Using desktop, computer-simulated virtual environments (VEs), the authors conducted 5 experiments to investigate blocking of learning about a goal location based on Shape B as a consequence of preliminary training to locate that goal using Shape A. The shapes were large 2-dimensional horizontal figures on the ground. Blocking of spatial learning was found when the initially trained Shape A was presented in the context of auxiliary shapes that were anticipated to be irrelevant to goal localization. When Shape A was initially presented in the absence of these auxiliary shapes, no evidence of blocking between shapes was apparent. The results are discussed in terms of the similarity between spatial and other forms of contingency learning, the operation of a specialized geometric module, and changes in attention as a consequence of discrimination learning. (Contains 7 figures, 1 table and 1 footnote.)