Abstract Two experiments with human adults investigated the extent to which the transfer of function in accordance with nonarbitrary versus arbitrary stimulus relations may be brought under contextual control. Experiment 1 comprised four phases. Phase 1 consisted of multiple‐exemplar training to establish discriminative functions for solid, dashed, or dotted lines. Phase 2 trained and tested two equivalence classes, each containing a 3D picture, a solid, a dashed, and a dotted form. During Phase 3, a discriminativ…