Jealousy or Injustice? Exploring the Evolution of 'Fairness' in Canines

1. Introduction

1.1 The Familiar Scene and the Real Question

The "jealous dog" is one of the most confidently asserted claims in everyday dog talk, and it is worth taking seriously precisely because it is so intuitive. The scientific task is to ask what is actually happening when a dog reacts to a housemate's treat: a social comparison, a felt injustice, or simple frustration at missing out. These are different psychological processes with different implications, and telling them apart is the whole point (attributing a rich emotion from the outside is exactly what to be cautious about).

1.2 How to Read the Evidence

Two cautions frame the field. First, the interpretation runs along a spectrum from rich ("a sense of fairness") to lean ("socially facilitated frustration"), and the same behavior can support either reading, so the leaner account is preferred until the richer one is earned. Second, the literature is genuinely mixed: some paradigms elicit the effect robustly, others fail to, and part of the science is figuring out which conditions produce it (the general problem of measuring behavior reliably). This article flags where the evidence is solid and where it is contested.

1.3 Why the Word Fairness Does the Damage

Fairness in human usage implies a norm, a judgment against it, and a sense that something ought to be otherwise. None of that is measured in any study described here.

The term used in the research is inequity aversion, and it is deliberately narrower: a negative response to an unequal distribution of outcomes. It carries no claim about norms, judgment or moral evaluation, and everything in this article should be read in that sense. "Fairness" appears here as the popular label for the phenomenon, not as a translation of the human concept.

What is measured is that a dog stops working sooner under one condition than another. The gap between that observation and the word attached to it is the whole subject of this article.

1.4 Why the Question Is Still Worth Asking

Dismissing the topic as anthropomorphism would be as much of an error as accepting the strong reading. Something happens in some versions of this paradigm, it appears in wolves as well as dogs, and it is not explained by any single obvious mechanism, even though pooled data across accept/reject tasks do not support it as a general effect (see 5.2).

An unresolved effect is more interesting than a settled one, and the honest article says so rather than picking a side.

2. The Seminal Discovery: Range et al. (2009)

The scientific study of fairness in dogs began in earnest with a 2009 study by Friederike Range and colleagues at the University of Vienna, published in PNAS. The design was elegantly simple: pairs of dogs were asked to give a paw under different reward conditions (Range et al., 2009).

2.1 The Paw-Test Paradigm

Two dogs sat side by side and were asked alternately to give a paw, while the experimenter varied the reward: both dogs rewarded, neither rewarded, or one rewarded while the other performed the same action for nothing. The results were striking. With both rewarded, cooperation continued; with neither rewarded, both eventually stopped, but gradually. In the unequal condition, the unrewarded dogs stopped much faster and showed stress signals — scratching, yawning, lip-licking, and gaze avoidance.

2.2 What the Dogs Were — and Were Not — Responding To

The authors read this as more than simple frustration at no food. Tested alone, without a rewarded partner, dogs persisted far longer when reward ceased; the presence of a rewarded partner sharply accelerated refusal, which they took to imply a social comparison rather than mere disappointment, although chapter 3 sets out why the design leaves the leaner reading open (a violated expectation in a social context). But the limits were just as telling. Dogs were not sensitive to reward quality — one getting sausage and the other bread provoked no refusal — and they did not object when a partner got food "for free" while they had to work, as long as both got something. Capuchins in the founding primate study did respond to differences in reward quality (Brosnan & de Waal, 2003), although the primate data have been contested (Bräuer, Call & Tomasello, 2009). Whatever dogs have, it is a narrower thing than the primate version.

2.3 Why the Original Design Was Clever

Using an action the dogs already performed willingly — giving a paw — removed the need for training within the experiment and gave a clean measure: how many times the dog complies before stopping.

Simple measures are a strength. They are also what makes the result compatible with several explanations, because a single number carries no information about why the dog stopped. Later studies added stress signals and gaze, which help and do not settle the question either.

2.4 What the Reward Quality Result Adds

Dogs respond more strongly to receiving nothing than to receiving something of lower value (Range et al., 2009), which distinguishes them from the capuchin result, where quality differences produced refusals (Brosnan & de Waal, 2003).

