Abstract: Brain lateralization is a phenomenon widely reported in the animal kingdom and sensory laterality has been shown to be an indicator of the appraisal of the stimulus valence by an individual. This can prove a useful tool to investigate how animals perceive intra- or hetero-specific signals. The human-animal relationship provides an interesting framework for testing the impact of the valence of interactions on emotional memories. In the present study, we tested whether horses could associate individual human voices with past positive or negative experiences. Both behavioural and electroencephalographic measures allowed examining laterality patterns in addition to the behavioural reactions. The results show that horses reacted to voices associated with past positive experiences with increased attention/arousal (gamma oscillations in the right hemisphere) and indicators of a positive emotional state (left hemisphere activation and ears held forward), and to those associated with past negative experiences with negative affective states (right hemisphere activation and ears held backwards). The responses were further influenced by the animals' management conditions (e.g. box or pasture). Overall, these results, associating brain and behaviour analysis, clearly demonstrate that horses' representation of human voices is modulated by the valence of prior horse-human interactions.

Abstract: This field study tested the hypothesis that domestic horses living under putatively challenging-to-welfare conditions (for example involving social, spatial, feeding constraints) would present signs of poor welfare and co-occurring pessimistic judgement biases. Our subjects were 34 horses who had been housed for over 3 years in either restricted riding school situations (e.g. kept in single boxes, with limited roughage, ridden by inexperienced riders; N = 25) or under more naturalistic conditions (e.g. access to free-range, kept in stable social groups, leisure riding; N = 9). The horses' welfare was assessed by recording health-related, behavioural and postural indicators. Additionally, after learning a location task to discriminate a bucket containing either edible food ('positive' location) or unpalatable food ('negative' location), the horses were presented with a bucket located near the positive position, near the negative position and halfway between the positive and negative positions to assess their judgement biases. The riding school horses displayed the highest levels of behavioural and health-related problems and a pessimistic judgment bias, whereas the horses living under more naturalistic conditions displayed indications of good welfare and an optimistic bias. Moreover, pessimistic bias data strongly correlated with poor welfare data. This suggests that a lowered mood impacts a non-human species' perception of its environment and highlights cognitive biases as an appropriate tool to assess the impact of chronic living conditions on horse welfare.