Anim Cogn DOI 10.1007/s10071-010-0340-y
SHORT COMMUNICATION
Are dogs (Canis familiaris) misled more by their owners than by strangers in a food choice task? S. Marshall-Pescini • E. Prato-Previde P. Valsecchi
•
Received: 16 January 2010 / Revised: 1 July 2010 / Accepted: 2 July 2010 Ó Springer-Verlag 2010
Abstract Dogs are highly skilled in understanding a large variety of human social cues and use them appropriately to solve a number of different cognitive tasks. They rely on human signals even when these are contradictory or misleading and ultimately prevent them from correctly solving a task. In the following two experiments, we investigated whether the owner and a stranger differently influenced dogs’ choices in food discrimination tasks. In Experiment 1, 48 dogs were tested in 3 different conditions: (1) choice between a large and a small amount of dog pellets with no demonstration; (2) choice between a large and a small amount of dog pellets after having witnessed the owner/ stranger favouring the small quantity; (3) choice between two single food pellets after observing the owner/stranger choosing one of them. In Experiment 2, 48 dogs could choose between two foods of different palatability: in Condition 1, dogs chose between a slice of sausage and a dry pellet with no demonstration. In Condition 2, the same choice was available but with a person (owner/stranger) showing a preference for the dry pellet. In Condition 3, dogs chose between a single dry pellet and 8 slices of sausage, with the person (owner/stranger) showing a preference for the pellet. In both experiments, dogs conformed to the human’s indications even though these led to the selection of the less advantageous option (i.e. the smaller
S. Marshall-Pescini (&) E. Prato-Previde Dipartimento di Scienze e Tecnologie Biomediche, Sezione di Psicologia, Universita` di Milano, Via Fratelli Cervi, 93, 20090 Segrate, MI, Italy e-mail:
[email protected] P. Valsecchi Dipartimento di Biologia Evolutiva e Funzionale, Sezione di Biologia Animale, Universita` di Parma, Viale G.P. Usberti, 11/A, 43100 Parma, Italy
amount of food in Experiment 1 or the low quality food in Experiment 2). However, the owner and the stranger did not differently influence the dogs’ behaviour. Results show that dogs are willing to follow a person’s indication even when this is visibly (if perhaps only mildly) counterproductive to them and that they are socially prepared to rely equally on cues given by the owner and an unfamiliar friendly person. Keywords Dog Inter-specific communication Owner Stranger Food discrimination task
Introduction Studies on social cognition in domestic dogs have shown that dogs are highly sensitive to a large variety of human social cues and use them appropriately to solve a number of different cognitive tasks. In particular, dogs appear sensitive to such goal-directed actions as pointing, head turning, nodding, bowing, glancing and a human placing a token on the target location in object choice and food location tasks (Agnetta et al. 2000; Hare et al. 1998; Miklosi and Soproni 2006; Soproni et al. 2002). The tendency in dogs to respond to such cues appears particularly strong since they follow a pointing gesture to an empty food location, despite having seen the correct one being baited (Szetei et al. 2003). The strong influence that human gestures and communicative signals have on dogs’ performance in cognitive tasks was further confirmed in a previous study by our group using a food discrimination task (Prato-Previde et al. 2008). In a free choice between two plates containing a large and a small food quantity, dogs would predominantly choose the large, but when their owners showed a manifest preference for the smaller one,
123
Anim Cogn
they conformed their choice to the owners’ indication even though this choice was visibly counterproductive. Interestingly, those dogs that scored higher in the ‘attachment’ factor of a temperament evaluation questionnaire (C-BARQÓ, Hsu and Serpell 2003) followed their owners’ indications significantly more in the counterproductive condition than the other dogs. This suggests the existence of a relationship between the effectiveness of human indication and the nature of the bond with the person providing the cues. Although there is clear experimental evidence that dogs form a preferential relationship with their owner (Prato-Previde et al. 2003; Palmer and Custance 2007), very few studies have employed the owner instead of an unfamiliar researcher as the person giving the social cues during testing (e.g. Kubinyi et al. 2003). Two studies compared the effects of owner versus stranger demonstrations employing the detour task (Pongra´cz et al. 2001, 2004) reporting no preferential learning from either one or the other. However, in the extinction and in the reversal learning of a response to the pointing gesture in an object choice task, the familiarity of the person who emits the cues plays an important role (Elgier et al. 2009). The aim of this study was to compare a stranger’s and the owner’s influence on dogs’ food choice adopting the experimental paradigm of the previous study (PratoPrevide et al. 2008). In experiment 1, the choice was between two plates containing the same quality of food (i.e. commercial dry dog pellets) and indications were given by either the owner or the stranger; in experiment 2, the choice was between two plates, one containing dog pellets (standard palatability) and the other containing slices of sausage (high palatability), thus increasing the cost of the dogs’ choice. This could potentially make the role of the owner in directing the dogs’ choice more relevant compared to that of the stranger. Based on the evidence that dogs show social attachment to the owner and considering our previous results (Prato-Previde et al. 2003, 2008), we predicted that dogs would be influenced more strongly by the owner than by a stranger.
