The stimulus–response learning model and the constructivist learning model are experimentally compared based upon probabilistic psychometric measurements of the judged relative priorities of 12 global environmental issues. Respondents are divided into an experimental group and a control group. The experimental group is provided with an incisive description of the issues. Both groups provide pairwise comparison judgments of the importance of the issues. The judgments are analysed using the probabilistic geometric modelling algorithms known as PROSCAL. The results confirm the constructive learning model. Some implications for environmental education, policy and management are suggested.