Edited by:
Reviewed by:
*Correspondence:
This article was submitted to Comparative Psychology, a section of the journal Frontiers in Psychology.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
Certain aspects of social life, such as engaging in intergroup conflicts, as well as challenges posed by the physical environment, may facilitate the evolution of quantity discrimination. In lack of excessive comparative data, one can only hypothesize about its evolutionary origins, but human-raised wolves performed well when they had to choose the larger of two sets of 1–4 food items that had been sequentially placed into two opaque cans. Since in such paradigms, the animals never see the entire content of either can, their decisions are thought to rely on mental representation of the two quantities rather than on some perceptual factors such as the overall volume or surface area of the two amounts. By equaling the time that it takes to enter each quantity into the cans or the number of items entered, one can further rule out the possibility that animals simply choose based on the amount of time needed to present the two quantities. While the wolves performed well even in such a control condition, dogs failed to choose the larger one of two invisible quantities in another study using a similar paradigm. Because this disparity could be explained by procedural differences, in the current study, we set out to test dogs that were raised and kept identically as the previously tested wolves using the same set-up and procedure. Our results confirm the former finding that dogs, in comparison to wolves, have inferior skills to represent quantities mentally. This seems to be in line with
The ability to discriminate between different quantities is thought to be advantageous for any social species (
Experimental studies examining dogs’ abilities under more controlled conditions also suggest that dogs’ quantity discrimination skills may be limited in regard to relying on mental representations of various amounts.
Interestingly, wolves, the closest living relatives of dogs (
These two canine studies are comparable to a certain degree, because both studies made sure that the animals needed to rely on the food quantities instead of choosing simply based on the amount of time needed to present the two sets of quantities. Obviously, inserting four pieces of food sequentially takes longer than dropping two pieces of food, which potentially allows the subjects to base their choices on temporal cues (
It is still possible, however, that the better performance of the wolves in our study compared to the dogs in the
If we can exclude these explanations, the difference found between dogs and wolves raises interesting evolutionary questions. Although addressing broader evolutionary issues is currently difficult because few studies on other canine species have been conducted (see
Here, in order to test the representation-based abilities of dogs to discriminate different quantities and to compare them to wolves, we tested 13 dogs using the same experimental set-up that brought positive results in wolves. Importantly, since the 13 dogs have not only been trained and tested but also raised and kept under the same conditions as the previously tested wolves, we can assume that differences in socialization and previous experience cannot account for differences in the performance of the animals.
No special permission for use of animals (dogs) in such socio-cognitive studies is required in Austria (Tierversuchsgesetz 2012 – TVG 2012). The relevant committee that allows running research without special permissions regarding animals is: Tierversuchskommission am Bundesministerium für Wissenschaft und Forschung (Austria).
The 13 dogs (
Information on the subjects participating in this study.
Subject | Origin | Litter | Pack | Age | Sex | Participation |
|||
---|---|---|---|---|---|---|---|---|---|
Training | Test | Time | Stone | ||||||
Kilio | Paks, Hungary | 1 | 1 | 3 | M | x | x | x | x |
Meru | Velence, Hungary | 2 | 1 | 2.5 | M | p.p. | n.p. | n.p. | n.p. |
Nia | Paks, Hungary | 3 | 1 | 1.5 | F | x | x | x | x |
Bashira | Paks, Hungary | 4 | 1 | 2.5 | F | p.p. | n.p. | n.p. | n.p. |
Asali | Siofok, Hungary | 5 | 2 | 2.5 | M | x | x | x | x |
Bora | Györ, Hungary | 6 | 2 | 1.5 | F | p.p. | n.p. | n.p. | n.p. |
Nuru | Paks, Hungary | 7 | 3 | 1.5 | M | x | x | x | x |
Layla | Györ, Hungary | 6 | 3 | 1.5 | F | p.p. | n.p. | n.p. | n.p. |
Zuri | Paks, Hungary | 7 | 3 | 1.5 | F | x | x | x | x |
Rafiki | Tengelic, Hungary | 8 | 4 | 3 | M | p.p. | n.p. | n.p. | n.p. |
Hakima | Paks, Hungary | 4 | 4 | 2.5 | M | x | x | x | x |
Maisha | Paks, Hungary | 1 | 4 | 3 | M | x | x | x | x |
Binti | Siofok, Hungary | 5 | 4 | 2.5 | F | x | x | x | x |
At the time of testing, the dogs and wolves were closely matched in age (average age dogs: 2.2 years; average age wolves: 2.6 years) and had comparable experience in regard to behavioral studies. All animals had participated in the same cognitive tests prior to this study ranging from personality to social learning tasks. Also, all animals participated in a quantity discrimination task when they were 1 year or younger, where they had to choose between 1 and 8 pieces of cheese put in front of them on two plates. That study aimed at testing the influence of a human on the choice of the animal (see