That is awkward for the fairness reading. A dog comparing its treatment against another's should notice a worse reward; a dog responding to its own reward rate dropping to zero should not (where frustration is treated on its own terms). The founding capuchin study reported the opposite pattern, although the primate data are themselves contested (see 7.2).

3. What the Paradigm Can and Cannot Isolate

3.1 The Design in One Sentence

Two dogs, the same requested action, one rewarded and one not, and the measure is how long the unrewarded dog keeps working. Everything in this article rests on that arrangement (Range et al., 2009).

3.2 What Changes Between Conditions

When the partner is rewarded and the subject is not, several things differ at once. The subject's own reward rate has dropped. A conspecific is eating nearby. The situation now predicts something different from what it predicted a moment ago.

A fourth belongs on the list and is easy to leave off: something interesting is happening next to the dog. A partner being handed food draws attention on its own, and attention directed elsewhere changes what the subject does without any comparison being involved. Any of those could produce the observed refusal, and the design does not separate them.

3.3 The Frustration Reading

A dog that has been rewarded and then is not is in extinction, and extinction produces exactly this: refusal, latency, disengagement. The presence of a partner receiving food adds salience without requiring any comparison (where what was learned turns out not to be erased).

Explaining the result this way needs nothing beyond ordinary learning, which is why it remains the default alternative the richer account has to displace. Preferring it is a working rule about explanations rather than a claim that dogs are incapable of comparison.

3.4 What Would Have to Be Shown

For social comparison specifically, the subject's behavior would have to depend on what the partner receives rather than only on what the subject receives. Conditions in which the subject's own reward is held constant while the partner's varies are the informative ones.

Where such comparisons have been run, the picture is mixed rather than clean, which is the honest description of this literature (McGetrick & Range, 2018). A mixed picture after more than fifteen years is itself a finding about how hard the question is, and the review is a better guide to the state of play than any single study on either side.

3.5 The Control Nobody Can Run

The decisive control would present the same drop in reward rate with no partner present and no other change. Achieving that is harder than it sounds, because removing the partner also removes a source of movement, sound and attention that affects the subject independently.

Every version of this control alters more than one variable, which is why the debate has not closed. Designing a partner that is present but not socially salient is close to a contradiction in terms, which is why the machine-dispenser version described later is the more promising route.

4. Interpreting the Findings: Jealousy, Fairness, or Frustration?

Range's findings drew headlines about canine fairness and jealousy; scientists urged more caution.

4.1 The Case for Social Comparison

The strongest point for a "fairness-like" reading is social specificity: dogs balked only when a partner was rewarded and they were not, not when merely unrewarded alone. It has been suggested that the sensitivity might extend beyond food to other valued resources, such as human attention, although this has not been tested systematically. Evolutionarily, the biologist Marc Bekoff has framed inequity sensitivity as adaptive for cooperative canids, whose group success can depend on each individual pulling its weight — so intolerance of unfairness helps sustain cooperation (Bekoff, 2004).

4.2 The Case for Frustration and Social Facilitation of Extinction

Critics offered a leaner account. Alexandra Horowitz raised a key methodological point: in the original paw test, dogs were always rewarded on the first trial to avoid total frustration, which made the subsequent no-reward phase an extinction condition (Horowitz, 2012). A rewarded partner may simply have made the absence of reward more salient, speeding extinction through social facilitation rather than any grasp of fairness. Horowitz's own study had dogs watch a fair versus an unfair trainer and then choose whom to approach — and dogs preferred the over-rewarding trainer, showing no aversion to the one who had been generous to a control dog, and no preference against the under-rewarding one. Her conclusion: when stakes were high, dogs tracked reward quantity more than fairness — suggesting the response is about maximizing personal gain, closer to frustration than to justice.

4.3 The Question of Conceptual Understanding

Underlying the debate is whether dogs are even capable of a fairness concept. To match primates, a dog would have to compare its own effort and reward with a partner's and then make a meta-comparison — a cognitively demanding operation (the kind of higher-order comparison dogs are not clearly capable of) reported mainly in humans and, more contentiously, some primates. The parsimonious alternative is that dogs hold expectations built from experience and are distressed when a visible social partner receives what they do not — better described as socially facilitated disappointment than as a sense of justice.

4.4 Why Both Readings Survive

The social comparison account and the frustration account predict the same behavior in the standard paradigm, which is why more than fifteen years of work has not chosen between them. That is not a failure of the researchers; it is a property of the design, and the researchers say so themselves in the review literature.

