Sensitivity to inequity is thought to be an important mechanism for recognizing undesirable cooperative partners and thus crucial for the evolution of human cooperation []. This link may not be unique to humans, as cooperative non-human primates also react to unequal outcomes [], whereas non-cooperative species do not []. Although this hypothesis has not been tested in non-primate species, studies revealed that pet dogs show a limited form of inequity aversion, responding to reward, but not quality inequity []. It has been proposed that this primitive form of inequity aversion was selected for during domestication and thus absent in their ancestors, wolves. Alternatively, wolves, which hunt, raise pups, and defend their territory cooperatively, are similarly inequity averse as non-human primates, or at least to the same degree as pet dogs. Testing similarly raised and kept pack-living dogs and wolves, we found both to be inequity averse when their partner was being rewarded but they were not for performing the same action. Additionally, both wolves and dogs reacted to receiving a lower-quality reward than their partner. These results suggest that the inequity response found in pack-living dogs and wolves is comparable to that observed in non-human primates; results from studies on pet dogs may be confounded by the dogs’ relationship with humans. Consequently, our results suggest that inequity aversion was present already in the common—probably cooperative—ancestor of wolves and dogs and thus support the hypothesis of a close link of cooperation and inequity aversion.

Results

7 Hare B.

Tomasello M. Human-like social skills in dogs?. 8 Hare B.

Wobber V.

Wrangham R. The self-domestication hypothesis: evolution of bonobo psychology is due to selection against aggression. 9 Range F.

Virányi Z. Tracking the evolutionary origins of dog-human cooperation: the “canine cooperation hypothesis.”. 10 Mech L.D. The Wolf: Ecology and Behavior. 11 MacNulty D.R.

Tallian A.

Stahler D.R.

Smith D.W. Influence of group size on the success of wolves hunting bison. 12 Mech L.D.

Boitani L. Wolves: Behavior, Ecology, and Conservation. 3 Brosnan S.F. A hypothesis of the co-evolution of cooperation and responses to inequity. 4 Brucks D.

Essler J.L.

Marshall-Pescini S.

Range F. Inequity aversion negatively affects tolerance and contact-seeking behaviors towards partner and experimenter. 6 Range F.

Horn L.

Virányi Z.

Huber L. The absence of reward induces inequity aversion in dogs. 1 Brosnan S.F.

de Waal F.B.M. Monkeys reject unequal pay. 13 Clutton-Brock J. Origins of the dog: domestication and early history. Multiple hypotheses have been suggested to account for why pet dogs (Canis familiaris) show a form of inequity aversion. First, dogs may have acquired the capacity to respond to inequity due to an enhancement of their socio-cognitive abilities brought about by domestication. More specifically, by evolving alongside humans, dogs are thought to have gained unique abilities to successfully cooperate with both humans and conspecifics []. Based on this hypothesis, dogs’ closest living relatives, wolves (Canis lupus), should not exhibit inequity aversion. Alternatively, the “canine cooperation hypothesis” suggests that rather than dogs gaining these abilities during the course of domestication, they maintained such abilities from their ancestors, and hence they should be shared with wolves []. This hypothesis is supported by the fact that the wolves’ social ecology remains highly dependent on cooperation with conspecifics for activities such as puppy raising, territory defense, and hunting [], and in primates at least, cooperation and inequity aversion appear to be closely linked []. Compared to primates, however, pet dogs show a reduced sensitivity to inequity in that they show inequity aversion when receiving no reward next to a rewarded partner [] but do not respond to differences in reward qualities as some primates do []. This reduced sensitivity to inequity of pet dogs in comparison to non-human primates may be due to (1) canines showing in general a more primitive form of inequity aversion, predicting that wolves also do not react to quality inequity; (2) the domestication process [], which selected for unconditional cooperation with humans and perhaps a change in intraspecific cooperation, predicting that although wolves react to quality inequity, dogs do not, even if they have the same experiences as wolves; or (3) socialization effects, which, due to the close interaction of pet dogs with their human caregivers, lead to ignoring unfavorable conditions to please the humans, predicting that dogs raised in packs with a lot less interactions with humans would, like wolves, react to quality inequity.

4 Brucks D.

Essler J.L.

Marshall-Pescini S.

Range F. Inequity aversion negatively affects tolerance and contact-seeking behaviors towards partner and experimenter. 5 Range F.

