Animal Behaviour xxx (2011) 1e7 Contents lists available at ScienceDirect Animal Behaviour journal homepage: www.elsevier.com/locate/anbehav Social eavesdropping in the domestic dog S. Marshall-Pescini a, *, C. Passalacqua a, A. Ferrario a, P. Valsecchi b,1, E. Prato-Previde a a b Sezione di Psicologia, Dipartimento di Scienze e Tecnologie Biomediche, Università di Milano Dipartimento di Biologia Evolutiva e Funzionale, Università di Parma a r t i c l e i n f o Article history: Received 21 December 2010 Initial acceptance 15 February 2011 Final acceptance 22 February 2011 Available online xxx MS. number: 10-00880 Keywords: Canis familiaris cooperation dog eavesdropping food sharing social evaluation Eavesdropping on third-party interactions has been observed in a number of species and is considered an important source of information in decision-making processes relating to fighting and mate choice. Human beings, however, use publicly available information flexibly in many different contexts including assessing others’ altruistic tendencies, which may in turn inform their choice of the most appropriate cooperative partner. We assessed whether dogs, Canis familiaris, were capable of discerning a generous versus selfish food-sharing interaction between humans, and investigated which communicative cues (voice versus gestures) may be more salient for them. Importantly a control condition was included to ascertain whether it was in fact the interaction between individuals as opposed to the direct actions of the actors that the dogs evaluated. We found that the dogs were capable of eavesdropping on human food-sharing interactions, and vocal communication was particularly important to convey the human’s cooperative versus noncooperative intent. Ó 2011 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved. Social eavesdropping, that is, the use of information gathered by observing interactions between others (Bonnie & Earley 2007; Parejo & Aviles 2007) has been observed in a plethora of species from primates to invertebrates (reviewed in Danchin et al. 2004; Valone 2007). However, most studies have shown the use of eavesdropping in specific contexts such as dominance assessment/ fighting and mate choice, leading some authors to suggest that these abilities are domain specific and qualitatively different from the generalized ability to evaluate others socially that is observed in humans (Nowak & Sigmund 1998, 2005; Subiaul et al. 2008). In humans, social evaluation based on third-party interactions influences cooperative decisions, in that a person will more willingly cooperate with someone who is known to have been generous to a third party (Wedekind & Milinski 2000; Milinski et al. 2002). Furthermore, infants as young as 6e10 months prefer individuals who help rather than hinder a third party (Hamlin et al. 2007). To a certain extent cooperative behaviour is a trait we share with our closest living relatives. Spontaneous cooperation and choice of the best cooperator based on previous direct food-sharing experiences have been shown in chimpanzees, Pan troglodytes (Melis et al. * Correspondence: S. Marshall-Pescini, Sezione di Psicologia, Dipartimento di Scienze e Tecnologie Biomediche, Università di Milano, Via Fratelli Cervi 93, 20090 Segrate (MI), Italy. E-mail address: [email protected] (S. Marshall-Pescini). 1 P. Valsecchi is at the Dipartimento di Biologia Evolutiva e Funzionale, Università di Parma, Via Usberti 11/A, Parma 43100, Italy. 2006a, b). There is also some evidence that chimpanzees may assess an individual’s altruistic/food-sharing tendency based on third-party interactions (Russell et al. 2008; Subiaul et al. 2008). These results have led authors to consider the ability to evaluate generous actions towards third parties as a key element in the evolution of human cooperation (Nowak & Sigmund 1998, 2005). However, a study by Bshary & Grutter (2006) showed that client fish are more likely to invite interaction from cleaner fish that they had witnessed being cooperative (i.e. cleaning and not feeding selfishly) with another client fish. Thus, it seems that assessing another individual’s generosity or fair behaviour is an ability that, like others, will evolve if useful for survival, for example in a cleaner/client fish mutualisitc relationship. Arguably, the ability to assess food-sharing tendencies both directly (on past experience) and indirectly (through ‘eavesdropping’) should be important for any social species that depends on cooperative/mutualistic interactions, and/or where interactions with unfavourable partners may be costly. Dogs, Canis familiaris, in relation to humans, would seem to fit both these criteria. They show forms of cooperation (Boitani & Ciucci 1995; Naderi et al. 2001; Bauer & Smuts 2007; Bonanni et al. 2010), may be averse to ‘inequity’ (Range et al. 2009), easily follow human communicative cues (reviewed in Miklosi & Soproni 2006) and may even perform counterproductive actions to follow a person’s indications (Prato-Previde et al. 2008; Marshall-Pescini et al. 2010). However, dogs’ understanding of third-party interactions has been scarcely explored. Consolation (third-party-initiated 0003-3472/$38.00 Ó 2011 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.anbehav.2011.02.029 Please cite this article in press as: Marshall-Pescini, S., et al., Social eavesdropping in the domestic dog, Animal Behaviour (2011), doi:10.1016/ j.anbehav.2011.02.029 2 S. Marshall-Pescini et al. / Animal Behaviour xxx (2011) 1e7 loser-directed affiliation) has been shown in dogs (Cools et al. 2008), suggesting they are capable of recognizing the outcome and individual roles of conspecifics in third-party conflicts, and also in a conspecific play context (Ward et al. 2009). Furthermore, Rooney & Bradshaw (2006) showed that dogs could assess a potential human play partner, according to his/her behaviour towards another dog (Rooney & Bradshaw 2006). Dogs preferentially chose the winning partner (whether human or canine) in a tug-of-war game; however, they preferred the losing partner if the tug-of-war was not preceded by play signals hence suggesting a competition over resources rather than a playful interaction. Finally, Kundey et al. (2010) looked at whether dogs could understand a third-party interaction between humans involving competition over food. Results showed that dogs preferentially ate food placed close to a person who had previously allowed a human (or mechanical) recipient take it from her, as opposed to food adjacent to a person who had removed it when the recipient reached out for it. Thus, taken together these results suggest that dogs, like many other species, can eavesdrop on third-party conflicts when these involve resources, but they can also assess a social partner in a playful context. Similarly to the study by Kundey et al. (2010), we were also interested in assessing the dog’s ability to eavesdrop on an interaction between humans. However, our first, specific aim was to ascertain whether it was indeed the interaction between the actors that dogs used to evaluate the human partners, as opposed to the direct behaviours exhibited by the actors themselves. It has been noted that social eavesdropping studies often lack this important control condition (Bonnie & Earley 2007). Thus, to address this issue, we included a ‘ghost control group’ where the actors’ behaviour was exactly the same as that exhibited in the experimental group, but, crucially, no recipient was present. The second aim of the study was to start teasing apart what kinds of communicative cues allow dogs to understand the generous versus selfish attitude of humans in the third-party interaction. To this aim we reduced the visibility of the food during the interaction and rather emphasized the communicative behaviour of the donors, who signalled their generous/cooperative or selfish/noncooperative intent towards the beggar using either a combination of gestural and vocal cues (experimental group) or each element separately (voice versus gestures groups). METHODS Subjects We recruited 100 dogeowner dyads through personal contacts, advertisements in parks and veterinary surgeons. The dog sample consisted of 44 males and 56 females whose ages ranged from 1 to 10 years (mean 4.2 2.7 years); 68 were pure breeds and 32 mixed breeds (see Appendix for details). Most dogs (67 individuals) had participated during the previous year in a study involving a food choice task (procedures similar to Prato-Previde et al. 2008). Given the amount of time elapsed and the different nature of the test we considered that this previous experience would not affect their behaviour; however, to control for these differences, test-naïve and experienced dogs were equally distributed between experimental and control groups. This research complies with the Italian laws on animal welfare. Procedure All testing took place in a relatively bare (4 3 m) test room of the laboratory Canis Sapiens of the University of Milan. The chosen area was unfamiliar to the dogs. The behaviour of the dog during 1.5 m 2.5 m Figure 1. Schematic depiction of the experimental set-up. The owner and dog were always in the same position whereas the generous and selfish donors swapped places (unseen by the dog) prior to each new observation trial. the whole procedure (observation and testing) was videorecorded using a wide-angle video camera positioned on a tripod located in one corner of the test area (see Fig. 1). Prior to testing, the owner was asked to enter the test area, with his/her dog, and the dog was allowed to explore the environment freely while the experimenter briefly described the procedure to the owner. The procedure involved two phases: (1) observation and (2) test. Although the observation phase varied according to group allocation, the test was identical for all groups. Observation Phase Experimental group: the owner was asked not to interact with the dog during the experiment and simply remain seated, keeping the dog on the leash between his/her legs. Two chairs were placed equidistant and to either side of the seated owner (Fig. 1). Two people, both unfamiliar to the dog, entered the test room holding two identical bowls containing sausages (strong smelling and appetizing to dogs) and (in a separate compartment) a small quantity of cereal (which was used to feed the human beggar). They both simultaneously approached the dog and allowed it to inspect the interior of both bowls, thus making sure the dog had both seen and smelled the contents. The dog, however, was not allowed to take food from the bowls. The researchers then sat in their allocated positions, and simultaneously started eating the cereal. A third person (the beggar) entered the room, approached one of the researchers, knelt down and peered inside the bowl while