The pool: how the Morris water maze began

Opening

Picture a round pool, about as wide as a small kitchen, filled with water that's been turned milky white. Somewhere under that cloudy surface, a centimetre or so down, sits a little platform. You can't see it. Neither can the rat.

Pop the rat in at the edge. It paddles, it bumps along the wall, it tries the middle. Then, more or less by accident, its feet touch something solid. It climbs up, shakes off, and has a good look round the room. A poster on one wall. A lamp in the corner. The door.

Next day, same pool, different starting point. This time it swims almost straight there. And that's the whole trick. The platform is invisible, so the rat has to remember where it is, using the room around it.

What the water maze is

This is the Morris water maze, and labs have been running it for more than forty years. How did a pool of cloudy water become one of the most trusted tests of memory in neuroscience? And what can it really tell us?

The parts of the pool

Start with the parts. There's a circular tank, usually made of plastic or steel, and filled about halfway with water. The water is made opaque, so nothing below the surface shows. Inside sits an escape platform, small enough to miss, and sunk just below the waterline. For rats, a modern protocol might use a tank around two metres across, with a platform about ten centimetres wide.

That ratio matters more than you might think. A big pool and a small platform make a search area hundreds of times larger than the target. A lucky swim can't solve that. The animal needs a plan.

One widely used protocol gives a worked example. A tank about two hundred and ten centimetres across, with a square platform ten centimetres on a side, gives a search area roughly three hundred and fifty times the size of the target. Halve the platform's width, and the ratio climbs to nearly fourteen hundred.

Around the pool, on the walls of the room, there are cues. They might be bold shapes, a shelf, a cupboard, even the experimenter's chair. These are called distal cues, because they're far away, outside the pool itself. The platform stays in one place, so its position can be worked out from those landmarks.

A single trial is simple. The animal is lowered into the water facing the wall, at one of several start points. The start point changes from trial to trial, so the animal can't just learn a fixed set of turns. It swims until it finds the platform, or until a time limit runs out, at which point it's gently guided there. It gets a short rest on the platform, which gives it time to take in the view. Then it's dried off and the next trial begins.

Repeat that four times a day, for several days, and you have the classic learning curve. Early on, an animal might take a minute to find the platform. By the end of training, a good swimmer might get there in a few seconds.

Where it came from

The idea came from Richard Morris and colleagues, in the early nineteen eighties. Morris was then at the University of St Andrews in Scotland. At the time, the go-to spatial tests were dry mazes, often with food as the reward. Those tests worked, but they came with baggage. Hungry animals had to be kept on restricted diets. And the animals could, in principle, follow smell trails or other local clues rather than truly remembering a place.

Morris wanted something cleaner. Water was the answer. Rats are capable swimmers, but they'd rather be out of the water than in it, so escape is its own reward. No food deprivation needed. And in a pool, there's no floor to carry a scent trail.

His first paper on the task appeared in 1981, in the journal Learning and Motivation. Its title says it all: spatial localization does not require the presence of local cues. In other words, a rat could find a spot it couldn't see, hear or smell, just by knowing where it was relative to the room. That first pool was about one point three metres across.

The following year came the result that made the task famous. Writing in Nature in 1982, Morris worked with Garrud, Rawlins and O'Keefe, and they tested rats with damage to the hippocampus. That's a curved structure deep in the brain, already suspected of building maps of space. The rats with hippocampal lesions were impaired at finding the hidden platform. Place navigation was impaired, to borrow the paper's own title. Yet the same rats could still learn to swim to a platform they could see. Their swimming, eyesight and motivation were fine. What they'd lost was the map.

Two years after that, a methods paper set out ways of developing the procedure. It described tracking a hooded rat's position automatically, using video, without fixing lights to the animal. It also showed how the same pool could test working memory and simple non-spatial discrimination. That's how a home-made pool turned into a standard tool.

What the task shows

What does the task actually show? At its heart, it's a test of place learning. The animal learns where something is, rather than which turn to take or which smell to follow. And that kind of learning leans heavily on the hippocampus.

The first measure researchers used was escape latency. That's simply the time from entering the water to reaching the platform. Latency falls over days of training, and a slower fall suggests weaker learning.

But latency has a weakness. A fast swimmer can stumble on the platform sooner than a slow swimmer who knows exactly where it is. Path length fixes some of that. It's the distance swum before reaching the goal, and it doesn't depend on speed.

Picture two rats on day four. One takes twenty seconds and swims eight metres. The other also takes twenty seconds, but swims only four metres, slowly and deliberately. Same latency, very different stories. Video tracking is what makes that difference visible.

Then there's the probe trial, usually given a day after training ends. The platform is removed, and the animal gets a fixed time to search, often thirty or sixty seconds. If it remembers the spot, it spends far more time in that part of the pool than chance would predict. Since the pool is often divided into four quarters, chance is about twenty-five percent of the time in the right one.

Reviewers have also pointed out a practical strength. Escape from water is relatively unaffected by differences in activity levels or body weight. That makes it useful for comparing animals that differ in ways a dry maze might struggle with. A standard protocol, written up in 2006 by Charles Vorhees and Michael Williams, notes that useful results can come from an individual animal in as few as six days. They also describe the task as closely linked with hippocampal synaptic plasticity and with N M D A receptor function, two pillars of memory research.

There are clever variations too. Move the platform to the opposite side, and you test reversal learning, which asks whether the animal can let go of an old answer. Make the platform smaller, and you test how precise the memory is.

Where the pool can mislead

For all that, the pool can mislead. Swimming is a skill as well as a motivation. An animal that's unwell, tired, or can't see properly will look forgetful when it isn't. That's why a visible platform test matters. A flag or a marker shows where the platform is, and the curtains can be drawn to hide the room cues. If an animal can't manage that version, its trouble isn't spatial memory. The protocol authors note that this control is all too often left out.

Stress is another factor. Water is a strong motivator, which is part of the appeal. But it can also make some animals panic, float, or cling to the wall. Mice, in particular, aren't natural swimmers the way rats are, so their behaviour in the pool can reflect stress as much as memory.

Timing matters as well. If the probe trial comes straight after training, it may only capture recent memory. Probe trials are also a little like extinction trials. Give too many, and the animal may learn that the platform has gone.

And latency alone, as we heard, can hide what the animal is really doing. Some animals learn the platform's distance from the wall and simply circle at that radius. They get faster, but not because they've learned a place. Spotting that takes path analysis, not a stopwatch.

Back to the pool

Back to that milky pool, and the rat that swims almost straight to a platform it can't see. The beauty of the task is that the answer lies outside the water, in the posters and lamps and doors around it. The animal has to build a picture of the room and place itself in it. Forty years on, that simple demand still makes the pool one of the clearest windows we have on how brains map the world.

The video tracking that turned those swims into numbers has a long history of its own. hvsimage.com has more on how the analysis works. You'll find the papers we mentioned, and a transcript, at water-maze.com.