Quantity discrimination in canids: Dogs (Canis familiaris) and wolves (Canis lupus) compared

Accepted Manuscript Title: Quantity discrimination in canids: Dogs (Canis familiaris) and wolves (Canis lupus) compared Authors: Miletto Petrazzini Maria Elena, Clive D.L. Wynne PII: DOI: Reference:

S0376-6357(17)30191-2 http://dx.doi.org/10.1016/j.beproc.2017.09.003 BEPROC 3504

To appear in:

Behavioural Processes

Received date: Revised date: Accepted date:

22-4-2017 9-8-2017 5-9-2017

Please cite this article as: Elena, Miletto Petrazzini Maria, Wynne, Clive D.L., Quantity discrimination in canids: Dogs (Canis familiaris) and wolves (Canis lupus) compared.Behavioural Processes http://dx.doi.org/10.1016/j.beproc.2017.09.003 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Quantity discrimination in canids: dogs (Canis familiaris) and wolves (Canis lupus) compared

Miletto Petrazzini Maria Elena1* and Clive D. L. Wynne2 1

University of Padova, Department of General Psychology, Padova, Italy.

2

Arizona State University, Department of Psychology, Tempe, AZ, U.S.A.

*Contact: Phone: +39 0498276957; fax: +39 0498276600; e-mail: [email protected]

Highlights  Dogs and wolves were presented with a spontaneous choice between two sets containing different quantities of food items  Four numerical contrasts with increasing numerical ratio were presented: 1 vs. 3 (0.33 ratio), 2 vs. 4 (0.5), 2 vs. 3 (0.67) and 3 vs. 4 (0.75)  Dogs’ accuracy was affected by the numerical ratio, while no ratio effect was found in wolves  These results suggest the existence of different quantitative abilities in dogs and wolves

1

ABSTRACT Accumulating evidence indicates that animals are able to discriminate between quantities. Recent studies have shown that dogs’ and coyotes’ ability to discriminate between quantities of food items decreases with increasing numerical ratio. Conversely, wolves’ performance is not affected by numerical ratio. Cross-species comparisons are difficult because of differences in the methodologies employed, and hence it is still unclear whether domestication altered quantitative abilities in canids. Here we used the same procedure to compare pet dogs and wolves in a spontaneous food choice task. Subjects were presented with two quantities of food items and allowed to choose only one option. Four numerical contrasts of increasing difficulty (range 1-4) were used to assess the influence of numerical ratio on the performance of the two species. Dogs’ accuracy was affected by numerical ratio, while no ratio effect was observed in wolves. These results align with previous findings and reinforce the idea of different quantitative competences in dogs and wolves. Although we cannot exclude that other variables might have played a role in shaping quantitative abilities in these two species, our results might suggest that the interspecific differences here reported may have arisen as a result of domestication.

KEYWORDS: Quantity discrimination, spontaneous choice, dogs, wolves

2

1.

INTRODUCTION

Recent studies show that the ability to discriminate quantities is widespread in the animal kingdom. Comparing quantities is important in many contexts, such as optimising food intake, choosing the best social group to reduce predation risk, and regulating intergroup interactions based on number of opponents (reviewed in Geary et al., 2014). However, the discrimination between “more” and “fewer” items could be achieved relying on non-numerical cues (e.g., overall space occupied by the sets) that co-vary with numerosity (Cantlon & Brannon, 2007; Henik, 2016), rather than on absolute numbers (i.e., exactly 2 vs. 4 items). Thus, numerical cognition studies commonly distinguish between “quantity discrimination”, that refers to the condition in which animals are able to discriminate between more and fewer items but without evidence of a use of numbers, and “numerical discrimination”, the case in which animals make use of numerical information when the physical attributes are controlled for (Agrillo, 2014). Although quantitative abilities have been studied in many species (for a review see Agrillo & Bisazza, 2014), few data are available on canids. Recent studies have shown that dogs are able to spontaneously discriminate small quantities of food items, but their performance decreases as the numerical ratio between the smaller and the larger set approaches 1, in agreement with Weber’s law (Ward & Smuts, 2007; Baker et al., 2012; Range et al., 2014; Miletto Petrazzini & Wynne, 2016) as observed in other vertebrates (for a review: Agrillo & Bisazza, 2014). Coyotes, like dogs, are able to select the larger of two visible quantities, with accuracy decreasing as the ratio increases (Baker et al., 2011). Interestingly, wolves were found to respond accurately independent of the ratio between the number of objects sequentially presented (Utrata et al., 2012), in contrast to dogs tested in the same paradigm (Range et al., 2014). Range et al. (2014) adopted the same experimental setup and procedure previously used with wolves in which subjects had to choose between two sets of food items (1 to 4) sequentially inserted in two opaque containers. In this way, the dogs could never have a global view of the entire contents of both sets and were required to enumerate the items 3

