Self-Awareness in Dogs: What the Experiments Can Show

1. What the Question Is Actually Asking

1.1 Self-Awareness Is Not One Thing

"Is my dog self-aware?" sounds like a single question with a yes or no answer. It is not. The term covers at least four capacities that can come apart, and the canine evidence differs sharply across them.

In comparative cognition, self-representation refers to information an animal can use about itself. It does not necessarily imply conscious self-awareness.

Body awareness is having a usable representation of one's own physical extent — knowing, before trying, whether a body will fit through a gap or whether it is standing on the thing it needs to move.

Self-recognition is identifying a representation of oneself as oneself: the mirror case, and its olfactory equivalent.

Memory of one's own actions is being able to recall what one did, as opposed to recalling what happened.

Metacognition is having access to the state of one's own knowledge — knowing that one does not know something, and acting on that.

An animal can have the first without the second. The canine literature suggests precisely that, and the interesting part of this field is the pattern of what is present and what is absent rather than a verdict.

1.2 Why This Is a Different Question From Emotion

The question of whether a dog's internal states are consciously felt and the question of whether a dog has a representation of itself are related but separate. The first asks what accompanies a state; the second asks whether there is a modelled self in the picture at all (emotions in dogs).

The distinction matters because the experiments differ. Emotion research manipulates a condition and measures a response. Self-representation research sets up a situation that can only be solved by using information about oneself, and checks whether the animal solves it. The second design is in some ways the more tractable of the two, because a correct solution is observable in a way that a feeling is not.

1.3 What This Article Covers

This article sets out why the mirror test is the wrong instrument for this species and what its failure does and does not tell us, what the olfactory alternatives have found, the body-awareness experiments that form the strongest line of evidence, what has been shown about memory for one's own actions and about acting on uncertainty, and what the whole pattern amounts to — including where the popular reporting of it goes wrong.

2. The Mirror Test and Why It Is the Wrong Instrument Here

2.1 Where the Test Came From

Gallup's 1970 study established the paradigm. Chimpanzees with mirror experience were anaesthetized, marked with dye at a location they could not see without a mirror, and then observed. They directed exploration at the mark when the mirror was present. The inference was that the animal understood the reflection as itself.

The design is elegant because it requires no verbal report and has a clear pass criterion. It became the standard test for self-recognition across species, and the list of species that pass it is short.

2.2 Dogs Do Not Pass It

Dogs have not passed the mirror self-recognition test. Horowitz states this plainly as the starting point for her own work: domestic dogs have been found skillful at social cognitive tasks and even some metacognitive tasks, and have not passed mirror self-recognition. The metacognitive part of that sentence is itself a finding worth holding onto (metacognition in dogs).

The temptation is to read that as an answer. It is not, for three reasons that all concern the instrument rather than the animal.

2.3 Three Reasons the Negative Result Is Only Partly Informative

The modality is wrong. The test assumes that visual self-inspection is a thing the animal does. For a dog, the primary channel for identity information is olfactory, not visual (canine olfaction, vision in dogs). Testing dogs primarily through visual self-recognition may underestimate their abilities because the task relies on a sensory modality that is not dominant in canine social and individual recognition.

The motivation is wrong. The mark test works because a chimpanzee finds an unexpected mark on its own face interesting enough to investigate. Nothing establishes that a dog cares about a mark on its head, and what a dog attends to is not what a primate attends to (how dogs sense their world). Failure to investigate is failure to be motivated, which is not failure to recognize.

The criterion is all-or-nothing. A pass requires a specific behavior toward a specific stimulus. Self-representation, if it is modular, might be present in a form the test cannot detect — which is the argument the body-awareness researchers make explicitly.

A negative result from a test the subject has no reason to engage with is weak evidence. This is a general point about comparative cognition and worth holding beyond this topic: the absence of a behavior in a task designed around another species tells you about the task (cognitive abilities in dogs).

