This study compared learning performance during and following AR and non-AR topographic map instruction and practice Two-way ANOVA testing indicated no significant differences on a posttest assessment between map type and spatial ability. Prior learning activity results revealed a significant performance difference between AR and non-AR treatment conditions. The potential intervening variable of spatial ability was ruled-out as a rival explanation for the score differences between the two treatment conditions, although the novelty of the Augmented Reality Sandbox may have contributed somewhat to higher scores for the AR treatment group. We propose that the Augmented Reality Sandbox's AR affordances offer new directions for instructional design of interfaces that promote learning through embodied interaction and multisensory feedback.