presented in the two containers rather than using non-numerical cues (e.g., overall volume of food) to choose the larger option. Control for handling time was also set up to rule out the possibility of dogs using the length of time required to insert the different number of food items into the containers. Dogs proved able to select the larger amount of food only up to 0.5 ratio and their performance was strongly affected by the numerical ratio, contrary to wolves. These results might suggest the existence of different quantitative abilities between dogs and wolves. Several studies showed that numerical acuity of non-human animals is affected by the procedure used (Agrillo & Bisazza, 2014). However, even when animals are tested in identical conditions, other factors intrinsic to the tested species may prevent a fair comparison (Agrillo et al., 2012). To date, there is increasing consensus in the scientific community about the importance of replication studies. Although considered less publishable than original research papers (Neuliep & Crandall, 1993), replication is one of the main principles of science and is essential to validate previous findings before drawing inappropriate conclusions about a species (for a discussion see Agrillo & Miletto Petrazzini, 2012). In this study we tested dogs and wolves with the same paradigm as Range et al., (2014), but using a different procedure (simultaneous and not sequential food presentation) to determine whether the results previously obtained are not simply an artifact of the methodology employed. We recorded subjects’ spontaneous tendency to select the array with more food items. The same numerical ratios were presented to both species (0.33, 0.50, 0.67 and 0.75): if quantitative skills are similar in dogs and wolves, the same impact of numerical ratio is expected in the performance of the two species.

2.

MATERIALS AND METHODS

2.1. Subjects 4

Five pet dogs and seven human-socialized wolves participated in this study (Table 1). Dogs were living with their owners and tested individually in a familiar room at Camp Marlin Doggie Daycare. All wolves were housed in outdoor enclosures at Wolf Park, Battle Ground Indiana, USA and were tested individually in a familiar enclosure.

2.2. Experimental procedure On each trial, one experimenter (E1) baited two plates with a different number of food treats while a second experimenter (E2) held the subject at a distance of 1.5 m from the midline between the plates. After baiting, E1 approached the subjects and placed both plates on the ground 1 m apart and equally distant from the subject. Subsequently E1 returned to the starting position and after 5 s instructed E2 to release the subject so that it could make its choice (by approaching one plate and eating its contents). Once the subject had chosen one plate, the other was quickly removed by E1. In order to avoid any potential cueing, E1 never looked at the subject while placing the two plates. In a pre-trial session we presented 1 vs. 0 food treats for four consecutive trials with item position counterbalanced over trials to screen for side bias (Miletto Petrazzini & Wynne, 2016). Two more trials were added if subjects did not select the item in all trials. Only subjects that selected the item at least twice on each side were admitted to the experiment. Two wolves (Wolfgang, Wotan) were excluded at this stage because of a 100% side bias. The final sample consisted of five dogs and five wolves. The food used was soft dog treats for dogs and summer sausage for wolves. Before testing the dogs, the palatability of the food was evaluated by offering the subjects a few pieces. For all wolves, we used summer sausage based on the suggestion of the Wolf Park staff as a suitable treat for the wolves. Both dogs and wolves were motivated by the food treats used, readily eating all food offered. The food used with both species was cut into equal-sized (1.5 x 1.5 cm) pieces and arranged in two white plastic plates (22 cm Ø). In this way the total amount of food (i.e., volume) 5

