Kangaroos display gazing and gaze alternations during an unsolvable problem task

Abstract

Domestication is generally assumed to have resulted in enhanced communication abilities between non-primate mammals and humans, although the number of species studied is very limited (e.g. cats, Felis catus; dogs, Canis familiaris; wolves, Canis lupus; goats, Capra hircus; horses, Equus caballus). In species without hands for pointing, gazing at humans when dealing with inaccessible food during an unsolvable task, and in particular gaze alternations between a human and the unsolvable task (considered forms of showing), are often interpreted as attempts at referential intentional communication. We report that kangaroos, marsupial mammals that have never been domesticated, actively gazed at an experimenter during an unsolvable problem task (10/11 kangaroos tested), thus challenging the notion that this behaviour results from domestication. Nine of the 10 kangaroos additionally showed gaze alternations between the unsolvable task and experimenter. We propose that the potential occurrence of these behaviours displayed towards humans has been underestimated, owing to a narrow focus on domestic animals, as well as a more general eutherian research bias.

1. Introduction

Domestication is believed to have resulted in animals having enhanced interspecific communication abilities with humans [1–4]. Referential intentional communication occurs when a signaller attempts to direct the attention of a recipient to the desired goal through persistent pointing or showing [1,3,5] and can vary depending on the recipient's attentiveness [5]. Gazing and gaze alternation (forms of showing) are often interpreted to have a referential communication function with humans, in a similar manner to pointing by humans and non-human primates, as they may serve to attract the attention of an observer towards an object [1,4]. This type of communication is considered a prerequisite for mind reading (e.g. understanding intentions) because it represents a form of mental planning [1,5]. Because most of the research on behaviours that are interpreted as animal–human communication has been carried out on domestic animals, it is unclear whether they are found among non-primate mammals that have not experienced a long period of domestication. Finding those behaviours in other species would suggest that the potential for animal–human communication is far more widespread, given the right conditions.

The cognitive abilities of marsupials such as kangaroos have rarely been investigated [6]. Herein we studied the ability of kangaroos to express human-directed gazing behaviour, as well as gaze alternations, when faced with an unsolvable problem task [1,4]. Kangaroos are large, herbivorous, social mammals that are also often kept in captivity and thus become habituated to humans. We focused our research on captive Kangaroo Island kangaroos (Macropus fuliginosus fuliginosus; subspecies of western grey kangaroos) because they are known to be docile and interactive with humans. We also tested small numbers of eastern grey (Macropus giganteus) and red (Macropus rufus) kangaroos.

2. Methods

The study was conducted at three locations in Australia: (i) Australian Reptile Park (ARP, https://reptilepark.com.au), (ii) Wildlife Sydney Zoo (WSZ, https://www.wildlifesydney.com.au) and (iii) Kangaroo Protection Co-operative (KPC, https://kpcwildlifesanctuary.org.au).

We tested 16 habituated kangaroos (12 Kangaroo Island, 2 eastern grey and 2 red kangaroos), in their home enclosures between 07.00 and 18.00 during November 2019, over eight separate days (10 subjects at ARP, 3 at WSZ and 3 at KPC). Thirteen females and 3 males were used; mean age = 4.6 years; age range = 18 months to 12 years. Kangaroos were distinguishable by their body conformations, fur markings and microchip numbers. They were selected based on their willingness to approach the experimenters. All kangaroos were naive to the task and had not been used in any previous cognitive research.

