Ethical permit
Ethical permission for conducting this study was obtained from The Institutional Committee of Eötvös Loránd University (N. PE/EA/692-5/2019). All experiments were carried out in accordance with relevant guidelines and regulations for both owners as volunteers and their animals participating in the experiments. Owners gave informed consent to participate in the study with their animals. The study is reported in accordance with ARRIVE guidelines. Informed consent was obtained to publish images in an online open-access publication.
Subjects
We tested 42 dog puppies of various breeds (mean ± SD age = 13.4 ± 1.7 weeks old), 39 kittens (13.8 ± 2.6 weeks old) and 8 wolf pups (12 ± 2.8 weeks old), 5 of which belonged to the same litter. All subjects of all species were socialized, lived in human families, and experienced the typical life of family companion animals. The wolves were hand-raised by their caregivers after being separated from their mothers in the first 12 days from birth, independently from the present study. They were bottle-fed and like the dog puppies and kittens, lived in the owner’s family, thus were in close contact with them during their early development (3 wolf pups were raised and lived in human families in Mexico, were registered by their breeders and were kept according to the regulations of the Secretaría de Medio Ambiente y Recursos Naturales; 5 wolf pups originated from the Horkai Animal Training Center at Gödöllő, Hungary, and were raised in human families in Hungary. For further details, see Ujfalussy et al.31). Living in their human families, they regularly met strangers and conspecifics, similarly to the other two species included in this study.
Testing site
Depending on the owners’ availability, the tests were carried out at the Department of Ethology at Eötvös Loránd University, or at the owner’s home (subjects tested at the Department: N = 32 of 42 dog puppies; N = 20 of 39 kittens; and N = 5 of 8 wolf pups). The Experimenters (C.F., A.T., S.U.) carried out all the tests, both at the Department and at the owners’ homes.
Objects used for testing (Fig. 1)
Figure 1 Objects used for testing (Box at the top, Wobbler Kong at the bottom). Full size image
Two unfamiliar objects were used: (1) a white plastic box (ca. 12 × 26 × 17 cm); and (2) a Wobbler Kong (19 × 12 cm). We ensured that the objects were unfamiliar to the subjects by asking the owners if they might have had any previous experience with them and whether they had these objects at home. The objects were allocated randomly to subjects. The same object was used in all trials for a given subject.
Protocol
Before the start of the tests, the subjects were allowed to get acquainted with the demonstrator, while the test procedure was explained to the owners. The subjects were free in the experimental room (or in the designated test area when tested at home), while the experimenter and the owner talked and briefly greeted the subject when approached. This familiarization lasted until the explanation of the protocol was over (ca. 10 min).
Exploration trial
Before receiving trials with demonstrations (see below), the subjects were allowed to explore the target object. The aim of the free exploration trial was to observe the subjects’ spontaneous behaviour in the absence of a demonstration and served also as a control for social facilitation, i.e. to exclude the effect of the mere presence of the demonstrator as an explanation for action matching32.
At the beginning of the free exploration trial, the owner held the subject on the floor by gently holding it at its chest, at 1.3 m from the target object. The experimenter crouched down within reaching distance from the object and ignored the subject by looking down at her/his knees. The owner released the subject and allowed it to explore for 25 seconds. After the subject was released, we monitored and noted whether the subject touched the object and observed the body part (nose or paw) used for this interaction.
Trials with demonstrations
Immediately after the free exploration trial, subjects were exposed to trials, which included a demonstration made by a human experimenter.
In these trials, a pre-determined action was demonstrated (Fig. 2). We used two actions for the demonstrations: touching the object with hand or touching the object with nose. To minimize the possibility of our tests becoming confounded by typical actions, the action that the subjects had eventually performed during the previous free exploration trial was contrasted by the different action demonstrated in the experimental trials. For instance, if the subject touched the object with its nose during the free exploration trial, then the demonstrated action in the forthcoming experimental trials was planned as a hand action and vice versa. If the subject had not interacted with the object during the free exploration trial, then the action to be demonstrated in the following experimental trials was randomly determined (either nose or hand).
Figure 2 Experimental setup. In the free exploration trial and in the experimental trials, the demonstrator is within reaching distance from the object. The owner gently holds the subject at its chest, in front of him/her. In the trials with demonstration, the demonstrator makes eye contact and uses ostensive communication before and during the demonstration. Full size image
The subject was held by the owner similar to the free exploration trial (Fig. 2). The experimenter was kneeling in front of the target object, within reaching distance from it. She first looked at the subject and called its attention by pronouncing its name and/or using attention grabbing vocal signal (e.g. smacking her lips).