correlated with numerosity and the set with the larger number of food items also had the greatest amount of edible food (e.g., the ratio between the amount of food in 1 vs. 3 contrast was equal to 0.33). As a consequence, discrimination could be based on either or both numerical and nonnumerical information. The spatial arrangement of the food items was changed across trials to prevent the subjects from using pattern recognition when making their choices. To this aim, the experimenter arranged the individual pieces as pseudo-randomly as possible, but two pieces never touched each other. During the quantity discrimination task, all subjects were presented with four numerical contrasts with increasing numerical ratio: 1 vs. 3 (0.33 numerical ratio), 2 vs. 4 (0.50), 2 vs. 3 (0.67) and 3 vs. 4 (0.75). Both wolves and dogs underwent a total of 28 trials (7 for each contrast), across 4 sessions, with numerical comparisons intermixed across trials. Subjects were not fed for at least 4 hours prior testing. Stimuli were presented in a pseudo-random sequence with the larger/smaller stimulus never presented more than twice in a row on the same side. The left–right position of the stimuli was counterbalanced over trials. We performed both individual and group analyses. Individual performance was analysed through binomial tests on frequencies of correct choices. For group analyses we computed the proportion of choices for the larger quantity over the smaller quantity. This variable was arcsine square root transformed (Sokal & Rohlf 1995). To assess whether the overall proportion of choices of the larger food quantity was different from that expected by chance, we used a one-sample t test. To compare the discrimination abilities of wolves and dogs, we run a linear mixed model (LMM) with species and numerical ratio as fixed factors, and subject as a random factor. The effect of the numerical ratio within both species was examined with a linear mixed model (LMM). All the statistical tests were two-tailed, α was set up at 0.05 and data were analysed using SPSS 24.0.

6

3.

RESULTS

Both wolves and dogs’ preference for the larger quantity of food items was above chance level (mean ± std. dev., respectively: 0.73 ± 0.10, one sample t test, t(4) = 4.57, P = 0.01; 0.69 ± 0.10, t(4) = 4.09, P = 0.01). The LMM model on the accuracy of wolves and dogs as a function of ratio revealed a main effect of numerical ratio (F

(3,8)

= 10.39, P = 0.004) but not a significant effect of

species (F(1,8) = 0.20, P = 0.70). The interaction was marginally significant (F(3,8) = 3.18, P = 0.08). As shown in Fig. 1, wolves’ ability to discriminate between quantities was not affected by the ratio between the two sets (LMM: F (3,4) = 3.10, P = 0.15). Wolves performed above chance level in all numerical contrasts (all Ps < 0.05). At the individual level, three wolves showed a statistically significant preference for the larger set of items, one wolf showed a tendency for the larger quantity and the other one did not select it more than chance (Table 2). Dogs’ accuracy was significantly influenced by the numerical ratio (F

(3,4)

= 26.08, P = 0.004)

and the subjects significantly selected the set with more food items only when presented with 1 vs. 3 (t(4) = 3.93, P = 0.02) and 2 vs. 4 items (t(4) = 3.14, P = 0.03) but not when presented with 2 vs. 3 (t(4) = 1.36, P = 0.24) and 3 vs. 4 items (t(4) = 0.26, P = 0.81). At the individual level, three dogs showed a statistically significant preference for the larger set of items, and the other two dogs did not select it more than chance (Table 2).

4.

DISCUSSION

We found that both dogs and wolves chose the larger of two quantities, but their performance was differently affected by the numerical ratio. These results parallel those obtained by Utrata et al. (2012) and Range et al. (2014) who showed that wolves, but not dogs, were able to select the larger set of items sequentially presented independently of the ratio between the two quantities to be compared.

