Developing and using ontologies in behavioural science: addressing issues raised [version 2; peer review: 2 approved, 1 approved with reservations]

Ontologies are ways of representing aspects of the world in terms of uniquely defined classes of ‘entities’ and relationships between them. They are widely used in biological science, data science and commerce because they provide clarity, consistency, and the ability to link information and data from different sources. Ontologies offer great promise as representational systems in behavioural science and could revolutionise descriptions of studies and findings, and the expression of models and theories. This paper discusses issues that have been raised about using ontologies in behavioural science and how these can be addressed. The issues arise partly from the way that ontologies represent information, which can be perceived as reductionist or simplistic, and partly from issues to do with their implementation. However, despite the simplicity of their structure, ontologies can represent complex entities that change over time, as well as their inter-relationships and highly nuanced information about them. Nevertheless, ontologies are only one of many ways of representing information and it is important to recognise when other forms are more efficient. With regard to implementation, it is important to build ontologies with involvement from the communities who will be using them. Far from constraining intellectual creativity, ontologies that are broadly-based can facilitate expression of nuance, comparison of findings and integration of different approaches and theories. Maintaining and updating ontologies remain significant challenges but can be achieved through establishing and coordinating communities of practice.