3. The Olfactory Route

3.1 One Dog and Some Yellow Snow

The first attempt to examine the question in a sensory modality that matters particularly to dogs came from an observational study. Bekoff spent five winters moving urine-marked snow from place to place and recording how a single dog, his own, responded to his own urine compared with that of other males and females. He tracked sniffing duration and marking frequency.

It is an N of one, conducted by an owner on his own animal, and the author presents it as a pilot. Its value is that it asked the right question in the right sensory channel two decades before anyone built a controlled version, and it is a useful reminder of how far a single case can and cannot carry an argument (operationalizing behavior for measurement).

3.2 The Olfactory Mirror

Horowitz built that controlled version. The logic is a direct translation of the mark test: if a dog processes its own odor as self-referential information, then modifying that odor should produce the kind of extra interest that a mark on the face produces in a chimpanzee.

Dogs were presented with canisters holding odor stimuli, and investigation time was recorded. The critical comparison was the dog's own urine against its own urine with an additional odor accompanying it. Thirty-six dogs took part in the main study and twelve in a follow-up.

Three results, in Horowitz's own terms. Dogs investigated their own odor for longer when it had an additional odor with it than when it did not. Dogs spent longer investigating the odor of other dogs than their own odor, which is what ecological validity would predict. And in the second experiment, subjects spent longer with the modified stimulus than with the added odor presented by itself — which rules out novelty alone as the explanation.

3.3 What That Establishes, in the Author's Own Wording

Horowitz's conclusion is carefully phrased and worth quoting for its restraint: such behavior implies a recognition of the odour as being of or from "themselves."

That is a claim about the dog treating an odor as referring to itself. It is not a claim that the dog has a concept of self, an autobiographical identity, or an inner observer. The paper's stated achievement is more modest and more useful: it translates the mirror paradigm for a species whose primary modality is olfaction, and shows that natural sniffing behavior can be brought into the lab.

3.4 The Gap Between the Finding and the Headlines

This is the point in the topic where the popular coverage and the literature separate most visibly. Sniff-test results have been reported as dogs having been shown to possess self-awareness, or a conscience, or self-consciousness. The underlying finding is a difference in how long a dog sniffs a canister.

Two things are worth separating. Differential investigation of a modified self-odor is a real, replicated, ecologically sensible result, and it is better evidence about dogs than a failed mirror test. And it does not establish self-consciousness, because nothing in the design distinguishes "this smell is mine" from "this smell is me". The first requires only that the animal has learned its own scent signature. The second is the interesting claim, and the paradigm does not reach it.

Anyone citing the sniff test should be able to say which of those two they mean.

4. Body Awareness: The Strongest Line of Evidence

4.1 The Body as an Obstacle

The most convincing canine work on self-representation abandons recognition altogether and asks something more basic: does a dog use information about its own body when solving a physical problem?

Lenkei, Faragó, Zsilák, and Pongrácz built a task in which the dog's own body is the obstacle. Thirty-two dogs completed all conditions, out of fifty-four initially tested. The dog stood on a small mat and had to pick up an object and give it to its owner. In the test condition the object was attached to the mat — so lifting it was impossible while the dog's own weight held the mat down. The dog had to get off the mat.

The control conditions are what make the study work. In the main control the object was attached to the ground, so leaving the mat would not help. A further condition had the experimenter tug the mat, to rule out startle. A fourth had the object freely available, to confirm the dogs could do the task at all.

Dogs left the mat more often and sooner in the test condition than in the main control: 27 of 32 in the test, 14 of 32 in the control. The authors describe this as the first convincing evidence of body awareness through understanding the consequence of one's own actions in a species where no higher-order self-representation capacity had previously been found.

4.2 Knowing How Big You Are

A second line asks whether dogs use information about their own dimensions when selecting a route.

Lenkei, Faragó, Kovács, Zsilák, and Pongrácz addressed this directly. Pongrácz, Dobos, Faragó, Kubinyi, and Lenkei then extended it to strategy choice with sixty-eight family dogs between fifty and sixty-five centimetres at the withers. Dogs faced a transparent fence with two ways to a treat: through a doorway, two metres away, or around the fence, seven metres. The door was either too small at twenty-one by twenty-one centimetres, adequate at sixty-eight by sixty-eight, or closed.