Leitner K.

Virányi Z. The influence of the relationship and motivation on inequity aversion in dogs. 6 Range F.

Horn L.

Virányi Z.

Huber L. The absence of reward induces inequity aversion in dogs. 4 Brucks D.

Essler J.L.

Marshall-Pescini S.

Range F. Inequity aversion negatively affects tolerance and contact-seeking behaviors towards partner and experimenter. Figure 1 Set-Up View from One Enclosure Show full caption Fencing between the two enclosures allowed for each individual to fully see their partner’s behavior. The same setup was used for the dogs (left) and the wolves (right). See also Figures S1 and S2 Table 1 Summary of Conditions Condition Subject Partner Social Conditions Equity test (ET; baseline) LVR LVR Quality inequity (QI) LVR HVR Reward inequity (RI) no reward HVR Food control (FC) HVR shown, LVR given HVR shown, LVR given Asocial Conditions Assessment control (AC) LVR LVR a In the asocial conditions, the partner’s site was empty; however, the normal procedure was carried out. The food was moved to the partner’s side and was then re-placed into the food bowl. No reward control (NR) no reward HVR a In the asocial conditions, the partner’s site was empty; however, the normal procedure was carried out. The food was moved to the partner’s side and was then re-placed into the food bowl. LVR, low-value reward; HVR, high-value reward. We aimed to test these hypotheses using an inequity aversion task based on earlier studies on pet dogs [] by testing similarly raised and kept pack-living wolves (n = 9) and dogs (n = 10) at the Wolf Science Center, to determine whether they show similar responses to inequity. Two animals were placed in separate enclosures, each equipped with buzzer apparatuses placed against the fencing (see Figure 1 ). The experimenter alternately pointed to the two buzzers and asked the respective animal to press with a paw to receive a reward. Depending on the condition, they were given a reward (or not) (see Table 1 ). In order to keep the test as similar as possible between conditions, in asocial conditions, the experimenter acted as if there were a partner in the enclosure, including giving commands to and showing the reward briefly to the partner enclosure; however, after the food rewards were presented, they were discretely moved back into the bowl. The bowl contained a large amount of both food rewards, such that over the course of the session, as foods were removed, the bowl remained sufficiently full. Rewards—a high-value (HVR) and a low-value (LVR) reward—were determined prior to testing based on the preferences of the animals. Each session lasted a maximum of 30 trials per animal. If an individual stopped performing the task, the experimenter gave up to ten commands (“press,” or “come” if the animal had moved away, up to ten times each). If the individual refused to press or return to the apparatus, the session was ended. Moreover, as we were interested in whether the inequity condition experienced by the subject directly affected subsequent interactions with its partner and the experimenter, we implemented an “interaction test” previously used with pet dogs []. After each test condition, both the subject and the partner were moved to a neutral enclosure where they had the choice to interact with each other and/or the experimenter.

14 Brosnan S.F.

Talbot C.

Ahlgren M.

Lambeth S.P.

Schapiro S.J. Mechanisms underlying responses to inequitable outcomes in chimpanzees, Pan troglodytes. Each session consisted of one condition, and only one session per day was conducted for each subject. In each condition, we changed exactly one factor in regard to the baseline equity condition (quality inequity: food quality of the partner’s food; food control: shown HVR but given LVR [to test for frustration effects]; reward inequity: subject unrewarded; assessment control: presence of the partner). To further control that a possible response in the reward inequity condition was due to the mere absence of food and not the social context, we ran a further condition for comparison (the no reward condition) in which, like in the inequity, the subject did not receive food, but this time the partner was absent (see Table 1 for a summary of the conditions). Conditions were randomized, except that no-reward conditions (i.e., RI and NR; see Table 1 ) were never consecutive, to prevent animals from losing motivation. Since the dominance relationship is known to affect responses to inequity in primates [], we considered both the dominance relationship (dominant-submissive) and rank distance between partners. See STAR Methods for more information on all methods.

