Cats Parallel Great Apes and Corvids in Motor Self-Regulation – Not Brain but Material Size Matters

Bobrowicz, Katarzyna; Osvath, Mathias

doi:10.3389/fpsyg.2018.01995

ORIGINAL RESEARCH article

Front. Psychol., 22 October 2018

Sec. Cognition

Volume 9 - 2018 | https://doi.org/10.3389/fpsyg.2018.01995

Cats Parallel Great Apes and Corvids in Motor Self-Regulation – Not Brain but Material Size Matters

$\r\nKatarzyna Bobrowicz*$ Katarzyna Bobrowicz^*

Mathias Osvath

Cognitive Zoology Group, Department of Philosophy and Cognitive Science, Lund University, Lund, Sweden

The inhibition of unproductive motor movements is regarded as a fundamental cognitive mechanism. Recently it has been shown that species with large absolute brain size or high numbers of pallial neurons, like great apes and corvids, show the highest performance on a task purportedly measuring this mechanism: the cylinder task. In this task the subject must detour a perpendicularly oriented transparent cylinder to reach a reward through a side opening, instead of directly reaching for it and bumping into the front, which is regarded as an inhibitory failure. Here we test domestic cats, for the first time, and show that they can reach the same levels as great apes and corvids on this task, despite having much smaller brains. We tested subjects with apparatuses that varied in size (cylinder length and diameter) and material (glass or plastic), and found that subjects performed best on the large cylinders. As numbers of successes decreased significantly when the cylinders were smaller, we conducted additionally two experiments to discern which properties (length of the transparent surface, goal distance from the surface, size of the side opening) affects performance. We conclude that sensorimotor requirements, which differ between species, may have large impact on the results in such seemingly simple and apparently comparable tests. However, we also conclude that cats have comparably high levels of motor self-regulation, despite the differences between tests.

Introduction

Motor self-regulation overrides unproductive motor movements triggered by salient stimuli in the environment (Beran, 2015). Such behavioral inhibitions are thought of as basic inhibitory abilities within the larger family of executive functions (Diamond, 2013). This mechanism is fundamental in the sense that poor motor self-regulation will lead to difficulties in executing other, more cognitively demanding, tasks. In a sense, without any motor self-regulation one gets stuck in the immediate sensorimotor moment.

This underlying character of motor self-regulation has made it appealing for species comparisons. For example, is there a correlation between these basic skills and performance on other cognitive tasks? A large-scale study by MacLean et al. (2014) comparing 567 individuals from 36 species, found a correlation between (mainly) absolute brain size and motor self-regulatory performance on two tasks: the cylinder task and the A-not-B test (2014). The cylinder task, which was administered to the largest number of subjects and species in the study, consists of a transparent cylinder, containing a reward in the middle, with openings at both ends. To successfully retrieve the reward the subject must refrain from immediately reaching for it through the shortest perceivable route (thereby bumping into the transparent surface) and instead make a detour to either side opening. The best performing taxon on this task was the primates, with great apes reaching top scores. A later study showed that corvids from the genus Corvus are as successful as great apes at solving the cylinder task, despite vastly smaller absolute brain sizes (Kabadayi et al., 2016). On the other hand, within birds the corvids have the largest number of pallial neurons and they show cognitive complexity in various studies (Olkowicz et al., 2016).

It is perhaps not surprising that cognitively sophisticated species find tasks like these simple, however, it is noteworthy that many other species fail to such degrees. In other words, why do so many animals show poor performance in motor self-regulation tasks, despite the fact that it appears to be an essential form of inhibition in the complex lives of mammals and birds (e.g., when stalking prey or foraging at great heights)? We hypothesize that the cylinder task may measure additional factors that may not in an immediate sense be related to motor self-regulation, such as the sensorimotor adaptations of different animals.

We investigated the performance of domestic cats (Felis catus) on different versions of the cylinder task. Cats arguably benefit from pronounced motor self-regulation due to their reliance on stalking and stealth during hunting, and hence the need to avoid alerting their prey in the wrong moment even when it can be clearly seen by the cat. Cats are also avid climbers and would therefore benefit from motor self-regulation when negotiating narrow supports high above ground. On the other hand, cats have an average mammalian relative brain size, and their absolute brain size is comparably small (Roth and Dicke, 2005; Finarelli, 2006), as is their number of neurons and neuronal density (Jardim-Messeder et al., 2017).

Retrieving a relatively small item from a cylinder likely requires a certain level of sensorimotor proficiency, e.g., hand-limb coordination and degrees of mobility in relevant joints. To test whether sensorimotor requirements influence performance on the cylinder task, we presented cats with two different cylinder sizes: large or small (Experiment 1). The larger cylinders allowed for more behavioral options such that the cats could use either their head (mouth) or paws to retrieve the reward, and consequently did not require as precise motor planning as the small ones, where the cats were restricted to using a paw (see Figure 1). Moreover, the different sizes resulted in that the reward was positioned at different distances from the transparent surface, i.e., in a large cylinder the surface was further away from the reward. Previous studies have shown that the cats and some other species performed better in similar detour tasks if the reward was partially occluded, e.g., if the barrier was made from a mesh instead of a fully transparent surface (cats: Schiller, 1950, other species: Kabadayi et al., 2017). We therefore presented each cylinder size in two different materials, glass and plastic, which reflect light differently. If the animal perceives the barrier to a greater extent, even if it is transparent, then this might benefit motor self-regulation.

