Abstract
Animals are capable of representing different scale spaces from smaller to larger ones. However, most laboratory animals live their life in a narrow range of scale spaces like homecages and experimental setups, making it hard to extrapolate the spatial representation and learning process in large scale spaces from those in conventional scale spaces. Here, we developed a 3-m diameter Barnes maze (BM3), then explored whether spatial learning in the Barnes maze (BM) is calibrated by scale spaces. Spatial learning in the BM3 was successfully established with a lower learning rate than that in a conventional 1-m diameter Barnes maze (BM1). Specifically, analysis of exploration strategies revealed that the mice in the BM3 persistently searched certain places throughout the learning, while such places were rapidly decreased in the BM1. These results suggest dedicated exploration strategies requiring more trial-and-errors and computational resources in the BM3 than in the BM1, leading to a divergence of spatial learning between the BM1 and the BM3. We then explored whether prior learning in one BM scale calibrates subsequent spatial learning in another BM scale, and found asymmetric facilitation such that the prior learning in the BM3 facilitated the subsequent BM1 learning, but not vice versa. Thus, scale space calibrates both the present and subsequent BM learning. This is the first study to demonstrate scale-dependent spatial learning in BM in mice. The couple of the BM1 and the BM3 would be a suitable system to seek how animals represent different scale spaces with underlying neural implementation.