Dear Fraser, 

You can check my paper out. We though used a linear maze with 11 possibilities for errors. If you test the same with Morris water maze, it should not make a difference or you can copy my maze and build it. 

Article Impact of age and situation-specific experience on maze lear...

Dear Fraser,

I'm not entirely sure which idea would assist you most as I'm not sure what your exact goal is, but here are a few ideas;

If you aren't already, I would recommend implementing the "Atlantis" trial (on-demand platforms) with your water maze design in addition to the previous suggestion. You can find the link attached. This should be beneficial for removing confounds and ensuring that you are reliably measuring the spatial memory of your subjects. By showing that the subjects are able to know the location of the platform even when it is not there it should assist in your goals.

Additionally, longer extinction periods are something I'd recommend experimenting with. Even as much as a week of no testing can be enough to ensure a decline in learning to the point that the layout of the maze will be forgotten. If you implement significantly longer extinction periods to allow for re-learning and choose to assess re-acquisition speed and properties you may be able to tell how learning is changing over the aging period.

Lastly, if you use a reversal design where you essentially place the platform on the opposite side of the maze to assess flexibility in learning and search patterns. This would disrupt the learning process of the maze layout, particularly if it is "reversed" more than once across the course of your experiment. 

I hope this helps!

http://www.nature.com/nprot/journal/v1/n2/abs/nprot.2006.116.html

I also suggest both the Atlantis rising platform and multiple platform locations.  I would use more than two locations, though, to reduce the risk of the mice picking up on the platform always being on the opposite side from last time.  Further, i would use different random sequences for each mouse, because if the mice have a tendency to search near the last platform location (which I would expect to happen initially), then they are likely to find a position 90 degrees to one side more quickly than 180 degrees away.  If each mouse has its own randomised sequence, that effect should average out..

Just to be clear, you would need multiple trials per day.  Across the first few days, you can expect the mice to learn some non-specific things, like not sticking to the walls, and how long they have to stay in one place to make the platform rise.  That will generate learning across days even with different platform locations, not only within days.  Once that has stabilised, you should be able to see effects of age in how much time is needed in the first trial of the day, and/or in how much they speed up from the first to the last trial of the day.

I don't know, though, how you would control for purely motoric decline.  Perhaps have a group for which you keep the platform in the same location throughout, so that you just measure swim speed?  Those mice are likely to spend less time in the water, though, so if initial speed is the same but they tire more easily, you wouldn't see that.  Unless you have two platforms, give the mice two seconds on a platform, then sink it and let them go for the other.  Then match for swimming time.  or something along those lines.  Then you have the difference that the swim speed test group gets a rest.  So alternatively, take more time to let the platform rise, perhaps yoked to a mouse in the memory test group.  That risks the mice just floating in place, though.

Disclaimer: I only ever trained rats in the water maze, and it was quite some time ago.