The dogs preferred the large door over the detour and chose the detour when the small door would not have worked — a decision about route rather than a reaction to an obstacle (planning and inhibitory control). The authors conclude that dogs relied on body-size awareness when both an open door and the detour were available, and describe this as the first evidence that dogs use body-size representation when choosing among several problem-solving strategies.

One incidental finding is worth noting for its own sake: head shape, measured as cephalic index, influenced the decision pattern, with shorter-headed dogs approaching the openings more often — which the authors relate to differences in retinal structure and visual field (brachycephaly).

4.3 Not a Breeding Artifact

A reasonable objection to all of this is that the dogs tested were purebred animals selected for working tasks, so what looks like self-representation might be a breed-specific skill.

Dobos and Pongrácz tested that by running the paradigm exclusively on mixed-breed dogs: forty-five adults, twelve excluded from analysis, at a three-metre transparent fence with adjustable openings. The dogs preferred the shortcut and either hesitated or took the detour when only the small door was open. A demonstration by a human produced a modest effect — dogs that watched longer in one trial chose the door less often in the next — but it was small next to their reliance on their own body assessment.

Their results indicate that body awareness is not restricted to particular breed groups. For a corpus that also contains the finding that breed explains about a tenth of individual behavioral variation, that is a consistent result rather than a surprising one (the genetics of dog behavior).

4.4 Experience Shapes It

The most recent addition asks whether experience influences aspects of body awareness, which is the question a practitioner would ask first.

Starić, Arnauer, Marshall-Pescini, and Range compared agility-trained with untrained pet dogs: fifty-nine dogs in total, forty-one completing both experiments, of which twenty-two were agility-trained and nineteen untrained. In the first experiment, openings in a fence varied in size and in shape — horizontal, vertical, circular. In the second, openings were covered with materials of differing solidity: a plastic plant, paper, and wood.

Both groups assessed size accurately, preferring larger openings, and both groups handled solidity, choosing the softer barriers more often. Shape was where the groups diverged: results were less conclusive overall, and the agility-trained dogs discriminated shapes better. Training also changed speed in opposite directions depending on the task, accelerating size-based choices and slowing solidity-based ones.

The authors' conclusion is that life experience may influence aspects of body awareness in dogs. That is a modest claim from a modest sample, and it provides first evidence that some parts of this capacity are not entirely fixed (behavioral flexibility).

4.5 Why the Researchers Are Careful

The body-awareness group is notably disciplined about what its results mean, and the discipline is instructive.

They describe body awareness as a neuro-ontogenetic precursor to higher-order self-representation — a building block, not the finished thing. They explicitly do not claim that dogs possess the higher-order self-representation associated with passing the mark test, and they do not claim a unified self-awareness comparable to that attributed to primates or elephants. What they argue for is a bottom-up, ecologically valid approach to a self-representation that is modularly constructed.

That framing is the single most useful thing to take from this literature. The question is not whether dogs have "it". The question is which components are present, and body awareness is the best studied and best evidenced of them.

5. Memory for One's Own Actions

5.1 Recalling What You Did, Not What Happened

A third component concerns the past. Remembering that a door opened is one thing; remembering that I opened it is another, because the second requires the action to have been encoded as one's own.

Fugazza, Pongrácz, Pogány, Lenkei, and Miklósi tested this in ten dogs using the "Do as I Do" method, in which a dog is trained to reproduce an action a human has just demonstrated. The design then turns that around: instead of asking the dog to copy a human, it asks the dog to reproduce its own preceding action, after a delay. The delays tested were zero seconds, twenty seconds, one minute, and one hour.

The authors conclude that dogs can recall their own past actions, and that this indicates a more complex representation of a key feature of the self than had previously been attributed to dogs. They emphasize that the effect appeared when encoding was incidental — when the dog was not being asked to remember — rather than only under formal training conditions.

5.2 What to Hold Onto and What Not

Ten dogs, all trained in a specialized imitation protocol, is a narrow base. These are not typical pet dogs; "Do as I Do" training is uncommon and the dogs who have it are a selected group.

