposals of multiple geographical origins illustrate the distinction between plausible models and resolved history. Frantz and colleagues' 2016 combination of genetic and archaeological evidence supported a dual-origin scenario, while subsequent genome-wide work has refined lineage relationships and made simple origin stories less tenable. Bergström and colleagues' 2026 study of early European dogs recovered genomic information from ancient canid remains, including a dog dated to approximately 14,200 years ago at Kesslerloch, and reported continuity of ancestry with later dogs. Population ancestry inference does not read prehistoric social behavior directly; it constrains when divergent dog lineages existed and whether geographically sampled remains represent dog or wolf ancestry Frantz et al. (2016) Bergström et al. (2026).
Archaeological proximity between humans and canids, comparative skeletal morphology and genetic ancestry provide different classes of evidence. Morphological classification of early specimens may be uncertain, and populations can undergo introgression after initial divergence. A genetic separation predating agriculture also does not prove whether attraction to human presence, dietary adaptation, cooperative opportunities or active human selection was the first driver. These causal pathways can have operated in different sequence and strength across populations. Consequently, a psychological model should not turn a genomic divergence date into a precise date for the emergence of an ability to follow a pointed finger Bergström et al. (2020) Bergström et al. (2026).
Population history remains essential to cognitive interpretation because present-day wolves are surviving relatives rather than a direct sample of the ancestral wolves involved in domestication. Modern dogs also represent populations shaped by later selection, migration, functional breeding and differential contact with humans. Controlled modern comparisons therefore estimate differences among sampled living populations, under their rearing and testing arrangements. Historical causal inference needs an additional link from those differences to ancestral selective environments. Ancient genomes can clarify phylogenetic structure, but the behavioral bridge must be investigated through development, selection experiments, comparative ecology and explicit model comparison Bergström et al. (2020) Frantz et al. (2016) Hare et al. (2005).
Evolutionary Hypotheses and Competing Causal Models
The domestication hypothesis proposes that interactions with humans altered dog behavior and cognition by selection. One version emphasizes direct selection for sensitivity to human communicative acts, especially gestures indicating resources. A second emphasizes indirect effects: selection for reduced fear, aggressive response or greater tolerance increases the time and attention that an animal can devote to human actions. A third emphasizes pre-existing cooperative and perceptual skills in wolves, subsequently recruited into new relationships through changed social motivation and development. These versions make partly distinct predictions and can act simultaneously rather than comprising a single mutually exclusive pair Hare et al. (2002) Hare et al. (2005) Range & Virányi (2015).
The direct-selection model predicts early competence on human-specific communication tasks even with limited training, heritable variation under appropriate designs, and performance differences exceeding those attributable to attention or fear alone. The correlated-selection model predicts shifts in communicative behavior when selection targets reduced avoidance rather than cue-following accuracy. The cooperative ancestry model predicts that well-socialized wolves retain substantial abilities to read and cooperate with partners, though they may deploy them differently with humans. Evidence from genetically variable puppies, selected foxes and well-reared wolves gives empirical relevance to all three explanatory domains Bray et al. (2021) Hare et al. (2005) Udell et al. (2008) Range & Virányi (2015).
Developmental systems approaches locate causation in interactions between inherited dispositions and early relational ecology. A puppy that readily approaches people obtains more opportunities to observe hands, faces and food contingencies; these opportunities can amplify subtle initial tendencies into reliable adult behavior. Conversely, a wolf that initially avoids close human contact may accumulate fewer such experiences unless deliberate hand-rearing alters the developmental pathway. This is a feedback explanation with testable mediators, not a claim that genes merely determine personality or that experience can erase phylogenetic differences. It motivates longitudinal measurement of approach latency, eye contact, reinforcement and subsequent gesture performance Gácsi et al. (2005) Salomons et al. (2021).
Comparative inference is strongest when each explanation predicts contrasting results under manipulations that isolate its mechanisms. Testing high-socialization wolves against low-socialization dogs, using human and conspecific demonstrators, and varying whether pointing is perceptually obvious or communicatively marked can reveal failures of oversimplified models. The empirical question is not whether domestication matters in general; it is which functional component was altered, by what developmental route, and in which ecological context. A mature account of dog–human cognition will integrate phylogenetic history, selective environments, temperamental approach, learning opportunities and the organization of current interactions Udell et al. (2008) Range & Virányi (2014) ManyDogs Project et al. (2023).
Human Pointing as a Comparative Object-Choice Paradigm
The object-choice paradigm typically presents two potential hiding places and supplies a human cue toward one. Performance is summarized as a proportion of choices toward the indicated target or through trial-level models that control chance, individual variation and experimental conditions. Its attraction is experimental tractability: the handler can control location, gesture type and feedback. Yet a point is simultaneously a visible arm movement, a spatial vector, an action performed by a social partner, and sometimes a communicative signal accompanied by gaze or vocal address. These elements need to be separated before the experiment can reveal how an animal interprets another species' conduct. A choice toward a container establishes discriminative cue use within the protocol; additional controls are required to establish referential understanding Hare et al. (2002) ManyDogs Project et al. (2023).
The contrast between proximal and distal points has become particularly informative. A finger placed directly beside a container may serve as a local enhancement stimulus, attracting attention to a location without communicating where something is hidden. A momentary distal point, in which the experimenter's arm returns before the animal makes a choice, requires some retention or continued representation of the indicated target. Even the distal condition can be solved through perceptual mapping and learned contingencies. Inferential strength improves with crossed control conditions: pointing at empty places, experimenter gaze directed away, noncommunicative arm movements, trials without informative rewards, and systematic exchanges of experimenter identity Virányi et al. (2008) Udell et al. (2008).
Task outcomes are highly sensitive to timing, distance, experimenter position, reward visibility, cue repetition and the dog's access to feedback. Tests differ in whether animals are held by unfamiliar assistants, allowed to orient freely, or exposed to pretraining, and whether a failure to approach is counted as a wrong answer. Animals may understand an indication but decline to approach a stranger, or may select by odor if containers are handled inconsistently. Comparing raw percent-correct scores across studies with these differing features risks equating different psychological and motivational tasks. Researchers increasingly prioritize transparent protocols, trial-level data and comparisons against both chance and well-specified alternative models ManyDogs Project et al. (2023) Bray et al. (2021).
A theoretically discriminating paradigm is therefore a collection of experimental contrasts rather than one stimulus. The strongest design establishes whether an animal tracks the informative direction of a cue after controlling local enhancement, whether performance transfers across people and settings, and whether communicative marking improves response independently of familiarity and arousal. The finding of above-chance choice is meaningful, but claims about language-like understanding or mental-state attribution require additional convergent evidence. The distinction preserves the considerable evidence of animal sensitivity to human gestures while disciplining explanations of the processes that generate this sensitivity ManyDogs Project et al. (2023) Hare et al. (2002).
Early Dog–Wolf Contrasts and the Socialization Debate
Hare and colleagues' 2002 experiments established one of the founding comparisons in this literature: domestic dogs, including puppies, used human communicative gestures to locate food, whereas wolves in an initial rearing comparison did less well. The researchers interpreted the pattern as consistent with cognitive effects of domestication. The historical importance of the work lies in its experimentally focused evolutionary hypothesis and its demonstration that ordinary dog performance could exceed the performance of a close nondomesticated relative under the tested conditions. That contrast alone could not eliminate differences in duration, quality and setting of human exposure, willingness to approach the experimenter, or cue-training history Hare et al. (2002).
Subsequent research made these potential explanations experimentally consequential. Miklósi and colleagues compared responses to human cues and behavior when an animal faced an unsolvable task. Dogs characteristically looked back at human partners more than similarly reared wolves in the tested conditions, suggesting differences in the organization of human-directed communication. Yet looking back can be shaped by learned experience with people as sources of assistance as well as temperamental approach. In dogs bred and housed for collaboration with humans, looking to a partner may have been reinforced frequently and become a reliable strategy. It therefore measures social orientation under task pressure, and the mechanisms generating that orientation remain a separate empirical problem Miklósi et al. (2003).
Virányi and colleagues' 2008 developmental comparisons showed a more nuanced picture: young hand-reared dogs used several pointing forms earlier in testing, while wolves made progress with experience and could acquire reliable use of particular gestures. Udell, Dorey and Wynne reported that highly socialized wolves could outperform dogs in following human social cues in their protocols. These findings are informative because they alter the balance of evidence available to strong versions of the hypothesis that wolves lack a relevant capacity. They support an account in which performance depends on rearing, testing experience, attention and motivation as well as domestication history Virányi et al. (2008) Udell et al. (2008).
