Neuroscience 2009 location:Chicago, USA date:17 - 21 October 2009
The Morris water maze is used extensively as a test of visuo-spatial learning and memory in rodents. The performance of humans in analogous virtual paradigms is of growing interest. Despite its popularity, however, the behavioural processes contributing to overall task performance in both species are still not clearly understood. We have developed a virtual Morris water maze for humans which attempts to ensure that specific learning processes observed in rodents are closely matched. Here, we compare the performance of humans and mice on a standard version of the task. In addition to commonly used measures that characterise general task performance such as trial time and path length, we quantified the performance of specific behavioural processes. Based on recent evidence, it appears that the direction of the initial movement trajectory is the outcome of a critical decision making process at the beginning of each trial. In both humans and mice we observed a significant reduction in the angular error of the initial movement trajectory (with respect to the true direction of the platform) over the course of training. On trials in which the platform was not successfully located during the initial movement trajectory, a search phase ensued. The average distance to the platform during the search phase was also significantly reduced over the course of training in both species, indicating an increase in spatial certainty of the platform location. The clear learning effects in both species revealed by the 'initial direction error' and 'average search distance to platform' suggest that the specific behavioural processes which underlie these variables are critical to overall task performance. An additional measure of spatial learning was obtained from human subjects, in
which they were required to indicate the location of the platform without accuracy feedback at various stages during training. This measure excludes task specific strategies which influence performance on regular search trials. We found that spatial accuracy on these trials also improved with training. Subsequent fMRI experiments in humans and lesion experiments in mice are planned to delineate the neural networks responsible for visuo-spatial learning.