Experiment 1 Method Subjects Dogs were recruited through personal contacts and advertisements in parks and veterinary surgeons. All dogs were kept for companionship, lived within the human household and had either no or only basic levels of training. None of the dogs had previously participated in cognitive studies. Forty-eight dogs were divided into two groups:
123
Group 1 (tested with the owner) consisted of 14 males and 10 females, whose age ranged from 6 months to 10 years (mean 5.1 years); Group 2 (tested with the stranger) consisted of 15 males and 9 females, whose age ranged from 6 months to 10 years (mean 5.4 years; see the Sect. ‘‘Appendix’’ for details of breeds represented). Procedure The testing took place in an outdoor enclosed testing area, approximately 25 m2, unfamiliar to the dogs at Parma University. Prior to testing, dogs were allowed to freely explore the area whilst the experimenter carefully instructed owners on the procedure. To be sure that the dogs would be sufficiently motivated to perform the food choice task, the owners were asked not to feed their dogs at least 4 h prior to testing. The palatability of the food used was always evaluated by offering the dog a few pieces prior to testing. We used dry food pellets of medium size for large-medium dogs and of mini size for small dogs (Purina, Friskies either ‘Adult Complete or ‘Mini Menu’; same composition and palatability), presented in two small white plastic dishes (10 cm in diameter). Two different quantities of food were used during testing: small, consisting of a single piece of food; and large, consisting of eight pieces of food. The behaviour of dogs during testing was video recorded. The procedure consisted of three conditions: Condition 1 – the free choice test between the large and small food quantity; Condition 2 – the choice between the large and small food quantity with the owner/stranger clearly showing a preference for the small one; and Condition 3 the choice between the two equally small quantities of food with the owner/stranger showing a preference for either one or the other (see below for the detailed description of each condition). All dogs received a total of 9 trials, i.e. three trials in each condition. The position of the large and small food quantity was counterbalanced both in Condition 1 and Condition 2. In Condition 3, the position of the person was also counterbalanced in the same manner. In all conditions, the dogs were on a leash and were held 1.5 m from the axis on which the plates were set; the plates were set 1 m apart. Between conditions, there was a 2-min interval during which the dog was offered a bowl of fresh water and was allowed to walk around freely. Outlined below is a detailed description of the three conditions. Condition 1 Free choice test between a large and a small food quantity. The owner held the dog’s leash and was instructed to avoid interacting/talking to it. The experimenter entered the arena in front of the dog and placed the two plates of food in the allocated spots, never facing the
Anim Cogn
dog and taking care never to look at either one or other of the plates. She backed off a few steps and waited until the dog had visually examined the content of both plates, then instructed the owner to drop the leash to allow the dog to move towards the plates. Whilst the dog fed from the chosen plate, the non-chosen one was removed by the experimenter. Condition 2 Choice between a large and a small food quantity with owner/stranger influence. Owner group The dog was held by an assistant, the owner was at her left side, and the experimenter entered the area presenting the plates to the dog as in Condition 1 and withdrew from the arena. After the experimenter left the arena, the owner walked between the two plates, turned towards the dog, approached the plate containing the small food quantity, picked up the piece of food and with an enthusiastic tone of voice, said: ‘‘Oh wow, this is delicious, lovely, so good!’’. Then he/she placed the piece of food back on the plate, backed off a few steps, without further looking at the dog. The assistant released the dog. As in Condition 1, the non-chosen plate was removed. Stranger group The dog was held by the owner, and the experimenter entered the arena presenting the plates to the dog as in Condition 1 and left the arena. The assistant with the same movements described for the owner, approached the plate containing the small food quantity, picked up the piece of food and with an enthusiastic tone of voice, said: ‘‘Oh wow, this is delicious, lovely, so good!’’. Then he/she placed the piece of food back on the plate and backed off a few steps without further looking at the dog. The owner released the dog. As in Condition 1, the non-chosen plate was removed. Condition 3 Choice between two single pieces of food with owner/stranger influence. The procedures for both the stranger and owner groups were identical to those described in Condition 2 with the only difference that the owner/ stranger approached either the right- or the left-hand plate following the experimenter’s instructions.