tapping on her arm. The behaviour of the donor varied according to her role. The selfish donor said ‘No’ in a firm voice while flicking her hand out from holding the bowl thus gesturing the beggar away. The generous donor said, ‘Have it’ in a friendly tone of voice, and placed a small bit of cereal in the beggar’s mouth. The beggar moved from one person to the other a total of six times, in a semirandom order, never begging more than twice from the same person and spending the same amount of time next to each donor. The selfish and generous donor continued eating throughout the observation phase taking care that they both ate the same amount of food. The Please cite this article in press as: Marshall-Pescini, S., et al., Social eavesdropping in the domestic dog, Animal Behaviour (2011), doi:10.1016/ j.anbehav.2011.02.029 S. Marshall-Pescini et al. / Animal Behaviour xxx (2011) 1e7 bits of cereal were too small to be visible when held in the hand and did not make a crunching sound when being eaten. The beggar then left the room, asking the owner to release the dog as soon as the door closed behind her. Ghost control group: in the observation phase for dogs in the ghost control group, no beggar was present; however, the selfish and the generous donors pronounced the same words and carried out the same gestures, following the same timing and order, as that for dogs in the experimental group. Thus the selfish donor said ‘No’ in a firm voice while flicking her hand out from holding the bowl, while the generous donor said ‘Have it’ in a friendly tone of voice, holding out a bit of cereal to an ‘invisible beggar’. The generous donor then held the cereal in the palm of her hand, and after dipping her hand in the bowl, brought the titbit to her mouth and ate it. The ghost condition was run to discern whether, in their response, dogs take into account the third-party interaction or simply use an avoidance strategy, keeping a distance from the selfish person using a word (‘No’) and tone of voice that they may have previously associated with a negative outcome. Gestures group: dogs in the gestures group saw the selfish donor brusquely push the beggar away from the bowl, and the beggar losing her balance and almost falling to the ground. The generous donor took the food out of the bowl and placed the cereal in the beggar’s mouth. Neither person said anything. In all other respects the procedure was identical to the experimental group. Voice group: dogs in the voice group saw both the selfish and generous donors always keep their hand on the bowl, except when bringing food to their own mouth. When the beggar approached, the selfish donor simply said ‘No’ in a harsh tone of voice and the beggar moved away. The generous donor said ‘Have it’ in a friendly tone of voice, at which point the beggar put her own hand into the bowl and removed a piece of cereal and ate it. In all other respects the procedure was identical to that described for the experimental group. Test Phase The selfish and generous donors remained seated and still, holding the bowl on their lap with both hands (so as not to use hand cues inadvertently, which may influence the dog’s choice), and looking down at the bowl (never at the dog). The dog was allowed free movement in the room for 20 s. If dogs tried to jump up and snatch the bowl from the researcher’s hands, the researcher held the bowl out of reach until the dog was back on the ground, and then resumed the position described above. No food was ever given to the dogs. After the test phase, the dog and owner were asked to leave the room, the selfish and generous donors swapped sitting places, and the owner was instructed to re-enter the room. Once the dog had settled down, another observation trial, identical to that described above, began. Overall, dogs witnessed three observation trials (each lasting approximately 60 s), followed each time by a test period lasting 20 s. Prior to each new observation trial, when the owner and dog left the test room, its owner, outside the test room to maintain the dog’s motivation with the test procedure, delivered a small slice of sausage to the dog. The researchers maintained their roles as selfish and generous donors throughout the experiment for each dog. However, their roles were counterbalanced across subjects and within each group. Furthermore, the order in which the beggar approached the selfish and generous donors was counterbalanced so that 50% of dogs in each group saw the beggar approach the selfish donor last (before the start of the test trial), and 50% saw the generous donor being approached last. 3 Analysis The Solomon (beta 091110, copyright 2006e2008 by András Péter, Eötvös Loránd University, Department of Ethology, Budapest, Hungary) software for behavioural analysis was used to code the frequency and duration of behaviour from the video. During the observation phase, the duration of the dog’s gazing behaviour was coded in terms of where/whom it focused on: (1) gaze at owner; (2) gaze at beggar; (3) gaze at selfish donor; (4) gaze at generous donor; (5) gaze other. Gazing was defined as the dog orienting its body and/or head towards a specific person, while remaining in its original location. During the observation phase, when the beggar was adjacent to the donors, it was not always possible to discern at whom the dog was looking; thus in this case the dog was considered to be