What the result does establish is that information about a dog's own previous action was available for retrieval. Something about that action was stored in a form that could be accessed an hour later without the dog having been instructed to store it. That is the same general picture as the work on what dogs retain from single episodes (episodic memory in dogs), and it is a component of self-representation rather than the whole of it.

6. Knowing What You Do Not Know

6.1 The Checking Paradigm

Metacognition is the component closest to what most people mean by self-awareness: access to the state of one's own knowledge. The standard test does not ask the animal to report anything. It creates uncertainty and looks for information-seeking.

Belger and Bräuer ran three consecutive experiments — with forty-eight, twenty-four, and twenty-five dogs — using two V-shaped fences with a gap at the point of the V. A reward was hidden behind one. The geometry allowed the dog either to commit by walking around a fence, or to gather more information first by checking through the gap. The key manipulation was whether the dog had seen the hiding or whether a curtain had blocked the view.

6.2 A Partial Positive

Dogs checked through the gap more often when they had not seen where the reward was hidden. That is the core result, and it is the behavior the paradigm was built to detect: adjusting information-seeking to one's own ignorance.

Several other predictions failed. Dogs did not check more often as the delay between hiding and choosing grew longer — they did become slightly less successful, so they were forgetting, but they did not compensate by checking. Reward quality did not affect checking either, although dogs were faster to fetch a high-quality food reward. Dogs rewarded with toys chose correctly more often than dogs rewarded with food.

The authors' summary is that dogs, like apes, seek additional information in uncertain situations, but that their behavior in uncertain situations is less flexible than that of great apes. They also note that the checking behavior was not learned during the experiment (metacognition in dogs).

6.3 Why the Failures Matter as Much as the Success

A capacity that responds to one manipulation and not to two others is a capacity with a specific shape, and the shape is the finding.

Dogs behaved differently depending on whether they had seen the hiding event. They did not track how long ago they had seen it, or how much the answer was worth. That pattern fits an animal with a coarse signal about the state of its own information rather than a graded model of its own reliability — which is exactly the sort of detail that gets lost when this is summarized as "dogs know when they don't know."

7. Self and Other

7.1 The Boundary the Experiments Keep Running Into

Two of the studies above are, on closer reading, about the line between self and other rather than about the self alone, and that is not incidental.

The action-memory design depends on it. A dog trained in "Do as I Do" has learned to map an action performed by a human body onto its own, structurally different body — a human standing and turning in a circle is not the same movement as a dog doing it. Turning that around to ask the dog to reproduce its own preceding action only makes sense if the dog was keeping the two apart in the first place (social learning in dogs).

The mixed-breed body-awareness study pits the two against each other directly. Dogs could either assess their own dimensions or copy what a human had just demonstrated, and self-assessment dominated: the demonstration produced only a modest effect. When a dog can solve a problem either by reference to itself or by reference to someone else, it reached first for itself.

7.2 Why That Cuts Against the Easy Story

Dogs are usually described as the species that outsources cognition to humans — reading gestures, checking faces, looking to a person when stuck (how dogs read human gestures). Much of that is well supported.

The body-awareness results sit awkwardly beside it, and usefully so. On a physical problem about the dog's own dimensions, a human demonstration was the weaker source of information. Which suggests the human-oriented picture is domain-specific rather than general: dogs look outward for information about what is going on socially, and inward for information about their own body.

7.3 What Is Still Missing Here

The component that would tie this together has not been established: whether a dog can treat another individual's viewpoint as different from its own in a way that requires modeling both. Copying, following and imitating can each be achieved without that (overimitation in dogs), and no design in the self-representation literature tests it.

8. What the Pattern Adds Up To

8.1 Modular, Not Unitary

Set the results side by side and a coherent picture emerges that is not the picture either camp in the popular argument expects.

Dogs use a representation of their own body to solve physical problems, reliably, across several independent designs and both purebred and mixed-breed samples. They treat their own odor as referring to themselves. They can retrieve what they themselves did an hour earlier. They seek information when they have not seen something. And they do not pass the mirror mark test.

