Once the fish had reached criterion, they were tested in a reversed maze so that the preferred spatial information fish used to navigate the maze could be determined. Fish could have either learned to use the plant landmarks positioned next to the open doors, or a sequence of turns such as “left door, then right door”.
For this reversed stage, doors that were previously deadends were
now opened and vice versa. Plants were moved such that they continued to be reliable cues indicating the location of the open door.
Fish using landmarks were expected to follow these and so swim
through the reversed maze relatively quickly making few errors.
However, fish using the sequence of turns for open and closed
doors would now presumably make many more errors as their
strategy would cause them to keep entering dead-ends.