Different findings across laboratories should motivate a specification of the experimental population rather than an ad hoc ranking of whole species. Hand-raising can entail human presence nearly continuously during sensitive developmental periods, whereas pet dogs may receive substantial but uneven contact and shelter dogs different routines. Wolves can exhibit human-directed familiarity with specific caretakers yet respond differently to strangers; domestic dogs likewise show large individual differences in fearfulness and sociability. Robust inference requires documenting age at first contact, daily exposure, handling, feeding procedures, test setting, prior training and trial-by-trial engagement. The experimentally supported conclusion concerns the joint role of evolutionary history and development, while assigning specific variance to each cause remains a design-dependent task Gácsi et al. (2005) Udell et al. (2008) Salomons et al. (2021).
Cross-Rearing, Sensitive Periods, and Developmental Calibration
Cross-rearing and intensive hand-raising seek to reduce the disparity in exposure between dogs and wolves. In Gácsi and colleagues' 2005 study, hand-reared dog and wolf pups were compared on their spontaneous interactions with humans during early development. The researchers documented species-specific differences even under intensive rearing, indicating that matched caregiving does not necessarily produce indistinguishable behavior. However, equivalence of an experimenter's planned care is not equivalence of experienced opportunities: two species may react differently to identical handling, seek different kinds of interaction, or require different exposure to produce the same degree of security. These possibilities require measured developmental mediators, not an assumption that matched schedules are matched experience Gácsi et al. (2005).
Sensitive-period reasoning helps explain why temporal organization matters. Early approach and reduced fear can create a developmental cascade in which exposure increases attention to hands, eyes and voices; attention facilitates learning from contingencies; reinforced learning then increases future approach. Animals entering that cascade at different ages may show divergence even with similar total hours of later training. Developmental studies should therefore distinguish age at initial exposure, hours and quality of contact, number of distinct humans encountered, reinforcement history, and onset of autonomous choice. A comparison conducted at 12 weeks can reflect accumulated developmental feedback rather than a single age-independent cognitive trait Gácsi et al. (2005) Salomons et al. (2021).
Longitudinal work is especially valuable because responsiveness and willingness to engage may follow different trajectories. A young dog can attend confidently to a stranger while a wolf may orient to a trusted caregiver and avoid unknown experimenters. Failure to choose in the latter case cannot be analyzed solely as a failure to process pointing. The most informative designs separate no-choice trials from incorrect selections, include independent fear and attention measures, and test identical cues with familiar and unfamiliar people. Hierarchical models can then ask whether early approach predicts later cue accuracy after controlling lineage, litter and testing history Virányi et al. (2008) Bray et al. (2021).
Developmental calibration also clarifies a common conceptual mistake about innate behavior. An early-emerging preference may have a heritable component and still be shaped by expected developmental interaction; early occurrence shows a compressed learning history, not its total absence. Conversely, the demonstration that a highly socialized wolf succeeds at an object-choice task does not entail that the wolf's developmental route was typical of historical wild populations. Evolutionary claims address the distributions of traits and experiences generated by population histories. Cross-rearing tests whether changed developmental input can alter outcomes, and each explanation adds a different part of the causal account Bray et al. (2021) Udell et al. (2008) Hare et al. (2002).
Puppy Development, Heritability, and Early Social Communication
Bray and colleagues' 2021 study investigated social-communicative development in 375 young dogs from an assistance-dog breeding population. Puppies around eight weeks of age responded to human social cues and oriented to human faces, and the authors estimated substantial heritability for measured variation in these traits. The study strengthened the case that individual differences in early human-directed communication have a genetic component in a defined breeding population. Its evidence is particularly notable because many pups were assessed before extensive individualized training, reducing one familiar source of task experience. Heritability estimates nonetheless refer to variation among the sampled dogs under the actual shared and variable environments, and should not be interpreted as a universal fixed percentage of behavioral causation Bray et al. (2021).
Salomons and colleagues' 2021 comparisons of dog and wolf puppies measured orientation toward people and performance across cognitive tasks. Dog puppies were generally more attracted to humans and showed stronger spontaneous performance on particular cooperative-communication measures, while wolf puppies could show comparable abilities in other cognitive domains. A focal scientific interpretation was that domestication altered the early organization of social engagement with humans. An important qualification is that species comparisons contain more than the domestication variable: present-day wolves and dogs differ in ancestry, developmental tempo, ecology and response to experimental environments. Strong causal attribution relies on convergence across designs rather than one cross-sectional species comparison Salomons et al. (2021).
Developmental data distinguish the existence of a capacity from the motivation to express it. A pup may detect the human's motion and still avoid approaching the human-centered apparatus; another may choose quickly because previous interaction has made people attractive. Independent measures of latency, gaze orientation, human approach and nonsocial object memory help map those possibilities. Genetic variation affecting fearfulness or engagement could generate apparent variation in communicative accuracy without directly encoding a representation of pointing. Conversely, when cue performance remains associated with heritable differences after relevant mediators are measured, a more specific selection hypothesis becomes plausible. These are empirically discriminable models, not semantic preferences Bray et al. (2021) Salomons et al. (2021).
Multi-population sampling is the next test of generality. Assistance-dog lines, village dogs, purpose-bred working dogs, companion dogs and free-ranging dogs experience different selective pressures and human ecologies. Estimates from a structured breeding pedigree need replication against genomic relatedness and household effects in differently recruited populations. Longitudinal measurement also allows tests of genetic correlations: do the variants associated with early approach predict adult cue following independently of reinforcement and environmental enrichment? Such research would refine the bridge between population selection and concrete social-cognitive processes Bray et al. (2021) ManyDogs Project et al. (2023).
Gaze, Ostension, and the Inference of Communicative Intent
Human communication commonly includes ostensive signals—such as addressing an individual by name or meeting its gaze—that announce an intention to inform. Experiments with dogs have asked whether these markers make a human point more likely to be followed. The question is theoretically important because a dog that privileges an ostensively introduced cue may be sensitive to the social relevance of an act. Yet an attentive signal also changes arousal, gaze direction, readiness to act and the predictability of reinforcement. Ostension therefore needs contrasts that equate perceptual salience and human presence while selectively changing communicative context ManyDogs Project et al. (2023).
The ManyDogs 1 collaboration tested pointing comprehension across 20 research sites in a sample of 455 dogs included in the principal analysis. Dogs followed pointing above chance under the compared conditions, while the study did not establish the predicted robust advantage for ostensive over nonostensive pointing. This result provides multisite support for gesture-following performance but puts pressure on an expansive interpretation in which particular attention-getting signals are necessary for understanding human reference. It also illustrates the methodological value of joint protocol design, sufficient power and independent laboratory settings when effects have previously been established from relatively small samples ManyDogs Project et al. (2023).
Pointing comprehension and communicative intent attribution need not be identical constructs. An individual can acquire a reliable rule that the human arm predicts food location, can evaluate which of two cues is more informative, and can orient to a person's face without representing the person's communicative mental state in the same manner as an adult human. The scientific challenge is to distinguish processing levels using intervention and generalization: an animal might spontaneously transfer a gesture to a novel object, respond to a person whose belief is inaccurate, or distinguish informative from deliberately misleading actions. Each additional condition carries its own perceptual, reinforcement and inhibition controls Hare et al. (2002) ManyDogs Project et al. (2023).
The appropriate synthesis is an experimentally bounded one. Dogs in several populations readily use human gestures, some wolves can learn or spontaneously use them under suitable developmental and testing conditions, and communicative markers do not always produce the theoretically expected increment. A comprehensive theory must consequently explain reliable cross-species communication without building its mechanism into the operational criterion. Differentiating communicative sensitivity from intention attribution leaves substantial room for sophisticated nonverbal interaction and for gradual evolutionary changes in how social attention is recruited Virányi et al. (2008) Udell et al. (2008) ManyDogs Project et al. (2023).
Experimental Selection for Tameness in Silver Foxes
The farm-fox selection experiment provides an especially revealing comparison because researchers deliberately bred foxes for responsiveness toward people over successive generations. Hare and colleagues in 2005 tested foxes from lines selected for tame behavior and compared their responses to human gestures. The selectively bred foxes demonstrated heightened performance on some human-directed social-cognitive tasks, despite the absence of an explicit breeding program for comprehension of pointing. The study supported a correlated-response hypothesis: selection affecting fear or affiliative approach can alter the developmental and behavioral prerequisites of social communication without selection directly targeting gesture comprehension Hare et al. (2005).
