Contributed By – Dr. Joanna Hockenhull
Kieson E, Felix C, Webb S, Charles I. Abramson CI (2020) The effects of a choice test between food rewards and human interaction in a herd of domestic horses of varying breeds and experiences. Applied Animal Behaviour Science 231: 105075 https://doi.org/10.1016/j.applanim.2020.105075
The use of reward-based training for horses, as an alternative to more traditional training practices based on pressure/release (negative reinforcement) is becoming more widespread. As usual when it comes to changes in knowledge and practices, we saw reward-based methods being widely used in other species first. Food rewards are often used in dog training, although alternatives such as access to a toy are also suggested based on what your individual dog finds reinforcing. We are also seeing food rewards being used as part of positive reinforcement training in horses, often alongside a clicker or vocal cue. Some people have expressed concerns about using food as part of their training practice, for example due to fears that their horse may start ‘mugging’ them for food or because their horse becomes over-excited once food is involved. As horses aren’t well known for finding play with a toy sufficiently motivating as a reward during training, equestrians are often advised to use scratching as an alternative to food rewards. But is receiving a scratch from a human rewarding for a horse? This study aimed to find out, while also exploring whether equine preference for food or scratches was associated with the history and experience of the horse, and the familiarity of the human scratcher.
Eleven horses were taught the meaning of three different symbols: an X represented food treatment, an O represented scratching by a human and a solid square represented patting (another often used human to horse contact intended as a reward) by a human. Both scratching and patting were restricted to the horses’ neck area, and a bamboo back scratcher was used for the scratching treatment to make it more consistent between researchers. The horses ranged in age from 8-20 years and were a mixture of breeds. There were 6 geldings and 5 mares, and the researchers knew the background of each horse. Test trials with a researcher familiar to the horses and an unfamiliar researcher took place over 6 test days. Trials took place in the horses’ home field, in visual contact with the remainder of their herd. The horse’s behaviour during the test trails was also observed.
The familiar researcher conducted the initial training to familiarise the horses with the symbols and their meanings. The researcher ended up having to train all of the horses to touch the symbols using the X symbol and food reward only, as the horses in the scratching and patting groups did not learn to touch the symbol when trained using these rewards. Once all horses were trained to touch the X symbol, the researcher changed the symbols to teach the horses that touching them resulted in a different reward depending on the symbol.
Subsequent discrimination and preference tests revealed that all 11 horses preferred food rewards over human contact (patting or scratching). The background and experience of the horse did not affect preference, neither did the familiarity of the researcher. When given the choice between touching the X symbol for food, or touching the O or the solid square for scratching or patting the horse always chose the X. When the O and solid square were presented together, the horses chose not to touch either symbol.
Behaviour indicating heightened arousal was observed in all horses when the O and solid square symbols were presented. There were consistent individual differences in the arousal behaviours each horse displayed, for example pawing, muscle tension, raised neck and biting the target. In addition, the researchers noted other behaviour associated with the testing regime. The horses all willingly entered the test area and approach the target without the need for any reinforcement, halters of leads. They also reported that some of the horses who were typically difficult to catch in the field actually became easier to catch after the first training session (which used food rewards).
The researchers concluded that in training conditions where horses are required to complete a specific task, food rewards are preferred by horses over human interaction (patting or scratching). They questioned whether the heightened arousal observed during the tests may have influenced the preferences seen, speculating that the horses were perhaps too aroused at that point to find human interaction rewarding.
The findings of this study are really interesting and demonstrate the power of using food as a reward in positive reinforcement training. But the results also raise a lot of questions in relation to how consistent these findings may be across different contexts (both training and otherwise), and whether reward preference may change with equine arousal levels.