Cognitive bias testing

A method developed in welfare science for assessing whether an animal is in a positive or negative affective state by measuring how they interpret ambiguous information.

An animal in a positive state tends to interpret ambiguous cues optimistically (treating them as predicting positive outcomes); an animal in a negative state interprets them pessimistically (treating them as predicting negative outcomes). The method provides a window onto affective state that does not depend on behavioural observation alone, and has become one of the most useful tools in modern welfare assessment.

The basic protocol typically involves training the animal to associate two distinct cues with two different outcomes (one positive, one negative), then presenting ambiguous cues that fall between the two trained cues and measuring the animal’s response. Animals in positive affective states tend to approach the ambiguous cues as if they predict positive outcomes; animals in negative states tend to treat them as if they predict negative outcomes.

Cognitive bias testing has been applied across species from rats and pigs to dogs and horses to several wild species. The cross-species generalisability of the method is one of its strengths: the underlying mechanism (affective state influencing interpretation of ambiguous information) appears to be conserved across mammals and probably across birds, allowing similar protocols to be used across species with appropriate adaptation.

The method is one of the more rigorous tools available for assessing affective state in non-verbal animals, because it directly measures something about how the animal is processing information rather than relying on indirect behavioural or physiological correlates. The trade-off is that the protocols are typically time-consuming to set up and require careful training and validation for each species.

The understanding of cognitive bias and its relationship to welfare continues to develop, and the methods continue to be refined. Current research is exploring whether cognitive bias measures can be standardised, whether they can be used as routine welfare indicators in applied settings, and how they relate to other measures of affective state.