Abstract
An animal's behaviour and its state, including its health and affective state, are dynamic and tightly coupled, influencing each other over time. Although both are relevant to the animal's welfare, there has been limited research on their dynamics in welfare studies. Here we aim to: (i) review evidence for feedbacks between state and behaviour that could have beneficial or detrimental consequences for farm animal welfare; (ii) propose ways in which an understanding of such feedbacks could be used to enhance welfare; and (iii) provide practical guidance. We include as state variables any features that could influence the costs and benefits of an animal's behavioural actions, including individual characteristics and aspects of its social environment. We find evidence supporting positive state-behaviour feedback loops in various livestock species, suggesting that these loops could be common in farm settings and have significant welfare implications, such as leading to abnormal behaviours and persistent negative affective states. We suggest (i) estimating within-individual feedback loops to extract individual characteristics for studying differences in welfare; (ii) identifying scenarios where change accelerated by positive feedbacks pushes an animal (or a group of animals) to a new state, also called tipping points; and (iii) generating positive feedback loops to elicit and maintain positive affective states. We end by encouraging use of dynamic models that integrate longitudinal data on animals' behaviour and state to enable exploration of their dynamics, and we provide a practical guide with annotated R code for support. Since the principles and ideas discussed here are relevant to any animals under human care, this approach could foster new perspectives for improving the welfare of all captive animals.