Abstract: Horses are capable of identifying individual conspecifics based on olfactory, auditory or visual cues. However, this raises the questions of their ability to recognize human beings and on the basis of what cues. This study investigated whether horses could differentiate between a familiar and unfamiliar human from photographs of faces. Eleven horses were trained on a discrimination task using a computer-controlled screen, on which two photographs were presented simultaneously (32 trials/session): touching one was rewarded (S+) and the other not (S-). In the training phase, the S+ faces were of four unfamiliar people which gradually became familiar over the trials. The S- faces were novel for each trial. After the training phase, the faces of the horses' keepers were presented opposite novel faces to test whether the horses could identify the former spontaneously. A reward was given whichever face was touched to avoid any possible learning effect. Horses touched the faces of keepers significantly more than chance, whether it was their current keeper or one they had not seen for six months (t = 3.65; p < 0.004 and t = 6.24; p < 0.0001). Overall, these results show that horses have advanced human face-recognition abilities and a long-term memory of those human faces.

Abstract: We tested the hypothesis that social defensive responses to the vocalisation of a predator still exist in horses. The recordings of a grey wolf, an Arabian leopard and a golden jackal were played to 20 Konik polski and Arabian mares. Durations of grazing, standing still, standing alert and the number of steps in walk and trot/canter were measured. In one-minute scans, the distances of the focal horse from the reference horse (DIST-RH) and from the nearest loudspeaker (DIST-LS) were approximated. The vocalisation of a leopard aroused the Arabians more than the Koniks (less grazing, stand-still and walk, more stand-alert and trotting/cantering). Koniks showed more relaxed behaviours to the leopard vocalisation (more grazing, stand-still and walk), but high alertness to the wolf playback (stand-alert, trotting/cantering). Spatial formation of the herd of Koniks showed tight grouping (lower DIST-RH) and maintaining distance from the potential threat (DIST-LS) in response to the wolf howling, while the Arabians approached the loudspeakers in linear herd formation when the leopard growls were played. Adult horses responded to potential predation by changing spatial group formations. This ability to apply a social strategy may be one of the explanations for the least number of horses among all hunted farm animal species.

Abstract: A human-animal relationship can be developed through subsequent interactions, affected by the positive or negative emotional valence of the proceeding one. Horses implement a process of categorization to classify humans with whom they interact as positive, negative, or neutral stimuli by evaluating the kind of approach and the nature of the contact. In these terms, human-animal interactions are emotionally charged events, eliciting specific emotional states in both subjects involved. Although the human-horse relationship has been mainly investigated through behavioral analysis, physiological indicators are needed for a more objective assessment of the emotional responses. Heart rate variability (HRV) is a commonly used autonomic nervous system (ANS) correlate estimating the sympathovagal balance as a psychophysiological marker of emotion regulation in horses. We have assumed that long-term positive relationships with humans may have a positive and immediate impact on the emotional arousal of the horse, detectable, via ANS activity, during the interaction. We analyzed horses' heartbeat dynamics during their interaction with either familiar or unfamiliar handlers, applying a standardized experimental protocol consisting of three different conditions shifting from the absence of interaction to physical contact. The ANS signals were monitored through an innovative non-invasive wearable system, not interfering with the unconscious emotional response of the animal. We demonstrated that horses appeared to feel more relaxed while physically interacting (e.g., grooming on the right side) with some familiar handlers compared to the same task performed by someone unfamiliar. The shift of the sympathovagal balance toward a vagal predominance suggests that the horses experienced a decrease in stress response as a function not only of the handler's familiarity but also of the type of interaction they are experiencing. These results constitute the objective evidence of horses' capacity to individually recognize a familiar person, adding the crucial role of familiarity with the handler as a paramount component of human-animal interaction. Our rigorous methodological approach may provide a significant contribution to various fields such as animal welfare while further investigating the emotional side of the human-animal relationships.