Progress requires conditions where the two predict different outcomes, and those are harder to build than they look.

4.5 The Conceptual Question Is Separate

Even if dogs do respond to a partner's outcomes, that would not establish that they hold a concept of fairness. Sensitivity to a difference and judgment about a difference are different capacities, and no animal paradigm tests the second.

Comparative psychology generally calls the first a precursor, which is a deliberately modest term. Precursor means it might be part of what a fuller capacity is built from, not that a smaller version of that capacity is present.

5. Refining the Paradigm: Context, Replication, and Causality

Since 2009, researchers have worked to pin down exactly which conditions elicit the effect — and whether it holds up.

5.1 The Role of Shared Resources

One study addressed a discrepancy: an earlier attempt using separate food bowls had struggled to reproduce the effect, prompting the idea that a single shared food source was necessary. Testing this directly with separate bowls, researchers still found the typical inequity response, showing that a shared source is not required and showing that the effect, where it appears, does not depend on a shared source (McGetrick et al., 2019).

5.2 Methodological Variability and the Replication Problem

Other work complicates the picture. Using a novel task requiring no prior training — dogs simply approached or declined food allocations — researchers found no inequity aversion in either dogs or dingoes (McAuliffe, 2021). Together with earlier mixed replications, this suggests the effect is context-dependent and may specifically require trained behaviors and performance expectations, rather than being a general response to any unequal allocation. This inconsistency is central: inequity aversion in dogs has been reported under some paradigms and not under others, and a comprehensive review concluded that the details of the task matter a great deal (McGetrick & Range, 2018).

In 2024 that inconsistency was given a verdict. A meta-analysis pooled the primary data from 23 studies using accept/reject tasks — 60,430 observations across 18 species — and found no evidence of inequity aversion in non-human animals within that task family (Ritov et al., 2024). This is the largest empirical test of the phenomenon to date, and it does not soften the picture: the effect that individual studies report does not survive aggregation.

Two things are worth keeping straight here. The analysis covers accept/reject paradigms specifically, which is the family the dog work largely belongs to — a dog withdrawing its paw is declining an offered distribution. And "no evidence" is not the same as "disproven": the finding is that the pooled data do not support the effect, not that a sense of fairness has been ruled out by some other route. But for the claim as it is usually made — that dogs object to unfair treatment, demonstrated by the paw task — this is the strongest counterweight currently available, and it comes from a group that includes the Vienna institute where much of the original dog work was done.

5.3 Individual Differences

Responses also vary between individuals. Horowitz found that ownership length, age, and prior cooperative-work experience predicted a dog's preference for the fair trainer, while breed did not — pointing to experience, especially the history of social interaction with humans (rather than breed, which is a weak predictor of behavior generally), as the shaping force. This fits a broader picture in which social expectations are learned, not fixed (how experience tunes behavioral tendencies).

5.4 Why Methodological Variability Matters Here

Studies differ in the action required, the reward used, the partner's familiarity, the number of trials and the measure reported. With that many degrees of freedom, agreement between studies is weaker evidence than it appears and disagreement is harder to interpret.

A review of the field states the position more carefully than most secondary accounts do (McGetrick & Range, 2018), and reading it is a better guide than any single experiment.

5.5 What Individual Differences Imply

Relationship and motivation influence whether a dog shows the effect (Range, Leitner & Virányi, 2012), which means the response is not a fixed species characteristic.

For a reader that is the practically useful finding: whether a given dog responds to a partner being rewarded depends on that dog and that relationship, not on a general property of dogs.

6. The Evolutionary Context: Cooperation in Canids

Inequity aversion makes most sense placed in an evolutionary frame. Dogs descend from wolves — highly social canids that hunt, raise pups, and defend territory cooperatively — and cooperation among non-kin poses a puzzle: why help others at a cost to oneself?

6.1 Inequity Aversion as a Cooperation Mechanism

One influential theory holds that inequity aversion evolved to sustain cooperation: an individual who consistently receives less from joint effort should withdraw, so a sensitivity to being short-changed — and a willingness to "protest" by refusing to continue — helps stabilize cooperative relationships over time.