The causal significance of this experiment rests on an intervention across generations. In ordinary dog–wolf comparisons the histories of the lineages are already intertwined with many ecological differences. In the fox experiment the investigator's breeding criterion was specified, creating a clearer relation between the selected trait and the measured behavior. Yet the result remains a population-level association following breeding rather than identification of a singular neural or genetic mediator. Selection can affect several physiological and behavioral systems simultaneously, and highly approachable foxes may acquire different everyday experiences with human keepers before formal testing Hare et al. (2005).
Phenotypic covariation under selection has multiple causal interpretations. Selection for tameness might alter stress responsiveness and create more opportunities to explore unfamiliar gestures; it might favor attention to humans through shared developmental regulation; or it might affect a broader suite of correlated behaviors. Discrimination among these possibilities requires independent measures of approach, vigilance, sensory attention, learning rate, endocrine responses and cue following across generations and control lines. Experiments that expose groups to matched cue experience can then ask which developmental mediators remain relevant. The fox results make the mechanism question testable rather than settling it by definitional assertion Hare et al. (2005) Bray et al. (2021).
The fox evidence generalizes most directly to the possibility of correlated evolution under one specified selection regime. It does not establish that domestic dogs underwent the same historical trajectory or that human interaction automatically produces a uniform cognitive phenotype in every domestic species. Selection intensity, population structure, life history and ecology differ sharply among foxes, dogs, goats, pigs and horses. The experimental lesson is that a social trait may evolve through selection on motivational and affective prerequisites; the historical transfer to dogs and other species requires corroborating comparative and population evidence Hare et al. (2005) Bergström et al. (2020).
Fear, Temperament, Exploration, and Human Approach
Fear of unfamiliar people is a central confound in human-directed cognition experiments because most task protocols require approaching an experimenter or an apparatus positioned close to a person. An animal that keeps its distance may fail to expose its information-processing capacities to the test. Domestication-related changes in approach and tolerance can therefore produce improvements in recorded task performance even when general spatial or associative capacities differ little. Experimental designs should measure approach and vigilance separately, include low-threat and familiar-partner conditions, and analyze no-choice behavior explicitly Gácsi et al. (2005) Salomons et al. (2021).
Temperament also alters learning opportunity over the life course. Animals willing to remain near humans will experience more reward contingencies and fine-grained movements, potentially accelerating their sensitivity to subtle cues. This developmental amplification creates statistical mediation between inherited predisposition and adult gesture performance. By measuring approach before any cue training and then tracking how often individuals interact with people, researchers can assess whether approach predicts later social learning. Such a pathway could account for species differences without supposing that every feature of communicative cognition was itself a direct target of selection Hare et al. (2005) Bray et al. (2021).
Motivational factors also complicate comparisons inside a species. Food-deprived and satiated animals may engage differently; reward preferences vary across individuals; and a caretaker's presence may either facilitate exploration or inhibit it through anticipation of punishment or restraint. Traits measured as willingness to wait, persistence or attention can correlate with cue following without describing the interpretation of the cue. Manipulating reward value and physical access while retaining the same gesture permits estimation of which variance belongs to social engagement and which belongs to discrimination accuracy. Causal clarity depends on these mundane controls Udell et al. (2008) ManyDogs Project et al. (2023).
Temperament is itself socially embedded. Repeated positive human contact can reduce avoidance, and individuals may form selective relationships rather than generalized trust toward everyone. Studies must therefore specify the handler, test administrator and familiar partner instead of treating 'the human' as an interchangeable stimulus. This distinction is particularly important when comparing hand-reared wolves familiar with a small set of caretakers with dogs accustomed to multiple people. Evidence of a social-cognitive difference should survive reasonable transformations of familiarity and response costs before being interpreted as a lineage-wide functional incapacity Gácsi et al. (2005) Udell et al. (2008).
Communicative Gaze and the Unsolvable-Task Paradigm
The unsolvable-task paradigm gives an animal a problem it can initially solve and then makes the same problem inaccessible. Investigators measure whether it persists alone, alternates attention between the apparatus and a human, or seeks contact with a familiar person. Miklósi and colleagues' 2003 dog–wolf comparison made this method prominent by identifying a difference in looking back toward humans. The design operationalizes the organization of help-seeking and interspecies attention under frustration. It does not, by itself, prove that an animal attributes knowledge or benevolent intentions to the human Miklósi et al. (2003).
Alternating gaze may provide a communicative function when it recruits an observer to act, and the observable outcome can be tested by comparing trials where an attentive person responds to gaze with trials where the person ignores it. However, gaze can also arise as part of scanning, frustration regulation, habit or sensitivity to possible reinforcement. A strong inference requires evidence that it flexibly changes with the availability and attentional state of a helpful partner, and that the animal differentiates a knowledgeable observer from an irrelevant bystander when simpler cues are controlled. These designs keep communicative function, learning history and mental-state attribution as distinct questions Miklósi et al. (2003) ManyDogs Project et al. (2023).
The opportunity to develop such gaze patterns differs across domestic and nondomestic animals. Pet dogs commonly encounter humans as providers of food, access and aid, whereas a wolf's experimental opportunities for soliciting assistance may depend on intensive rearing and husbandry. A look-back measure can thus record developmental ecology in addition to evolved motivation. Repeating the task with familiar and unfamiliar handlers, equalizing successful opportunities for assistance, and quantifying distress or persistence would strengthen causal comparisons. A negative result is ambiguous when the individual remains engaged with the apparatus and has no practical expectation that a human will intervene Udell et al. (2008) Gácsi et al. (2005).
The wider significance of the unsolvable task is that social cognition can be understood as an interactional strategy rather than a collection of isolated abilities. A dog and a human coordinate when the dog's attention changes the person's behavior, which changes the dog's next action. This closed-loop organization can be measured through temporal sequences instead of one aggregate gaze count. Future models should analyze the order and contingency of looks, vocalizations, body orientation and human responses. The explanatory target becomes a coupled system while retaining independent measurements of each partner's behavior Miklósi et al. (2003) Nagasawa et al. (2015).
Conspecific Learning, Wolves, and the Cooperation Hypothesis
Comparative research must ask whether a competence apparently specialized for humans reflects a change in general social-cognitive capacity or a redistribution of attention between partner species. Wolves possess complex within-species social lives involving coordination, conflict and affiliative interaction. The Canine Cooperation Hypothesis advanced by Range and Virányi argues that dog–human cooperation may have recruited capacities already present in wolf social organization. Under this framework, domestication could have modified tolerance, affiliation and the identities of relevant partners without inventing flexible social learning from the beginning Range & Virányi (2015).
Range and Virányi's 2014 imitation experiments compared dogs and wolves learning from conspecific demonstrators. They reported stronger imitation performance among wolves in the specified task, demonstrating that one cannot infer general social-learning superiority from dogs' ease of interaction with humans. Species may perform differently depending on whether the demonstrator is a conspecific or human, how behavior is rewarded and whether tasks align with natural social expectations. The finding illustrates the need for factorial comparisons of species and demonstrator identity rather than isolated rankings Range & Virányi (2014).
Cooperation requires further differentiation between coordinated behavior and the mental processes that sustain it. A pair can act jointly because each responds to the other's movement, because learned routines stabilize a sequence, or because individuals monitor partners' roles and flexibly adapt to unexpected disruptions. Studies involving apparatuses that demand simultaneous actions can identify whether an animal waits for a partner and adjusts behavior when the partner becomes unavailable. A human–dog performance advantage in such a design may reflect a history of joint activity, motivation to interact, or species-specific attention; alternative mechanisms require separate controls Range & Virányi (2015).
The cooperative ancestry account complements rather than erases evidence for domestication. Selection may change how an existing competence is deployed and which social partners are salient, while intensive rearing may reveal latent capabilities in wolves. Historical evolution, present-day functional repertoire and experimentally observed preference should therefore be represented as different variables. This layered formulation predicts circumstances in which dogs outperform wolves with unfamiliar human cues, wolves outperform dogs with conspecific models, and thoroughly socialized wolves approach dog-level performance on selected human tasks. Such conditional predictions are more informative than a universal species ranking Range & Virányi (2014) Udell et al. (2008) Salomons et al. (2021).
Cooperative Action, Partner Monitoring, and Joint Outcomes
Human–animal cooperation spans routine husbandry, navigation, assistance, coordinated hunting and experimental tasks. The psychological components vary across contexts: some interactions require recognizing an informative gesture; others require waiting for a partner, inhibiting an independent response or adjusting behavior to another individual's ongoing action. A scientific synthesis should therefore distinguish cue responsiveness, dyadic coordination and goal structure. The mechanisms supporting each may diverge even within a single domesticated species. A dog can be excellent at following a human point yet show limited flexibility when an unfamiliar partner changes the required action sequence Range & Virányi (2015) Hare et al. (2002).
