• Animal welfare goes beyond absence of suffering. • We identify 3 features of the environment to be optimised. • The situations faced by animals should be mostly pleasant. • The environment should offer some complexity in space or time. • The environment should allow the animal to express its agency. Animal welfare was initially conceived as the absence of negative experiences. More recently, the importance of positive affects was recognised. This suggests a continuum from very poor welfare, where suffering predominates, to very good welfare, where positive affects predominate, and that the intrinsic hedonic value of the environment, i.e. whether it is pleasant or not, determines the level of welfare an animal can reach on this continuum. In this paper we question whether environments with a high intrinsic hedonic value are enough to allow good welfare. For instance, pleasant but monotonous environments can induce boredom, which is likely to affect animals negatively. By contrast, complex environments can encourage engagement. In addition, the possibility to act freely and efficiently plays a role. Environments that limit behavioural expression or make behaviour inefficient can lead to frustration and apathy. By contrast, environments that give freedom of action and make these actions efficient encourage animals to exert their agency – through choices, control, problem solving, free action, or exploration. We argue that intrinsic hedonic value, complexity, and possibility for action all contribute to welfare. However, a constant high level of hedonic value may result in a decrease in the hedonic value perceived by the animal. An overly complex environment or one that affords too wide a variety of actions may overwhelm animals. Therefore, rather than being maximised, hedonic value, complexity, and possibility for action need to be optimised to maximise animal welfare. Where the optimum sits varies between individuals depending on factors such as species, age, physiological status, or individual traits.
Veissier et al. (2026) studied this question.