looking at the donor to which the beggar was closest (the beggar spent an equal amount of time close to each donor). During the test phase, the duration of the following behaviours was coded: (1) interact with selfish donor; (2) interact with generous donor; (3) interact with owner; (4) other (any other behaviour not described above); (5) out of sight. The dog’s first approach behaviour was also coded as (1) approach generous donor, (2) approach selfish donor or (3) ambiguous approach. An approach/interaction was defined as the dog moving towards a specific person and either stopping in front of her (within arm’s length) and adopting a static position (sitting, standing, lying down) oriented to that person, or interacting with the person in a more direct manner (pawing, nudging with nose, jumping up with both forepaws on the donor’s lap, rubbing body against the donor’s legs, etc.). For the first-choice data, an approach was considered ambiguous if a dog moved towards the centre of the room, equidistant from both donors, with a body orientation either towards the owner or towards some other undetermined location, or if it moved towards one of the donors, but then sat oriented towards the other, or if it moved around the room but then returned to a location adjacent to the owner. A second coder analysed 35% of trials for the duration of gazing behaviour in observation trials and interaction behaviour in test trials. Spearman correlations for both behaviours were high (observation trials: N ¼ 90: gaze at generous donor: rS ¼ 0.90, P < 0.001; gaze at selfish donor: rS ¼ 0.93, P < 0.001; test trials: N ¼ 90: interact with generous donor: rS ¼ 0.93, P < 0.001; interact with selfish donor: rS ¼ 0.90, P < 0.001). Furthermore, the dog’s first approach on 30% of trials was coded by a second observer resulting in a substantial level of agreement (Kappa ¼ 0.81). To assess the potential group differences in the preference for gazing at the generous versus selfish donors during the observation phase, a gaze preference index was calculated: gaze at generous donor/(gaze at generous þ gaze at selfish donor) 100. To evaluate the potential preference for each donor in the test phase, we calculated the percentage of overall test time spent interacting with the generous and the selfish donors and compared this within each group (one-sample t test) and between groups (univariate ANOVA followed by Tukey post hoc comparisons). Chi-square and binomial tests were used to compare first-choice data. Statistical tests were two tailed and the P value was set at 0.05. Data conformed to assumptions for the use of parametric statistics. RESULTS We excluded 16 dogs from the study because during either the observation phase or the test phase they hid under/behind the owner, or showed other signs of stress and gave no attention to the events during the observation phase. This left 20 dogs in the experimental group (seven males and 13 females, mean 4.5 2.8 years), 24 in the ghost control group (12 males and 12 females, Please cite this article in press as: Marshall-Pescini, S., et al., Social eavesdropping in the domestic dog, Animal Behaviour (2011), doi:10.1016/ j.anbehav.2011.02.029 4 S. Marshall-Pescini et al. / Animal Behaviour xxx (2011) 1e7 mean 3.7 2.4 years), 20 in the gestures group (six males, 14 females, mean 4.4 3 years) and 20 in the voice group (10 males, 10 females, mean 4 2.8 years). During the observation phase, a significant difference emerged between groups in the dogs’ looking preference towards the generous donor (experimental: 62% (mean 64 s); ghost control: 56% (mean 50 s); voice: 61% (mean 66 s); gestures: 51% (mean 47 s); F3,80 ¼ 5.05, P ¼ 0.003) with dogs in the experimental and voice groups looking longer at the generous donor than dogs in the gestures group (post hoc: experimental versus gestures: P ¼ 0.007; voice versus gestures: P ¼ 0.01) and no other difference occurring (post hoc: experimental versus ghost control: P ¼ 0.2; experimental versus voice: P ¼ 0.9; Fig. 2). To evaluate whether dogs showed a clear preference in their choice of donor from the start, we compared the dogs’ first choice between the generous and selfish donors (thus excluding dogs that performed an ambiguous approach) in all three trials for each group. Binomial tests (two tailed; test value set ¼ 0.5) showed that dogs in the experimental group significantly preferred the generous donor throughout (test 1: P ¼ 0.007; test 2: P ¼ 0.049; test 3: P ¼ 0.0002); however, this preference did not occur in either the ghost control group (test 1: P ¼ 1; test 2: P ¼ 1; test 3: P ¼ 0.7) or the gestures group (test 1: P ¼ 0.3; test 2: P ¼ 0.8; test 3: P ¼ 1). In the voice group, no significant preference for the generous donor occurred in the first test (P ¼ 0.3); however, it approached and then reached significance in the second and third tests (test 2: P ¼ 0.057; test 3: P ¼ 0.03). We carried out two separate between-group comparisons in accordance with the aims of the study. (1) To assess whether dogs were capable of distinguishing between a generous and selfish donor on the basis of a third-party interaction, we compared the experimental with the ghost control group and found that significantly more dogs chose the generous donor first in the experimental group (c22 ¼ 3.98, P ¼ 0.046; Table 1). (2) To assess which human communicative cues dogs used more to evaluate the interaction, we