Abstract: Understanding the factors that influence horse learning is critical to ensure horse welfare and rider safety. In this study, data were obtained from horses (n=96) training to step backwards through a corridor in response to bit pressure. Following training, learning ability was determined by the latency to step backwards through the corridor when handled on the left and right reins. Additionally, horse owners were questioned about each horse's management, training, behavior and signalment (such as horse breed, age and sex). Factors from these four broad domains were examined using a multiple logistic regression (MLR) model, following an Information Theoretic approach, for associations between horses' behavioral attributes and their ability to learn the task. The MLR also included estimates of the rider's ability and experience as well as owner's perceptions of their horse's trainability and temperament. Results revealed several variables including explanatory variables that correlated significantly with rate of learning. Horses were faster at backing, a behavioral trait, when handled on the right (t = 3.65, df = 94, P < 0.001) than the left side. Thoroughbred horses were slower at completing the tests than other breeds of horses when handled on the left side (LM, F1,48=4.5, P=0.04) and right side (LM, F1,45=6.0, P=0.02). Those in regular work, a training factor, did not learn faster than their unworked counterparts on the right rein but completed the task faster on the left rein (F1,44=5.47, P=0.02). This may reflect differences in laterality and habituation effects. In contrast, more anxious horses were faster at completing the test when handled from the right (Spearman, r=-0.22, P=0.04). It is possible that these horses have an increased arousal level when interacting with handlers, resulting in more engagement with the lesson, accounting for the improved performance results. The findings of this study will help clarify how horse behavior, training and management may influence learning and how their application may optimize learning outcomes. Future equine behavior assessment and research questionnaires should include items that assess these qualities.

Abstract: By definition, punishment makes a response less likely in the future. Because horses are largely trained by negative reinforcement, they are susceptible to inadvertent punishment. Delays in the release of pressure can make desirable responses less likely and thus punish them. This study examines the correct use of negative reinforcement and identifies a continuum between poorly timed negative reinforcement and punishment. It explores some of the problems of non-contingent punishment and the prospect of learned helplessness and experimental neurosis. It concludes by introducing the concept of ethical equitation.

Abstract: There is evidence that more than 47% of the sports horse population in normal work may be lame, but the lameness is not recognized by owners or trainers. An alternative means of detecting pain may be recognition of behavioral changes in ridden horses. It has been demonstrated that there are differences in facial expressions in nonlame and lame horses. The purpose of this study was to develop a whole horse ethogram for ridden horses and to determine whether it could be applied repeatedly by 1 observer (repeatability study, 9 horses) and if, by application of a related pain behavior score, lame horses (n = 24) and nonlame horses (n = 13) could be differentiated. It was hypothesized that there would be some overlap in pain behavior scores among nonlame and lame horses; and that overall, nonlame horses would have a lower pain behavior score than lame horses. The ethogram was developed with 117 behavioral markers, and the horses were graded twice in random order by a trained specialist using video footage. Overall, there was a good correlation between the 2 assessments (P < 0.001; R2 = 0.91). Behavioral markers that were not consistent across the 2 assessments were omitted, reducing the ethogram to 70 markers. The modified ethogram was applied to video recordings of the nonlame horses and lame horses (ethogram evaluation). There was a strong correlation between 20 behavioral markers and the presence of lameness. The ethogram was subsequently simplified to 24 behavioral markers, by the amalgamation of similar behaviors which scored similarly and by omission of markers which showed unreliable results in relation to lameness. Following this, the maximum individual occurrence score for lame horses was 14 (out of 24 possible markers), with a median and mean score of 9 (±2 standard deviation) compared with a maximum score of 6 for nonlame horses, with a median and mean score of 2 (±1.4). For lame horses, the following behaviors occurred significantly more (P < 0.05, chi-square): ears back, mouth opening, tongue out, change in eye posture and expression, going above the bit, head tossing, tilting the head, unwillingness to go, crookedness, hurrying, changing gait spontaneously, poor quality canter, resisting, and stumbling and toe dragging. Recognition of these features as potential indicators of musculoskeletal pain may enable earlier recognition of lameness and avoidance of punishment-based training. Further research is necessary to verify this new ethogram for assessment of pain in ridden horses.