For training and testing, kangaroos were physically separated from their familiar conspecifics but remained in the same enclosure, where they maintained visual, auditory and olfactory contact with one another. Kangaroos were tested individually with the first experimenter (Experimenter 1) using a piece of food to entice the subject to an area free from disturbance. Here, Experimenter 1 had set up the testing equipment, by placing a square wooden board (1 × 1 m) with a transparent plastic box lid screwed into its centre onto the ground (figure 1). The plastic box could then be either loosely or securely fastened to the lid depending on the trial (table 1). For all trials, Experimenter 1 was positioned approximately 1 m from the board's edge, facing the kangaroo. This experimenter was responsible for placing a small food reward into the box and maintained position throughout the trials. Kangaroos were highly motivated by the food reward, which varied according to the testing location (e.g. one piece of sweet potato or carrot at ARP, a few dried corn kernels at WSZ and macropod grass pellets at KPC). A second experimenter (Experimenter 2) video recorded all observations and prevented other kangaroos from entering the test area using two garden lattices (0.9 × 1.8 m). Neither experimenter physically interacted with the focal kangaroo during training nor testing, although in some cases, the subject approached Experimenter 1 during the test trial. Owing to the challenges of working with animals that could not be led or restrained, we focused on determining the ability of kangaroos to express the core human-directed gazing behaviour towards Experimenter 1, as well as gaze alternations between the ‘unsolvable problem task' and that experimenter.

Figure 1.

Test trial; downturned closed plastic box with food inaccessible. (a) Kangaroo attempting to gain access to food. (b) Kangaroo gazes at the experimenter.

Table 1.

Description of training and testing trials to which the kangaroos were individually exposed.

trial type	trial	description of procedure
training	1 and 2	Experimenter 1 placed food into upturned box that was loosely sitting on top of the lid: food accessible
	3 and 4	Experimenter 1 placed food into downturned box that was loosely sitting on top of the lid: food accessible
	5 and 6	Experimenter 1 placed food into downturned box that was loosely placed into the grooves of the lid: food accessible
testing	7	Experimenter 1 placed food into downturned box that was securely fastened into the grooves of the lid: food inaccessible

Subjects were exposed to seven trials; six consecutive training trials followed by a single test trial (table 1). Eleven (out of 16; 9 females and 2 males) subjects successfully retrieved the food during training and proceeded to testing. During the test trial (unsolvable task), kangaroos still had visual and olfactory cues to the food owing to the transparency of the box and small holes cut into the plastic. Combined, the seven trials lasted no longer than 10 min per kangaroo. Training trials commenced when Experimenter 1 had moved back to the starting position after placing the food in the box. Each training trial ended once the kangaroo had accessed the food, with the subsequent trial beginning once the kangaroo had finished eating and looked back towards the experimenter. The test trial started once Experimenter 1 had moved back to the starting position after placing the food in the box and ended either when the kangaroo had moved away from the test apparatus, or after a maximum duration of 2.5 min. Therefore, the test duration depended on the behaviours of the subjects. This approach of having unfixed trial duration was also necessary in a similar study of free-ranging dogs [7]. For test trials, all subjects were given the food reward at the end.

Videos were analysed using BORIS v.7.9.7 (http://www.boris.unito.it/) [8]. The latency until first behavioural occurrence and frequency of the following were quantified: (i) box interactions: kangaroo establishes contact with the box, including sniffing, nosing or pushing it with face, nose or claws (behaviour was only coded when food was in the box); (ii) experimenter interactions: kangaroo turns and lifts head to gaze towards the experimenter, or approaches and scratches the experimenter, and (iii) gaze alternations: kangaroo alternates gaze between box and experimenter. The proportion of time showing (i) box and (ii) experimenter interactions relative to the total trial duration was also assessed, to standardize the outcome between kangaroos that had different trial durations.

Six of the 11 training and test trials were double-coded by K.H.O and A.C.G, resulting in a high Kohen's kappa coefficient for inter-observer reliability (all κ ≥ 0.85). All statistical analyses were performed in R Studio v 1.2.5042 [9]. For training (trials 1 to 6), mean values were calculated for the proportions of box and experimenter interactions, and rate of gaze alternations. Individual kangaroo responses were then graphically compared between training and testing for the proportions of box and experimenter interactions (electronic supplementary material). The presence or absence of gaze alternations was also described for training and testing (electronic supplementary material).

