New Behavioural Reflection Test measures reasoning in moral and epistemic dilemmas

The Cognitive Reflection Test (CRT) has long been a standard for measuring how people override intuitive conclusions, but its effectiveness is eroded by widespread familiarity. Researchers from UNSW Sydney present the Behavioural Reflection Test (BRT) in a new arXiv paper. The BRT combines a low-exposure, four-item bespoke CRT (bCRT) with open-ended reasoning tasks set in morally and epistemically charged scenarios—e.g., whether to believe a controversial news source or how to balance ethical trade-offs. This design aims to capture reflective thinking in environments dense with both reliable and misleading information.

In a study of 473 online adults, higher bCRT scores strongly predicted evidence-sensitive, ethically driven decisions and a preference for high-quality sources. Participants scoring higher used language that was more emotionally engaged, risk-attentive, and economical—associations the classic CRT failed to recover. The BRT showed convergent validity and added item information beyond mean ability. With a median completion time of just 11.8 minutes, the open-ended BRT offers a practical, scalable assay for researchers studying human reasoning, misinformation resilience, and decision-making under ambiguity. For tech professionals, this tool could inform how AI systems model human judgment or how platforms detect cognitive biases in user behavior.