One of the most frequently used spatial learning task in rodents is the Morris water maze. Image analysis systems have made it more convenient to quantify the performance of animals. However, to define the search / swim patterns of free swim trials in this task as persistent or random, a large set of parameters has to be chosen: like total distance swum, distance and time in quadrants, crossings of and latencies to former and "possible" platform locations in other quadrants, subdivisions of the maze into inner circle and rim along the walls, velocity, etc. These data sets have to be compared to finally describe a swim pattern as persistent, i.e. focused to the former platform location, or random.

During the past years we have used the water maze mainly to characterize the role of corticosteroid hormones in learning and memory. Specifically free swim trials proved to be an excellent tool to differentiate the involvement of hippocampal mineralocorticoid and glucocorticoid receptors, MR and GR, respectively [1]. A balanced activation of both receptor types is reflected by short latencies to platform (memory) and a persistent search pattern (strategy). In contrast, animals with suppressed MR functions showed a distorted and random search pattern, but still short latencies. This dissociation between memory and strategy and the required extended description of free swim trials listed above prompted us to search for more economic ways. Furthermore, we were interested if behavior during free swim trials before spatial training might be of predictive value for memory and strategy after spatial training.

To answer these questions, we used swim patterns of mice of several studies. Estimating the distance between the former platform location and each sample of the animal’s position (EthoVision 1.70, Noldus Information Technology), distance to zone platform and possible platform positions proved to be the parameter to differentiate persistent from random search patterns. To allow comparison between experiments with e.g., different sampling or maze sizes, we established and validated that a percentage of distance to platform < 16 reflects persistent search. Random search could be further subdivided into a circling pattern close to and at a certain distance from the walls. Thus, the two parameters latency to zone and distance to zone are sufficient to describe memory and strategy used in free swim trials. Furthermore, we did not find any relationship between behavior in free swim trials before and after spatial training. Our next goal is to quantify the "small loop" swim pattern often seen in mice. In terms of latency and distance measures, this ‘looping’ can be as efficient as the "spatial" solution during training trials. Taken together, behavioral image analysis systems do not only increase the amount of data output, but their flexibility also allows to extract the essentials in an efficient way.

Supported by ISAO 679756-0270 and EC BiotecPL960179.

References

1. Oitzl, M.S.; de Kloet, E.R.; Joels, M.; Schmid, W.; Cole, T.J. (1997). Spatial learning deficits in mice with a targeted glucocorticoid receptor gene disruption. European Journal of Neuroscience, 9, 2284-2296.