One methodological challenge is that human experimenters often provide subtle prompts while attempting to test spontaneous cooperation. Tone, gaze, body posture and reward timing can structure the animal's behavior even when explicit instruction is withheld. Double-blind or automated cue delivery can help where technically feasible, though fully automated systems may reduce ecological validity by replacing a social partner with a device. Studies should report exactly what the human was permitted to do, how responses were scored and whether blind coders confirmed the predicted pattern. Comparisons of human and conspecific partners can then clarify transfer across species ManyDogs Project et al. (2023) Range & Virányi (2014).
Cooperative behavior also changes through repeated interaction. A reliable partner creates opportunities to predict contingencies and refine coordination; a partner who fails to respond can reduce solicitation or encourage independent problem solving. Within-subject designs that manipulate partner reliability and permit recovery after disruption can reveal whether animals update expectations about particular individuals. Such evidence would extend beyond one-off pointing accuracy to an account of relationship-specific prediction and flexible joint regulation. The relevant causal outcome is change in coordinated behavior as a function of partner history Miklósi et al. (2003) Nagasawa et al. (2015).
At a broader scale, the coevolution of human institutions and domestic animal behavior created repeated niches for interspecies cooperation. Work roles, feeding practices, handling routines and cultural norms shape the exposure of individual animals to human intentions and gestures. These social histories matter even when the proximate experiment holds cue type constant. Future research should integrate behavioral trials with ethnographic or longitudinal records of everyday human–animal interaction, using shared operational measures of partner monitoring and successful coordination. The evolutionary theory remains accountable to observable mechanisms in contemporary relationships Range & Virányi (2015) Bergström et al. (2020).
Attachment-Like Bonds and Comparative Separation–Reunion Designs
Attachment is an organized pattern of relationship-specific behavior that includes seeking proximity, using a familiar partner as a source of security, and adjusting exploration or stress regulation as partner availability changes. The adaptation of Ainsworth's Strange Situation framework to dogs offered an operational route to studying such patterns. In structured episodes, an owner and a stranger enter, leave and return, allowing researchers to examine selective proximity and behavioral change. Topál and colleagues' 1998 investigation showed that adult dogs' responses could be studied as differentiated caregiver-related patterns, establishing a major empirical foundation for human–animal attachment research Topál et al. (1998).
The procedure has interpretive conditions that differ across species and social environments. Separation distress can arise from isolation, novelty, frustration or expectation of routine rather than attachment alone. Likewise, a calm animal may be securely regulated or simply unresponsive to the situation. Researchers therefore interpret multiple behaviors jointly: orientation and approach to the familiar person, differential response to a stranger, changes in exploration, behavior at reunion and repeated observations outside the test. Valid comparisons should establish that experimental episodes carry comparable functional meaning for the animals being compared Topál et al. (1998) Vitale et al. (2019).
Domestic dogs' sensitivity to particular human partners is a strong demonstration of relational organization without implying that their relationship is developmentally identical to an infant's relationship with a caregiver. The attachment construct transfers only to the extent that its components show convergent predictive validity in the target species. In dogs, owner-specific behavior can arise through extended cohabitation, dependence on humans for resources, repeated comforting interaction and species-level dispositions toward social engagement. Determining the relative contributions requires longitudinal records, caregiver substitutions and control for training and housing Topál et al. (1998) Gácsi et al. (2001).
Comparative attachment research can connect social cognition to behavioral regulation. An animal's response to a pointing gesture may depend on whether the gesturing person is familiar and trusted, while a familiar partner's presence may lower vigilance and increase exploration. Such links suggest interaction between relational security and information use, but each causal path must be tested rather than inferred from concurrent changes. Controlled tests that vary caregiver presence while holding cue properties and rewards fixed can evaluate the effect of relational context on observed cognitive performance Topál et al. (1998) ManyDogs Project et al. (2023).
Caregiver Specificity, Rescue Settings, and the Formation of New Bonds
Attachment-like behavior has been observed beyond dogs raised continuously in one household. Gácsi and colleagues' 2001 study examined adult dogs housed in rescue centers and their capacity to form new human relationships. The work is important because it introduces relational plasticity: existing social histories and institutional conditions can shape the formation of selective interactions with people. It argues against treating one early developmental relationship as a complete explanation of every adult social bond. At the same time, the speed and quality of bond formation should be interpreted within the specific handling and repeated-contact procedures of the study Gácsi et al. (2001).
Rescue and shelter populations are methodologically valuable because their social histories may differ from those of household pets, but these histories are often incompletely documented. Abandonment, changes in handlers, uncertainty of feeding schedules and novel environments can all affect activity, vigilance and proximity seeking. Recruitment into a shelter study also creates selection effects, since animals that tolerate handling may be more likely to be tested. A responsible analysis records time since arrival, prior known caregiving, frequency of staff contact, behavioral screening and the physical context of reunion episodes Gácsi et al. (2001) Topál et al. (1998).
The ability to form a new bond should be distinguished from a universal preference for humans. Selectivity depends on recognizing particular individuals and changing behavior in relation to their presence; broad sociability is a propensity to approach people generally. These traits can be correlated without being identical. An animal may approach all strangers readily yet show little caregiver-specific regulation, while another may exhibit selective trust after cautious initial behavior. Studies combining social-approach assessments with separation–reunion procedures can test whether the dimensions covary across individual developmental histories Gácsi et al. (2001) Vitale et al. (2019).
Caregiver specificity provides a bridge to the science of interspecies cooperation. Familiar people become reliable predictors of consequences, which can increase attention to their communication and change how uncertainty is managed. That effect may be adaptive in domestic settings and experimentally measurable by alternating gestures from trusted and unfamiliar humans under otherwise matched conditions. A critical interpretation demands evidence of the directional pathway; finding that attached dogs perform well in cue tasks does not prove that attachment caused accuracy, because socialization and motivation may be common determinants Gácsi et al. (2001) ManyDogs Project et al. (2023).
Oxytocin, Mutual Gaze, and Physiological Feedback
Endocrine research has added a physiological dimension to the study of human–dog relationships. Nagasawa and colleagues in 2015 reported an association between mutually sustained gaze during human–dog interaction and changes in oxytocin in dogs and their human partners, together with experimental components exploring oxytocin-related feedback. The proposed positive-loop account links social attention and neuroendocrine processes with reciprocal affiliative behavior. Its major value is to put bidirectional interaction, rather than animal behavior alone, at the center of the model. Physiological measurement can show that social exchange is accompanied by biological change, but interpreting such change requires careful separation of correlation, manipulation and generalization Nagasawa et al. (2015).
Oxytocin is involved in several behavioral and physiological processes; its presence cannot be read as a unique molecular marker of love, attachment or mutual understanding. Hormone concentrations depend on sampling matrix, assay validity, sampling time, stress and other contextual factors. Intranasal administration studies pose additional questions about dose, central access and the conditions under which behavioral effects emerge. A social gaze association may be influenced by owner behavior, dog familiarity, arousal and prior reinforcement, even when pharmacological interventions identify a causal contribution under tested conditions. Interpreting the entire evolutionary history of a relationship from one endocrine response exceeds what the experiment can identify Nagasawa et al. (2015).
The physiological feedback hypothesis makes distinct predictions that can be tested. If mutual gaze changes subsequent human responsiveness and dog behavior, time-resolved interaction records should show a sequence of endocrine and behavioral changes that is robust to controls for simple presence and physical contact. If attachment-like interactions influence stress regulation, researchers should measure both social behavior and valid physiological outcomes during separation, reunion and novel challenges. Within-subject repeated measures are preferable to inferring a trait from one post-interaction sample. It is also necessary to consider whether animals that spontaneously make eye contact have different developmental histories from animals that do not Nagasawa et al. (2015) Topál et al. (1998).
The comparative question remains open across wolves and other domesticated species. A physiological association that occurs frequently in familiar human–dog pairs could result from a selected readiness for human engagement, long-term relational experience or both. Parallel protocols in socialized wolves, cats, horses and goats, adjusted for species-specific physiology and interaction forms, would test which components are broadly mammalian and which depend on particular domestication histories. A scientifically productive account treats oxytocin as one possible mediator within a multi-level social system, not a sufficient single-cause theory of human–animal bonds Nagasawa et al. (2015) Vitale et al. (2019).
Emotional Signaling, Relational Regulation, and Welfare
Social cognition in human–animal relationships is often studied in emotionally charged contexts. Human tone, posture, approach trajectory and handling can signal threat or reassurance, while animals' movement, gaze, vocalization and proximity changes shape human behavior in return. Such signals may function through learned prediction, immediate affective contagion, perceptual discrimination or flexible social expectations. Distinguishing these routes requires more than a successful behavioral response: it requires controls for sensory cues, prior association and context. The same animal may read a familiar person's intent accurately while responding cautiously to an unfamiliar experimenter Miklósi et al. (2003) Topál et al. (1998).