FIGURE 1

FIGURE 1. The two different sizes of cylinders used in the study and the retrieval behaviors they afford. (A) The large cylinder allows the cat to retrieve the reward directly with its mouth, but also by its paw. (B) The small cylinder only allows food retrieval by a paw. Note that both cylinders in the figure are made from plastic.

In each condition we largely followed the methods of the recent studies on the cylinder task in order to obtain comparable results (MacLean et al., 2014; Kabadayi et al., 2016).

Thereafter, we conducted two control experiments (Experiment 2 and 3) with the same subjects. The cylinders in Experiment 1 differed in terms of: (1) length of transparent surface, (2) distance between the surface and the reward, and (3) opening size. Experiment 2 and 3 tested which of these factors had the largest influence on the performance in Experiment 1.

Experiment 1: the Joint Effect of Size and Material

Materials and Methods

Subjects

We tested eight adult cats, three of which were females (see Table 1). Their ages ranged between 2 and 13 years. The cats were kept as pets in two different households (four cats from each household). All subjects lived both in- and outdoors in a rural environment and had daily experience with transparent surfaces, such as windows.

TABLE 1

TABLE 1. Individual information for the tested cats.

Apparatuses

We used four different transparent cylinders, and corresponding opaque cylinders, of glass and plastic. Two cylinders – one glass and one plastic – were large enough for the cats to retrieve the food with their mouths (18.5 cm in diameter and 25 cm in length), and two – one glass and one plastic – were small so as to encourage food retrieval by a paw (9.5 cm in diameter and 14 cm in length). See Figures 1, 2.

FIGURE 2

FIGURE 2. The four cylinders and their dimensions (Experiment 1). The reward was inserted either into a large-size cylinder (A,C) or a small-size cylinder (B,D). Two cylinders – one plastic (A,B) and one glass (C,D) – were manufactured within each size.

Procedure

Before the subject was tested on a transparent cylinder, it was familiarized with the task on a same-sized opaque cylinder, that is, each subject received familiarization with each size within the respective experiment. The subject always watched the experimenter place the food item inside the cylinder. In the familiarization phase the cat received five trials where it could retrieve the reward from any of the two side openings. The trial started when the cat was approximately one meter from the broad side of the cylinder and the experimenter inserted the reward in the middle of the cylinder. The side from which the reward was inserted was pseudo-randomized. The subject had to perform four correct detours out of five trials before testing commenced on the corresponding size a day later. All subjects completed the trials on the first try.

On the following day the cats were tested on the transparent cylinder. The procedure was identical to the familiarization phase. All cats were first tested on the large glass cylinder (N = 10 trials per condition), followed by the small plastic one, and thereafter by the small glass and the large plastic cylinder, with varying time intervals in between (see Supplementary Table 1). The reason for this order was that we hypothesized that the large cylinders would be easier to succeed in, and any carry over effect to the small cylinders would make the comparison more conservative (the number of individuals was too low to make counterbalanced test orders statistically meaningful). A failure was recorded if the cat touched the front of the cylinder with its snout or paw before entering the side opening and retrieving the reward. That is, if the animal oriented toward the visible reward from the front of the cylinder and touched its surface, it failed the trial. Regardless of failures, the subjects were eventually allowed to retrieve the reward. There was no time limit for completing a trial and none of the subjects left before retrieving the reward. The tests were conducted indoors in a well-lit room. The experimenter was familiar to the cats.

Statistics

Two-way ANOVA for 2¹ factorial design was used to estimate the effect of size and material on the success rate. The results were further confirmed in best general linear model selection with the individual score as a response variable, and two predictor variables: size and material. To test for the learning effect between the second and the first half of trials, a paired Wilcoxon signed rank test with continuity correction was used. All statistical analyses were conducted in R (v.3.3.2, the R Foundation for Statistical Computing²). Significance level was set at 0.05.

Results

In all failed trials, the cats directly oriented toward the front of the transparent barrier and touched it with its snout or paw, i.e., they did not ever first move toward a side opening. The success rates differed between conditions (see Figure 3 for overall results). Percentages were calculated for each individual and then a group average of these percentages was calculated. The highest average group level of success, 98.75%, was found in the large glass cylinder. This was followed by 97.5% success in the large plastic cylinder. In the small glass cylinder the subjects succeeded in 83.75%, and in the small plastic cylinder they reached 73.75% success.

FIGURE 3

FIGURE 3. A comparison of overall performance on the four cylinders in Experiment 1 (percentage of N = 10 trials correct).

The main effect of material [F(1,32) = 3.982, p = 0.056] and the interaction effect [F(1,32) = 1.524, p = 0.227] on the individual success rates were not significant, while the main effect of size was significant [F(1,32) = 22,462, p < 0.001] (see Supplementary Figure 1). Post hoc comparison using the Bonferroni correction confirmed the significance of the main effect of size on the success rate (p < 0.001).

The number of failures did not decrease over trials in any barrier (see Supplementary Table 2). We did not detect a learning curve over trials in any of the four conditions. A paired Wilcoxon signed rank test with continuity correction showed no significant difference between the success rate in the first five and the second five trials in any of the four conditions (small plastic: p = 0.1427, small glass: p = 0.4568, large plastic: p = 1, and large glass: p = 1). Overall, the number of failures was highest in the third and the 10th (last) trial (19%), and lowest in the first trial (3%).