6.2 The Wolf Evidence: It Predates Domestication

This theory makes a testable prediction, and the test has now been run. If inequity aversion was created by domestication (bred into dogs to cooperate with humans), wolves should lack it; if it is an older cooperative adaptation, wolves should share it. Testing equally raised wolves and dogs, researchers found that both refused to keep working when a partner was rewarded and they were not — and the wolves' response was at least as strong as the dogs' (Essler et al., 2017). That suggests the sensitivity may predate domestication rather than being a product of it. Common inheritance is one explanation; similar social ecology, similar motivational systems and similar learning mechanisms would produce the same pattern without it. This is the keystone of the evolutionary argument, and what it supports is narrower than it is often made to sound: that the response is not a product of domestication, not that it is a deep cooperative legacy, which a comparison of hand-raised animals cannot show.

6.3 The Human Factor

Domestication did, however, position dogs to cooperate with a different species. Millennia of working alongside people — herding, hunting, guarding, assisting — may have amplified or refined their sensitivity to social contingencies, including expectations about how humans dole out rewards. The strong dog–human bond, supported in part by oxytocin (the neurochemistry of the dog–human relationship), creates the very context in which such expectations form.

6.4 Distinguishing Inequity Aversion from Resource Guarding

A clean distinction matters here. A dog growling over a bone or guarding its bowl is not responding to unfairness; it is defending a valuable resource — securing its own access, not comparing outcomes. Competition for a teat in a large litter is partly innate and is about winning access. Refusing to cooperate while watching another be rewarded for the same work is qualitatively different: it is often read as social comparison, however basic, although chapter 3 sets out the leaner reading. Conflating the two — reading ordinary resource competition as "jealousy" — is a common error.

7. Which Findings Come From Which Species

7.1 A Comparative Literature by Design

Unlike most topics in this collection, the comparative material here is the point rather than a substitute for canine data. The question is whether inequity aversion is shared across species, so the primate and canid work belongs in the argument.

7.2 Where It Started

The paradigm entered comparative psychology through capuchin monkeys rejecting unequal pay (Brosnan & de Waal, 2003), and the ape data have been contested since (Bräuer, Call & Tomasello, 2009).

That the founding result has been disputed within its own order is worth knowing before the canine version is treated as settled.

7.3 The Canid Comparisons

Wolves show the effect as well, which argues that it does not depend on domestication (Essler, Marshall-Pescini & Range, 2017). Dingoes have been compared with domestic dogs directly (McAuliffe, 2021), and a shared food source has been shown not to be necessary for the effect (McGetrick et al., 2019).

Those three studies do real work: each removes a candidate explanation without requiring the richer interpretation to be true. That is the pattern of a healthy research program — narrowing the space of explanations rather than accumulating support for a favorite.

7.4 The Pooled Negative Result

A 2024 meta-analysis pooled the primary data from 23 accept/reject studies across 18 species and found no evidence for inequity aversion within that task family (Ritov et al., 2024). Negative results in this area matter disproportionately, because the effect is defined by a difference between conditions and a null means the difference did not appear.

It is not a single negative result but the largest test available, and the dog work belongs largely to the paradigm it covers. It does not show that no dog responds to a partner's reward in any design, and it does not reach paradigms outside the accept/reject family. What it does is shift the burden: after it, the positive canine studies are the ones that need explaining.

7.5 What the Comparative Picture Supports

Where the effect appears, it appears in wolves as well as dogs, and it was first described in capuchins, although pooled accept/reject data do not support it as a general effect. What that something is — sensitivity to a partner's outcomes, or a lower tolerance for one's own reward rate dropping in company — is what the comparative work has not resolved.

Both readings are compatible with every result described here, which is the situation the next chapter addresses. Neither is anthropomorphic and neither is deflationary; they are two ordinary hypotheses that the available designs cannot separate.

8. What Dogs May Actually Understand

8.1 A Precursor to Fairness, Not Fairness Itself

If the effect is real, the most defensible reading is that dogs show a fairness-related sensitivity, possibly a precursor to fairness: a basic behavioral sensitivity to certain inequities, especially when they are the ones losing out. On that reading they notice when a social partner gets what they do not, and that noticing produces negative affect and behavioral change — but it does not require an abstract concept of "fair" versus "unfair." Notably, it has not been shown to extend to advantageous inequity (dogs getting more than a partner), and in the trainer-choice study it did not override the pull of a more generous trainer (Horowitz, 2012).