The order of presentation of the three conditions (Condition 1, 2, 3 in sequence) remained constant since in the previous study no difference had emerged between two presentation orders (Prato-Previde et al. 2008). Data analysis Given the dogs’ choice could be determined without ambiguity, the scoring was carried out from videotapes by a single observer who also checked that dogs had truly looked at both plates before being released. We recorded the number of times a dog chose the large food quantity in Condition 1 and 2 and the number of times it followed the owner’s indication in Condition 2 and 3 (i.e., dogs scored 0 to 3 in each condition). Since the data were not normally distributed, non-parametric tests were used. We used the Mann–Whitney test for between-groups analysis, the Wilcoxon test for within-groups analysis, the one-sample sign test to assess whether dogs as a group performed differently from chance within each condition. Results and discussion No difference emerged between owner and stranger influence on the dogs’ performance in either Condition 2 or 3 (Mann–Whitney test n1 = 24, n2 = 24; Condition 2: z = 0.99, P = 0.32; Condition 3: z = 0.24, P = 0.81). Thus, data of the two groups were pooled for further analysis. No age or sex differences emerged in any condition. In the free choice condition (Condition 1), 79% of dogs chose the large quantity more often than the small one (i.e. two or three times; Table 1): thus, dogs as a group chose the large quantity significantly more often than the small quantity (one-sample sign test, n = 48, P \ 0.0001). After observing the person (owner or stranger) expressing a preference for the small food quantity (Condition 2), the dogs’ choice changed and 52% of dogs chose the small quantity more often than the large following the person’s choice. In this condition, the group of dogs
Table 1 Number of dogs choosing the best option in Condition 1 (the larger quantity in Exp. 1 and the best quality food in Exp. 2), and following the person (owner or stranger) in Condition 2 and 3 Condition 1 0/3 Experiment 1 Experiment 2
1/3
Condition 2 2/3
3/3
0/3
Condition 3
1/3
2/3
3/3
0/3
1/3
2/3
3/3
Owner
0
5
7
12
4
8
10
2
0
3
9
12
Stranger
0
5
8
11
0
11
9
4
0
3
10
11
Owner Stranger
1 0
6 7
8 11
9 6
5 1
2 3
6 7
11 11
8 9
4 6
8 5
4 4
123
Anim Cogn
performed at chance level in the discrimination task (onesample sign test, n = 48, P = 0.89). The comparison between Condition 1 and Condition 2 showed that dogs chose the large quantity of food significantly more often when they could choose freely than when the person showed a preference for the small food quantity (Wilcoxon signed-rank test, n = 48, z = 4.14, P \ 0.001). In Condition 3, when both plates contained a single piece of food, 87.5% of dogs chose the person’s preferred plate more often than the other one. As a whole, the group of dogs significantly followed the person’s choice (onesample sign test, n = 48, P \ 0.0001). Furthermore, dogs followed the person’s choice significantly more in Condition 3 than in Condition 2 (large vs. small food quantity; Wilcoxon signed-rank test, n = 48, z = 4.01, P \ 0.001). Overall results are comparable with previous findings showing that dogs are highly influenced by humans in a neutral situation and are also significantly influenced by them in the counterproductive condition. However, interestingly, the owner and the stranger had comparable effects on the dog’s performance, suggesting that, at least in this task, the human influence is independent of the relationship bond.
