Simple SummaryGood animal welfare requires minimizing suffering and promoting positive experiences. To achieve this, we need reliable indicators of animals’ psychological states. In humans, different moods and emotions (“affects”) are associated with changes in visual attention (“attention bias”). We review studies investigating whether attention biases are also indicators of affect in animals. Although research is limited, evidence for affect-driven attention biases has been found in several species, especially primates and livestock. These studies are discussed in relation to tasks developed for measuring attention in humans. We identify additional findings from human psychology that might be applied to animals, particularly species not studied before, and conclude that affect-driven attention bias is a promising welfare indicator. However, it may be more useful for studying responses to specific stimuli, rather than general wellbeing. With further study, we hope these findings contribute to fulfilling society’s ethical obligations towards animals.AbstractAttention bias describes the differential allocation of attention towards one stimulus compared to others. In humans, this bias can be mediated by the observer’s affective state and is implicated in the onset and maintenance of affective disorders such as anxiety. Affect-driven attention biases (ADABs) have also been identified in a few other species. Here, we review the literature on ADABs in animals and discuss their utility as welfare indicators. Despite a limited research effort, several studies have found that negative affective states modulate attention to negative (i.e., threatening) cues. ADABs influenced by positive-valence states have also been documented in animals. We discuss methods for measuring ADAB and conclude that looking time, dot-probe, and emotional spatial cueing paradigms are particularly promising. Research is needed to test them with a wider range of species, investigate attentional scope as an indicator of affect, and explore the possible causative role of attention biases in determining animal wellbeing. Finally, we argue that ADABs might not be best-utilized as indicators of general valence, but instead to reveal specific emotions, motivations, aversions, and preferences. Paying attention to the human literature could facilitate these advances.