Short communication
A note on the effect of a palatable food reward on operant conditioning in horses

https://doi.org/10.1016/j.applanim.2007.02.010Get rights and content

Abstract

We investigated the effect of two food types (hay or pellet) on operant conditioning of 12 horses. One group was trained using pelleted concentrate (Group P) and the second (Group H) was trained using cut timothy hay as rewards. We investigated the effect of the food type on the time needed to complete the operant conditioning. There was no difference between the groups in the time needed to complete the operant conditioning (P = 0.78). Following completion of the operant conditioning, 10-min trials were conducted on five consecutive days. On the 2nd and 3rd days the food rewards (P and H) were switched for the two groups, and on the 4th and 5th days the food rewards were returned to the original type. When the food reward was changed from P to H, the mean number of rewards obtained decreased significantly in Group P (the 1st day (121.8) versus the 2nd day (45.4) and the 3rd day (39.4), P < 0.05, respectively) and tended to decrease in Group H (the 3rd day (120.4) versus 4th day (80.8), P = 0.06). The change induced the extinction of operant responses in three horses in Group P. When the food reward was changed from H to P, the mean number of rewards obtained increased significantly in Group P (the 3rd day (39.4) versus the 5th day (105.4), P = 0.10) and Group H (the 1st day (62.2) versus the 3rd day (120.4), P < 0.01). These results suggest that the palatability of the food reward may not improve the time required to train horses, but changing the food reward from a more palatable to a less palatable food may decrease the number of operant responses in horses, and extinguish them in some horses that were initially conditioned using a more palatable food. These results also suggest that whatever food reward is initially used in operant conditioning protocols in the horse may be important if it is necessary to subsequently change the food reward.

Introduction

Operant conditioning is often employed in the management of horses where a trainer uses positive and negative reinforcers to encourage a horse to perform the desired behavioural response. A positive reinforcer (e.g. food) is used after the desired behavioural response, whereas a negative reinforcer (e.g. pain) is removed when the animal performs the desired behavioural response (Waran et al., 2002). Successful training could influence horses’ utility and economic value, and it is, also, important for the welfare of horses to investigate the effect of positive, rather than negative reinforcers. However, there are only limited scientific studies on operant conditioning in horses and further research in this area is required (Nicol, 2002). Currently only a few studies relating to positive reinforcement (Myers and Mesker, 1960) and primary reinforcers (e.g. food, Houpt et al., 2004) for horses have been conducted.

In this study, we used a pelleted concentrate and cut timothy hay as food rewards during operant conditioning. The pelleted concentrate used in this study contained molasses, for which horses have a preference over timothy hay (Thorne et al., 2005). Therefore, we hypothesized that pelleted concentrate might be a palatable food and, as a reward, have a greater beneficial effect on operant conditioning compared to cut timothy hay. In our first experiment we investigated the effect of the palatability of the food reward on the time required to complete an operant conditioning task. It has been suggested that, during operant conditioning, animals can learn to have an expectation that pushing a button leads to a food reward (Toates, 2004). In our second experiment, we investigated the effect of changing the food reward, after the initial operant conditioning had been completed, on the operant responses of the horses. It was hypothesized that the change of reward would affect the horses’ expectation that had been induced by the operant conditioning and change their response rates.

Section snippets

Animals and management

Twelve thoroughbred female horses ranging from 2 to 5 years old (2 years old: n = 3; 3 years old: n = 4; 4 years old: n = 3; 5 years old: n = 2) were used at the Equine Research Institute of the Japan Racing Association in Utsunomiya, Japan. They were kept individually in box stalls (3.6 m × 5.2 m). They were turned out into a pasture or a paddock from 09:00 h to 13:00 h. Each received a pelleted concentrate and oat mixture (2.75 kg, fresh weight) and cut alfalfa hay (300 g, fresh weight) at 06:00 h and 16:00 h,

Experiment 1

There was no difference in the mean time needed to complete the operant conditioning between the two groups of horses (P: 46.6 min (S.E. 14.6), H: 41.3 min (S.E. 10.8), t = 0.29, P = 0.78) or in the mean number of rewards obtained during operant conditioning (P: 230.3 (S.E. 51.6), H: 137.7 (S.E. 18.8), t = 1.69, P = 0.12).