That combination is only paradoxical if self-awareness is one thing. Treated as a set of separable components, it says: the components tied to acting in a physical and olfactory world are present and measurable; the component that requires identifying a visual representation of oneself is either absent or untestable with the instrument available.

8.2 What Is Established

Body awareness in the sense of using one's own physical extent as information: supported by converging evidence from several paradigms and laboratories, with additional evidence suggesting that some aspects of it are influenced by experience.

Differential treatment of one's own odor: established, with controls that rule out the obvious alternatives.

Retrieval of one's own recent actions: shown, in a small and specially trained sample.

Information-seeking under uncertainty: shown, in a narrow form, with several related predictions failing.

8.3 What Is Not Established

That dogs have a concept of themselves as an entity persisting through time. That they can take their own perspective as an object of thought. That any of the components above is consciously experienced. And that the olfactory results demonstrate self-consciousness, which is the claim most often attached to them.

On the consciousness question this article ends where the emotion literature does: one position holds that it cannot be demonstrated for any species, including our own; another holds that dogs do consciously experience their states and the useful question is which; and a third treats the identification of consciousness markers as a research program rather than a closed door. Nothing in the self-representation work settles it, and the body-awareness researchers do not claim it does.

9. What Follows for Practice

9.1 Body-Awareness Work Is Worth Doing

This is the concrete payoff, stated at the strength the evidence allows. Agility-trained dogs discriminated shapes of openings better than untrained dogs, and the authors take life experience to be a plausible contributor. That makes deliberate body-awareness work a defensible training goal rather than a fashion — not a demonstrated intervention effect.

What follows is unremarkable in method and better founded than it was: work over, under, through and onto varied surfaces and gaps; let the dog solve rather than lure it through; vary the dimensions so the dog has to assess rather than repeat. The claim being made here is modest — one study, fifty-nine dogs, correlational with respect to training history — but it points the same way as the rest of the literature.

9.2 A Dog That Misjudges a Gap Is Not Being Careless

Dogs assess size accurately in these studies and shape less well, and shape improves with experience. A dog that clears a jump cleanly but catches itself on a narrow vertical gap is showing exactly the profile the research describes, not a lapse of attention.

The same applies to dogs that misjudge after a body change — weight gain or loss, a cast, a coat, a harness with protruding parts. If the representation is built from experience, it needs updating when the body changes, and that update is training rather than something the dog simply knows.

9.3 Checking Behavior Is Information

A dog that hesitates and gathers information before committing is doing something the metacognition literature treats as a positive finding. In a training context the same behavior is often read as distraction or as the dog being unsure of the cue.

The distinction worth making in the moment is between a dog that is checking — orienting, sampling, then acting — and a dog that is stuck. The first is a functioning process and interrupting it is counterproductive. The second is a signal that the task is above threshold (behavioral assessment).

9.4 Do Not Build a Training Rationale on Self-Awareness

Claims that a dog "knows what it did", "understands that it was wrong", or "is aware of its role" are not supported by anything above, and two of them run into the guilt literature directly (the guilty look).

Memory for one's own action, demonstrated in ten specially trained dogs, is not the same as evaluating that action against a standard. Nothing in this field licenses a method that depends on the dog appraising its own behavior morally.

9.5 Where It Starts, and Where It Gets Rebuilt

If body awareness is built from experience rather than issued at birth, then the periods in which a dog's body is changing fastest are the periods in which the representation is being built and rebuilt. Puppies grow into a body they have not had before, and adolescents do it again at a different rate (the sensitive period, adolescence).

Nobody has traced that trajectory, so what follows is inference rather than finding. But it makes the ordinary developmental advice more concrete: varied surfaces, heights, gaps and footing during growth are not merely enrichment, they are the input a body model is built from (the neurobiology of play).

The same logic applies at the other end. An older dog whose proprioception is declining, or a dog recovering from injury, is working with a model that no longer matches the body, and the mismatch is a plausible contributor to clumsiness that gets read as carelessness (exercise and behavior).

