This method paper details a protocol to test spatial working memory in mice using a semi-automated radial 8-arm maze (RAM). The RAM is a partially enclosed apparatus with 8 horizontal, equally spaced arms radiating from a central hub, from which access to each arm can be controlled individually by servo-controlled motorized doors. Animals start in the central hub and are allowed to explore the maze for a food reward at the end of each arm or selected arm. The RAM task was originally designed for rats, but we have adapted the protocol for mice, for example, by including more habituation steps. In our protocol, all arms are initially baited with sweetened condensed milk, and mice are admitted sequentially into four pseudo-randomly selected arms to collect the rewards ("forced run") before opening all doors together to allow the mice to run freely and find the remaining four rewards ("free run"). A 5 s delay is introduced between the forced and free runs to challenge working memory; an error is recorded if the mouse enters any previously visited arm during the free run. The task is complete when all rewards are recovered. After 6 days of habituation and 9 days of maze training, male C57BL/6 mice regularly achieve ≥ 80% daily success rate score, defined as 4/(4+E), where E is the number of errors. This semi-automated task could, in principle, be combined with in vivo monitoring methods such as electrophysiology, multiple-photon microscopy, or calcium imaging.
Shimizu et al. (Tue,) studied this question.