Experiment 2 The experimental paradigm was modified so as to make the choice even more counterproductive for dogs when following the person’s indications. To this aim, both the quality and the quantity of food was varied: thus in the most extreme case, dogs had to choose between a dry food pellet and eight slices of sausage, with either the owner or the stranger indicating the food pellet as the best option. We hypothesized that increasing the cost of the dogs’ choice, a difference would emerge between owner and stranger influence, with dogs being prepared to follow their owner’s suggestions more willingly than a stranger’s. Method Subjects Dogs’ recruitment and criteria selection were the same as in Experiment 1. Forty-eight naive dogs were divided into two groups: Group 1 (tested with owner) consisted of 11 males and 13 females, whose age ranged between 6 months to 11 years (mean 4.2 years) Group 2 (tested with the stranger) consisted of 14 males and 10 females, whose age ranged between 6 months to 11 years (mean 3.9 years)
123
Procedure and data analysis The procedure was identical to that described in Experiment 1 except for the content of the plates. In Condition 1, dogs were given a choice between a plate containing a single dry food pellet and another plate containing a single slice of sausage (of comparable size and colour); in Condition 2, the same choice was available to the dog, but in this case, the person (owner/stranger) showed a preference for the dry food pellet; in Condition 3, the dogs could choose between a single food pellet and 8 slices of sausages, but the person showed a preference for the single food pellet. Data analysis was carried out as for Experiment 1. Results and discussion No difference emerged between owner and stranger influence on the dogs’ performance in either Condition 2 or 3 (Mann–Whitney test n1 = 24, n2 = 24, Condition 2: z = 0.61, P = 0.53; Condition 3: z = 1.03, P = 0.3). Thus, data of the two groups were pooled for further analysis. No age and sex effect was found in any condition. In the free choice condition (Condition 1), 71% of dogs chose the sausage more often than the dry pellet (i.e. two or three times; Table 1). Thus, dogs as a whole chose the sausage significantly more often than the dry pellet (onesample sign test, n = 48, P \ 0.01). After observing the person expressing a preference for the dry pellet (Condition 2), the dogs’ choice changed and 72% of dogs chose the dry food pellet following the person’s preference. In this condition, the group of dogs performed significantly above chance in the direction of the person’s indication (one-sample sign test, n = 48, P \ 0.001). The comparison between Condition 1 and Condition 2 showed that dogs chose the sausage significantly more often when they could choose freely than when the person showed a preference for the dry food pellet (Wilcoxon signed-rank test, n = 48, z = 4.41, P \ 0.001). In Condition 3, when one plate contained eight pieces of sausage and the other one single food pellet, 56% of dogs preferentially chose the sausage plate: in this condition, the group of dogs performed at chance level in the discrimination task (one-sample sign test, n = 48, P = 0.47). Furthermore, a comparison between Condition 2 and Condition 3 revealed that dogs followed the person’s choice significantly less in Condition 3 when the choice was between following the person’s suggestion and eating a single dry food pellet or ignoring the person and obtaining eight bits of sausage (Wilcoxon signed-rank test, n = 48, z = 3.71, P \ 0.001). To understand whether the presence of a palatable food as the sausage made the choice more relevant to the dogs and the indication of the humans more counterproductive
Anim Cogn
Fig. 1 Dogs’ choices matching the owner’s and the stranger’s in the most counterproductive conditions of Experiment 1 (large vs small quantity of dry food) and Experiment 2 (large quantity of sausages vs single dry food pellet)
for them, we compared the dogs’ performance in Experiment 1 and 2. In the free choice task of Experiment 1 (one vs eight food pellets), 79% of dogs chose the larger quantity, whereas in Experiment 2 (one food pellet vs one slice of sausage), 71% of dogs chose the sausage; thus in both cases, there was indeed a clear preference for what researchers imagined would be considered the ‘best choice’ by the dogs. When comparing the two most counterproductive conditions across studies (i.e. Experiment 1, small vs large quantity of dry pellets and Experiment 2, one dry pellet vs eight slices of sausage), no significant difference emerged in the probability of dogs following their owner (Mann–Whitney test n1 = 24, n2 = 24, z = 0.25, P = 0.8); but dogs were, though not significantly, less inclined to follow the stranger when having to give up lots of sausages for a single dry pellet than when they were giving up the large quantity of dry pellets for the smaller one (Mann–Whitney test n1 = 24, n2 = 24, z = 1.9, P = 0.054; Fig. 1).
