Abstract: This study evaluates the horse (Equus caballus) use of human local enhancement cues and reaction to human attention when making feeding decisions. The superior performance of dogs in observing human states of attention suggests this ability evolved with domestication. However, some species show an improved ability to read human cues through socialization and training. We observed 60 horses approach a bucket with feed in a three-way object-choice task when confronted with (a) an unfamiliar or (b) a familiar person in 4 different situations: (1) squatting behind the bucket, facing the horse (2) standing behind the bucket, facing the horse (3) standing behind the bucket in a back-turned position, gazing away from the horse and (4) standing a few meters from the bucket in a distant, back-turned position, again gazing away from the horse. Additionally, postures 1 and 2 were tested both with the person looking permanently at the horse and with the person alternating their gaze between the horse and the bucket. When the person remained behind the correct bucket, it was chosen significantly above chance. However, when the test person was turned and distant from the buckets, the horses’ performance deteriorated. In the turned person situations, the horses approached a familiar person and walked towards their focus of attention significantly more often than with an unfamiliar person. Additionally, in the squatting and standing person situations, some horses approached the person before approaching the correct bucket. This happened more with a familiar person. We therefore conclude that horses can use humans as a local enhancement cue independently of their body posture or gaze consistency when the persons remain close to the food source and that horses seem to orientate on the attention of familiar more than of unfamiliar persons. We suggest that socialization and training improve the ability of horses to read human cues.

Abstract: Dogs are known to consistently follow human pointing gestures. In this study, we asked whether dogs “automatically” do this or whether they flexibly adjust their behavior depending upon the reliability of the pointer, demonstrated in an immediately preceding event. We tested pet dogs in a version of the object choice task in which a piece of food was hidden in one of the two containers. In Experiment 1, Phase 1, an experimenter pointed at the baited container; the second container was empty. In Phase 2, after showing the contents of both containers to the dogs, the experimenter pointed at the empty container. In Phase 3, the procedure was exactly as in Phase 1. We compared the dogs’ responses to the experimenter’s pointing gestures in Phases 1 and 3. Most dogs followed pointing in Phase 1, but many fewer did so in Phase 3. In Experiment 2, dogs followed a new experimenter’s pointing in Phase 3 following replication of procedures of Phases 1 and 2 in Experiment 1. This ruled out the possibility that dogs simply lost motivation to participate in the task in later phases. These results suggest that not only dogs are highly skilled at understanding human pointing gestures, but also they make inferences about the reliability of a human who presents cues and consequently modify their behavior flexibly depending on the inference.

Abstract: Abstract— We present a study focused on a quantitative estimation of a human-horse dynamic interaction. A set of measures based on magnitude and phase coupling between heartbeat dynamics of both humans and horses in three different conditions is reported: no interaction, visual/olfactory interaction and grooming. Specifically, Magnitude Squared Coherence (MSC), Mean Phase Coherence (MPC) and Dynamic Time Warping (DTW) have been used as estimators of the amount of coupling between human and horse through the analysis of their heart rate variability (HRV) time series in a group of eleven human subjects, and one horse. The rationale behind this study is that the interaction of two complex biological systems go towards a coupling process whose dynamical evolution is modulated by the kind and time duration of the interaction itself. We achieved a congruent and consistent
statistical significant difference for all of the three indices. Moreover, a Nearest Mean Classifier was able to recognize the three classes of interaction with an accuracy greater than 70%. Although preliminary, these encouraging results allow a discrimination of three distinct phases in a real human-animal interaction opening to the characterization of the empirically proven relationship between human and horse.

Abstract: Group housed horses at a stud farm/riding stable in Belgium were observed on 17 days between 21 February and 25 April 2008, totalling 54hr25min of detailed data. The original group consisted of 8 Irish Cob mares, 1 Warmblood mare, 1 Arabian gelding and 2 Arabian mares. The group had been established in December 2007. During the course of the study 5 horses were removed from the group and 2 foals were born. 3 highly pregnant mares were housed adjacent to the group for part of the period. Horses were regularly used for lessons. Available surface area differed with the group on pasture at the end. Continuous all occurrence sampling of 10 agonistic and 2 affiliative behaviours was carried out for all group members present. Overall the group showed a frequency of 44.75 agonistic interactions per hour and 11.25 affiliative per hour. Of those agonistic interactions 46.3% were threats while 47% were less active interactions (displacement, being avoided), leaving only 6.7% more aggressive interactions ( mainly biting, some kicking and chasing). The effect on acting agonistically was not significant for age (p=0.1591) and borderline significant for density (p=0.0627). The analysis of the frequency of affiliative interactions showed there is no significant effect of age (p=0.1865) or density (p=0.7923). Agonistic and affiliative interactions were not significantly correlated (p=0.72). Affiliative behaviour a horse received showed a borderline effect (p=0.0787) on agonistic behaviour, as did the interaction between received agonistic and affiliative interactions (p=0.0725). Received agonistic interactions had a borderline negative effect (p=0.0656) on affiliative behaviour. A dominance hierarchy was calculated based on agonistic interactions using Empirical Bayes’ estimates based on Poisson regression with random effects. Agonistic behaviour expressed to other horses was significantly effected by relative rank (p=0.0243). Overall horses tended to be 3.7 times more aggressive towards lower ranking horses compared to higher ranking horses. Affiliative behaviour shown to other horses was not significantly influenced by the rank of the social partner (p=0.7915). Some individuals did show a significant effect whereby they showed more affiliative behaviour towards lower ranking individuals. This study was a small project to look at a practical situation of riding horses being kept in group housing. The frequent changes in group composition and available surface made it possible to look at agonistic and affiliative interactions in such circumstances. This is useful as instability in group composition is often used as main reason not to keep horses in social groups. The results from this study showed a borderline effect of density on agonistic behaviour. In reality it was also influenced by practical details, like a narrow paddock with only 2 hay crates on the smallest surface. Rank in a dominance hierarchy, based on agonistic behaviour, had a significant effect on the agonistic behaviour expressed towards higher or lower ranking horses. No injuries or escalating fights were observed. This study shows it is possible to keep a group of riding horses in a social context without excessive aggression.