A dog’s personality and temperament is often a distinct factor in the creation of human-animal relationships. Favourable personality traits among dog breeds often differ depending on situation. For example, working dogs go through rigorously intense training regimes to determine their suitability for specific working roles; however in some cases, dogs nearing ‘graduation stage’ still display undesirable personality trait and cannot progress further in their role. Additionally, behavioural problems, including fear and anxiety, are one of the primary reasons for relinquishment of dogs at rescue centres. 

Therefore, Amirhosseini et al. (2024) have attempted to apply Artificial Intelligence (AI) and Machine Learning (ML) techniques to the classification of canine personalities using behavioural data. The behavioural data was derived from a University of Pennsylvania School of Veterinary Medicine database- the Canine Behavioural Assessment and Research Questionnaire (C-BARQ). 

Five personality types were identified in the paper: 

1. Excitable/ Hyperattached

2. Anxious/ Fearful 

3. Aloof/ Predatory

4. Reactive/ Assertive

5. Calm/ Agreeable

Four ML models were used to predict the personality traits based on the data, and provided predictions with an accuracy of 99%.

The availability of such a method to assess temperament in a dog at an early stage may slimline the selection process for working roles, match dogs to owners for optimum relationships, and provide catered approaches to behavioural problems with the animal’s underlying personality type in mind. Further research is needed to determine how these findings can be best applied in practice.