compared the first choice carried out by the voice and gestures groups; however, no significant difference emerged (Table 2). In the test phase, we further compared groups on the percentage of trial time spent interacting with the generous donor. A significant difference emerged (experimental: 45%; ghost control: 15%; voice: 27%; gestures: 21%; F3,79 ¼ 9.14, P < 0.001) with dogs in the experimental group spending more time interacting with the generous Table 1 Number (and percentage) of dogs whose first approach was towards either the generous or selfish donor, or was considered ambiguous, in the experimental and ghost control group Test 1 Test 2 Experimental Ghost Generous 15 (75) donor Selfish donor 3 (15) Ambiguous 2 (10) Experimental Ghost Experimental Ghost 10 (42) 13 (65) 9 (38) 5 (21) Test 3 4 (20) 3 (15) 8 (33) 13 (65) 7 (30) 9 (37) 0 7 (35) 4 (17) 6 (25) 14 (58) donor than dogs in all other groups (post hoc: experimental versus ghost control and experimental versus gestures: P < 0.001; experimental versus voice: P ¼ 0.03; Fig. 3). Conversely, dogs in the ghost control group spent significantly more time interacting with the selfish donor (experimental: 2%; ghost control: 11%; voice: 2%; gestures: 9%; F3,79 ¼ 3.67, P ¼ 0.015) than dogs in the experimental and voice groups (post hoc: ghost control versus experimental: P ¼ 0.045; experimental versus voice: P ¼ 0.046). To assess further whether the dogs within each group exhibited a preference for the generous versus the selfish donor, we compared the percentage of time spent interacting with each actor. Dogs in the experimental (mean: generous: 45%; selfish: 2%; t20 ¼ 7.17, P < 0.001) and voice (mean: generous: 27%; selfish: 2%; t20 ¼ 5.95, P < 0.001) groups spent significantly more time with the generous than the selfish donor, but this was not the case for dogs in the ghost control (mean: generous: 4%; selfish: 11%; t24 ¼ 0.72, NS) and gestures group (mean: generous: 21%; selfish: 9%; t20 ¼ 1.8, P ¼ 0.08). DISCUSSION The first aim of the study was to assess whether dogs could eavesdrop on a food-sharing interaction between humans, where the generous versus selfish intent was expressed by communicating with the beggar. Given that a clear preference for the generous donor emerged in the experimental group (in all three tests), that significantly more dogs in the experimental than the ghost control group approached the generous donor first and spent more time interacting with her, we can conclude that dogs are effectively using third-party interactions to gain information on the 100 90 80 % Trial time 70 60 50 40 30 20 10 0 Experimental Ghost control Voice Gestures Group Figure 2. Gaze preference index (time spent looking at generous/generous þ selfish donor 100): mean percentage of time spent looking at the generous (dark grey) and selfish (light grey) donors during the observation phases for each group. Please cite this article in press as: Marshall-Pescini, S., et al., Social eavesdropping in the domestic dog, Animal Behaviour (2011), doi:10.1016/ j.anbehav.2011.02.029 S. Marshall-Pescini et al. / Animal Behaviour xxx (2011) 1e7 behaviours with no recipient present; however, this kind of ‘ghost’ condition has been fruitfully employed in the context of other social-learning studies (Hopper 2010), and, we think, could be successfully adopted in studies involving humanedog third-party interactions. Our results, however, are also consistent with the possibility that dogs are not so much showing a preference for the person interacting generously, but rather seek to avoid the person who behaved more brusquely with the third party. We do not think this is in fact the case, since in the gestures group the selfish donor’s behaviour was more aggressive than in the other two groups (she did not just flick a hand but actually pushed the beggar away so that the latter fell back). Given studies showing that dogs tend to read human aggressive behaviour in terms of bodily movement (Vas et al. 2005), we would have expected a strong avoidance of the selfish individual in this group if dogs had been showing an avoidance of the selfish donor rather than a preference for the generous donor. However, this possibility cannot be totally excluded and further studies will be necessary to discriminate between competing hypotheses. The second aim of the study was to start investigating which kind of communicative cues dogs use to interpret third-party interactions between humans in a food-sharing context. To this aim we started by assessing whether dogs would rely more on the humans’ vocal or gestural communication towards the beggar. Although dogs in neither group showed a preferential approach to the generous donor in the first test, in the voice group the dogs’ preference for the generous donor gradually emerged in the second and reached significance in the third test. Furthermore, overall, dogs in the voice group spent significantly more time interacting with the generous than the selfish donor, whereas no such preference was shown by dogs in the gestures group. Finally, in the observation phase, whereas dogs in both the experimental and voice groups spent more time looking at the generous donor (suggesting a preference already at this stage), dogs in the gestures group did not. These results show that the use of gestures alone was insufficient for dogs to discriminate between the selfish and generous intent