Dogs can discriminate human smiling faces from blank expressions

Transcription

1 Ani Cogn (11) 14: DOI 10.7/s ORIGINAL PAPER Dogs can discriinate huan siling aces ro lank expressions Miho Nagasawa Kensuke Murai Kazutaka Mogi Takeui Kikusui Received: 15 July 10 / Revised: 21 January 11 / Accepted: 14 Feruary 11 / Pulished online: 26 Feruary 11 Ó Springer-Verlag 11 Astract Dogs have a unique aility to understand visual cues ro huans. We investigated whether dogs can discriinate etween huan acial expressions. Photographs o huan aces were used to test nine pet dogs in two-choice discriination tasks. The training phases involved each dog learning to discriinate etween a set o photographs o their owner s siling and lank ace. O the nine dogs, ive ulilled these criteria and were selected or test sessions. In the test phase, 10 sets o photographs o the owner s siling and lank ace, which had previously not een seen y the dog, were presented. The dogs selected the owner s siling ace signiicantly ore oten than expected y chance. In susequent tests, 10 sets o siling and lank ace photographs o persons unailiar to the dogs were presented (10 ales and 10 eales). There was no statistical dierence etween the accuracy in the case o the owners and that in the case o unailiar persons with the sae gender as the owner. However, the accuracy was signiicantly lower in the case o unailiar persons o the opposite gender to that o the owner, than with the owners theselves. These results suggest that dogs can learn to discriinate huan siling aces ro lank aces y looking at photographs. Although it reains unclear whether dogs have huan-like systes or visual processing o huan acial expressions, the aility to learn to Electronic suppleentary aterial The online version o this article (doi:10.7/s ) contains suppleentary aterial, which is availale to authorized users. M. Nagasawa K. Murai K. Mogi T. Kikusui (&) Departent o Anial Science and Biotechnology, Azau University, Fuchinoe, Sagaihara, Kanagawa-ken , Japan e-ail: kikusui@azau-u.ac.jp discriinate huan acial expressions ay have helped dogs adapt to huan society. Keywords Dog Huan acial expression Two-choice discriination tasks Visual discriination Introduction Knowing the state o others eotions is extreely iportant or social anials to decide their susequent ehaviors and it thereore aects survival. There are any sensory cues or the assessent o others eotions, and the type o cue that is iportant varies with species. For huans, acial expressions extreely coplicated visual cues play a vital role. It is possile to live in harony with other eers o society y reading their eotions ro acial expressions and adjusting one s ehaviors accordingly (Ekan and Friesen 1975). Dogs (Canis ailiaris) are also very social anials; they have any visual counication ethods in coon with wolves (Canis lupus), including postures and acial expressions, which indicate doinance status, aggression, and ear (Arantes 1987; Bradshaw and Nott 1995). In recent years, scientists have egun to ocus on social visual cognitive ailities in dogs interactions with huans. For exaple, i a huan throws a all or a dog to etch and then turns his ack, the dog alost always rings the all ack around the huan s ody in order to drop it in ront o his ace (Hare et al. 1998; Gacsi et al. 04). Dogs can also understand the relationship etween the direction in which huans are acing or gazing and their attentional state (Call et al. 03; Viranyi et al. 04; Schwa and Huer 06). Regarding a dog s individual recognition o huans, recent ehavioral studies have shown that dogs generate their internal

2 526 Ani Cogn (11) 14: representation o the owner s ace when they hear the owner calling the (Adachi et al. 07), and in visual paired-coparison tasks dogs looked at the pictures o novel huan aces longer than at ailiar huan aces who was not dog s owner (Racca et al. 10). These studies show that dogs have a air understanding o huan aces and the role that they play in social interaction. Although it reains unclear whether dogs can interpret huan eotions y looking at their aces, it will e valuale to clariy whether they can discriinate huan acial expressions. In Japanese onkeys (Macaca uscata), the study o the recognition o acial expressions o huans suggests that it is ore diicult or Japanese onkeys to detect the oveent o eyerows in the huan ace; they do not use this eature or intraspecies counication (Kanazawa 1996). Considering that onkeys unale to discriinate huan expressions not used in their interspecies counication in nature, we can predict that