Animal welfare is part of measurement validity. An experiment that relies on repeated separation, unexpected restraint or social isolation may change the very variables that it purports to measure, including attention, willingness to approach and reward valuation. Welfare-sensitive designs minimize unnecessary aversive exposure, use appropriate stopping criteria, monitor stress-relevant behavior and explain why the procedure is necessary. These precautions protect animals while also reducing uncontrolled motivational variation. Testing animals in a familiar environment and incorporating familiar handlers can improve engagement, though researchers must balance this against the need to avoid cueing by caregivers Topál et al. (1998) ManyDogs Project et al. (2023).
Social experience can also alter the animal's ecological role. Working dogs, livestock, equids used in human services and companion cats interact with people under dramatically different constraints. Human access to resources, mobility, opportunities for retreat and consistent handling influence which communicative behaviors are adaptive or reinforced. Data gathered in a food-choice test may predict only a narrow portion of those everyday relationships. Ecologically valid observation requires defining the behavior of interest in ordinary practice and connecting it to experimentally controllable variables without assuming that laboratory success automatically entails better welfare Kaminski et al. (2005) Proops & McComb (2010) Vitale et al. (2019).
An integrative welfare agenda asks whether communication methods improve mutual predictability and reduce avoidable conflict. Outcomes might include fewer failed handling episodes, more voluntary participation and stable behavior during necessary care. Any intervention intended to increase human-directed attention must be evaluated for unintended demands, such as suppressing species-typical avoidance when withdrawal would be appropriate. In this framework, social cognition research contributes to human stewardship precisely by acknowledging the agency and ecological needs of the animals whose behavior it studies Range & Virányi (2015) Nagasawa et al. (2015).
Goats and Human-Directed Object-Choice Performance
Domestic goats provide an important test of the claim that responsiveness to human gestures is a distinctive outcome of canine domestication. Kaminski, Riedel, Call and Tomasello's 2005 experiments examined whether goats used social cues in choosing between locations containing a food reward. Goats used some human-given cues, including pointing, under the reported conditions, extending evidence for functional responsiveness beyond the dog–wolf comparison. The study cautions against equating a similar experimental success with an identical psychological mechanism. Goats have different foraging ecology, domestication history and routine handling experiences, and the salience of a gesture may differ from that in canine tests Kaminski et al. (2005).
An object-choice task in goats needs species-appropriate control of neophobia, approach to the testing device and food motivation. Animals managed by people may already associate the handler's movements with food delivery or movement through gates. A point may therefore inform location through well-learned contingencies without requiring mental-state attribution. Comparing familiar handlers with strangers, controlling the visibility of reward placement, and distinguishing hand-touch cues from distal indicating gestures would help specify the component that transfers across settings. This is also why recording which cue forms work matters more than labeling goats globally capable or incapable of human communication Kaminski et al. (2005).
Goats create an evolutionary contrast with dogs because their ancestral social organization, feeding niche and selection histories differ. If responsiveness appears across both species, several explanations remain available: repeated human exposure, general mammalian social attention, selection for tolerance of handling, or convergent changes shaped by domestication. Demonstrating the same observable choice does not allow researchers to discriminate among these alternatives. A stronger program compares wild relatives, minimally handled domestic animals and extensively socialized individuals, while holding task demands appropriate to each species. Evolutionary convergence should be inferred from matched functional patterns supported by developmental and comparative evidence Kaminski et al. (2005) Hare et al. (2005).
The scientific significance of the goat case is the separation between species-specific evolutionary history and cross-species functional possibility. Domestication in different lineages can place animals in recurrent situations in which human actions predict valuable outcomes. Such ecologies can generate a shared practical vocabulary of movement and attention even when perceptual priorities differ. A comparative social cognition framework therefore asks how different species arrive at reliable cross-species coordination and which mechanism is responsible in each, instead of presuming a canine blueprint Kaminski et al. (2005) Proops & McComb (2010).
Horses and Recognition of Human Attentional States
Horses interact with humans in settings where posture, orientation and access to visual information have immediate practical consequences. Proops and McComb's 2010 study tested how domestic horses use human-given cues in relation to another person's attentional state. It documented sensitivity to relevant configurations of human attention in experimental settings. This provides a comparative case in which response to humans need not depend on a conspicuous pointing gesture. The psychological question concerns which observable human attributes the horse discriminates and whether that information guides its own choice or solicitation of attention Proops & McComb (2010).
Human attention is not directly observable as a mental state; experiments manipulate perceptual indicators such as body orientation, gaze and visibility. A horse might respond to the geometry of those indicators, to a history of rewarded interactions, or to a representation of what a person can see. These hypotheses predict differences when a human's visible face is incongruent with bodily orientation, when familiar cues are selectively occluded, or when an attentive actor cannot actually access information. Controls must preserve the animal's normal perceptual access and avoid inadvertent changes in handler presence or reward expectation Proops & McComb (2010).
Individual histories are especially influential in equine samples. Riding-school horses, pastured animals with minimal handling and horses trained for high-frequency human interactions encounter different contingencies. These environments can shape approach to strangers and responsiveness to subtle bodily signals. Breed, age and handling history should be measured alongside experimental accuracy, with training practices documented rather than treated as incidental background. A result from a small convenient set of familiar horses may establish a capacity in the tested population without determining its developmental distribution across all domestic horses Proops & McComb (2010) ManyDogs Project et al. (2023).
The horse literature broadens the research object from reading intentional pointing to identifying how animals adjust behavior in response to a partner's apparent availability. That adjustment may influence safe handling and cooperative routines, but applied claims demand independent evaluation. Researchers can link experimental measures of attention discrimination to observation of everyday interactions and test whether reliable human signaling improves communication without increasing aversive control. This approach integrates ecological validity, social learning and welfare while keeping claims about mental-state reasoning proportionate to the evidence Proops & McComb (2010).
Pigs, Attention, Impulsivity, and Human-Given Cues
Domestic pigs provide a useful contrast because they are intelligent, exploratory omnivores whose daily experience of human interaction varies widely with husbandry. Nawroth, Ebersbach and von Borell's 2013 work assessed juvenile pigs' sensitivity to the attentive states of humans and emphasized how impulsivity can affect choice behavior. Their 2014 object-choice study then tested pigs' use of human-given cues. Together these studies make clear that the animal's regulatory capacity and the task's response demands influence the observable expression of social-cognitive competence Nawroth et al. (2013) Nawroth et al. (2014).
Impulsivity is a methodological issue whenever an animal must inhibit an immediate approach, wait for a gesture or select one of two competing food locations. A choice made before the cue is completed can count as incorrect despite intact discrimination when the animal attends. Experimental interpretation therefore improves when investigators code approach latency, pre-cue movements and failure to wait, and compare trials of different temporal demands. Such measures can reveal whether apparent performance differences are attributable to attention, inhibitory control or the informativeness of the human signal Nawroth et al. (2013).
Pig studies also challenge categorical narratives about selection for companion-like relationships. Livestock species may acquire effective communication with people through routine feeding, movement and handling even without selective breeding for intimate social partnership. This possibility suggests that domestication-related changes in human social engagement are distributed across multiple ecological routes. Yet selective breeding for productivity, different rearing systems and often restricted human contact create broad heterogeneity. Researchers cannot explain performance in pigs using the same developmental template applied to household dogs without measuring relevant differences Nawroth et al. (2014) Hare et al. (2005).
An integrated comparative design would contrast pigs with varying exposure to individual caretakers, use unfamiliar and familiar human demonstrators, and add nonsocial cue controls matched for movement and timing. Observing an increase in response after structured socialization would estimate the role of experience in the sampled setting; persistent population differences could motivate further genetic and ecological studies. Independent replication remains necessary because individual livestock samples and procedures are unlikely to provide stable species-wide rankings. The value of the pig literature is to articulate mechanisms that are invisible in a narrow canine-versus-wolf design Nawroth et al. (2013) Nawroth et al. (2014).
Domestic Cats and Caregiver-Specific Social Bonds
Cats are often used rhetorically as a contrast to dogs, but controlled studies have examined their ability to form selective human relationships. Vitale, Behnke and Udell's 2019 report adapted a secure-base test and classified patterns of attachment behavior in cats toward human caregivers. The experiments found that some cats exhibited behaviors consistent with security-related social organization, supporting the proposition that caregiver-directed bonds extend beyond dogs. The result warrants attention because cats differ from dogs in domestication ecology and typical human management; comparable relational outcomes do not require identical selective histories Vitale et al. (2019).