9.6 What You Can Say to an Owner

Accurately, and without overclaiming: a dog has a working sense of its own body and uses it to decide how to get somewhere; it can tell its own scent from another dog's; it can remember what it just did; and it will go and look when it does not know something. What it does not have, as far as anyone can show, is a picture of itself as a person has one.

That is both more interesting and more useful than either "dogs are self-aware" or "dogs just react" (behavior is not emotion).

10. Summary at a Glance

Self-awareness is at least four capacities, not one. Body awareness, self-recognition, memory of one's own actions, and metacognition can come apart — and in dogs they do.

Dogs fail the mirror test, and that is largely uninformative. The instrument assumes visual self-inspection, assumes the animal cares about a mark on its face, and has an all-or-nothing criterion.

The olfactory version was built for the species. Thirty-six dogs investigated their own odor longer when it had been modified, spent longer on other dogs' odor than on their own, and a second experiment with twelve dogs ruled out novelty as the explanation.

The author's own claim is narrow. The behavior implies recognition of the odor as being of or from the dog itself — not a concept of self.

Body awareness is the strongest line. With the body as obstacle, 27 of 32 dogs left the mat when it was necessary against 14 of 32 in the control condition, with three control conditions ruling out startle and incapacity.

Dogs choose strategies by body size. Sixty-eight dogs preferred a large door to a seven-metre detour and switched to the detour when the door was too small at 21 by 21 centimetres.

It is not a breed artifact. Forty-five mixed-breed dogs showed the same reliance on body assessment, with a demonstration by a human having only a modest effect.

Experience may play a part. Among fifty-nine dogs, agility-trained animals discriminated the shape of openings better; both groups judged size and material solidity accurately.

Dogs can retrieve their own recent actions. Ten dogs trained in an imitation protocol reproduced their own preceding action after delays up to an hour, including when encoding was incidental.

Metacognition is present in a narrow form. Dogs checked more when they had not seen the hiding, but did not check more as delays grew or as reward value rose, and were less flexible than great apes.

Self beat other in a head-to-head. Given the choice between assessing their own dimensions and copying a human demonstration, mixed-breed dogs relied mainly on their own body assessment.

The researchers call body awareness a precursor. They explicitly do not claim higher-order self-representation or a unified self-awareness comparable to primates.

11. Research Gaps

No one has tested whether the components correlate. Body awareness, odor self-recognition, action memory and uncertainty-checking have each been examined in separate samples. Whether a dog strong on one is strong on another is unknown, and that is the question that would show whether "self-representation" is one thing in dogs at all.

The olfactory paradigm cannot distinguish "mine" from "me". Differential sniffing of a modified self-odor is compatible with a learned scent signature and no self-concept whatever. A design that separates the two has not been proposed.

Samples are small and often selected. Thirty-two, forty-five, fifty-nine, ten dogs. The action-memory work rests on animals trained in a rare protocol.

Development is unmapped. If body awareness is built from experience, it must have a trajectory across puppyhood and adolescence, and nobody has traced it.

The training evidence is correlational. Agility-trained dogs differ from untrained dogs, and agility-trained dogs also differ in who chose to do agility with them. A controlled training study has not been done.

Body change has not been studied. Whether and how quickly a dog updates its body representation after weight change, injury, or wearing equipment is an obvious question with practical consequences, and it appears to be open.

12. Conclusion

The honest answer to "is my dog self-aware?" is that the question needs splitting before it can be answered, and that once split it has four answers rather than one.

Dogs demonstrably use information about their own body to solve physical problems. That model is good enough to pick between a shortcut and a detour, it works in mixed-breed dogs with no working selection behind them, and agility-trained dogs judged the shape of openings better than untrained ones. Dogs treat their own scent as theirs. They can retrieve what they did an hour ago. They look for information when they lack it, in a coarse rather than graded way.

They do not pass the test that the term "self-aware" is popularly attached to — and that test was built for a visual primate that cares about its own face.

