ABSTRACT
The submerged platform or Morris water escape task is widely used to study genetic variation in spatial learning and memory, but interpretation is sometimes difficult because of wall hugging, jumping off the platform, floating or non-spatial swim strategies. We modified the task by introducing four wide arms into the circular tank and adding features that reduced, eliminated, or compensated for several competing behaviors. Three versions of the 4-arm task were evaluated in detail, and the third version yielded good results for six of eight inbred strains. Furthermore, the 4-arm task could be scored adequately without computerized video tracking. Although performance on the 4-arm task was generally superior to the Morris maze, the extent of the improvement was strain dependent. Two strains with retinal degeneration (C3H/HeJ, FVB/NJ) performed poorly on both the Morris and 4-arm mazes, whereas C57BL/6J and DBA/2J did well on both mazes. A/J performed poorly on the Morris task but became very proficient on the 4-arm maze, despite its strong tendency to hug the walls of the tank. The BALB/cByJ strain, on the other hand, exhibited the best probe trial performance on the Morris maze but was very slow in acquiring the 4-arm task. We conclude that no single task can reveal the full richness of spatially guided behavior in a wide range of mouse genotypes.