Secure-base procedures depend on careful operationalization. Investigators infer relationship-specific regulation from changes in exploration and social contact across separation and reunion episodes, and must distinguish it from general novelty reactions. Differences in the cat's immediate environment, hiding opportunities, tolerance of handling and activity rhythms can strongly affect the observed pattern. The classification scheme should be evaluated for repeated-measures stability, inter-rater agreement and association with independent behavioral outcomes. An attachment label becomes scientifically useful when its observable components predict behavior across relevant settings, rather than when a procedure simply assigns categories Vitale et al. (2019) Topál et al. (1998).
Cross-species comparisons should be cautious about one-to-one transfer of behavioral indicators. Cats may express proximity and social attention through postures and movement patterns unlike those of dogs. A brief encounter near the caregiver may be meaningful in one species and routine in another. Testing a common relational function thus calls for species-appropriate coding while retaining comparable experimental contrasts: familiar versus unfamiliar person, caregiver present versus absent, and altered exploration in a novel environment. This balances functional comparison with the specificity of each animal's behavioral repertoire Vitale et al. (2019).
The cat evidence enriches the central argument that domestication and individual experience jointly organize interspecies relationships. An animal can develop a stable expectation about a particular human without having the same inclination to orient to strangers' hand gestures as a typical dog. Attachment-like regulation and human gesture comprehension are therefore distinguishable research constructs; their correlation should be established empirically, not presupposed. This distinction opens a richer comparative account of social cognition as a family of interacting capabilities rather than one domestication-produced package Vitale et al. (2019) Hare et al. (2002).
Comparative Synthesis Across Domestic Species
Dog, goat, horse, pig and cat studies reveal a recurring phenomenon: animals living in human-modified environments can extract useful information from aspects of human behavior and develop partner-specific patterns of interaction. The relevant responses include object choice after pointing, adjustment to apparent human attention, caregiver-regulated exploration and solicitation of aid. These abilities share an interactional setting but need not share a single evolutionary mechanism. Cross-species synthesis should therefore map the function of the behavior, the cue employed, the response demanded and the developmental history rather than translate every success into a general score of intelligence Kaminski et al. (2005) Proops & McComb (2010) Nawroth et al. (2014) Vitale et al. (2019).
The strongest inferences come from convergence across complementary evidence. Selection experiments in foxes suggest that modifying human-directed temperament can shift performance on social tasks; puppy genetics links variation in early canine social communication to heritable differences; wolf studies demonstrate substantial effects of human exposure; and other domestic species show alternative routes to social responsiveness. None of these sources alone establishes a complete historical pathway, but together they define a model with inherited dispositions, developmental learning and ecological partner relevance as distinct causal dimensions Hare et al. (2005) Bray et al. (2021) Udell et al. (2008).
Comparative methodology must also confront asymmetries in research effort. Dogs have received intensive study and increasingly coordinated replication, while some findings in goats, pigs, horses and cats derive from narrower populations or small sets of paradigms. Unequal study density can create an illusion that one species possesses more complex cognition simply because more aspects have been tested. A scientific comparison should report uncertainty and the extent of cross-laboratory confirmation by species, while allowing that different tasks are feasible and ethically appropriate for different animals ManyDogs Project et al. (2023) Kaminski et al. (2005) Nawroth et al. (2013).
Domestication can be conceived as a family of historical processes with recurrent functional consequences, rather than one cognitive transformation repeated identically in every lineage. Human proximity changes selective and developmental opportunity, and recurring interaction generates channels for information transfer. The empirical agenda is to specify which channels arise spontaneously, which require socialization or training, how much variation is heritable, and how stable successful coordination remains outside controlled experiments. A mature comparative account integrates species-specific differences into a common causal framework while retaining sufficient precision to explain individual behavior Bergström et al. (2020) Salomons et al. (2021) Vitale et al. (2019).
Measurement Validity and the Identification of Cognitive Mechanisms
Construct validity is the central methodological challenge of domestication-related social cognition. A behavioral label such as pointing comprehension, attachment or cooperation refers to a cluster of observations and hypothesized processes. Validation requires more than demonstrating above-chance accuracy: a measure should respond appropriately when its presumed cause is manipulated and remain discriminable from neighboring constructs. An animal can show successful pointing use through spatial association while displaying little evidence about an experimenter's communicative mental state. Therefore a construct should be operationally defined at the granularity supported by the task, and the inference from performance to mechanism should be explicit Hare et al. (2002) ManyDogs Project et al. (2023).
Convergent designs ask whether the same individual performs similarly across complementary tasks that share the proposed construct but differ in sensory and motivational demands. Discriminant designs ask whether the measure can be separated from general exploration, olfactory discrimination, food motivation or human approach. In a pointing test, deliberately misleading perceptual cues and controls for reward odor can identify simple alternatives; changing whether the person is familiar can reveal socialization effects. In attachment-like procedures, repeated caregiver-specific tests and independent measures of stress regulation provide stronger evidence than a single separation episode ManyDogs Project et al. (2023) Topál et al. (1998).
Measurement invariance across species is a stronger demand still. An identical gesture and timing protocol does not ensure an identical cognitive problem for animals with different perceptual fields, attention priorities, locomotor behavior and social motivations. Researchers can seek functional equivalence through species-appropriate response methods, while preserving a common experimentally manipulated causal relation. Both identical and adapted designs are informative when investigators document what has changed. Statistical models should account for individual and site variation rather than treat trials as independent observations from the same homogeneous species Kaminski et al. (2005) Proops & McComb (2010) ManyDogs Project et al. (2023).
Mechanism identification ultimately requires interventions and model comparisons. One model may specify perceptual attraction to human arm movement; another may posit a learned prediction about food locations; a third may posit recognizing the communicative relation between an agent and a referent. These models can make similar predictions in the simplest trials, so the discriminating data lie in novel or contradictory conditions. A well-supported explanation will predict an informative pattern of successes, failures and learning transitions across such tests. This standard keeps cognitive theory tied to observable causes without narrowing social cognition to one favored vocabulary Virányi et al. (2008) ManyDogs Project et al. (2023).
Sampling, Population Structure, and Reproducibility
Comparative psychology historically relied on small local samples of animals with unusually intensive human contact. Such samples can be valuable for identifying possible capacities but cannot, on their own, estimate how widespread a trait is among companion dogs, village dogs, wolves or other species. Behavioral responses cluster by litter, facility, caregiver, breed and testing history, creating correlated observations. Studies should make recruitment and exclusions transparent, include statistical terms for nested data when justified, and report uncertainty around individual and population estimates. Broad statements about a species require populations sampled across relevant ecological and institutional variation Bray et al. (2021) ManyDogs Project et al. (2023).
Replication is particularly important when a theoretically striking result may depend on small procedural changes. Differences in pointing duration, where the arm remains after gesturing, how the dog is addressed, and whether rewards are replaced following an error can alter difficulty. Registered multisite protocols allow researchers to see which effects recur and which depend on particular apparatuses or local populations. Yet standardized protocols also risk imposing one task design on populations for whom it is less natural. Investigators should report standardized core trials alongside justified species- or population-specific adaptations ManyDogs Project et al. (2023).
Published evidence can be distorted by outcome selection and selective reporting. An experimenter may test multiple gestures and publish the most favorable one; a study with uncertain results may be less likely to appear in the literature. Statistical uncertainty, preregistered hypotheses and disclosure of all tested conditions reduce this risk. Systematic synthesis would need a declared search strategy, study inclusion criteria, independent bias evaluation and analysis of selective publication. The present monograph integrates identifiable original studies narratively and therefore makes no claim that its selected studies exhaust the field or produce pooled prevalence of canine abilities ManyDogs Project et al. (2023).
Independent replication must extend beyond point following. The importance of attachment, help-seeking and physiological interaction findings justifies shared protocols that measure repeated behavior, site characteristics and preexisting human contact. For endocrine measures, assay and sample handling protocols require coordination across facilities. Robustness should be assessed under both laboratory and ecologically meaningful settings rather than indexed solely by a single p value. The practical reward is clearer knowledge of which mechanisms generalize and which rely on a distinctive local human–animal relationship Nagasawa et al. (2015) Vitale et al. (2019) Topál et al. (1998).
