What information is used by rats to update choices in the radial-arm maze?

This study examined the information processed by rats in the radial maze. In Experiment 1, performance in the standard straight-arm radial maze was compared with performance in a angle-arm maze which required a right turn at the end of each radial arm. The results showed that the performance of rats initially trained in the angle-arm maze was much more affected when later tested on the straight-arm maze than was the performance of rats initially trained on the straight-arm maze and then moved to the angle-arm maze. In Experiment 2, two groups of rats were trained in the angle-arm maze and then subjected to two transfer tests during which the location and/or direction of the correct goal arms was manipulated. Both transfer tests produced an increase in errors, but the increase was much more pronounced in rats required to learn a new set of baited locations than in rats tested on the same set of baited locations and required to use new initial travel directions. Together, these experiments demonstrate that rats make complementary use of information about both initial travel direction and goal location.