it will e diicult or dogs to discriinate huan acial expressions, as their acial structure diers greatly ro ours. On the other hand, dogs coe into contact daily and ore closely with huans than do onkeys, and they generally act appropriately according to the dierent huan ehavioral cues (Vas et al. 05). Thereore, dogs ay learn to associate huan expression with a speciic outcoe, and they ay e ale to recognize the oveents o huan acial parts. To deterine whether dogs can discriinate etween huan acial expressions and can generalize huan acial expressions, we conducted a two-choice discriination task. Range et al. (08) used this sae task to show the aility o dogs to visually classiy natural stiuli. We used photographs o siling huan aces as positive stiuli in this study ecause o the inding that dogs looked at their owners aces longer when the owners elt happy than when they were sad (Morisaki et al. 09). Materials and ethods Sujects The sujects were our Larador Retrievers and ive Standard Poodles. The age o the sujects was ± onths (ean ± SE). Inoration aout the dogs and their owners is shown in Tale 1. All dogs were pets owned y university sta and students. The dogs oten cae to the university with their owners and interacted with a variety o people. They were aintained on a noral diet, which was not changed during the test period. The dogs had undergone asic oedience training and were naïve to the experiental task. Stiuli The presentation stiuli were color photographs o huan aces. A suary o the stiuli is shown in Tale 2. The training phases In the irst training phase, a set o ace (positive stiulus: S?) and ack-o-the-head photographs (negative stiulus: S-) o one particular ale person, who was a university student and usually had soe contact with each dog 2 5 days per week in the university, was presented to all sujects (Fig. 1a). In the second training phase, a set o each owner s siling ace (S?) and a lank ace Tale 1 Inoration on dogs and their owners Dogs Owners Nae Breed Sex Age (onths) Gender Age The period o owning (onths) Charley SP Neutered 11 Feale 9 Cuppa LR Male 70 Male Fleecia LR Neutered 74 Feale Jasine SP Feale 11 Male 39 9 Kai LR Neutered 71 Male Cody SP Male 102 Male 39 Anita SP Spayed 86 Male Punch SP Neutered Male Fook LR Neutered 70 Feale The ive dogs aove the dash line et the learning criteria in the training phases and proceeded to the test phase SP standard poodle, LR larador retriever

3 Ani Cogn (11) 14: Tale 2 Experiental procedure and presentation stiuli Phase Session Stiulus photograph Nuer o trial First training phase A set o the ront ace (S?) and ack o the head (S-) o the ailiar ale (Fig. 1a) Second training phase A set o each owner s siling ace (S?) and lank ace (S-; Fig. 1) Test phase OW session 10 sets o siling ace (S?) and lank ace (S-) o each owner (Fig. 1c) UPSG session 10 sets o siling ace (S?) and lank ace (S-) o 10 unailiar persons o the sae gender as the owner (Fig. 1d) UPOG session 10 sets o siling ace (S?) and lank ace (S-) o 10 unailiar persons o the opposite gender as the owner (Fig. 1d) 30 trials/session 30 trials/session 10 trials/session 10 trials/session 10 trials/session Control phase PC session (positive control) A set o the owner s siling ace (S?) and lank 10 trials/session ace (S-) SS session A set o sae siling aces o the owner a 10 trials/session BB session A set o sae lank aces o the owner a 10 trials/session AB session A set o angry (S?) and lank ace (S-) o the owner (Fig. 1e) 10 trials/session a The positive stiuli were selected randoly Fig. 1 The stiulus photographs. a The sae ailiar person s ace (S?) and ack o the head (S-; the irst training phase) The owner s siling (S?) and lank (S-) ace (the second training phase) c The owner s siling (S?) and lank (S-) ace (OW session in the test phase) d The unailiar person s siling (S?) and lank ace (S-; UPSG and UPOG sessions in the test phase) e The owner s angry ace (S?) and lank ace (S-; AB session in the control phase) a c d e photographs (S-) was presented (Fig. 1). All owners were photographed under the sae conditions. The test phase This phase coprised three kinds o session. In the irst session, 10 sets o the siling ace (S?) and lank ace (S-) photographs o each owner were presented (OW session). These particular photographs were new to the suject; the owners changed clothes and wore glasses or hats or each photograph (Fig. 1c). In the second session, 10 sets o siling ace (S?) and lank ace (S-) photographs o 10 unailiar persons o the sae gender as the owner were presented (a set o photographs per unailiar person, UPSG session; Fig. 1d). In the last session, 10 sets o siling ace (S?) and lank ace (S-) photographs o 10 unailiar

