Temporal consistency of judgement biases in bumblebees

Judgement bias tasks are increasingly used to assess affective states in animals, yet the extent to which they might reflect transient states or stable traits remains unclear. Here, we tested bumblebees (Bombus terrestris) in an active choice task across three repeated sessions to assess individual consistency in the absence of any manipulation. Bees were trained to associate each of two colours with either a high or a low reward, presented in separate chambers. During testing, they were presented with ambiguous colours. Bees were more likely to choose the high-reward chamber and to choose more quickly in response to colours closer to the positive colour. The latency to choose the cues showed significant and moderate repeatability across sessions, suggesting a stable, trait-like underlying component. In contrast, the repeatability of the chamber choices was negligible, indicating that such responses might be largely state-dependent and influenced by situational factors. These findings suggest that judgement biases, particularly as assessed through an active choice task, reflect states affected by external factors. Active choice tasks may help disentangle stable behavioural traits from transient affective states in invertebrates.