As soon as the subject looked at her, the experimenter performed the demonstration. In the case of nose demonstration, the experimenter touched the object with her nose two times. In the case of the hand demonstration, the experimenter touched the object with her hand two times. The experimenter provided ostensive cues also while performing the demonstration by alternating her gaze between the subject and target object and saying sentences like “Look at what I am doing!”, “This is so nice!”. After demonstration, the demonstrator looked down and ignored the subject.
The subject was released by the owner and was free to explore for 25 seconds, or until it interacted with the target object. If the subject interacted with the object, the trial was stopped and the owner gently called the subject back to the starting position for the next trial, while briefly verbally praising it.
In the following two trials, the same procedure was repeated by the experimenter who demonstrated the same action on the same object.
After these 3 trials, the subjects received a short break (approximately 5 to 10 minutes) during which they were taken out of the experimental room, and were then tested in another 3 similar experimental trials with the same target object. The only difference was the movement of the object during the demonstration: in “move” trials, the demonstration resulted in moving the object, while in “touch” trials the object did not move. Order of move and touch trials was randomized between the first and the second set of three repeated trials.
We also carried out three ‘Ghost trials’ in which the object was moved remotely by using a transparent fishing line. This was done to observe whether the movement of the object attracted the subjects’ attention and stimulated interaction with it. Subjects were randomly allocated to this condition or to the experimental one, unless owners were available for repeated testing, in which case subjects were tested in all conditions.
Data collection and analysis
Experimental trials were video recorded. From the recordings, the following behavioural variables were coded, using Solomon Coder (beta 15.03.15, © András Péter):
Latency to looking at the demonstration: time measured from the start of the trial, until the head of the subject was oriented towards the object;
Latency to interacting with the object: time measured from when the subject was released, until it physically interacted with the object using either nose or paw;
Body part used to interact with the object (binary response for nose and paw, separately): Nose, if the subject interacted with the object by touching it with nose; Paw, if the subject interacted with the object by touching it with paw; Both paw and nose if they touched the object with their nose and paw at the same time.
From the above coded variables, we then derived Action matching (a binary response), defined as the subject interacting with the object using its paw when the demonstration was performed with hand/paw and interacting with the object using nose when the demonstration was performed with nose, including the cases when both paw and nose were used at the same time.
Statistical analyses
We used the R statistical environment (v. 4.1.2; R Core Team 2021) to analyze our data. For each trial, we coded whether and how fast a given behaviour occurred. Latencies to looking at the demonstration and interacting with the object (separate response variables), were analyzed in Cox Mixed Models (‘CMM’, R package ‘coxme’33). In CMMs, time (in seconds) spent until the focal behaviour was performed (i.e. until looking at the demonstration or interacting with the object) was the response, and the focal behaviour was the terminal event. Individuals that did not perform the focal behaviour were treated as censored observations. Probability of action matching, body part used during free exploration and paw use (separate binary response variables) were analyzed using binomial Generalized Linear Mixed Models (‘GLMM’ henceforth, R package ‘lme4’34). The analysis of paw use was based on our analysis of the body part used by the subjects when interacting with the object in the free exploration trial; this showed that they almost invariably interacted with the object spontaneously with their nose. Therefore, in models of paw use we analyzed whether the demonstration of a hand/paw action in experimental trials resulted in an increase in paw actions (thus, in contrast to their preference; see ‘increase in paw use’ below).
Both in GLMMs and CMMs, the following fixed effects were analyzed:
species (factor with 3 levels)
trial number (factor with 3 levels)
condition (factor with two levels: ghost or experimental)
testing site (factor with two levels: department or home)
touch or push (factor with two levels: ‘touch’: the demonstration did not result in object movement or ‘push’: the demonstration resulted in object movement)
body action used for demonstration (factor with two levels: nose or hand – included only in models of paw use).
In addition to the above fixed effects, all models included subject ID as a random effect. The effects of explanatory variables were analyzed by likelihood ratio tests, and non-significant explanatory variables were excluded from the final models based on a stepwise model selection. We also provide parameter estimates (for GLMMs) and hazard ratios (exp[β], with 95% CI, for CMMs) between levels of a given significant fixed factor.
The final models of interacting with object included both trial and touch or push (see “Results”). The latter variable, however, had no value for the free exploration trial, hence we could not investigate whether subjects behaved differently in trials that were preceded by a demonstration as opposed to the free exploration trial. Therefore, for interaction with object, we also carried out separate analyses for each species, focusing only on the effect of trial.
We applied the same approach for analyzing paw use; following the analysis including all three species that was limited by low occurrence of interacting with the object in kittens, we also carried out separate analyses for each species to reveal more details for dogs and wolves.