of the donors, and that the voice was the more salient cue for understanding the interaction. However, dogs in the Table 2 Number (and percentage) of dogs whose first approach was towards either the generous or selfish donor, or was considered ambiguous, in the voice and gestures group Test 1 Generous donor Selfish donor Ambiguous Test 2 Test 3 Gestures Voice Gestures Voice Gestures Voice 10 (50) 5 (40) 5 (10) 10 (50) 5 (40) 5 (10) 10 (50) 8 (25) 2 (25) 11 (55) 3 (15) 6 (30) 5 (25) 6 (30) 9 (45) 12 (60) 3 (15) 5 (25) 5 potential likelihood of sharing food with a human stranger. The lack of a preference for either donor in the ghost control condition suggests that the effect observed in the experimental condition is not simply due to a preference or avoidance of the behaviours exhibited by the donors per se (e.g. avoiding the selfish donor because she shows brusque movements and a harsh tone of voice), but rather it is the interaction between the donors and the beggar that provides important information that the dogs can use. However, another possibility that must be considered is that the difference in behaviour of dogs in the experimental versus ghost control groups is simply due to different patterns of attention being given to the donors during the observation phase. It may be that the presence of the beggar increased the salience of the donor’s cues by focusing the dog’s attention. This is unlikely, since dogs in both the experimental and ghost control groups showed a similar preference for looking at the generous donor, and it was only in the gestures group, despite the beggar being present, that dogs showed equal attention to the generous and selfish donors. Thus, attention to the donors does not seem to have been affected by the presence/ absence of the beggar. As highlighted by Bonnie & Earley (2007), social eavesdropping studies do not always control for the possibility that the preference for a specific individual is due to the outcome of an interaction with a social partner rather than having seen them manifest certain behaviours during that interaction, which would have been equally attractive had they been manifested without the presence of the social partner. This may of course be difficult to achieve when working with animals that cannot be asked to reproduce specific ** 60 ** 50 % Trial time 40 ** 30 20 10 0 Experimental Ghost control Voice Gestures Group Figure 3. Mean percentage and SE of time spent interacting with the generous (dark grey) and selfish (light grey) donors during the test phase for each group. **P < 0.01. Please cite this article in press as: Marshall-Pescini, S., et al., Social eavesdropping in the domestic dog, Animal Behaviour (2011), doi:10.1016/ j.anbehav.2011.02.029 6 S. Marshall-Pescini et al. / Animal Behaviour xxx (2011) 1e7 experimental group, where the actors combined voice and gestures to convey the generous versus selfish acts, showed the strongest preference for the generous donor (significantly more so than dogs in the voice group), suggesting that the more communicative cues available to the dogs, the better their discriminatory abilities. Previous studies have shown that dogs are capable of obtaining information through third-party interactions in a conflict situation between conspecific (Cools et al. 2008) and that dogs are also capable of flexibly choosing with whom to interact depending on the outcome of a playful versus competitive interaction between a human and a dog (Rooney & Bradshaw 2006). In the latter study, dogs preferred either the winner or the loser depending on the context, without showing a bias towards the human or dog. Thus humans and dogs were ‘treated alike’ and chosen based solely on their role in the interaction. Kundey et al. (2010) went a step further, showing that dogs can also evaluate humans based on observing a competitive interaction over a food resource. However, it is not altogether clear from Kundey et al’s. (2010) study whether in fact dogs evaluated the interaction between actors and recipients, or simply chose the generous person (or avoided the possessive one) based on the specific behaviours exhibited towards the food resource (as opposed to the behaviours exhibited towards the recipient). In fact, in all the cases in which dogs discriminated between actors they saw a large bone either being withdrawn from the attempted grasp of the recipient or being ignored by the previous handler. Given the absence of a control condition in which the exact same behaviours were manifested but no recipient was present, it cannot be excluded that dogs simply avoided the person manipulating the food in a possessive manner, or chose the person who placed food on the ground and then no longer showed interest in it. Although the results of that study are interesting, since they show that dogs are capable of discriminating between different human behaviours towards a food resource, it may not necessarily involve an understanding of a third-party interaction. Potential support for this argument also comes from the fact that, in our study, where food was not visible and thus dogs had to evaluate the situation based on the behaviours directed towards the beggar, they seemed to find it difficult to discriminate between the actors’ generous versus selfish intent when this was expressed with gestures alone, whereas in Kundey et al’s. (2010) study the gestures directed towards the