Figure 2 Number of Trials Completed by Condition Show full caption Data are represented as mean ± SE. For both dogs and wolves, there were fewer trials completed in the reward inequity condition than in the equity condition (F = −6.698, df = 59, p < 0.001; Figure 2 Data S1 ), showing that both wolves and dogs performed fewer trials when they were unrewarded in the presence of a rewarded partner. Importantly, wolves and dogs also completed fewer trials in the reward inequity than in the asocial no reward condition (F = −0.478, df = 18, p < 0.001; Data S1 ), indicating that the presence of the partner, not the lack of reward, is driving their refusals in the reward inequity condition. In other words, dogs and wolves respond to the inequity in the social context and not solely to the reward distribution. In contrast to previous studies on pet dogs, both wolves and dogs also completed fewer trials in the quality inequity condition than in the equity condition (F = −1.998, df = 59, p = 0.050; Figure 2 Data S1 ). Though there was no significant effect of species (F = −1.490, df = 14, p = 0.158), it appears that the wolves responded more strongly than the dogs (see Figure 2 ). We did not find any difference between the equity condition and our food control condition ( Data S1 ).

Furthermore, we investigated how many commands, normalized by number of trials, were given by the experimenter in each condition as a measure of their resistance. Wolves and dogs both were given more commands in the reward inequity condition compared to the equity condition (F = 2.176, df = 64, p = 0.033; Data S1 ), but there was no difference between the reward inequity and no reward conditions (F = 1.675, df = 17, p = 0.112; Data S1 ). We found a species interaction for the quality inequity condition (species × QI: F = 2.753, df = 64, p = 0.008; Data S1 ). Although wolves were given more commands in the quality inequity condition compared to the equity condition, this was not the case for the dogs (wolves, QI versus ET: F = 3.415, df = 30, p = 0.002; dogs, QI versus ET: F = 0.688, df = 34, p = 0.496; Data S1 ). Thus, in order to continue working in the quality inequity condition (compared to the equity condition), the experimenter had to prompt the wolves significantly more times per trial than the dogs, suggesting that the willingness to work in that condition was lower in the wolves than in the dogs. This finding, paired with the appearance that wolves responded more strongly to the quality inequity condition, suggests that there may be a difference between dogs and wolves in how they respond, or how strongly they respond, to the quality inequity condition. We did not find any difference between the equity condition and the food control condition for number of commands per trial ( Data S1 ).

Figure 3 Number of Trials Completed by Rank Distance between Partners for Reward Inequity, No Reward, and Equity Conditions We found an effect of rank on the number of trials completed. Regardless of species, the more dominant animals were to their partner (i.e., the larger the rank distance between individuals in the dyad), the fewer trials they completed in the reward inequity condition compared to the equity condition and the fewer trials they completed in the reward inequity condition compared to the no reward condition (RI versus ET: F = −2.364, df = 59, p = 0.021; RI versus NR: F = −2.491, df = 15, p = 0.025; Data S1 Figure 3 ). This suggests that the more dominant an individual is to their partner, the more sensitive they are toward the reward inequity. Additionally, there was an interaction between species and rank distance for the rate of stress behaviors (e.g., lip licking; see Table S1 ) in the social test conditions (species × rank distance: F = 3.101, df = 13, p = 0.008; Data S1 ). Further analysis showed that the more subordinate dogs had a tendency for a higher rate of stress behaviors seen in the test conditions; this was not the case in wolves (dogs, rank distance: F = −2.340, df = 6, p = 0.058; wolves, rank distance: F = 0.765, df = 7, p = 0.469; Data S1 ). However, it is not clear that the stress seen in the subordinate dogs is due to working next to a more dominant partner, because there was no effect of condition between the assessment control condition and the equity condition (ET versus AC: F = −0.841, df = 60, p = 0.403; Data S1 ).

The test conditions also affected the subject’s subsequent interactions with both the experimenter and the partner. Dogs took longer to go to the experimenter (F = 2.184, df = 27.83, p = 0.038; Data S2 ) and spent less time with the experimenter (F = −2.225, df = 35.33, p = 0.033; Data S2 ) in the quality inequity condition compared to the equity condition. Independent of condition, there was an effect of rank distance, where the more subordinate a dog was to their partner, the lower their latency to go to the experimenter (F = −4.646, df = 6.080, p = 0.003; Data S2 ). For wolves, we found that subjects went to the experimenter faster in the quality inequity compared to the equity condition (F = −2.261, df = 32, p = 0.031; Data S2 ), but there was no difference in time spent with the experimenter between the test conditions (F = 1.398, df = 32, p = 0.172; Data S2 ).