Exploring the limits of spatial memory in rats, using very large mazes.

In Experiment 1, rats foraged for food in six successive phases with 8, 16, 24, 32, 40, and 48 arms attached in random locations to a large radial maze. The percentage of novel choices appeared to be determined more by spatial proximity than by number of arms. In Experiment 2, rats foraged for food in four successive phases with 8, 16, 24, and 48 arms attached to the maze in spread-out or tight configurations. Performance was poor in the tight configurations regardless of the number of arms. Performance was excellent in the 8-arm spread-out condition but declined as 16 and, then again, 24 arms were added. Thus, spatial separation, not number of locations, was the chief determinant of performance in the first two experiments. In Experiment 3, in successive phases, 8, 16, 24, 32, 40, 48, 16, and 8 food towers were set in a circle on the floor, with the spatial separation between adjacent towers held constant at 33 cm. The percentage of novel choices declined as 8 towers became 16 and did not change again with 24, 32, 40, or 48 towers in place but then increased again as 16 towers became 8. In Experiment 4, in successive phases, 8, 16, 24, and 32 food towers were set in a circle, with the spatial separation between adjacent towers held constant at 66 cm. The percentage of novel choices declined as 8 towers became 16 and again as 16 towers became 24 but did not decline further. These data were discussed in terms of the fundamental problems posed by variations in the number of food locations in the pursuit of the limit of spatial memory in rats.