The experimental apparatus, consisting of a wooden table (170 cm × 40 cm × 60cm) with two opaque plastic cans (
The experiment consisted of a training and a test phase (including two controls). The basic procedure was the same for all trials (except for training level 1): for each session the dogs were brought in a testing enclosure separated from the rest of the pack. The experiment began after the trainer had positioned herself about 1.5 m in front of the apparatus holding the dog on its collar. During the entire training and test procedures, the trainer had her eyes closed and her head held down so that she could see none of the experimenter’s actions. The experimenter prepared the required amount of food (or stones) and inserted both of her filled, closed hands into the holes above the opaque cans. Then she showed the dog one item at a time holding it visibly between two fingers while the rest of the food items were still hidden in her closed palm. Next, the experimenter called the dog’s name to get its attention and once it looked, the experimenter placed the item either onto the table in front of the can (
After showing her second empty hand to the dog, the experimenter gave a signal to the trainer who released the dog giving a short command (“go”). Although in the training phases the animals had to step with their forepaw on the buzzer to provide an acoustic signal to clearly indicate their choice, in the testing phase choices were also counted if the dog used the buzzer without producing a signal, stepped on the wooden panel to which the buzzer was attached to, or touched the fence on the side of the can with its nose for at least 3 sec. This was implemented to avoid missing or misinterpreting the dog’s first choice by waiting too long for it to use solely the buzzer. If a correct choice was made, the experimenter released the items chosen by the subject by pulling out the plastic bar from under the can. If the items had been placed on the table (training steps 2, 3, and 5), the experimenter inserted them into the can from where they could slide into the enclosure. Incorrect choices were not rewarded except in training steps 3 and 4 (see below for details).
Three professional animal trainers were involved in testing the dogs. Moreover, two experimenters conducted the testing in the first five vs. the last 3 months. Since the experimenters were hidden behind the apparatus (see experimental set-up), we expected that this would not influence the performance of the animals. Nevertheless, we checked for a potential effect of experimenters by integrating them as a factor in the statistical analyses but we found no significant effect (
Yellow cheese pieces (Gouda, 1 cm × 1 cm × 1 cm) were used as reward and black stones of a comparable size were used in the control trials. Only one session per day was conducted with 1–2 days elapsing between sessions.
During the five-step training phase, the animals were familiarized with the apparatus and the procedure. In
In
In
In
However, since the stones were always added on the side with the fewer pieces of cheese, the animals could have easily chosen the bigger cheese amount in this control experiment by avoiding the sound the stones made when being dropped into the respective can. Therefore, in the second control experiment, we added an extra stone to both sides (e.g., 3 vs. 1: one can contained three pieces of cheese and one stone and the other can contained one piece of cheese and three stones).
Each control experiment consisted of four sessions of six trials each and tested only the following three (cheese) quantity pairs: 1 vs. 2, 1 vs. 4, 2 vs. 3. Accordingly, we had a pair with a small distance and a large ratio between sets (1 vs. 2), a pair containing a large distance and a small ratio between sets (1 vs. 4) and a pair with a large ratio and an intermediate distance (2 vs. 3).
Initially we examined the performance of dogs in the training phase and compared them to the wolves (based on data from
In the test and control sessions, we analyzed whether non-quantity and quantity properties influenced the animals’ performance by calculating non-linear mixed effect (nlme) models using binomial distribution and including subjects and sessions as random factors. We tested the following non-quantity factors: (1) the side where the larger cheese quantity had been placed (‘side_larger quantity’), (2) the order of placing the two sets (larger amount placed first, ‘order_first’), (3) session (sessions 1–8, ‘sess’) and (4) we tested for the influence of the experimenter’s identity. In regard to the quantity factors we included the ratio of the two presented sets (‘ratio’: 0.25, 0.33, 0.5, 0.66, 0.75). Finally, species was included as last factor to test for differences between wolves and dogs. If the non-quantity factors proved to have no influence on the animals’ performance they were excluded from further analyses.
Furthermore, by comparing the data to chance level with one-sample
The data were analyzed using the statistical software
Eight of the 13 dogs that participated in this study (61.5 %) passed all training steps and were tested in the quantity discrimination test and control experiments (see
Number of trials every subject needed to reach criterion and to enter the next step (step 2–5) in training phase.
