Validating attention bias as a novel measure of affect in sheep

Abstract

These datasets were generated during the PhD candidature of Jessica Monk between 2016 to 2019. Data were generated across 5 experiments, each of which aimed to further develop, refine and validate a novel method for assessing emotional states in sheep as a measure of animal welfare. The attention bias test is a 3 min test which records behavioural responses of sheep to a potential predator threat. Key behaviours recorded included duration of vigilance, looking behaviours and latency to feed. The first experiment examined the impact of pharmacologically-induced stress on judgement bias and attention bias in sheep. The study found no clear indication that elevated cortisol concentrations impacted on cognitive biases in sheep. The second experiment examined the impact of pharmacologically induced anxiety-like and calm-like states on a refined attention bias test in sheep, finding that the test can be shortened, a habituation period removed and that it is sensitive to anxiety-like and calm-like states. The third experiment introduced a modified method for the assessment of attention bias, and examined the impact of pharmacologically-induced anxiety-like and depression-like states in sheep. The modified method could differentiate these states. The fourth experiment examined the influence of pharmacologically induced positive affective states on the modified attention bias test but found no clear effect on attention bias. The final experiment examined the repeatability of the attention bias test across 3 time periods using the same animals from the Sheep CRC information nucleus flock. Key measures of attention had low repeatability and are suggested to be readily influenced by emotions and moods. Measures of vigilance and zones crossed had moderate repeatability and are suggested to be more heavily influenced by temperament or personality traits.