4 528 Ani Cogn (11) 14: persons o the opposite gender to that o the owner were presented (UPOG session; Fig. 1d). These unailiar persons were university students who had never et the suject dogs, and they were photographed at various places. The control phase This phase was coposed o our sessions. In the irst session, a set o siling (S?) and lank ace (S-) photographs o each owner was presented as a positive control (PC session). In the second session, the sae siling aces o the owner were presented siultaneously (SS session), and in the third session, the sae lank aces o the owner were presented (BB session). In these sessions, the S? was chosen randoly. In the inal session, photographs o the owner s angry (S?) and lank (S-) ace were presented (AB session, Fig. 1e). Apparatus The experients were conducted in a roo ( ) at Azau University, which was ailiar to the suject dogs. Two A4 clear PVC olders were attached to the wall at the height o each dog s head (aout aove the ground). The distance etween the two olders was 0.1. The stiuli were presented as A4 photographs inside these olders in order to prevent dogs ro getting olactory cues. Beore presentation o the stiuli, the experienter ade the dogs sit and wait at the waiting position 1 away ro the wall. The dogs were not on a leash during the experients. In order to prevent the dogs ro oserving the exchange o stiulus photographs, the partition ( ) was set etween the dogs and wall, and it was reoved ater the exchange o stiulus photographs. In order to control possile social cues when dogs selected the photographs, the experienter stood 1.5 ehind the dogs and then cued the to select photographs (Fig. 2). When dogs selected S? photographs, they were rewarded with coercial dog treats. Procedure First, the dogs were trained to touch the experienter s hand with their nose at a cue, through a shaping procedure in which operant conditioning, using positive reinorceent (ood reward) and negative punishent (longer interval), was conducted. I the dogs did touch the experienter s hand with their noses, treats were given to the at the sound o a clicker, and i dogs did soething else, they were coanded to sit and stay or s. The dogs were then trained to touch the photograph o the experienter s hand in the clear older on the wall. Ater the dogs were accustoed to touching the clear older with their 1.5 noses, they were allowed to proceed to the irst training phase. The procedure o experients is shown in Tale 2. The irst training phase One positive (the ront ace o a ailiar person) and one negative (the ack o the head o the sae person) training stiuli were presented siultaneously. Two stiulus photographs were placed in the clear olders on the wall, and their positions were changed randoly once in each trial. Each session consisted o 30 trials. I dogs touch the S? directly, a reward was provided iediately, accopanied y a click sound. I dogs touched the S-, the experienter ade the dog sit and stay or s and then a correction trial was conducted. In a correction trial, once the stiulus photographs were hidden y the partition, we again presented the in identical positions as eore. Until dogs touch the S?, trials were repeated with photographs in identical positions. The dogs were transerred to the second training phase ater they relialy perored on a level at or eyond the learning criteria. The criterion o the irst training phase required C21 correct irst choices in 30 trials in three consecutive sessions. The second training phase Partition ( ) Fig. 2 The experiental design. The photographs were placed in 2 PVC olders attached to the wall, side-y-side, 10 c apart. The experienter ade the dog sit 1 away ro the wall and then oved 1.5 ehind the dog eore cueing it to select photographs A set o photographs o the owner s ace one siling ace (S?) and one lank ace (S-) was used repeatedly as the discriination stiulus. The procedure was the sae as or the irst training phase. The learning criterion o the second training phase required C24 correct irst choices in 30 trials