Subject | Step 2 |
Step 3 |
Step 4 |
Step 5 |
Overall |
---|---|---|---|---|---|
Kilio | 111 | 27 | 20 | 7 | 165 |
Meru | 57 | 21 | 147 | n.p. | |
Nia | 56 | 21 | 13 | 7 | 97 |
Bashira | n.p. | ||||
Asali | 86 | 35 | 7 | 7 | 135 |
Bora | 111 | 42 | 126 | n.p. | |
Nuru | 81 | 21 | 35 | 7 | 144 |
Layla | 129 | n.p. | |||
Zuri | 85 | 49 | 21 | 8 | 163 |
Rafiki | 126 | n.p. | |||
Hakima | 103 | 20 | 27 | 15 | 165 |
Maisha | 99 | 14 | 49 | 23 | 185 |
Binti | 72 | 28 | 6 | 7 | 113 |
The remaining eight dogs did not differ from the wolves in the number of trials to reach criterion in training step 2, 4, and 5 (Mann–Whitney-
Again, neither in the time nor in the stone control experiments was the dogs’ performance influenced by the session (NLMEsess: time control:
Quantity discrimination test. Overall, dogs and wolves chose the larger quantity significantly above chance (one sample
However, when looking at each numerical pair separately, it turns out that, while wolves were significantly above chance at all tested ratios, dogs performed above chance level only at ratios at or below 0.5 (see
Performance of dogs and wolves according to the ratios tested in the quantity discrimination test.
Quantity discrimination test |
Time control |
Stone control |
|||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Ratio | Sets | Dog | Wolf | Dog | Wolf | Dog | Wolf | ||||||
0.25 | 1:4 | t9 = 3.32 | p = 0.009 | ||||||||||
67.19% | 77.50% | 71.88% | 71.25% | 53.13% | 69.74% | ||||||||
0.33 | 1:3 | textitp = 0.005 | |||||||||||
75.00% | 71.60% | ||||||||||||
0.50 | 1:2 | ||||||||||||
71.88% | 73.42% | 39.06% | 65.82% | 54.69% | 60.53% | ||||||||
0.50 | 2:4 | ||||||||||||
75.00% | 68.75% | ||||||||||||
0.67 | 2:3 | t7 = -0.36 | |||||||||||
53.13% | 62.96% | 48.44% | 62.96% | 51.56% | 71.05% | ||||||||
0.75 | 3:4 | t9 = 3.82 | |||||||||||
46.88% | 68.35% |
Overall, we found that the dogs as well as the wolves chose the larger quantity above chance level in the quantity discrimination test. However, a closer look at the performance of the subjects according to the tested ratios revealed that while wolves performed above chance even with high ratios (e.g., 0.67 and 0.75), in these conditions dogs performed at chance level.
Our dog results are in line with previous studies on canines suggesting that at least dogs and coyotes have problems discriminating between two small quantities of high ratios (0.75 and 0.66) if these are invisible at the moment of the choice (
Interestingly, in contrast to dogs, wolves were able to discriminate between high ratios of small quantities when tested in the same task (
First, canines have an extraordinary sense of smell compared to humans, which could theoretically allow them to discriminate at least quantities of lower ratios based on olfaction. However, there is little indication that canines would rely on their olfactory cues in such a situation if not specifically trained to do so. For example, dogs do not rely on their sense of smell in two choice tasks if visual information is provided by a human experimenter (
Second, the observed difference between wolves and dogs could be due to dogs having worse eyesight than wolves (
Yet another explanation could be that the dogs were less motivated than the wolves to solve the task. However, the dogs did solve several combinations in the quantity discrimination test and in the lowest ratio of the first control condition. Moreover, since both wolves and dogs were reared and kept under similar conditions, differences in socialization or reinforcement, that have been shown to influence performance in problem solving skills (
Finally, it is possible that due to domestication, dogs possess a weaker ability to mentally represent and discriminate quantities in comparison to wolves. More specifically,
The observed differences between wolves and dogs in our task are not unusual. Differences in performances between closely related species have also been reported for elephants for instance. African elephants are affected by the numerical ratio (
In conclusion, our study supports results by
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
The Wolf Science Center was established by Zsófia Virányi, Kurt Kotrschal, and Friederike Range and we thank all our helpers who made this possible and thus indirectly supported this research. We thank Marianne Heberlein for helping with the statistical analyses. We thank Jennifer Essler as well as three reviewers and the editor for comments on an earlier draft of the manuscript. The project was financially supported by Austrian Science Fund (FWF) project P21244-B17. Writing has received funding from the European Research Council under the European Union’s Seventh Framework Program (FP/2007–2013)/ERC Grant Agreement n. [311870]′ to the first and from the WWTF project CS11-026 to the last author. We further thank many private sponsors including Royal Canin for financial support and the Game Park Ernstbrunn for hosting the Wolf Science Center.