food were easily understood even when accompanied by almost no social cues. Our results raise a number of interesting questions concerning both the complexity and the potential origins of dogs’ ability to understand human communicative intent. One argument is that humans differ from other species in their ability to evaluate others socially because we are capable of forming a reputation about other individuals based on a number of different parameters, thus showing an abstract social-reasoning skill which can be applied in many contexts. This is considered very different from the ‘domain-specific’ ability of eavesdropping in the context of mate selection or fighting shown in other species (Subiaul et al. 2008). Although still in its infancy, studies on eavesdropping in dogs so far seem to suggest that they are capable of judging a human’s social attractiveness in at least three different contexts (playful or competitive: Rooney & Bradshaw 2006; Kundey et al. (2010); resource sharing: current study) showing a certain degree of flexibility. It may be that dogs learn about human social cues exhibited directly towards them, and towards others, and are capable of abstracting knowledge from these situations to assess their partners’ potential value (Call 2001). As to the origins of dogs’ ability to understand human communicative intent, dogs, like their wolf, Canis lupus, ancestor, could be considered a cooperative species (although see CluttonBrock 2009, for definitions of different forms of cooperation), but reports of cooperation in feral populations are relatively rare (Pal 2005; Cafazzo 2007; Bonanni et al. 2010) and there are few experimental studies investigating this topic directly. The ability to assess a human being’s generosity in a food-sharing context might be a direct transfer of the ability to assess other dogs’ food-sharing tendencies; however, no studies have investigated this issue, nor is there any evidence that more tolerant dogs in a food-sharing context are preferentially chosen as social/cooperative partners (a link that has been shown in chimpanzees: Melis et al. 2006a, b). Alternatively, assessing another individual’s generosity in a food-sharing context may be a specialized skill that dogs have developed in relation to humans, as part of the changes brought about by the domestication process (Hare et al. 2002; Miklósi et al., 2003; Hare & Tomasello 2005; Gacsi et al. 2009) and further enhanced by the experience of frequent encounters with unknown humans, a peculiarity that is thought to increase the importance of reliance on indirect as opposed to direct reputation formation (Bshary & Grutter 2006; Subiaul et al. 2008). The requirement to interact frequently with conspecific and heterospecific strangers is a trait that dogs share with humans and that may have had a crucial role in shaping this species’ cognitive and communicative abilities. Studies conducted to understand dogs’ (direct and indirect) assessment of human and conspecific partners are still too scarce to make firm conclusions; however, it may be that given the importance humans have for dogs’ survival, the latter have evolved a sensitivity to specific human features, allowing them to discriminate between potentially useful and harmful social partners, both directly and indirectly. It is, none the less, important to note that, given a dog’s preference for multiple communicative cues (voice and gestures) to detect the generous versus selfish stance of the donors, it is likely that indirect assessment of human generosity is substantially mediated by a learned component. Only once dogs are capable of understanding our species’ behavioural outputs in different contexts can they abstract social knowledge from these situations (Call 2001). Acknowledgments This research was supported by doctoral and postdoctoral funds from the University of Milan to C.P. and S.M.-P. We are grateful to all dogs and owners for participating as volunteers. We thank Marco Colombetti for drawing Fig. 1 and Simona Cafazzo, Roberto Bonanni, Zsofi Viranyi, Frederike Range and Juliane Kaminski for comments on the manuscript. We also thank the editor and two anonymous referees for helpful suggestions and comments. References Bauer, E. B. & Smuts, B. B. 2007. Cooperation and competition during dyadic play in domestic dogs, Canis familiaris. Animal Behaviour, 73, 489e499. Boitani, L. & Ciucci, P. 1995. Comparative social ecology of feral dogs and wolves. Ethology Ecology & Evolution, 7, 49e72. Bonanni, R., Valsecchi, P. & Natoli, E. 2010. Pattern of individual participation and cheating in conflicts between groups of free-ranging dogs. Animal Behaviour, 79, 957e968. Bonnie, K. E. & Earley, R. L. 2007. Expanding the scope for social information use. Animal Behaviour, 74, 171e181. Bshary, R. & Grutter, A. S. 2006. Image scoring and cooperation in a cleaner fish mutualism. Nature, 441, 975e978. Cafazzo, S. 2007. Dinamiche sociali in un gruppo di cani domestici (Canis lupus familiaris) liberi in ambiente suburbano. Ph.D. thesis, University of Parma. Call, J. 2001. Chimpanzee social cognition. Trends in Cognitive Sciences, 5, 388e393. Clutton-Brock, T. 2009. Cooperation between non-kin in animal societies. Nature, 462, 51e57. Cools, A. K. A., Van Hout, A. J. M. & Nelissen, M. H. J. 2008. Canine reconciliation and third-party-initiated postconflict affiliation: do peacemaking social mechanisms in dogs