The researchers who produced the strongest evidence here describe what they found as a precursor to higher-order self-representation and decline to claim more. That restraint is the right model for talking about it. The finding that a dog takes its own body extent into account when deciding does not need inflating into consciousness to be worth knowing; it is directly useful, it was hard to demonstrate, and it is trainable.

Key Insights

Split the question first. Body awareness, self-recognition, action memory and metacognition are separate capacities, and dogs score differently on each.

The failed mirror test says more about the mirror than about the dog. Wrong modality, no motivation, all-or-nothing criterion.

The sniff test shows "this scent is mine", not "this scent is me". Anyone citing it should know which claim they are making.

Body awareness is the best-evidenced component. Several paradigms, several labs, purebred and mixed-breed samples, all converging.

Your dog knows roughly how big it is — and judges shape worse than size. Which is why narrow vertical gaps catch out dogs that clear jumps cleanly.

It may be trainable. Agility-trained dogs discriminated opening shapes better, which the authors relate to life experience — a correlational finding, not a controlled training study.

A changed body may need a re-learned representation. Weight change, a cast, a bulky harness: if experience feeds the model, a changed body is a reason to rebuild it. Untested, but it follows.

Checking is not dithering. A dog that gathers information before committing is doing the thing the metacognition studies were designed to detect.

Dogs track whether they know, not how well they know. They checked more when they had not seen — and not more as time passed or stakes rose.

Dogs look outward for social information and inward for their own body. A human demonstration was the weaker source when the problem was about the dog's own dimensions.

Remembering an action is not judging it. Nothing here supports a method that relies on the dog evaluating its own behavior.

References

Bekoff, M. (2001). Observations of scent-marking and discriminating self from others by a domestic dog (Canis familiaris): Tales of displaced yellow snow. Behavioural Processes, 55(2), 75–79. https://doi.org/10.1016/S0376-6357(01)00142-5

Belger, J., & Bräuer, J. (2018). Metacognition in dogs: Do dogs know they could be wrong? Learning & Behavior, 46(3), 398–413. https://doi.org/10.3758/s13420-018-0367-5

Dobos, P., & Pongrácz, P. (2025). Body awareness does not need a pedigree: Mixed-breed dogs rely more on self-representation than social learning in a spatial task. Animals, 15(3), 432. https://doi.org/10.3390/ani15030432

Fugazza, C., Pongrácz, P., Pogány, Á., Lenkei, R., & Miklósi, Á. (2020). Mental representation and episodic-like memory of own actions in dogs. Scientific Reports, 10, 10449. https://doi.org/10.1038/s41598-020-67302-0

Gallup, G. G. (1970). Chimpanzees: Self-recognition. Science, 167(3914), 86–87. https://doi.org/10.1126/science.167.3914.86

Horowitz, A. (2017). Smelling themselves: Dogs investigate their own odours longer when modified in an ‘olfactory mirror’ test. Behavioural Processes, 143, 17–24. https://doi.org/10.1016/j.beproc.2017.08.001

Lenkei, R., Faragó, T., Kovács, D., Zsilák, B., & Pongrácz, P. (2020). That dog won’t fit: Body size awareness in dogs. Animal Cognition, 23, 337–350. https://doi.org/10.1007/s10071-019-01337-3

Lenkei, R., Faragó, T., Zsilák, B., & Pongrácz, P. (2021). Dogs (Canis familiaris) recognize their own body as a physical obstacle. Scientific Reports, 11, 2761. https://doi.org/10.1038/s41598-021-82309-x

Pongrácz, P., Dobos, P., Faragó, T., Kubinyi, E., & Lenkei, R. (2023). Body size awareness matters when dogs decide whether to detour an obstacle or opt for a shortcut. Scientific Reports, 13, 17899. https://doi.org/10.1038/s41598-023-45241-w

Starić, D., Arnauer, L., Marshall-Pescini, S., & Range, F. (2026). Influence of agility training on body-size and object solidity perception in pet dogs. PLOS ONE, 21(1), e0338647. https://doi.org/10.1371/journal.pone.0338647