8.2 Social Expectation, Not Moral Judgment

What dogs appear to run on is a system of learned social expectation: once a pattern is established ("I give paw, I get food"), violations are aversive, especially when another individual keeps getting the expected outcome. This is not a judgment about what the other deserves; it is a distress response to a violated expectation in a social setting — consistent with a cognitive architecture superb at patterns, contingencies, and social cues but not built for abstract reasoning about rights or justice.

8.3 Distress, Not Resentment

The stress signals in these studies — scratching, yawning, lip-licking — indicate genuine emotional arousal; dogs are not indifferent to the situation. But the affect is better described as distress or frustration than as resentment or moral outrage. The distinction sometimes drawn is between "constructive envy" — wanting the same reward — and "destructive envy," wishing the other ill; if anything, dogs' behavior fits the former. What they appear to feel is I want that too, not they should not have it.

9. What Would Settle It

9.1 Vary the Partner's Reward Alone

The most direct test holds the subject's reward constant and varies only what the partner receives. Any change in the subject's behavior under those conditions cannot be attributed to its own reward rate.

Designs of this kind exist and are harder to run than the standard version, which is part of why the standard version dominates. They also require more trials per dog, and dogs stop cooperating with repetitive tasks long before the statistics are satisfied.

9.2 Use a Non-Social Comparison

Replacing the partner with a machine that dispenses food, or with an inanimate event of similar salience, tests whether the social character of the partner matters. If the effect survives without a conspecific, social comparison is not required to explain it.

9.3 Report Individuals, Not Means

Individual variation in this paradigm is substantial, and a group difference is compatible with a minority of dogs driving the whole effect. Reporting the distribution would tell readers whether they are looking at a species-typical response or at a subset of dogs behaving distinctively.

That is a reporting decision rather than a new experiment, and it would improve the literature immediately.

9.4 Pre-Register the Analysis

Given the number of measures available — latency, refusals, stress signals, gaze — and the number of conditions, the scope for analytic flexibility is large. Stating in advance which measure counts would strengthen positive and negative results alike, and would make the recent null result easier to weigh against the earlier positive ones.

None of these four requires new technology, and none has become standard in this field. The first two would need new experiments; the second two are decisions about how existing work is reported.

10. Research Gaps and Methodological Challenges

The honest state of the field is one of a real but fragile effect.

Replication is inconsistent. The effect appears robustly in some paradigms and vanishes in others (McAuliffe, 2021), and it may depend specifically on trained behaviors and performance expectations rather than being general (McGetrick & Range, 2018).

The pooled data do not support the effect. A 2024 meta-analysis of 23 accept/reject studies — 60,430 observations across 18 species — found no evidence of inequity aversion in non-human animals within that task family (Ritov et al., 2024). This does not rule out a sense of fairness by other routes, but it is the largest test available, and the dog work belongs largely to the paradigm it covers.

Rich versus lean interpretation is unresolved. Social comparison, extinction accelerated by social facilitation, and simple reward-maximizing can all produce the observed refusals, and the design details determine which explanation fits.

Individual variation is large. Experience, age, and cooperative history shape responses more than breed does, so group-level findings map loosely onto any individual dog.

Emotion is inferred, not observed. Stress behaviors indicate arousal, but labeling the internal state "jealousy," "injustice," or "frustration" is an interpretation the behavior alone cannot settle.

The paradigm cannot separate the candidate explanations. Reward omission, a conspecific eating nearby and a change in what the situation predicts all occur together, and the standard design does not isolate them.

The founding primate result is disputed. Token-exchange data in apes have been contested (Bräuer et al., 2009), and a 2024 meta-analysis of 23 accept/reject studies across 18 species found no pooled evidence for the effect (Ritov et al., 2024).

Methodological variability is high. Action, reward, partner familiarity, trial number and outcome measure all differ between studies (McGetrick & Range, 2018), which weakens both agreement and disagreement.

Individual data are rarely reported. A group difference is compatible with a minority of dogs producing the whole effect, and the distribution would settle that without a new experiment.

