Abstract: The millennia-old practices of horse training markedly predate and thus were isolated from the mid-twentieth century revelation of animal learning processes. From this standpoint, the progress made in the application and understanding of learning theory in horse training is reviewed including a discussion of how learning processes are employed or otherwise under-utilised in training. This review describes the process of habituation and the most commonly applied desensitisation techniques (systematic desensitisation, counter-conditioning, overshadowing, response prevention) and propose two additional techniques (approach conditioning and stimulus blending). The salience of different types of cues, the interaction of operant and classical conditioning and the impact of stress are also discussed. This paper also exposes the inflexibility and occasional inadequacy of the terminology of learning theory when translated from the research laboratory situation to the practical setting in horse training. While learning theory provides a rich toolbox for riders and trainers, the training process is subject to the simultaneous use of multiple learning processes. In addition, learning/behavioural outcomes and trained responses are not just the result of simple stimulus-response based interactions but are further shaped by arousal, affective and attachment states. More research is needed in these areas. For the field of equitation science to progress and to improve clarity and use of learning processes, changes in nomenclature are required. In particular, the use of the terms 'positive' and 'negative' as descriptive labels in both reinforcement and punishment modalities are unacceptably misleading for everyday use. These labels inhibit the understanding and recognition of the learning processes that these terms supposedly represent, yet the learning processes they describe are vital for horse riders, handlers and trainers to understand. We therefore propose that these labels should be re-labelled more appropriately as 'addition' or 'subtraction' reinforcement/punishment. This would enlighten trainers on the correct application of learning theory, and safety and welfare benefits for people and horses would follow. Finally it is also proposed that the term 'conflict theory' be taken up in equitation science to facilitate diagnosis of training-related behaviour disorders and thus enable the emergence of improved training practices. The optimal use of learning theory should be established as a fundamental principle in equestrian education.

Abstract: The emotion fear promotes the fitness of wild animals. In a farm environment, exaggerated fear, e.g., in horses, can cause several problems. Therefore, knowledge about fear in horses helps to prevent or to handle potential fear-inducing situations. The present study investigated which behavioural fear responses can be observed during exposure of horses to a novel stimulus, whether these behavioural responses are correlated with physiological changes, and whether and how specifically these changes are reduced after habituation training to one of the novel objects. Our data shows that behavioural and physiological fear responses in horses are correlated, are reliable to observe and to measure, and appear in a typical chronological order. Furthermore, after habituation-training to an object, the fear response to this object is specifically attenuated whereas the fear response to another object remains.

Abstract: Horses’ fear reactions are hazardous to both horses and human beings, but it is not clear whether fear is influenced more by training or by other factors such as genetics. The following study was designed to detect differences between young, untrained (U) and older, well-trained (T) horses of dressage (D), show-jumping (J), and mixed (M) genetic lines with regard to intensity of reaction and ease of habituation to a frightening stimulus. In five consecutive trials, 90 horses were exposed to a standardized fear-eliciting stimulus where intensity and duration of the reactions were recorded. Repeated measures analysis showed that flight reactions by J were less intense (p < 0.05) than those by D or M regardless of training status or age. Habituation to the stimulus over time was not significantly (p > 0.1) different between the disciplines, as indicated by similar slopes for all measurements, but reaction vigour declined faster for T than for U. These findings indicate that there may be a genetic basis for less strong, though not shorter-lasting, fear reactions in J compared to D or M lines of horses. Research including the estimation of genetic correlations between traits related to fearfulness and to performance would be required to verify this assumption.

Abstract: Habituation to frightening stimuli plays an important role in horse training. To investigate the extent to which horses generalise between different visual objects, 2-year-old stallions were habituated to feeding from a container placed inside a test arena and assigned as TEST (n=12) or REFERENCE horses (n=12). In Experiment 1, TEST horses were habituated to six objects (ball, barrel, board, box, cone, cylinder) presented in sequence in a balanced order. The objects were of similar size but different colour. Each object was placed 0.5m in front of the feed container, forcing the horses to pass the object to get to the food. TEST horses received as many 2min exposures to each object as required to meet a habituation criterion. We recorded behavioural reactions to the object, latency to feed, total eating time, and heart rate (HR) during all exposures. There was no significant decrease in initial responses towards a novel object with increasing object number, indicating that habituation was stimulus-specific. REFERENCE horses were exposed to the test arena without objects in the same period, and reactions of TEST and REFERENCE horses were subsequently compared in a fear-test (2m2.5m rubber mat placed under the feed container, forcing the horses to step on the mat to get food). There were no significant differences between the treatment groups, i.e. previous habituation of TEST horses to six visual objects did not reduce responses in a fear-test involving visual and tactile stimulation. Due to the lack of generalisation in Experiment 1, we designed a supplemental experiment (Experiment 2), in which REFERENCE horses were exposed to the same six objects except that object colour was kept constant. We found a significant reduction in response (behaviour and HR) with increasing object number, indicating that horses generalise between similarly coloured objects of varying shape. We conclude that a high degree of object similarity, e.g. identical colouring, appears to be crucial for object generalisation in horses.

Abstract: Nonreinforced preexposure to a conditioned stimulus impairs subsequent conditioning with that stimulus. The goal of these studies was to assess the extent to which acquisition performance could be predicted from preexposure performance using a correlational approach. For both preexposure and autoshaping, four measures of performance were computed, including overall average lever pressing, lever pressing in the initial session, percentage change in lever pressing, and slopes. These measures were correlated in a large sample of rats trained in an autoshaping situation. None of the three measures of autoshaping performance was consistently predicted by any of the three measures of preexposure performance. These results are consistent with the view that latent inhibition is not reducible to long-term habituation.