Humans can accurately categorise negative but not positive emotional facial expressions in horses

Recognising emotional facial expressions plays a key role in communication, both within and between species. Many non-human animals, including horses, discriminate and react to emotional human facial expressions. This raises the question of whether humans also consider some animals’ facial expressions when determining their emotional states. To address this, the present study aimed to assess human ability to categorise horses’ facial expressions according to their valence (positive/negative) and arousal (high/low), across eight distinct situations likely to elicit emotions (e.g., social isolation, going to a food bucket), and whether prior experience referring to the level of contact with horses improved this ability. An online task, in the form of a quiz, was conducted in which human participants were asked to categorise photographs of horses’ facial expressions based on perceived emotional valence and arousal. Results showed that participants (n = 930) performed well in the valence categorisation of expressions displayed in situations likely to elicit negative emotions (e.g., social isolation: 90 % of correct categorisation, sudden stimulus: 91 %), and that experience with horses improved performance in these cases. However, participants had greater difficulty in categorising the valence of facial expressions emitted in positive contexts (e.g., grooming: 42 %, going to a food bucket: 59 %), and experience, based on their level of contact, did not consistently enhance performance. Low arousal context (resting in the sun: 93 %) was well recognised, while categorisation accuracy for high arousal contexts (e.g., going to a food bucket: 55 %, sudden stimulus: 96 %) was more variable. These findings suggest that humans have a limited ability to recognise horses’ emotional states based on facial expressions, particularly for positive emotions, highlighting the need for increased awareness and caution when interpreting them. Accurate recognition of animals’ emotional facial expressions is therefore crucial, as it contributes directly to the broader assessment of their welfare.