11. Practical Implications for Living with Dogs

11.1 Managing Multiple-Dog Households

Because dogs may react when rewards are distributed unequally, it is worth being mindful of how rewards are distributed among multiple dogs. Perfect equality is neither necessary nor realistic — dogs are not counting individual treats — but a pattern where one dog is consistently rewarded while another does the same work for nothing can produce distress and reduced cooperation in the overlooked dog (and chronic social stress has real costs).

11.2 Beyond Food: Attention and Affection

The suggestion that the sensitivity extends to attention fits owners' experience of "jealousy" when affection goes elsewhere. Though not yet systematically established, the underlying principle — dogs form expectations about social interactions and react when those are violated in a social context — plausibly applies to valued resources including human attention.

11.3 Training Implications

The research reinforces reward-based training that creates clear, consistent contingencies: specific behaviors reliably producing specific outcomes (how reward-based learning builds stable expectations). This is not a mandate to treat every dog identically at every moment, but an argument for awareness that dogs are social observers who register what others receive (and that supporting emotional wellbeing beats suppression).

11.4 What Not to Conclude in a Household

A dog that reacts when another is given attention is not necessarily making a comparison, and describing it as jealousy imports a judgment the evidence does not support. It is also unhelpful, because jealousy suggests the dog should be corrected rather than that the situation should be arranged differently, and correction is the response least likely to change anything here.

The observable is that proximity to a resource plus another dog receiving it produces a response, which is a description a household can act on (where guarding is treated on its own terms).

11.5 Why the Distinction From Guarding Matters

Resource guarding concerns retaining something the dog has. The inequity paradigm concerns not receiving something another is receiving. Both produce tension around resources and they call for different handling.

Confusing them leads to guarding protocols applied to a dog that is not guarding anything, which produces no improvement and a household that concludes the plan failed. The distinguishing question is simple: does the dog have the resource, or is it watching another dog receive it?

12. Summary at a Glance

The finding is a difference in persistence — Dogs stop performing a requested action sooner when a partner is rewarded for the same action and they are not (Range et al., 2009).

The paradigm changes several things at once — The subject's reward rate drops, a conspecific eats nearby, and the situation's predictions change; the design does not separate them.

Frustration explains the same data — Reward omission produces refusal and disengagement without any comparison being made, which is why it remains the default alternative.

The effect does not appear to be a product of domestication — Wolves show it too (Essler et al., 2017), and a shared food source is not necessary (McGetrick et al., 2019).

Quality of reward matters less than presence — Dogs respond to receiving nothing more than to receiving something worse, which distinguishes the canine pattern from the capuchin result (Range et al., 2009; Brosnan & de Waal, 2003).

The founding primate result is itself contested — The token-exchange data in apes have been disputed (Bräuer et al., 2009), and a 2024 meta-analysis of 23 accept/reject studies across 18 species found no pooled evidence for the effect (Ritov et al., 2024).

Individual variation is large — Relationship and motivation influence the response (Range, Leitner & Virányi, 2012), and group means conceal how many dogs show it.

No study has demonstrated moral judgment — What is measured is a change in persistence, and everything beyond that is interpretation.

13. Conclusion

"Do dogs feel jealousy or a sense of fairness?" resists a simple answer, and the honest one, after more than fifteen years of research, is more interesting than either extreme. In the original paw test, dogs stopped cooperating sooner and showed stress signals when a partner was rewarded and they were not (Range et al., 2009), and the effect has been reported in some later paradigms and not in others (McGetrick et al., 2019; McAuliffe, 2021; McGetrick & Range, 2018). Where it appears, it is limited: no sensitivity to reward quality, no demonstrated objection to advantageous inequity, and no override of a more generous trainer (Horowitz, 2012). Wolves show it at least as strongly as dogs, which suggests it is not a product of domestication (Essler et al., 2017). Against all of this stands the largest test available: pooling 23 accept/reject studies across 18 species, a 2024 meta-analysis found no evidence of inequity aversion within that task family (Ritov et al., 2024). What dogs have, then, is not a moral principle, and whether it is a social comparison at all or frustration at a violated expectation in company is what the available designs do not establish. They notice when outcomes differ from their expectations, especially when another benefits and they do not. That noticing matters to them — and, in how we manage and care for them, it should matter to us (much as emotional experience shapes canine behavior more broadly). The next time your dog gives that "offended" look when a housemate gets the treat, you are not seeing spite or moral judgment, but a dog registering that its own expectation has not been met in a social world it actively interprets.