5 Ani Cogn (11) 14: over our consecutive sessions. The dogs that achieved this proceeded to the test phase. Each session o oth training phases included a 10-in reak. A session lasted approxiately in, and 2 4 sessions were conducted in 1 day, depending on the state o the dogs. Sessions were conducted 2 or 3 days in a week. The test phase The test phase coprised three kinds o session. To exaine whether the dog could discriinate the various photographs o the siling ace o the owner, we presented the 10 sets o siling and lank aces o the owner (OW session). The OW session consisted o 10 trials, and the sae photographs were not presented again. Then, whether the dog could discriinate the siling aces o unailiar persons was tested (UPSG session and UPOG session). These sessions also consisted o 10 trials. First, the OW session was conducted, then the UPSG, and UPOG sessions. These procedures were the sae as those o the training phases except or the nuer o trials, and the correction trials were not conducted. A session took approxiately 5 in, and one session was conducted per day. The control phase The control phase was conducted in order to veriy that: (1) the experienter did not provide cues to the sujects unawarely and (2) the dogs did not learn to choose the photograph o not a lank ace. The control phase consisted o our kinds o session. The procedure or all sessions was the sae as or the test session, and each session consisted o 10 trials. The irst session was a positive control session (PC session: siling (S?) and lank ace (S-) o owner). Next, or purpose (1), an SS session (the sae siling aces were presented siultaneously) was conducted, ollowed y a BB session (the sae lank aces were presented). The S? in the SS and BB sessions was decided randoly. I dogs selected S? eyond chance level even though there were no visually correct stiuli, they ight understand the experienter s unconscious cues to select S?. The last session was an AB session, in which the owner s angry (S?) and lank ace (S-) were presented or purpose (2). I dogs selected angry aces eyond the chance level, it eant that they ight not have selected S? y using the iage o the siling expression during the test phase. A session took approxiately 5 7 in, and one session was conducted in 1 day. Statistics In the test phase and the control phase, the coparisons o the correct response rate o all dogs and the chance level in each session were analyzed y a inoial test. The dierences etween the edians o response rates o sessions per phase were analyzed y a Kruskal Wallis test, and i signiicant dierences were shown, a Mann Whitney U test was used or the post hoc analysis. In the post hoc analysis, the signiicance level o 1.67%, in the case o the coparison o three sessions, and the signiicance level 0.8%, in the case o the coparison o our sessions, are equivalent to 5% (Bonerroni correction; SPSS v.17.0). Results were expressed as edians (interquartile range). The ages and nuer o trials in the training phases o dogs were expressed as ean ± SE. Results In the irst training phase, ive o the nine dogs (our ales and one eale, 47.4 ± 14.9 onths old) et the learning criterion. The ean o trials needed to eet the criterion was 4.6 ± 0.5 trials (Fig. 3a). The other our dogs showed an extreely strong lateral ias when selecting stiuli. Three dogs (Cody, Punch, and Fook) selected the let photograph with aove 90% (27/30 trial) at the irst ive sessions. Thereore, we judged that these dogs did not eet the criteria in the irst training phase. A ourth dog (Anita) oten selected the right photograph ut eventually stopped selecting the stiuli. Thus, they did not pass the irst training phase. In the second training phase, all ive dogs et the criterion; however, the oldest dog (Fleecia) needed 2 3 ties ore trials (41 trials) than did the other dogs (12, 14, 19 trials; Fig. 3). The ean nuer o trials needed to eet the criterion was.0 ± 5.4 trials. With the sall saple size used in this study, it was not possile to deterine whether these dierences were due to sex, age, reed, or to individual characteristics. In the test phase, the correct response rates in the OW session and in the UPSG session were signiicantly higher than the chance level y a inoial test (OW: ( 85)%, P \ 0.001, UPSG: (70 85)%, P \ 0.001). In the UPOG session, the correct response rate was 70(50 70)% and there was no signiicant dierence with the chance level (P = 0.12; Fig. 4a). As a result o the Kruskal Wallis test to copare the correct response rates o three sessions, a signiicant dierence was detected aong three sessions: v 2 (2) = 8.76, P = 0.01, and the edian correct response rate in the OW session was signiicantly higher than that in the UPOG session (Z =-2.74, P = 0.006; Fig. 4). There was no statistical dierence etween the edian correct response rate in the OW and in the UPSG session, and etween the UPSG and UPOG sessions (OW-UPSG: Z =-0.95, P = 0.34; UPSG-UPOG: Z =-2.13, P = 0.03).

