Two experiments were conducted to investigate kindergarteners ability to recognize common proportions across different instantiations. Both experiments varied between subjects the degree of concreteness of the instantiations used during training. In Experiment 1, when explicit training was given, participants who learned with either the concrete or generic material successfully transferred their knowledge to match common proportions involving novel objects. However, in Experiment 2 when no explicit instruction on proportion was given (participants were only shown two examples), only participants who were shown the generic examples successfully matched proportions with novel object. Participants who were shown concrete examples were unable to do so. These findings suggest that simple relations such as the concept of proportion can be picked up spontaneously from generic instantiations, while concrete instantiations do not promote this spontaneous structure learning.