A promising novel judgement bias test to evaluate affective states in dogs (Canis familiaris).

The judgement bias test represents one of the most applied tools to evaluate animals' optimistic/pessimistic attitude and to infer their emotional and welfare state accordingly. The judgement bias test (JBT) has been used several times with dogs (Canis familiaris), in most cases using a spatial test that evaluates the dog's attitude towards a bowl placed in ambiguous positions (located between two unambiguous trained positions associated with opposite outcomes). Results are contrasting and methodological and statistical caveats emerged: dogs struggled to learn the association between unambiguous positions and their outcomes, they hardly discriminated between adjacent locations and they might be influenced by researchers. Therefore, we propose a novel paradigm, aimed at easing the learning process and at achieving more reliable measures. Improvements of the novel paradigm are the increased difference between payoffs of trained locations, the reduction of the number of trials and of their length and the removal of the potential influence of researchers. Results showed that 98% of dogs reached the learning criterion and that their learning appeared more stable: dogs behaved differently between the two trained stimuli and the variability of responses towards these stimuli was lower than the one towards ambiguous stimuli. Behavioural analyses confirmed that dogs fully learned outcomes associated with trained stimuli and that they were hesitant towards ambiguous stimuli. Furthermore, dogs managed to successfully discriminate between each pair of adjacent locations. These results suggest that this protocol is a promising tool to assess judgement biases in dogs and to evaluate their affective state.