Experiment 2

Fig. 1 shows the number of rewards obtained in Groups P and H on the 5 days. In Group P, there was a difference in the mean number of rewards obtained among the 5 days (ANOVA: F4,16 = 

Discussion

The pelleted concentrate used in this study contained molasses, for which horses have a preference over timothy hay (Thorne et al., 2005). Moreover, as our two experiments were held after the horses had eaten timothy hay, it might be expected that horses would be more motivated to eat the pelleted concentrate than the hay because it is said that horses are motivated to eat a variety of foods (Ninomiya et al., 2004, Thorne et al., 2005). It has been also reported that horses pressed a panel for

Conclusions

Hay appeared to motivate horses to perform an operant response less than pelleted concentrate. However, there was no difference between hay and pelleted concentrate in the time it took to train the horses to perform the operant response. For horses trained with a palatable food, a change to one that is less palatable may induce frustration, resulting in extinction of, or at least a severe decrease in the operant response in some horses. It would appear to be important to consider the

Acknowledgement

We thank the staff of the Equine Research Institute, Japan Racing Association, for their support during this study.

References (11)

There are more references available in the full text version of this article.

Cited by (12)

  • Investigation of donkeys learning capabilities through an operant conditioning

    2022, Applied Animal Behaviour Science
    Citation Excerpt :

    Indeed, reinforcement can be positive or negative and, in either case, it is applied to increase the likelihood that one specific behaviour (born as a response to a stimulus) will recur in the future (Baragli et al., 2015; McLean and Christensen, 2017). Positive reinforcement consists in supplying the animal with something pleasant (Skinner, 1938) following the manifestation of the desired behavioural response (Lewis et al., 2021; McLean and Christensen, 2017; Ninomiya et al., 2007). For instance, it may be a particularly palatable feed or the possibility to have access to social interaction (in social species).

  • Hucul horses’ learning abilities in different learning tests and ue the association with behaviour, food motivation and fearfulness

    2021, Applied Animal Behaviour Science
    Citation Excerpt :

    We also recommend that future studies include behavioural indicators of frustration and ‘over-arousal’ caused by the use of food rewards during learning tests, which may act to impair attention and learning in highly food-motivated horses. It would also be interesting to include manipulation of food motivation, i.e. by measuring learning performance in hungry vs. satiated horses as well as altering the value of the reward (e.g. hay vs. concentrates; Ninomiya et al., 2007). Furthermore, as both motivation and fearfulness can affect learning, it is likely that for some animals, disturbing external environmental stimuli can overshadow the presence of food, while for other animals, the opportunity to obtain food may overshadow environmental stimuli.

  • The effects of a choice test between food rewards and human interaction in a herd of domestic horses of varying breeds and experiences

    2020, Applied Animal Behaviour Science
    Citation Excerpt :

    Within the original design, the study called for using patting and scratching as a potential reward for shaping two groups of horses to touch the target, none of the horses in either the scratching or patting groups responded to behavioral shaping did not succeed in learning to touch the target. This findings support current research showing that treats are more effective than grooming as a reward for training through positive reinforcement (Ninomiya et al., 2007; Sankey et al., 2010a). The horses in this study also demonstrated behaviors indicating physiological arousal (Table 1) which could have potentially influenced preference for human interactions.

  • Food motivation in horses appears stable across different test situations

    2018, Applied Animal Behaviour Science
    Citation Excerpt :

    In horses, Søndergaard et al. (2011) used a specially designed panel that opened a door by a set number of muzzle presses to measure motivation for social contact. Ninomiya et al. (2007a, 2007b) also used a device operated by muzzle pressing to measure the effect of reward palatability and satiation in horses. Despite the potential relevance of individual food motivation for animals’ performance in positively reinforced behavioural tests, there is currently no ‘golden standard’ for assessment of food motivation in horses and it is unknown whether food motivation is consistent across different test situations.

View all citing articles on Scopus
View full text