General discussion This study aimed at investigating whether communicative cues provided by a familiar (the owner) or an unfamiliar (the experimenter) person differentially influenced dogs’ performance in a food discrimination task. The role of familiarity between the person and the dog engaged in inter-specific communication has received little attention, despite the evidence that dogs develop a preferential relationship with the owner with whom they share life and experiences. Contrary to our prediction, our results did not highlight any differential effect of owner and stranger indications in directing dogs’ behaviour, both when facing a choice between large vs small quantity of the same food (Experiment 1) and when facing a choice between foods of different palatability (Experiment 2). In particular, in both
experiments, dogs were equally misled towards the less favourable choice (small quantity/less palatable food) by the cues received from the owner and the stranger, although comparing the two most costly conditions dogs seemed, though not significantly, less prone to follow a strangers’ indications. Similarly, Range et al. (2009) in a study on social attention towards a model demonstrating food-directed behaviour found that dogs tended to look longer at the model when he/she was the owner rather than a stranger. Elgier et al. (2009) showed that in the acquisition phase of an object choice task, dogs did perform equally well when either the owner or a stranger pointed to the baited container, but that a difference emerged in the extinction and in the reversal learning. Dogs that received the cue from their owners extinguished the response of going to the pointed container slower and learned the reversal response faster than dogs that received the cue from a stranger. On the contrary, in a detour task, Pongra´cz et al. (2001) found that owners and strangers were equally effective as demonstrators. Overall, these contradictory results suggest that the relationship of the dog with the person who emits the cues could influence dogs’ performance, but this influence might be context and task dependent. In our study, the lack of a differential effect of owner and stranger in directing dogs’ choices could be due to several factors: the testing condition was not stressful enough to trigger a differential attachment reaction, since the owner was constantly present throughout the test; all subjects were pet dog with no reason to neither fear nor mistrust humans, since humans in their lives are usually a source of food, comfort and protection, thus even a stranger may be rapidly considered a ‘friend’; the cost of following the misleading indications of the persons was not high enough to promote differential behaviour in response to the owner and to the stranger. Last but not least, the communication modality used by the persons included a combination of cues such as talking to the dog with an enthusiastic tone of voice whilst looking at it and simultaneously manipulating the food. A number of studies have indicated the importance of ostensive communication in the dogs’ learning process from humans. It has been shown that dogs are faster in socially learning a detour task if the human demonstrator talks to them, holds the desired object so it is visible to the dog and maintains eye contact throughout the demonstration than if the demonstration occurred without any communicative signals (Pongracz et al. 2004). Two studies have also suggested that dogs may be so sensitive to human communicative cues as to ignore their own individual perception of reality (Szetei et al. 2003; Topa´l et al. 2009). Thus, it is possible that in this study, the potentially different effects of owner and stranger were overshadowed by the strength of the communicative cues.
123
Anim Cogn Acknowledgments This research was supported by funds from the Universita` di Milano to Emanuela Prato-Previde and by funds from Universita` di Parma to Paola Valsecchi and MIUR (PRIN 2006). A special thank to Chiara Zanibelli for her invaluable help in data collection. Finally, we would like to thank all the owners and dogs that participated as volunteers. This research complies with the current Italian laws on animal welfare.
