Face-based perception of emotions in dairy goats

- We investigated the potential use of images of faces as cognitive bias stimuli.
- Goats react differently to pictures of faces taken in positive and negative situations.
- Attention is higher towards negative faces than towards positive or morphed ones.

Faces of conspecifics convey information about identity, but also gaze, and attentional or emotional state. As a cognitive process, face-based emotion recognition can be subject to judgement bias. In this study we investigated whether dairy goats (n = 32) would show different responses to 2-D images of faces of familiar conspecifics displaying positive or negative emotional states. We also examined the possible use of images of faces as stimuli in cognitive bias studies. The faces of four subjects were photographed in a positive and a negative situation. Three types of images of ambiguous facial expressions were then created using morphing software (75% positive, 50% positive, and 25% positive). In a test-pen, each goat was exposed for 3 s to each type of image, obtained from the same goat. All goats were shown non-morphed faces first, before being shown the three types of morphed faces, balanced for order. Finally, the first non-morphed face was shown again. Spontaneous behavioural reactions including ear postures (forward, backward and asymmetrical) and interactions with the screen (time spent looking or touching) were recorded during the 3 s. Results were analysed using REML with repeated measurements. Goats spent more time with their ears forward when the negative was shown compared to the positive(F4,121.3 = 2.51, P = 0.018), indicating greater interest in negative faces. Identity of the photographed goat influenced the time spent with the ears forward (F2,57.4 = 7.01, P = 0.002). We conclude that goats react differently to images of faces displaying different emotional states and that they seem to perceive the emotional valence expressed in these images. Response to morphed faces was not necessarily intermediate to response to negative and positive faces, and not on a continuum. Further study is thus needed to clarify the potential use of faces in cognitive bias studies.