Key Insights (Takeaways)

• In the original paw test, dogs showed a limited inequity response: in the paw test, unrewarded dogs quit faster and showed stress signals when a partner was rewarded, but not when merely unrewarded alone (Range et al., 2009). Unlike the capuchins in the founding primate study, they were insensitive to reward quality and to effort.

• The interpretation is contested. A leaner account holds that a rewarded partner simply accelerates extinction via social facilitation, and dogs actually preferred an over-generous trainer and tracked reward quantity over fairness (Horowitz, 2012) — closer to frustration and self-interest than to justice.

• The effect has been reported under some paradigms and not under others (McAuliffe, 2021; McGetrick & Range, 2018); it seems to depend on trained behaviors and performance expectations. A shared food source is not required (McGetrick et al., 2019), and individual experience, not breed, predicts responses.

• The strongest available test is negative. Pooling the primary data from 23 accept/reject studies across 18 species, a 2024 meta-analysis found no evidence of inequity aversion in non-human animals (Ritov et al., 2024) — which is not the same as disproof, but is the point at which "inconsistent" becomes a finding rather than a puzzle.

• The keystone evolutionary finding: equally raised wolves show inequity aversion at least as strongly as dogs, which suggests the sensitivity may predate domestication rather than being something humans bred in (Essler et al., 2017).

• The best summary is a fairness-related sensitivity rather than fairness itself: if the effect is real, dogs appear to run on learned social expectation, not moral judgment, and their reaction looks like distress ("I want that too"), not resentment ("they shouldn't have it"). Practically, be mindful of lopsided rewards among multiple dogs, and use clear, consistent, reward-based contingencies.

References

Bekoff, M. (2004). Wild justice and fair play: Cooperation, forgiveness, and morality in animals. Biology and Philosophy, 19(4), 489–520. https://doi.org/10.1007/s10539-004-0539-x

Bräuer, J., Call, J., & Tomasello, M. (2009). Are apes inequity averse? New data on the token-exchange paradigm. American Journal of Primatology, 71(2), 175–181. https://doi.org/10.1002/ajp.20639

Brosnan, S. F., & de Waal, F. B. M. (2003). Monkeys reject unequal pay. Nature, 425, 297–299. https://doi.org/10.1038/nature01963

Essler, J. L., Marshall-Pescini, S., & Range, F. (2017). Domestication does not explain the presence of inequity aversion in dogs. Current Biology, 27(12), 1861–1865. https://doi.org/10.1016/j.cub.2017.05.061

Horowitz, A. (2012). Fair is fine, but more is better: Limits to inequity aversion in the domestic dog. Social Justice Research, 25(2), 195–212. https://doi.org/10.1007/s11211-012-0158-7

McAuliffe, K. (2021). A comparative test of inequity aversion in domestic dogs (Canis familiaris) and dingoes (Canis dingo). PLOS ONE, 16(9), e0255885. https://doi.org/10.1371/journal.pone.0255885

McGetrick, J., Ausserwöger, S., Leidinger, I., Attar, C., & Range, F. (2019). A shared food source is not necessary to elicit inequity aversion in dogs. Frontiers in Psychology, 10, 413. https://doi.org/10.3389/fpsyg.2019.00413

McGetrick, J., & Range, F. (2018). Inequity aversion in dogs: A review. Learning & Behavior, 46(4), 479–500. https://doi.org/10.3758/s13420-018-0338-x

Range, F., Horn, L., Virányi, Z., & Huber, L. (2009). The absence of reward induces inequity aversion in dogs. Proceedings of the National Academy of Sciences, 106(1), 340–345. https://doi.org/10.1073/pnas.0810957105

Range, F., Leitner, K., & Virányi, Z. (2012). The influence of the relationship and motivation on inequity aversion in dogs. Social Justice Research, 25(2), 170–194. https://doi.org/10.1007/s11211-012-0155-x

Ritov, O., Völter, C. J., Raihani, N. J., & Engelmann, J. M. (2024). No evidence for inequity aversion in non-human animals: A meta-analysis of accept/reject paradigms. Proceedings of the Royal Society B, 291(2035), 20241452. https://doi.org/10.1098/rspb.2024.1452