The RWM is a swimming version of the radial-arm maze (RAM) that uses food reward. The Core has an 8-arm RWM. The RWM, like the RAM, tests either spatial trial-dependent, working memory or a combination of spatial working and trial-independent reference memory. In the classic RAM, rats must be food restricted and trained to eat some preferred food reward, then trained to find these rewards or baits within the maze until they learn to search all the arms and go to the very end of an arm to find the bait. After training, testing begins. For working memory, all arms are baited for each new trial. The animal must find all 8 rewards without repeating an arm visit from which it already took one of the rewards. The trial continues until all 8 rewards are retrieved or up to a maximum of 10 min. In the mixed working/reference memory version some arms are baited anew each day and some are never baited. In this version the animal’s task it to find the new rewards each day without repeating an arm entry and never enter the unbaited arms. Our version is a modified version of this that avoids food deprivation and training. One arm in the RWM is always the start and at the end of the other 7 arms is a hidden, submerged platform. On the first trial the animal can enter any arm and reach a platform upon which it is removed and that platform is removed so that on the next trial there are only 6 remaining platforms. The animal’s task is to find any other platform without reentering the one it found first. This continues for 7 trials until each one is found and each reentry to an arm without a platform is counted as an error. The task can be modified to be a mixed working/reference spatial memory task just as the RAM is. The working memory version typically requires 10-14 days for animals to become proficient with the task.