6 530 Ani Cogn (11) 14: a % correct response % correct response Nuer o trials in the irst training phase Charley Cuppa Fleecia Jasine Kai Nuer o trials in the second training phase Fig. 3 The correct response rates in the training phases. a The criterion or successully copleting the irst training phase was C21 correct irst choices in 30 trials in three consecutive sessions. The criterion or successully copleting the second training phase was C24 correct irst choices in 30 trials in our consecutive sessions In the control phase, the correct response rate in all dogs in the PC session was ( )% and it was signiicantly higher than the chance level, using the inoinal test (P \ 0.001). There were no signiicant dierences etween the correct response rates in other sessions and the chance level (SS: ( )%, P = 0.89; BB: 50(35 )%, P = 0.89; AB: 30(25 55)%, P = 0.12; Fig. 5a). The Kruskal Wallis test showed that there was a signiicant dierence aong the edians o the correct response rate o our sessions (v 2 (3) = 12.08, P = 0.007). The edian correct response rate in the PC session was signiicantly higher than that in other sessions (PC vs. SS: Z = 2.69, P = 0.007; PC vs. BB: Z =-2.66, P \ 0.008; PC vs. AB: Z =-2.66, P \ 0.008); however, other coparison showed no signiicant dierences (Fig. 5). Discussion We conducted a 2-choice task to deterine whether dogs could learn to discriinate etween a siling and a lank huan ace and generalize the siling expression. In the irst training phase, our o the nine dogs were rejected ecause o strong lateral ias. Dogs usually show lateralized ehavior (Wells 03), and dogs deonstrated a natural gaze ias toward the let visual ield like adult huans when looking at huan aces (Guo et al. 09). In uture study, using a two-choice task, the issue o lateral ias will need to e addressed. All the rest ive dogs et the learning criteria o the irst training phase and the second training phase and proceeded to the test phase. In the test phase, all ive dogs were ale, signiicantly ore oten than expected y chance, to discriinate their owners siling aces ro their lank aces. When shown photographs o unailiar persons, they were also ale to signiicantly ore oten discriinate siling aces ro lank aces o unailiar persons o the sae gender as their owners than the chance level. However, their correct response rate in cases o unailiar persons o the opposite gender to their owners did not dier signiicantly ro the chance level, and it was signiicantly lower than in the case o their owners. Although alost hal o the dogs were a ** ** a % correctresponse %correctresponse 0 Charley Cuppa Fleecia Jasine Kai 0 Median OW UPSG UPOG Fig. 4 The correct response rate in the test phase. a In the OW session, the correct response rate in all dogs was C%; however, in the UPOG session, the correct response rate tended to decrease ( in ars eans an unailiar ale; eans an unailiar eale). Only the correct response rate in the UPOG session was not eyond the chance level and was signiicantly lower than that in OW session. **P \ (vs. chance level; inoial test) a P = Kruskal Wallis test, Mann Whitney U test (Bonerroni correction)

7 Ani Cogn (11) 14: Fig. 5 The correct response rate in the control phase. a The correct response rates showed declining trends except or the PC session. Only the PC session was signiicantly eyond the chance level. **P \ (vs. chance level; inoial test) a P \ Kruskal Wallis test, Mann Whitney U test (Bonerroni correction) a % correct response %correctresponse ** a 0 Charley Cuppa Fleecia Jasine Kai 0 Median PC(silingvs.lankace) SS(silngvs.silingace) BB(lankvs.lankace) AB(angryvs.lankace) rejected during the selection procedure and the test saple was sall, these results suggest that dogs can learn to discriinate etween siling and lank huan aces conditionally. In this study, we ust consider the dierence in the correct response rate depending on the unailiar person s gender. The dogs in this study were noral house dogs and lived with the owner and a aily eer, except in the cases o two dogs whose owners lived alone. These two dogs also had suicient daily contact with huans o oth genders. Thereore, it is not reasonale to assue that these dogs were unailiar with persons o the opposite gender to their owner. It was shown that dogs ehave toward huans in a gender-dependent anner in case o the presentation o the huan whole ody (Lore and Eisenerg 1986) and have the concept o gender on the onitor (Takaoka et al. 09). Proaly, soe sutle dierences in the structure o the ace etween ales and eales ay pose an ipedient (Cellerino et al. 04), and it ay have een soewhat diicult or dogs to understand the siling ace as a coon eature in ales and