7

Replicating previous findings with a different procedure (simultaneous vs. sequential presentation of items) suggests that the different ratio sensitivity of dogs and wolves in quantitative tasks actually reflects an interspecific difference, instead of an artifact of the methodology adopted. One potential explanation for this difference involves the existence of different quantitative systems among vertebrates. Some authors suggest the existence of an Approximate Number System (ANS) involved in quantity representation across the whole numerical range for which accuracy is affected by the ratio (Gallistel & Gelman, 1992). On the other hand, other authors suggest a second mechanism that allows exact representation of small numbers (≤ 4), irrespective of the ratio, based on a system for tracking individual objects moving in space known as the Objects Tracking System (OTS) (Trick & Pylyshyn, 1994). This system would be particularly useful in social species, for example, to track individual members of a moving group or the presence of prey. Wolves, as social pack hunters, would have a strong need for the OTS to track their fellow hunters for effective coordination to maximise the attack success, since the location and the abundance of their prey changes seasonally and is not very predictable. Conversely, most dogs are scavengers, live close to humans’ settlements and scavenge on human refuse, a food source that is relatively seasonally constant and provided in predictable locations. Being able to discriminate quantities is essential to select the largest food patch, but possibly they are not motivated to select the larger quantity when the difference between the food options is small (Agrillo & Bisazza, 2014). Differences in the feeding ecology of the two species are strictly related to differences in their social organizations. In particular, wolves show high reliance on pack structure not only to forage but also to successfully defend the territory and raise pups. Dogs, on the other hand, are less reliant on pack members both for hunting and for pup-rearing. Such differences in their social ecology (feeding ecology and social organizations) are supposed to account for differences in dogs’ and wolves’ behaviour and cognitive skills (Marshall-Pescini et al., 2017). For instance, wolves have higher capacity for imitative learning from a conspecific to solve a problem compared to dogs 8

(Range & Virányi, 2015), and are more persistent and successful in manipulative tasks, showing complex behaviours corresponding to Stages 5 and 6 of human sensorimotor development which are associated with intentional accommodation and true representational thought whereas dogs’ behaviours are associated with Stages 3 and possibly 4 (Frank & Frank, 1985). Hence we cannot exclude that the different social ecology of wolves and dogs may also help to explain differences in their quantitative abilities. Alternatively, it might be argued that selective pressures during domestication altered the ability to discriminate between quantities in dogs as a result of their adaptation to the human environment where most needs are fulfilled by humans. This hypothesis, however, cannot explain why coyotes, a non-domesticated species, show the same ratio-dependence as dogs when tested with the spontaneous food choice task (Baker et al., 2012). Another explanation might be a difference in perception with wolves having better visual acuity than dogs (Miller and Murphy, 1995). This seems unlikely, however, because we used item sizes and distances to the experimenter similar to those used in previous experiments that reported dogs’ ability to discriminate between different quantities of items (Miletto Petrazzini et al., 2016; Ward & Smuts, 2007). Finally, it could be argued that the different performances observed here might be partly due to differences in sex ratio or age between the two groups. Contradictory results have been obtained with respect to sex differences in quantitative abilities. For instance, female voles are better able than male voles to distinguish between different numbers of over-marks from two scent donors (Ferkin et al., 2005) and female guppies are more accurate than males in discriminating between two shoals of conspecifics differing in number (Lucon-Xiccato et al., 2016). However, no difference in food quantity discrimination has been reported either in guppies (Lucon-Xiccato et al., 2015) or in salamanders (Uller et al., 2003). Based on mixed results it is difficult to draw any conclusion about a possible effect of sex in the present study. The same problem occurs for age 9

effect. A far as we aware, the only study that investigated spontaneous quantitative abilities at different ages in animals has been conducted in fish, showing that one-day-old guppies are as accurate as adults in discriminating between small groups (1-4) of conspecifics (Bisazza et al., 2010). At present, therefore, it is difficult to establish whether the different results observed in the species are due to an effect of age. The question remains whether dogs and wolves relied on the number of items or on nonnumerical cues (e.g., volume) when making their choices. However, it is worth remembering that we did not attempt to determine the exact nature of the processes underlying their discriminations, rather we aimed at comparing a more general quantitative competence in wolves and dogs using the same procedure. Further investigation controlling for non-numerical cues would answer this question. Studies on comparative cognition often have access to a limited number of individuals especially when working with exotic and non-domesticated species (Pepperberg, 2006: Perdue et al., 2012; Abramson et al., 2013; Vonk, 2014). Although this situation is often unavoidable, we are aware of the potential limits on statistical power and the representativeness of the sample tested to make inferences about populations. One way to overcome this limit is to replicate studies using both the same and different procedures to reach the more plausible explanation of cognitive abilities in nonhuman animals. In conclusion, our findings replicate those of Range et al. (2014) and Utrata et al. (2012), showing different ratio sensitivity between wolves and dogs in small-quantity discrimination. Future research, comparing different populations of dogs (pet-dogs and free-ranging dogs) and wild canids (wolves, coyotes, jackals and dingoes) may help determine whether this difference can be explained by different quantitative systems, domestication processes, or differences in their social ecology.