rival those of higher primates? Ethology, 114, 53e63. Danchin, E., Giraldeau, L. A., Valone, T. J. & Wagner, R. H. 2004. Public information: from nosy neighbors to cultural evolution. Science, 305, 487e491. Gacsi, M., Gyori, B., Viranyi, Z., Kubinyi, E., Range, F., Belenyi, B. & Miklósi, A. 2009. Explaining dog wolf differences in utilizing human pointing gestures: Please cite this article in press as: Marshall-Pescini, S., et al., Social eavesdropping in the domestic dog, Animal Behaviour (2011), doi:10.1016/ j.anbehav.2011.02.029 S. Marshall-Pescini et al. / Animal Behaviour xxx (2011) 1e7 selection for synergistic shifts in the development of some social skills. PloS One, 4, e6584, doi:10.1371/journal.pone.0006584. Hamlin, J. K., Wynn, K. & Bloom, P. 2007. Social evaluation by preverbal infants. Nature, 450, 557e559. Hare, B. & Tomasello, M. 2005. Human-like social skills in dogs? Trends in Cognitive Sciences, 9, 439e444. Hare, B., Brown, M., Williamson, C. & Tomasello, M. 2002. The domestication of social cognition in dogs. Science, 298, 1634e1636. Hopper, L. 2010. ‘Ghost’ experiments and the dissection of social learning in humans and animals. Biological Reviews, 85, 685e701. 10.111/j.1469185X.2010.00120.x. Kundey, S. M. A., De Los Reyes, A., Royer, E., Molina, S., Monnier, B., German, R. & Coshun, A. 2010. Reputation-like inference in domestic dogs (Canis familiaris). Animal Cognition, 14, 291e302. Marshall-Pescini, S., Prato-Previde, E. & Valsecchi, P. 2010. Are dogs (Canis familiaris) misled more by their owners than by strangers in a food choice task? Animal Cognition, 14, 137e142. Melis, A. P., Hare, B. & Tomasello, M. 2006a. Chimpanzees recruit the best collaborators. Science, 311, 1297e1300. Melis, A. P., Hare, B. & Tomasello, M. 2006b. Engineering cooperation in chimpanzees: tolerance constraints on cooperation. Animal Behaviour, 72, 275e286. Miklosi, A. & Soproni, K. 2006. A comparative analysis of animals’ understanding of the human pointing gesture. Animal Cognition, 9, 81e93. Miklósi, Á, Kubinyi, E., Topál, J., Gacsi, 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, 763e766. Milinski, M., Semmann, D. & Krambeck, H. J. 2002. Reputation helps solve the ‘tragedy of the commons’. Nature, 415, 424e426. Naderi, S., Miklósi, Á, Dóka, A. & Csányi, V. 2001. Co-operative interactions between blind persons and their dogs. Applied Animal Behaviour Science, 74, 59e80. Nowak, M. A. & Sigmund, K. 1998. Evolution of indirect reciprocity by image scoring. Nature, 393, 573e577. Nowak, M. A. & Sigmund, K. 2005. Evolution of indirect reciprocity. Nature, 437, 1291e1298. Pal, S. K. 2005. Parental care in free-ranging dogs (Canis familiaris). Applied Animal Behaviour Science, 90, 31e47. Parejo, D. & Aviles, J. M. 2007. Do avian brood parasites eavesdrop on heterospecific sexual signals revealing host quality? A review of the evidence. Animal Cognition, 10, 81e88. Prato-Previde, E., Marshall-Pescini, S. & Valsecchi, P. 2008. Is your choice my choice? The owners’ effect on pet dogs’ (Canis lupus familiaris) performance in a food choice task. Animal Cognition, 11, 167e174. 7 Range, F., Horn, L., Viranyi, Z. & Huber, L. 2009. The absence of reward induces inequity aversion in dogs. Proceedings of the National Academy of Sciences, U.S.A., 106, 340e345. Rooney, N. J. & Bradshaw, J. W. S. 2006. Social cognition in the domestic dog: behaviour of spectators towards participants in interspecific games. Animal Behaviour, 72, 343e352. Russell, Y. I., Call, J. & Dunbar, R. I. M. 2008. Image scoring in great apes. Behavioural Processes, 78, 108e111. Subiaul, F., Vonk, J., Okamoto-Barth, S. & Barth, J. 2008. Do chimpanzees learn reputation by observation? Evidence from direct and indirect experience with generous and selfish strangers. Animal Cognition, 11, 611e623. Valone, T. J. 2007. From eavesdropping on performance to copying the behavior of others: a review of public information use. Behavioral Ecology and Sociobiology, 62, 1e14. Vas, J., Topál, J., Gácsi, M., Miklósi, A. & Csányi, V. 2005. A friend or an enemy? Dogs’ reaction to an unfamiliar person showing cues of threat and friendliness at different times. Applied Animal Behaviour Science, 94, 99e115. Ward, C., Trisko, R. & Smuts, B. B. 2009. Third-party interventions in dyadic play between littermates of domestic dogs (Canis lupus familiaris). Animal Behaviour, 78, 1153e1160. Wedekind, C. & Milinski, M. 2000. Cooperation through image scoring in humans. Science, 288, 850e852. APPENDIX BREEDS OF PARTICIPATING DOGS One Alaskan malamute, one Australian collie, two basenjis, four beagles, one Belgian shepherd, one bichon bolognese, two bichon frises, one border collie, one borzoi, four boxers, one Brittany spaniel, two cocker spaniels, one corso, one Czechoslovakian wolfdog, one doberman, two drahtar, two German shepherds, four golden retrievers, one Jack Russell terrier, one Ibizan podenco, three Irish setters, seven labrador retrievers, one miniature schnauzer, one miniature poodle, two pinchers, one Portuguese water dog, two pugs, one rottweiler, two Scottish collies, two shiba inu, one standard poodle, two Siberian huskies, one terranova, two Yorkshire terriers, two vizla, one weimaraner, three whippets. Please cite this article in press as: Marshall-Pescini, S., et al., Social eavesdropping in the domestic dog, Animal Behaviour (2011), doi:10.1016/ j.anbehav.2011.02.029
© Copyright 2026 Paperzz