Goats do not use social information from conspecifics and humans in a manipulation task

Increasing knowledge on livestock behaviour will be helpful to define the best policies for animal welfare. The study of social learning is important for a better understanding of how the information's transmission from an animal to another happens. I investigated whether goats used horizontal transmission of information by observing conspecifics or a human in a manipulation task using a bidirectional control procedure: the goats had to learn how to slide a door to either the left or the right side to obtain a reward. Overall, 27 female dairy goats and 23 female dwarf goats were tested. The experiment has been conducted in August 2018 at the Agroscope Taenikon Institute, Switzerland. The video recordings were made available within the project "Impact of domestication on learning and cognitive capacities in goats (Capra hircus) – effects of long-term cognitive training on livestock welfare and husbandry" coordinated by Dr. Jan Langbein from FBN Dummerstorf and Dr. Nina Keil from Agroscope Taenikon. Goats were divided into three groups and the perceive one of the following three conditions: a “ghost” demonstration (the door opened by itself), a human demonstration, and a conspecific demonstration. Each goat was exposed to 10 trials. Each trial had three phases: motivational, observation and testing. Throughout the trials each goat could observe the same demonstrator in the observation phase. During the testing phase, the goats had 30 seconds to interact with the device: if they touched the device they had 30 seconds in addition to open the door and receive the reward located behind the device. The parameters scored were: 1. “subject's interaction with the door (touching, moving, opening)”; 2. “direction of door's first movement and opening”; 3. “latency of the door's touching or opening”. I analysed the influence of the 'breed' (dairy and dwarf goats) and 'condition' (“ghost”, conspecific and human demonstrators) using GLMMs and subsequent Likelihood ratio tests. Where there were found a significative p-value comparing the GLMMs in 'breed' it was used the Anova test (because I had two breeds to compare), if found a significative p-value in 'condition' case it was used the Tukey test (because I had three condition to compare) Results showed that 'breed' influenced the “subject's interaction with the door” and the “latency of the door's touching or opening”. The results suggest that the two breeds did not differ for the direction of the first door's movement as well as in the direction in which they open the door. 'Condition' influenced only the latency of touching the sliding door between human and conspecific observers. Overall, my results suggest that goats don't use social information from conspecifics and humans in a problem-solving task, but their motivation to interact with the device is likely influenced by the presence (or absence) of conspecifics.