eales. One possile explanation was that the relationship with the particular owner resulted in gender-dependent generalization y dogs. Dogs can or extreely ailiative relationships with huans, and it is possile or dogs and huans to estalish the kind o ond seen etween eers o the sae species (Nagasawa et al. 09). Thereore, dogs place the greatest ocus on inoration gained ro their owner, with who they have the ost interaction on a day-to-day asis, and they ight peror ore poorly at the discriination task involving an unailiar person o the opposite gender to their owner. Unortunately, it was diicult to clariy the generalization o huan acial expression y dogs ecause o sall saple in the present study. It has een shown that dogs are ale to categorize acoustic and visual stiuli (Hener 1975; Range et al. 08), and it has een suggested that they have a concept o huan gender (Takaoka et al. 09). Thereore, the aility o dogs in generalization o huan acial expression ay e ale to e proven y saple addition. Additionally, in the present study, we tested only two reeds. Woer et al. (09) suggested that dog s huan-like social skills diered aong reeds; it has yet to e veriied whether there are reed-dependent dierences in the aility to discriinate etween huan acial expressions. In the control phase, the correct response rates o the sessions showing the sae expressions siultaneously (siling ace and siling ace in the SS session and lank ace and lank ace in the BB session), and these rates were signiicantly lower than those o the positive control sessions. These sessions were conducted in order to exclude the inluence o the experienter s unaware cues to the dogs to select the correct stiuli, and dogs could not select the S? ecause there were no visually correct stiuli. Although we cannot exclude a possiility that the experienter s unaware ehavior in these sessions, these control sessions suggested that the dogs used only the stiuli presented to the, not unaware cues ro the experienter, to select the S? in the test phase. In order to exclude the inluence ro the experienter, the ore appropriate apparatus, such as a onitor with a touchsensitive switch and autoatic eeder, is needed or uture studies. In the session using owners angry and lank aces (AB session), with the angry ace serving as the S? in the control phase, there was no signiicant dierence ro chance. This shows that during the training phases, the dogs had not learned to avoid selecting a lank ace ut had learned to select a siling ace. Now, we have to also ocus on the poor perorance o the dogs in the second training phase (training to

8 532 Ani Cogn (11) 14: discriinate etween the siling (S?) and lank ace (S-). Dogs are extreely sensitive to huan ehavior, and they can acquire associative learning y taking cues ro the sutle oveents o huans (Udell et al. 10). However, the second training phase (owner s siling ace vs. lank ace) necessitated ore sessions in order to eet the criterion than did the irst training phase (the ailiar ale s ace vs. the ack o the head). This ay have een solely due to the dierence in the size o the targets used or discriination. However, another possile explanation is that it is undaentally diicult or dogs to recognize huan acial expressions. Kanazawa (1996) ound that Japanese onkeys could discriinate huan happiness aces ut could not discriinate etween huan anger/disgust and sad aces. Kanazawa deduces that as huan siles resele the griace that shows suission in onkeys, the onkeys ay have used this griace to discriinate the happiness ace. Dogs and huans are located uch urther away on the evolutionary tree than are onkeys and huans, and the structure o a dog s ace diers greatly ro ours. Yet, canines also use acial expressions, such as the oveent o outh, ears, and eye area (Fox 1970), in their counication, and they have lived closely with huans. Thereore, dogs ight alternatively e ale to recognize oveent in huan acial coponents. However, the act that the ive could discriinate etween siling and lank aces in the test phase ight largely e attriutale to the ovious dierences in the acial eatures, or exaple, whether teeth were visile in the siling condition. There have een dierent suggestions regarding dog s cognition o huan aces: Guo et al. (09) suggested that dogs ay use a huan-like gaze strategy or the processing o huan acial inoration, while Racca et al. (10) showed the possiility that the techniques dogs use or the visual processing o acial stiuli are dierent than those huans or other priates use. Further studies are needed to deterine which acial parts are involved, how visually processing dogs discriinate huan acial expressions, and whether dogs really do categorize