ManyDogs 1 as a Critical Replication Case
The ManyDogs 1 collaboration illustrates a transition from single-laboratory demonstration to coordinated inference. The consortium tested 455 dogs across 20 sites using a two-alternative object-choice paradigm and contrasted ostensive pointing, involving eye contact and name-calling, with a nonostensive control using a different attention signal. Dogs performed above chance in both conditions. The principal contrast did not show a statistically significant advantage of the ostensive condition, and the observed responses to contralateral momentary pointing were lower than some earlier estimates. These findings independently support a bounded capacity for following a human point while challenging claims that ostensive marking reliably enhances performance across varied tested populations ManyDogs Project et al. (2023).
The replication result must be interpreted at the level of its actual comparison. It does not demonstrate that dogs never interpret ostensive communication or that individual dogs cannot show effects in other paradigms. It establishes that the predicted group-level contrast did not replicate as a robust advantage under the coordinated design and sampled populations. Similarly, above-chance performance does not tell us whether the animals represented the human gesture as an intentional message. The theoretically productive outcome is narrower but stronger: communicative sensitivity must be decomposed into cue-following behavior and the additional psychological consequences of ostensive markers ManyDogs Project et al. (2023).
The comparison also shows why recorded sample sizes and analytic populations should be reported precisely. A related ManyDogs data resource describes a larger collected dataset, while the main 2023 paper reports the 455 dogs analyzed in the published replication. These values describe different scopes and should not be treated as contradictory estimates of the same analysis. Scientific claims about precision and effect size must be attached to the exact analysis and conditions that produced them. When discussing multisite findings, readers need to know which dogs, trials and exclusions underlie each result ManyDogs Project et al. (2023).
Extending the collaborative model to dog–wolf and cross-species experiments would be valuable but complex. Intensive wolf socialization is laborious, facilities are relatively rare, and welfare and experience cannot be made interchangeable across laboratories by fiat. Preregistered conceptual replication could therefore supplement literal protocol replication, testing the same mechanism through appropriately adapted procedures. The goal is a set of converging experiments that distinguish selection, exposure and cue properties while making differences in context visible and scientifically interpretable Udell et al. (2008) Virányi et al. (2008) ManyDogs Project et al. (2023).
Causal Inference Across Genes, Development, and Ecology
Domestication research uses at least four causal comparisons: genomic ancestry across archaeological time, lineages that differ in historical selection, animals with different early experience, and individuals experimentally exposed to altered signals or partner contingencies. Each design targets different mechanisms and carries different confounding risks. Ancient DNA can locate population divergence but cannot directly reveal an animal's response to a gesture. Fox breeding experiments manipulate selection on tameness but do not reconstruct prehistoric dog selection. Cross-rearing probes developmental plasticity, and cue interventions identify proximate processes. Their combination is an evidentiary strategy rather than a substitute for identifying what each design establishes Bergström et al. (2026) Hare et al. (2005) Udell et al. (2008).
An explicit causal diagram could represent selection history as affecting approach propensity and sensory-social attention; early human contact as affecting learned cue contingencies and caregiver security; and trial conditions as affecting the opportunity to display a response. Human contact may mediate an evolved inclination to approach, while experimenter familiarity can moderate the effect of a cue. Because the same behavior may be produced by multiple paths, apparent species differences in accuracy should be analyzed in relation to measured mediators and plausible unmeasured causes. Randomizing a cue tests a local causal effect, whereas comparing existing species does not randomize evolutionary history Bray et al. (2021) Salomons et al. (2021).
One especially consequential distinction is between genetic heritability and response to selection. A heritability estimate from early dog performance describes standing variation in a sample, not evidence that prehistoric human-directed pointing was directly selected. A breeding experiment that selects tameness and alters cue following supports correlated evolution within those lines, but does not identify the historical distribution of ancestral dog traits. Genomic ancestry gives phylogenetic context yet requires a genotype-to-phenotype bridge. An integrated theory should state the independent evidence required at each step before claiming an entire domestication pathway Bray et al. (2021) Hare et al. (2005) Bergström et al. (2020).
Ecological and cultural systems also change with the animals. Human handling, settlement, work practices and expectations about companion species shape which interactions are frequent and rewarded. Such environments can evolve across generations and influence both selection and development, producing feedback between human behavior and animal traits. Testing those pathways will require studies of free-ranging and working populations, historical records of husbandry, genetic data where appropriate, and prospective observation of social learning. Causal inference is strongest when evolutionary, developmental and interactional explanations are kept distinct yet linked by testable predictions Bergström et al. (2020) Range & Virányi (2015).
Human–Animal Social Cognition in the Artificial Era
Human–animal social cognition also clarifies a general research problem of the Artificial Era: how a social relationship is inferred from the behavior of a system whose organization differs from the observer's. Humans can rapidly interpret coordinated behavior as communication, responsiveness or understanding, while a scientific account must establish how the relevant signals are generated and used. Animal cognition research offers an instructive methodological discipline: define the behavior, compare mechanisms, distinguish historical development from present interaction, and evaluate generalization. This lesson bears on the study of human–artificial relations without converting biological domestication into a purported history of artificial systems ManyDogs Project et al. (2023) Topál et al. (1998).
In animal experiments, a subject's use of a human gesture can be produced through different combinations of perception, learning and social attention. In human interactions with artificial systems, similarly fluent outcomes may result from architectures with entirely different learning histories, optimization criteria and memory structures. A shared observable function therefore supports a functional comparison, while the mechanisms and welfare implications must be investigated in their own domains. Artificial systems should be studied with operational measures suited to their construction and interactions; conclusions about animal attachment, endocrine regulation or evolutionary selection cannot be transferred solely because the interaction appears reciprocal Nagasawa et al. (2015) Bray et al. (2021).
Human–animal research additionally foregrounds bidirectionality. A caregiver's behavior changes an animal's attention, and the animal's reaction changes the caregiver's next action. This dynamic is analytically relevant to human–artificial interaction, where a person's expectations and a system's outputs can co-shape repeated communication. The comparative object is the structure of contingent coordination across different kinds of participants. The biological cases show the importance of longitudinal, relationship-specific evidence and independent performance measures before broad psychological constructs are assigned Miklósi et al. (2003) Vitale et al. (2019).
Within the horizon From Homo to Artificial, this dossier supplies a precise comparative resource rather than an equivalence theory. It demonstrates how human-modified environments generate new interspecies social arrangements, how evolutionary and developmental pathways must be distinguished, and why successful coordination invites further questions about causal organization. Those principles can inform future research on Artificial social interaction while preserving the independent status and evidentiary requirements of animal psychology. The result is a methodological connection between two research programs, not a claim that domestication explains artificial cognition Range & Virányi (2015) ManyDogs Project et al. (2023).
Unresolved Questions and a Program for Future Research
One major unresolved issue is the mechanism underlying early canine responsiveness to human signals. Current developmental and behavioral-genetic studies suggest biological preparedness, whereas intensive wolf socialization and interspecies comparisons establish substantial contextual and experiential influence. To separate paths, future longitudinal experiments should recruit multiple dog populations, measure early approach and fear, systematically document human exposure and test cue transfer across different partners. Genomic pedigree analysis can be integrated with repeated measures rather than treated as an alternative to developmental observation Bray et al. (2021) Salomons et al. (2021) Udell et al. (2008).
A second issue is the validity of purported cognitive specializations. Does response to a point reflect representation of communicative intent, or can flexible associative learning and perceptual guidance account for the same behavior? Preregistered experiments should introduce conflicting social and nonsocial cues, vary informativeness without changing visual salience, manipulate partner knowledge, and examine transfer to tasks with different response demands. Such designs can compare explicit computational models of learning and social inference using out-of-sample predictions instead of relying on post hoc verbal interpretations Hare et al. (2002) ManyDogs Project et al. (2023).
A third issue concerns the ecological distribution of abilities. Evidence is uneven across dogs, wolves, horses, goats, pigs and cats, and some species have been studied through a limited range of paradigms. Collaborative multispecies projects should standardize explanatory contrasts while adapting apparatuses to species-appropriate perception, motivation and welfare. Researchers should publish null outcomes and complete procedural descriptions to make differing performance interpretable. Longitudinal outcomes of actual human–animal cooperation can be related to laboratory tasks to determine whether the measured constructs predict meaningful interaction outside the research setting Kaminski et al. (2005) Proops & McComb (2010) Nawroth et al. (2014) Vitale et al. (2019).
A fourth issue links evolutionary history with present-day social variation. Ancient genomic results increasingly establish the population structure of early dogs, yet psychological consequences are inferred from living descendants under contemporary ecological circumstances. Broader sampling of ancient genomes can refine historical hypotheses, while modern genetic and behavioral research tests the traits that differ among populations and the mechanisms producing them. The key outstanding challenge is to connect independently established levels of evidence without turning a demographic date or a behavioral association into an unwarranted claim of direct selection Bergström et al. (2020) Bergström et al. (2026).
