Animal Emotion and Welfare

: A Decision-Making and Computational Approach

Abstract

To assess an animal’s welfare, we need to be able to accurately measure their affective state. The judgement bias task uses decision-making under ambiguity as a measure of affective state and can be used in humans and non-human animals. Despite the potential for judgement bias to be used as a cognitive indicator of animal affective state, we currently lack an understanding of how exactly affective states bias decision-making. Here I explore how state and trait affect can bias decision-making in humans and rats. Using an online judgement bias task, I confirmed that self-reported negative trait affect negatively biases ambiguous decision-making in humans (Chapter 2). In a subsequent study, I showed that a short-term positive forest-bathing manipulation induced a more positive judgement bias compared to an urban environment manipulation (Chapter 3). Computational modelling of these data suggested that judgement bias was driven by a bias in evidence accumulation during the decision-making process rather than a pre-existing bias towards one response. The almost exclusive use of secondary reinforcers in human judgement bias tasks may limit the translation of findings to animal tasks which use primary reinforcers. To address this, I used both primary and secondary reinforcers with the same human participants, and with a trial-by-trial analysis approach I identified differences in reaction times and self-reported affective valence, but not overall responses (Chapter 4). Unfortunately, there were several issues with sufficiently training rats on a judgement bias task however plans for a probabilistic task variant and initial training data are discussed (Chapter 5). Finally, I explored how individual differences in rat trait affect relate to behaviour in an alternative patch foraging decision-making task; four measures of trait affect were identified through repeated testing of traditional measures of rat affect however, only one trait was significantly associated with behaviour in the task (Chapter 6).

Date of Award	7 May 2024
Original language	English
Awarding Institution	University of Bristol
Supervisor	Michael T Mendl (Supervisor), Iain D Gilchrist (Supervisor), Vikki M Neville (Supervisor), E S Paul (Supervisor) & John G Fennell (Supervisor)