expressions. Other possile reason is that the stiuli in this experient were two-diensional photographs and were contextually insuicient; that is, perhaps they did not recognize the iages in the photographs as huan aces. Thereore, ecause we thought that the presentation o the stiuli y two-diensional photographs did not arouse the dog s aversive ehavior, the owner s angry ace photographs were eployed as the stiuli in the control phase. As a result, the correct response rate in the AB session angry ace (S?) versus lank ace (S-) was not higher than chance level. In previous studies that used threediensional and whole huan odies as stiuli, dogs showed the coordinated reaction without training (Vas et al. 05; Morisaki et al. 09). So the presentation o visual stiuli to dogs requires ore consideration in uture studies. The present study indicates that dogs undergo a learning process under the experiental condition, which enales the to discriinate huan siling aces ro lank aces. Aong huans, the aility to accurately recognize other people s expressions and judge their eotions is a vital social skill. This study has shown that dogs that live closely with huans are also ale to recognize positive acial expressions, indicating that these dogs have acquired the social skills helpul to survive. The aility to learn to discriinate huan acial expressions ust have helped dogs to adapt to huan society. Reerences Arantes R (1987) The expression o eotions in an and canid. J Sall Ani Pract 28: Adachi I, Kuwahata H, Fujita K (07) Dogs recall their owner s ace upon hearing the owner s voice. Ani Cogn 10:17 21 Bradshaw JWS, Nott HMR (1995) Social and counication ehaviour o copanion dogs. The doestic dog: its evolution, ehaviour and interactions with people. In: Serpell JA (ed) The doestic dog, its evolution, ehavior and interactions with people, 11th edn. Caridge University Press, Caridge, pp Call J, Brauer J, Kainski J, Toasello M (03) Doestic dogs (Canis ailiaris) are sensitive to the attentional state o huans. J Cop Psychol 117: Cellerino A, Borghetti D, Sartucci F (04) Sex dierences in ace gender recognition in huans. Brain Res Bull 63: Ekan P, Friesen WV (1975) Unasking the ace: a guide to recognizing eotions ro acial clues. Prentice-Hall Englewood Clis, NJ Fox MW (1970) A coparative study o the developent o acial expressions in canids; wol, coyote and oxes. Behaviour 36:49 73 Gacsi M, Miklosi A, Varga O, Topal J, Csanyi V (04) Are readers o our ace readers o our inds? Dogs (Canis ailiaris) show situation-dependent recognition o huan s attention. Ani Cogn 7: Guo K, Meints K, Hall C, Hall S, Mills D (09) Let gaze ias in huans, rhesus onkeys and doestic dogs. Ani Cogn 12:9 418 Hare B, Call J, Toasello M (1998) Counication o ood location etween huan and dog (Canis ailiaris). Evol Coun 2: Hener H (1975) Perception o iologically eaningul sounds y dogs. J Acoust Soc A 58:S124 Kanazawa S (1996) Recognition o acial expressions in a Japanese onkey (Macaca uscata) and huans (Hoo sapiens). Priates 37:25 38 Lore RK, Eisenerg FB (1986) Avoidance reactions o doestic dogs to unailiar ale and eale huans in a kennel setting. Appl Ani Behav Sci 15: Morisaki A, Takaoka A, Fujita K (09) Are dogs sensitive to the eotional state o huans? J Vet Behav Clin Appl Res 4:49

9 Ani Cogn (11) 14: Nagasawa M, Kikusui T, Onaka T, Ohta M (09) Dog s gaze at its owner increases owner s urinary oxytocin during social interaction. Hor Behav 55: Racca A, Aadei E, Ligout S, Guo K, Meints K, Mills D (10) Discriination o huan and dog aces and inversion responses in doestic dogs (Canis ailiaris). Ani Cogn 13: Range F, Aust U, Steurer M, Huer L (08) Visual categorization o natural stiuli y doestic dogs. Anial Cognit 11: Schwa C, Huer L (06) Oey or not oey? Dogs (Canis ailiaris) ehave dierently in response to attentional states o their owners. J Cop Psychol 1: Takaoka A, Morisaki A, Fujita K (09) Interodal representation in dogs. J Vet Behav Clin Appl Res 4:54 55 Udell MAR, Dorey NR, Wynne CDL (10) What did doestication do to dogs? A new account o dogs sensitivity to huan actions. Biol Rev 85: Vas J, Topál J,Gácsi M, Miklósi A, Csányi V (05) A riend or an eney? Dogs reaction to an unailiar person showing ehavioural cues o threat and riendliness at dierent ties. Appl Ani Behav Sci 94: Viranyi Z, Topál J, Gácsi M, Miklósi A, Csanyi V (04) Dogs respond appropriately to cues o huans attentional ocus. Behav Process 66: Wells DL (03) Lateralised ehaviour in the doestic dog, Canis ailiaris. Behav Process 61:27 35 Woer V, Hare B, Koler-Matznick J, Wrangha R, Toasello M (09) Breed dierences in doestic dogs (Canis ailiaris) coprehension o huan counicative signals. Interact Stud 10:6 224