Social learning in horses : does the demonstration of a conspecific affect the ability to solve a detour task?

It has been difficult to find an effect of social learning in horses, perhaps more attention should be given to factors shown to affect it in other species, such as social dominance, group stability and experimental design. The aim of the present study was to investigate social learning by considering the effects of observing a trained dominant horse demonstrating a detour task. Seventeen horses (5 geldings and 12 females) of different breeds and ages took part in this study, all have been kept permanently on pasture in the same social group for the last 3 years. The dominance hierarchy order was defined through observation of dyadic interactions (focal animal sampling method): agonistic and submissive behaviours were ranked in a squared matrix. The highest rank horse was trained for detouring a U shaped fence to reach a bin with food. Horses were assigned to Control or Demonstrator Group using a randomization process stratified for gender, age and social rank. Single horses underwent to 10 detour trials divided in 3 sessions. Prior to testing, horses from Demonstrator Group were allowed to watch a conspecific detouring. We analysed the ability to solve the task and the latency to complete the detour. Mann-Whitney test was used to verify differences between groups. We found that demonstration did not improve the ability of the horses in solving the task: 4 subjects of the Control Group versus only 2 subjects of the Demonstrator Group reached the goal. The mean time to complete the detour was 58s (SD=23) for the control group and 49s (SD=31) for the Demonstrator Group (NS). Our results clearly did not prove social learning in horses, however some important considerations can be derived about factors affecting similar studies. Although living in a stable social group is important in social learning tests, horses managed in groups may be, as in our case, more motivated by social companionship than by food rewards. A spatial task is more likely to be transmitted than an operant task with less obvious proximate advantages to the performer but, as horses have short-term spatial recall of less than 10 s, the time between the demonstration and the trial could have been too long. Further studies should consider social reinforcements and better timing. It would be also interesting to assess the effect of the demonstration of a leader instead of the dominant one