Abstract: By definition, ethology is primarily the scientific study of animal behavior, especially as it occurs in a natural environment; applied ethology being the study of animal behavior in the human domain. The terms equine ethology and ethological training are becoming commonplace in the equestrian domain, yet they seem to be used with a conspicuous lack of clarity and with no mention of learning theory. Most of what we do to train horses runs counter to their innate preferences. This article summarizes the ethological challenges encountered by working horses and considers the merits and limitations of ethological solutions. It also questions the use of terms such as “alpha” and “leader” and examines aspects of learning theory, equine cognition, and ethology as applied to horse training and clinical behavior modification. We propose 7 training principles that optimally account for the horse's ethological and learning abilities and maintain maximal responsivity in the trained horse. These principles can be summarized as: (1) use learning theory appropriately; (2) train easy-to-discriminate signals; (3) train and subsequently elicit responses singularly; (4) train only one response per signal; (5) train all responses to be initiated and subsequently completed within a consistent structure; (6) train persistence of current operantly conditioned responses; and (7) avoid and disassociate flight responses. Adherence to these principles and incorporating them into all horse training methodologies should accelerate training success, reduce behavioral wastage of horses, and improve safety for both humans and horses.

Abstract: Horses are likely to exhibit aggression when meeting for the first time. Therefore, this study compared 3 methods for mixing horses to evaluate their effectiveness in reducing aggressive interactions: (1) mixing pairs of horses in a paddock (P, 10 minutes, 15 tests), (2) introducing 1 unfamiliar horse to a pair of familiar, resident horses in a paddock (PP, 10 minutes, 15 tests), (3) allowing limited physical contact between pairs of horses for a short period of pre-exposure in neighboring boxes (B, 5 minutes, 16 tests) before mixing them in a paddock (BP, 10 minutes 16 tests). A total of 16 Swedish Standardbred mares, aged 6-18 years (mean age ± SD: 11 ± 4.4), were included in the study. Half of the horses were familiar with each other (resident horses, n = 8), whereas the other half were bought in from a variety of sources (unfamiliar horses, n = 8). Social interactions, consisting of behaviors from the sender, the receiver, and the subsequent sender's response, were recorded continuously as frequencies. There were no differences in the frequencies of aggressive behaviors between the 3 mixing methods, including those aggressive behaviors in which physical contact had been attempted (kick, strike). Although resident horses were overall more aggressive (median number of aggressive behaviors per horse, 62; Q1, 36; Q3, 68.5) than unfamiliar horses (median per horse, 4; Q1, 2; Q3, 12.5) during all tests (U = 97, P = 0.003), none of the 62 tests needed to be terminated. Unfamiliar horses did not receive more aggression from resident horses in PP (mean per test ± SD: 5.1 ± 3.1) than in P (mean per test ± SD: 6.4 ± 4.9) (t = 0.63, P = 0.544). However, the behavior “attack” was more frequent in PP (median per test, 2; Q1, 0; Q3, 5) than in P (median per test, 0; Q1, 0; Q3, 1) (U = 282, P = 0.042), and “flee” was more frequent in PP (median per test, 6; Q1, 4; Q3, 8) than in P (median per test, 1; Q1, 0; Q3, 6) (U = 290, P = 0.018). Pre-exposure in boxes did not reduce aggression in BP (median per test, 7; Q1, 4.3; Q3, 11.8) as compared with P (median per test, 6; Q1, 2; Q3, 16) (U = 264, P = 0.767), but during pre-exposure in B tests, horses exchanged more nonaggressive (median per test, 2; Q1, 0.3; Q3, 4) than aggressive (median frequency of aggressive behavior, 0; Q1, 0; Q3, 1) (W = 71, P = 0.013) and mixed interactions (median per test, 0; Q1, 0; Q3, 1) (W = 92, P = 0.016) through the opening. Results suggest mixing an unfamiliar horse with 2 resident horses at the same time instead of one by one may be preferable. In this way, the total aggression received by the unfamiliar horse will potentially be less, even though aggressive interactions may be more intense.