10

FUNDING This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

COMPLIANCE WITH ETHICAL STANDARDS Ethical approval: This research was approved by the University of Florida Institutional Animal Care and Use Committee. Conflict of interest: The authors declare that they have no conflict of interest.

ACKNOWLEDGEMENTS We thank the staff at Wolf Park in Battle Ground, IN, the staff and director of Camp Marlin Doggie Day-care, Gainesville, FL, U.S.A., for their cooperation, and the dog owners who made their animals available to us.

REFERENCES Abramson, J.Z., Hernández-Lloreda, V., Call, J., Colmenares, F., 2013. Relative quantity judgments in the beluga whale (Delphinapterus leucas) and the bottlenose dolphin (Tursiops truncatus). Behav. Proc. 96, 11-19. Agrillo, C., 2014. Numerical and arithmetic abilities in non-primate species. In: Cohen Kadosh, R., Dowker, A. (Eds.), The Oxford Handbook of Numerical Cognition. Oxford University Press: Oxford, UK.

11

Agrillo, C, Bisazza, A., 2014. Spontaneous versus trained numerical abilities. A comparison between the two main tools to study numerical competence in non-human animals. J. Neurosci. Met. 234, 82-91. Agrillo, C., Miletto Petrazzini, M.E., 2012. The importance of replication in comparative psychology: the lesson of elephant quantity judgments. Front. Psychol. 3, 181. Agrillo, C., Miletto Petrazzini, M.E., Tagliapietra, C., Bisazza, A. 2012. Inter-specific differences in numerical abilities among teleost fish. Front. Psychol. 3, 483. Baker, J.M., Morath, J., Rodzon, K.S., Jordan, K.E., 2012. A shared system of representation governing quantity discrimination in canids. Front. Psychol. 3, 387. Baker, J.M., Shivik, J., Jordan, K.E., 2011. Tracking of food quantity by coyotes (Canis latrans). Behav. Proc. 88, 72-75. Bisazza, A., Serena, G. Piffer, L., Agrillo, C. 2010. Ontogeny of numerical abilities in guppies. PLoS ONE 5:e15516. Cantlon, J.F., Brannon, E.M. 2007. How much does number matter to a monkey (Macaca mulatta)? J. Exp. Psychol. Anim. Behav. Proc. 331, 32-41. Ferkin, M.H., Pierce, A.A., Sealand, R.O., delBarco-Trillo, J., 2005. Meadow voles, Microtus pennsylvanicus, can distinguish more over-marks from fewer over-marks. Anim. Cogn. 8, 182-189. Frank, H., Frank, M.G., 1985. Comparative manipulation test performance in ten-week-old wolves (Canis lupus) and Alaskan Malamutes (Canis familiaris): A Piagetian interpretation. J. Comp. Psychol. 99, 266-274. Gallistel, C.R., Gelman, R., 2000. Non-verbal numerical cognition: from reals to integers. Trends Cogn. Sci. 4, 59-65. Geary, D.C., Berch, D.B., Koepke, K.M., 2014. Evolutionary origins and early development of basic number processing. Elsevier, Amsterdam (Holland).