Conclusion
Domestication and human–animal social cognition form a research field concerned with how evolutionary history and human-modified social environments reorganize attention, communication, learning, cooperation and partner-specific regulation. The evidence establishes robust human-cue use in domestic dogs under many circumstances, meaningful early and heritable variation in canine social responsiveness, and major contributions from individual experience and rearing. Socialized wolves, selectively bred foxes and multiple other domestic species show why a single-cause account is inadequate. Successful interaction can emerge through different combinations of predisposition, development, motivation and task-specific information Hare et al. (2002) Hare et al. (2005) Bray et al. (2021) Udell et al. (2008).
The history of dog–wolf research illustrates the value of critical scientific revision. Early experiments motivated the idea of a domestication-specific social-cognitive shift; later comparisons revealed the influence of socialization and experience; and coordinated replication showed that the effects of some communicative signals depend more strongly on procedure than broad theories initially anticipated. These findings refine the object of explanation. A theory must predict which cue is followed, by which animal, in relation to which human, under what developmental history and experimental circumstances Virányi et al. (2008) ManyDogs Project et al. (2023).
Attachment-like interaction and physiological feedback widen the topic beyond gesture following. Caregiver-specific relationships, mutual attention and patterns of help-seeking reveal how social cognition develops within repeated dyadic exchange. Comparisons with goats, horses, pigs and cats extend the framework across divergent ecological and domestication histories. The comparative synthesis supports multiple routes to functional coordination and demands that behavioral similarity be connected to the particular mechanisms and populations in which it is observed Topál et al. (1998) Nagasawa et al. (2015) Kaminski et al. (2005) Vitale et al. (2019).
The general scientific conclusion is an integrated multilevel account with explicit empirical boundaries. Selection can alter dispositions toward humans, developmental experience calibrates expectations and learning, and moment-to-moment social situations determine how competence is expressed. Ancient genomics informs the timescale and structure of lineage history, while experiments establish the properties of observable interaction. The next advances will come from longitudinal, multisite and multispecies designs that link these levels through discriminating predictions. This foundation also supplies a disciplined comparative reference for understanding relational organization in the Artificial Era, where fluent social interaction likewise demands inquiry into underlying mechanism Bergström et al. (2026) Bray et al. (2021) ManyDogs Project et al. (2023).
References
- Bergström, A., Frantz, L., Schmidt, R., Ersmark, E., Lebrasseur, O., Girdland-Flink, L., Lin, A. T., Storå, J., Sjögren, K.-G., Anthony, D., Antipina, E., Amiri, S., Bar-Oz, G., Bazaliiskii, V. I., Bulatović, J., Brown, D., Carmagnini, A., Davy, T., Fedorov, S.,. Skoglund, P. (2020). Origins and genetic legacy of prehistoric dogs. Science, 370(6516), 557–564. https://doi.org/10.1126/science.aba9572
- Bergström, A., Furtwängler, A., Johnston, S., Rosengren, E., Breidenstein, A., Booth, T., McCabe, J. B., Peto, J., Williams, M., Kelly, M., Tait, F., Baumann, C., Radzeviciute, R., Barrington, C., Anastasiadou, K., Gilardet, A., Glocke, I., Sherman, M., Brativnyk, A.,. Skoglund, P. (2026). Genomic history of early dogs in Europe. Nature, 651, 986–994. https://doi.org/10.1038/s41586-026-10112-7
- Bray, E. E., Gnanadesikan, G. E., Horschler, D. J., Levy, K. M., Kennedy, B. S., Famula, T. R., & MacLean, E. L. (2021). Early-emerging and highly heritable sensitivity to human communication in dogs. Current Biology, 31(14), 3132–3136.e5. https://doi.org/10.1016/j.cub.2021.04.055
- Frantz, L. A. F., Mullin, V. E., Pionnier-Capitan, M., Lebrasseur, O., Ollivier, M., Perri, A., Linderholm, A., Mattiangeli, V., Teasdale, M. D., Dimopoulos, E. A., Tresset, A., Duffraisse, M., McCormick, F., Bartosiewicz, L., Gál, E., Nyerges, É. Á., Sablin, M. V., Bréhard, S., Mashkour, M.,. Larson, G. (2016). Genomic and archaeological evidence suggest a dual origin of domestic dogs. Science, 352(6290), 1228–1231. https://doi.org/10.1126/science.aaf3161
- Gácsi, M., Topál, J., Miklósi, Á., Dóka, A., & Csányi, V. (2001). Attachment behavior of adult dogs (Canis familiaris) living at rescue centers: Forming new bonds. Journal of Comparative Psychology, 115(4), 423–431. https://doi.org/10.1037/0735-7036.115.4.423
- 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., Brown, M., Williamson, C., & Tomasello, M. (2002). The domestication of social cognition in dogs. Science, 298(5598), 1634–1636. https://doi.org/10.1126/science.1072702
- Hare, B., Plyusnina, I., Ignacio, N., Schepina, O., Stepika, A., Wrangham, R., & Trut, L. (2005). Social cognitive evolution in captive foxes is a correlated by-product of experimental domestication. Current Biology, 15(3), 226–230. https://doi.org/10.1016/j.cub.2005.01.040
- Kaminski, J., Riedel, J., Call, J., & Tomasello, M. (2005). Domestic goats, Capra hircus, follow gaze direction and use social cues in an object choice task. Animal Behaviour, 69(1), 11–18. https://doi.org/10.1016/j.anbehav.2004.05.008
- ManyDogs Project, Espinosa, J., Stevens, J. R., Alberghina, D., Always, H. E. E., Barela, J. D., Bogese, M., Bray, E. E., Buchsbaum, D., Byosiere, S.-E., Byrne, M., Cavalli, C. M., Chaudoir, L. M., Collins-Pisano, C., DeBoer, H. J., Douglas, L. E. L. C., Dror, S., Dzik, M. V., Ferguson, B.,. Zylberfuden, S. G. (2023). ManyDogs 1: A multi-lab replication study of dogs' pointing comprehension. Animal Behavior and Cognition, 10(3), 232–286. https://doi.org/10.26451/abc.10.03.03.2023
- 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
- Nagasawa, M., Mitsui, S., En, S., Ohtani, N., Ohta, M., Sakuma, Y., Onaka, T., Mogi, K., & Kikusui, T. (2015). Social evolution: Oxytocin-gaze positive loop and the coevolution of human–dog bonds. Science, 348(6232), 333–336. https://doi.org/10.1126/science.1261022
- Nawroth, C., Ebersbach, M., & von Borell, E. (2013). Are juvenile domestic pigs (Sus scrofa domestica) sensitive to the attentive states of humans? The impact of impulsivity on choice behaviour. Behavioural Processes, 96, 53–58. https://doi.org/10.1016/j.beproc.2013.03.002
- Nawroth, C., Ebersbach, M., & von Borell, E. (2014). Juvenile domestic pigs (Sus scrofa domestica) use human-given cues in an object choice task. Animal Cognition, 17(3), 701–713. https://doi.org/10.1007/s10071-013-0702-3
- Proops, L., & McComb, K. (2010). Attributing attention: The use of human-given cues by domestic horses (Equus caballus). Animal Cognition, 13(2), 197–205. https://doi.org/10.1007/s10071-009-0257-5
- Range, F., & Virányi, Z. (2014). Wolves are better imitators of conspecifics than dogs. PLOS ONE, 9(1), e86559. https://doi.org/10.1371/journal.pone.0086559
- Range, F., & Virányi, Z. (2015). Tracking the evolutionary origins of dog–human cooperation: The Canine Cooperation Hypothesis. Frontiers in Psychology, 5, Article 1582. https://doi.org/10.3389/fpsyg.2014.01582
- 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
- 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), 219–229. https://doi.org/10.1037/0735-7036.112.3.219
- Udell, M. A. R., Dorey, N. R., & Wynne, C. D. L. (2008). Wolves outperform dogs in following human social cues. Animal Behaviour, 76(6), 1767–1773. https://doi.org/10.1016/j.anbehav.2008.07.028
- Virányi, Z., Gácsi, M., Kubinyi, E., Topál, J., Belényi, B., Ujfalussy, D., & Miklósi, Á. (2008). Comprehension of human pointing gestures in young human-reared wolves (Canis lupus) and dogs (Canis familiaris). Animal Cognition, 11(3), 373–387. https://doi.org/10.1007/s10071-007-0127-y
- Vitale, K. R., Behnke, A. C., & Udell, M. A. R. (2019). Attachment bonds between domestic cats and humans. Current Biology, 29(18), R864–R865. https://doi.org/10.1016/j.cub.2019.08.036