3. Results

Each trial differed in duration both within and between subjects (n = 11, mean ± s.d., training: 9.94 ± 3.56 s; testing: 60.2 ± 40.46 s, electronic supplementary material, table S1). During testing, 10/11 kangaroos gazed towards the experimenter (figure 1, electronic supplementary material, videos S1 and S2) at a rate of 0.093 ± 0.058 gazes per test (calculated as the frequency of gazing/total test duration). During testing, these 10 kangaroos interacted with the box at a rate of 0.067 ± 0.025 interactions per test (calculated as the frequency of box interactions/total test duration). Nine of the 10 kangaroos additionally showed gaze alternations between the unsolvable task and experimenter, at a rate of 0.06 ± 0.049 gaze alternations per test (calculated as the frequency of gaze alternations/total test duration).

4. Discussion

Domestication is often used to explain the enhanced communicative interactions of domestic animals with humans [1,4,10]. We investigated human-directed behaviour of kangaroos during an unsolvable problem task and found that they both gazed at the experimenter and displayed gaze alternations. Given that kangaroos have never been domesticated and are also marsupials, our results provide support for the hypothesis that in social species, phylogenetic prerequisites of responsiveness to visual stimuli from conspecifics, as well as conditioning and positive socialization, are likely to be more important than domestication for exhibiting human-directed behaviours. These behaviours have often been interpreted as referential communication towards humans; however, that suggestion requires further investigation in kangaroos [10]. In summary, given appropriate positive interactions, many social mammal species probably have the potential to modify behaviours that have evolved to communicate within their own species and instead display them in a manner that may suggest intentional human-directed communication.

The role of naturally occurring tameness should also be considered when animals engage in behaviours (e.g. gazing and gaze alternations) directed at humans, which could suggest intentional communication. Kangaroo Island kangaroos are known to be docile and curious, having evolved on an island lacking large predators. Selection for tameness has played a role in socio-cognitive changes in dogs and foxes (Vulpes vulpes), resulting in the ability to interact affiliatively with humans [11,12]. Nevertheless, even in kangaroo species that are not known for docility (eastern grey and red), we still found evidence for gazing behaviour, suggesting that these types of interactions are learned as a result of positive interactions with humans during ontogeny.

Overall, our results show that expanding research beyond the usual domestic species is required to resolve ongoing discussions about the effects of domestication on animal cognition [7,10,11,12,13]. These discussions cannot be resolved by simply testing domestic species along with tame individuals from their wild ancestors using the same task (e.g. dogs versus wolves, domestic goats versus wild goats). Future research on kangaroos should examine the visual cues and behaviours that they use for social interactions between conspecifics, for example head orientation and limb movements, and determine whether similar ones are used when interacting with humans. Now that we have established the occurrence of gazing and gaze alternations by kangaroos towards humans when faced with the unsolvable problem task, it will be possible to carry out more nuanced experiments akin to those that have been done with species that are more traditionally used in this type of research [4,7,13].

Ethics

Animal care and all experimental procedures were in accordance with the ASAB/ABS Guidelines for the Use of Animals in Research. Ethical approval was obtained from both the University of Roehampton (LSC 19/294) and Wildlife Sydney Zoo. Training and testing were non-invasive and did not cause any distress to the kangaroos.

Data accessibility

Data are available from the Dryad Digital Repository: https://dx.doi.org/10.5061/dryad.pnvx0k6k6 [14].

Authors' contributions

A.G.M conceptualized the study; A.G.M., K.H.O. and A.C.G developed the methodology; A.G.M., K.H.O. and A.C.G. conducted the fieldwork; A.C.G and K.H.O. analysed the data; A.G.M. and A.C.G wrote the original draft; A.G.M., K.H.O. and A.C.G revised and edited the final draft; A.G.M. acquired funding. A.G.M. and A.C.G contributed equally to writing the manuscript. All authors approved the final version of the manuscript and agree to be held accountable for the content of this paper.

Competing interests

We declare we have no competing interests.

Funding

The Association for the Study of Animal Behaviour (ASAB) funded the research with a grant to A.G.M.