Floating: when stillness looks like forgetting

Opening

The stopwatch says forty seconds. The path plot says almost nothing. The rat is in the pool, yes, but it is barely swimming. It drifts. It pauses. It floats with its nose up and its paws still. Latency looks terrible. Memory may be fine. Or the animal may simply have given up. Until you measure floating, you cannot tell which story you are watching.

What floating is

Floating, or inactivity, is the awkward pause in the Morris water maze. The animal is in the water, but it is not searching. It may tread lightly. It may hang near the wall. It may stop altogether for stretches of the trial. Why does that matter? Because almost every classic score assumes the animal is trying. Latency, path length and even some probe measures go wrong when that assumption fails.

How inactivity is scored

Start with what you see. A trial begins when the animal is released facing the wall. In a normal run it swims, turns, and explores. In a floating run the track sits still for seconds at a time. Speed drops toward zero. Distance barely grows. The clock keeps running. If your only number is escape latency, a floater and a lost searcher can look the same. Both take a long time. Only one was looking.

Software that records time-stamped positions can mark inactive bouts. You set a speed threshold. Below it, the animal counts as inactive. Above it, as active. You can then report the total time spent floating, the number of bouts, and when they happened. Early floating after release can mean shock or confusion. Late floating after a long search can mean fatigue. Floating right after a failed visit to the old platform spot can mean the animal has concluded the goal is gone.

Picture a probe trial with no platform. One mouse spends twenty seconds hugging the wall, then twenty exploring, then twenty floating. Another mouse searches the target quadrant for fifty seconds and floats for ten. Same trial length. Very different stories. Without an inactivity measure, both can collapse into one bland latency or one coarse quadrant percentage.

Floating is not the same as thigmotaxis. Hugging the wall is a place preference. The animal is still moving. Floating is a pause in movement. An animal can float in the centre. It can float by the wall. It can alternate bursts of swimming with stillness. That is why a wall measure alone cannot stand in for inactivity. You need both.

Where the measure came from

The maze itself never required animals to keep swimming. Richard Morris's early methods papers already treated the pool as a place-learning task, not a fitness test. Escape from water is the motivation. But motivation varies. Some strains float more. Some ages float more. Cold water, deep water, and long trials push animals toward stillness. So do drugs that sap drive or motor vigour. So does prior stress.

Labs noticed the problem long before anyone named a formal floating score. A path plot with long flat stretches is hard to miss. What was missing was a shared definition. How slow is inactive? How long must a pause last before it counts? Do you exclude the first second after a turn? Different groups answered differently, and papers became hard to compare.

Recent technical notes have tried to pin the measure down. One open note from HVS Image Software Limited defines inactivity, or floating, for Morris water maze analysis: how to implement it, how to read it, where it fails, and when to use it. The point is blunt. Poor cognitive scores must not be confused with non-cognitive drag. Fatigue, disengagement and motor slowing can all inflate latency. Floating read beside complementary measures helps keep those stories apart. The same logic is meant to travel from rodent pools into virtual reality spatial tasks in humans, where a participant can also stop exploring without the software noticing unless inactivity is scored.

Protocol guides still remind labs to watch for non-spatial behaviours. Vorhees and Williams, in their widely used 2006 Nature Protocols paper, treat swimming ability, vision and motivation as checks that sit beside memory scores. Floating belongs in that same family of checks. It is not a fancy optional extra. It is how you stop a tired animal looking like a forgetful one.

What floating shows

What does a floating score show that latency hides? First, it protects the claim of a memory effect. Suppose a drug group takes twice as long to find the platform. If that group also floats three times as much, you do not yet know whether the drug hurt memory or drive. If floating is unchanged and proximity to the goal is worse, the memory story is stronger.

Second, it cleans probe interpretation. Extinction probes can teach animals that escape is impossible. Some then float. Others keep searching. Group differences in floating can masquerade as group differences in spatial preference if you only read whole-trial averages.

Third, it helps with strain and sex comparisons. Albino rats, older mice, and some genetically altered lines differ in how willingly they keep swimming. A paper that reports floating beside path length and proximity is harder to misread than one that reports latency alone.

Worked numbers help. Imagine two groups of ten mice. Group A mean latency is twenty-five seconds. Group B is forty. Without floating, B looks impaired. Now add floating time. Group A floats for two seconds on average. Group B floats for eighteen. Active search time is then similar. The deficit was stillness, not a missing map. Or flip it. Both groups float for three seconds, but B's path stays far from the goal. Now the memory claim survives.

Floating also pairs with strategy classification. Direct paths, scanning, chaining and thigmotaxis describe how an animal searches when it is moving. Inactivity describes when it is not. A full behavioural picture needs both the shape of the path and the gaps in it.

Where it can mislead

Floating measures have limits. The speed threshold is a judgement. Set it too high and ordinary slow turns count as floating. Set it too low and genuine pauses vanish. Camera noise can create false micro-movements that break a real float into fake activity. Poor tracking of a dark animal on a dark pool can invent stillness that never happened. As with Atlantis platforms, the camera and the threshold are part of the experiment.

Floating also does not explain itself. An animal may float because it is cold, tired, nauseous, unmotivated, or strategically waiting. The score flags a problem. It does not name the cause. You still need temperature logs, visible-platform controls, open-field activity, and common sense about trial length.

And floating is not an excuse to ignore every long latency. Some animals truly do not know where to go. They swim the whole time and still fail. An inactivity measure that is near zero with a poor proximity score is exactly that case. The honest report shows both.

There is a last temptation. Once floating is easy to export, some papers may treat it as the star result. It shouldn't be. It is a safeguard and a companion. Latency, path length, proximity, heading and strategy still carry the spatial story. Floating tells you whether those scores were earned while the animal was actually trying.

Back to the quiet animal

Back to that quiet animal in the milky pool, paws still, clock running. The beauty of measuring floating is almost dull. No rising platform. No clever gadget. Just an admission that stillness is data. Forty years into the water maze, that admission still saves careers from overclaiming. A slow trial is not always a lost memory. Sometimes it is a pause. You only know if you count it.

You'll find the papers we mentioned, and a transcript, at water-maze.com. hvsimage.com has a note on floating, or inactivity, in Morris water maze analysis if you want the formal definition beside the path plots.