12

Henik, A. 2016. Continuous issues in numerical cognition: How many or how much. Academic Press, New York (NY). Lucon-Xiccato, T., Dadda, M., Bisazza, A., 2016. Sex differences in discrimination of shoal size in the guppy (Poecilia reticulata). Ethology 122, 481-491. Lucon-Xiccato, T., Miletto Petrazzini, M.E., Agrillo, C., Bisazza, A. 2015. Guppies discriminate between two quantities of food items but prioritize item size over total amount. Anim. Behav. 107, 183-191. Marshall-Pescini, S., Cafazzo, S., Virányi, Z., & Range, F., 2017. Integrating social ecology in explanations of wolf–dog behavioral differences. Curr. Opin. Behav. Sci. 16, 80-86. Miletto Petrazzini, M.E., Wynne, C.D., 2016. What counts for dogs (Canis lupus familiaris) in a quantity discrimination task? Behav. Proc. 122, 90-97. Neuliep, J.W., Crandall, R., 1993. Reviewer bias against replication research. J. Soc. Behav. Pers. 8, 21-29. Pepperberg, I. M. 2006. Grey parrot numerical competence: a review. Anim. Cogn. 9(4), 377391. Perdue, B.M., Talbot, C.F., Stone, A.M., & Beran, M.J., 2012. Putting the elephant back in the herd: elephant relative quantity judgments match those of other species. Anim. Cogn. 15, 955-961. Range, F., Virányi, Z., 2015. Wolves are better imitators of conspecifics than dogs. PLoS One 9:e86559. Range, F., Jenikejew, J., Schröder, I., Virányi, Z., 2014. Difference in quantity discrimination in dogs and wolves. Front. Psychol. 5, 1299. Sokal, R.R., Rohlf, F.J., 1995. Biometry: The principals and practice of statistics in biological research. W.H. Freeman, New York (NY). Trick, L., Pylyshyn, Z.W., 1994. Why are small and large numbers enumerated differently? A limited capacity preattentive stage in vision. Psychol. Rev. 101, 80-102. 13

Uller, C., Jaeger, R., Guidry, G., Martin, C., 2003. Salamanders (Plethodon cinereus) go for more: rudiments of number in an amphibian. Anim. Cogn. 6, 105-112. Utrata, E., Virányi, Z., Range, F., 2012. Quantity discrimination in wolves (Canis lupus). Front. Psychol. 3, 505. Vonk, J., 2014. Quantity matching by an orangutan (Pongo abelii). Anim. Cogn. 17, 297-306. Ward, C., Smuts, B.B., 2007. Quantity-based judgments in the domestic dog (Canis lupus familiaris). Anim. Cogn. 10, 71-80.

CAPTION

Fig 1. Proportion of wolves’ and dogs’ choices for the larger quantity as a function of the numerical ratio between the small and the large number of food items. Error bars indicate standard errors.

14

Table 1: Information of the subjects participating in the study. SUBJECTS Renki Marion Ayla Kailani Dharma Wolfgang Wotan Lada Tucker Hannah Maxuell Harley

CANID TYPE Gray Wolf Gray Wolf Gray Wolf Gray Wolf Gray Wolf Gray Wolf Gray Wolf Domestic Dog Domestic Dog Domestic Dog Domestic Dog Domestic Dog

AGE (years) 9 15 9 9 3 8 8 7 1 5 10 2

SEX M F F F F M M F M F M M

BREED ~ ~ ~ ~ ~ ~ ~ Pitbull/Boxer Great Pyrenese Labrador/Boxer Labrador retriever Golden retriever

15

Table 2: Proportion of choices for the larger quantity across all ratios for each individual dog and wolf (binomial tests).

SUBJECTS

CANID TYPE

Renki

Gray Wolf

Marion

Gray Wolf

Ayla

Gray Wolf

Kailani

Gray Wolf

Dharma

Gray Wolf

Lada

Domestic Dog

Tucker

Domestic Dog

Hannah

Domestic Dog

Maxuell

Domestic Dog

Harley

Domestic Dog

Proportion choices 0.75 P = 0.01 0.89 P < 0.001 0.71 P = 0.04 0.63 P = 0.25 0.68 P = 0.09 0.71 P = 0.04 0.54 P = 0.85 0.64 P = 0.18 0.75 P = 0.01 0.79 P = 0.004

16