Appendix Breeds of participating dogs Experiment 1: 2 Golden Retriever, 7 Labrador Retriever, 1 Flatcoated Retriever, 1 Chesapeake Bay Retriever, 1 Nova Scotia Duck-tolling Retriever, 1 Dobermann, 4 Terranova, 1 Border Collie, 1 Bernese Mountain Dog, 1 Shetland Sheepdog, 2 German Shepherd, 1 Scottish Collie, 1 Jack Russell, 1 Toy Poodle, 1 St. Bernard Dog, 1 Cocker Spaniel, 1 Spinone Italiano; 1 Australian Shepherd, 1 Afghan Hound, 1 Rottweiler and 19 mixed breed. Experiment 2: 3 Beagle, 1 Scottish Collie, 1 English Setter, 1 Argentinean Dogo, 1 Bernese Mountain Dog, 3 Dachshund, 4 Labrador Retriever, 2 Golden Retriever, 1 Dalmatian, 3 English Bulldog, 1 American Cocker Spaniel, 1 English Cocker Spaniel, 1 Miniature Schnauzer, 1 Miniature Poodle, 1 Czechoslovakian Wolfdog and 23 mixed breed.
References Agnetta B, Hare B, Tomasello M (2000) Cues to food locations that domestic dogs (Canis familiaris) of different ages do and do not use. Anim Cogn 3:107–112 Elgier AM, Jakovcevic A, Mustaca AE, Bentosela M (2009) Learning and owner-stranger effects on interspecific communication in domestic dogs (Canis familiaris). Behav Proc 81:44–49
123
Hare B, Call J, Tomasello M (1998) Communication of food location between human and dog. Evol Comm 2:137–159 Hsu Y, Serpell JA (2003) Development and validation of a questionnaire for measuring behavior and temperament traits in pet dogs. JAVMA 223:1293–1300 ´ , Csa´nyi V (2003) Dogs (Canis Kubinyi E, Topa´l J, Miklo´si A familiaris) learn from their owners via observation in a manipulation task. J Comp Psychol 117:156–165 Miklosi A, Soproni K (2006) A comparative analysis of animals’ understanding of the human pointing gesture. Anim Cogn 9:81–93 Palmer R, Custance D (2007) A counterbalanced version of Ainsworth’s strange situation procedure reveals secure-base effects in dog-human relationships. Appl Anim Behav Sci 109:306–319 ´ , Kubinyi E, Gurobi K, Topa´l J, Csa´nyi V Pongra´cz P, Miklo´si A (2001) Social learning in dogs. The effect of a human demonstrator on the performance of dogs (Canis familiaris) in a detour task. Anim Behav 62:1109–1117 ´ , Timar-Geng K, Csa´nyi V (2004) Verbal Pongra´cz P, Miklo´si A attention getting as a key factor in social learning between dog (Canis familiaris) and human. J Comp Psychol 118:375–383 Prato-Previde E, Custance DM, Spiezio C, Sabatini F (2003) Is the dog-human relationship an attachment bond? An observational study using Ainsworth’s strange situation. Behav 140:225–254 Prato-Previde E, Marshall-Pescini S, Valsecchi P (2008) Is your choice my choice? Owners’ influence on the pet dogs’ performance in a food choice task. Anim Cogn 11:167–174 Range F, Horn L, Bugnyar T, Gajdon GK, Huber L (2009) Social attention in keas, dogs, and human children. Anim Cogn 12:181–192 ´ , Topa´l J, Csa´nyi V (2002) Dogs’ (Canis Soproni K, Miklo´si A familiaris) responsiveness to human pointing gestures. J Comp Psychol 116:27–34 ´ , Topa´l J, Csa´nyi V (2003) When dogs seem to Szetei V, Miklo´si A lose their nose: an investigation on the use of visual and olfactory cues in communicative context between dog and owner. Appl Anim Behav Sci 83:141–152 Topa´l J, Gergely G, Erdo}hegyi A, Csibra G, Miklo´si A (2009) Differential sensitivity to human communication in dogs, wolves and human infants. Science 325:1269–1272
