Water maze glossary

Every field has its jargon, and the water maze has plenty. Here it is in plain words. If a term you need is missing, tell us.

A B C D E F G H I L M O P Q R S T V W

A

Acquisition: Acquisition is the training phase in which subjects learn to find the platform. Escape latency and path length usually fall across days; strategy measures show whether that fall reflects a map or a procedural shortcut. See also probe trial and reversal learning.
Active search: Active search is purposeful swimming that covers the pool rather than floating, wall-hugging or looping. It is the counterpart of inactivity (floating) and helps separate motivation from memory.
Allocentric and egocentric navigation: Allocentric navigation uses distal cues, such as large cues on the room walls, to find a place.
Allocentric strategy: An allocentric strategy uses the layout of distal room cues to locate a place in the pool. It is the map-like solution the water maze was designed to test. Contrast with an egocentric strategy. See also distal cues.
Analysis grid (open field zones): The analysis grid divides the arena into cells so you get time, path, entries and latency for each cell. Pick a size that puts each object in its own cell.
Angular span: Angular span is how many degrees of the pool circumference a subject covers while in a zone such as the thigmotaxis band or the chaining annulus. Continuous span criteria help distinguish genuine wall-hugging or chaining from a path that only briefly crosses the zone.
Annulus crossings: Annulus crossings count how often the subject enters a ring around the learned platform position in a probe. They are a little more forgiving than exact platform crossings when the animal searches nearby but not exactly on the spot.
Annulus (platform annulus): In many papers, "annulus" means a ring around the old platform position used in a probe. That is different from the chaining annulus, which is a ring at the platform's distance from the wall around the whole pool. See platform crossings and zones and the annulus.
Arm entries (multi-arm mazes): Arm entries count how often the subject entered each arm, using the Detection Area you choose. Arms are listed from the left arm clockwise.
Atlantis (rising) platform: The Atlantis platform sits below the surface and rises automatically when the subject reaches the target. Set the run-screen button to "USB on End" and choose an Auto-end option.
Auto-end: ‘Auto-end on platform’ decides whether a trial ends when the subject reaches the platform: ‘Yes’ ends on arrival, ‘Delay in Secs’ after a set stay, and ‘No’ means you end it with ‘END’.
Average and active speed: Average speed is path / trial duration and includes stationary time. Active speed leaves out slow floating periods.
Average proximity: Average proximity is the mean distance from the subject to the goal across a trial (also called Gallagher global). Lower means a tighter search. It is one of the Gallagher proximity measures and is especially useful in probes. See Gallagher proximity.

B

Barnes box: The Barnes box shows which maze the screen is running. You don't need to tick it yourself: it is ticked when you open the Barnes maze and unticked when you open the water maze.
Barnes errors and primary path: In the Barnes maze, errors are pokes into wrong holes. Primary path length is the distance to the first approach of the escape hole.
Barnes maze: The Barnes maze is a dry circular platform with holes around the edge, one of which leads to an escape box. It is often used as a less stressful relative of the water maze. Strategies include direct, serial and random search of the holes. See also Barnes errors and primary path.
Behaviour identification: Behaviour identification labels stretches of a trial that meet objective criteria for strategies such as direct path, chaining, thigmotaxis or target scanning. More than one label can apply in the same trial when the subject switches mid-swim. See search strategies.
.BOND file: Every VR trial is saved automatically as a JSON .BOND file plus an .hvs file. The .BOND file holds the time-stamped positions and all settings.
Broad spatial coverage: Broad spatial coverage describes a search that spreads across much of the pool rather than staying near the wall or looping at one radius. It can look busy on a path plot without being aimed at the goal. Related to scanning and random search.

C

Calibrated positions: Calibrated positions are places in the pool whose coordinates are stored at set-up (the platform and any other spots you mark). Close-encounter zones, Gallagher measures and exact crossings can then be computed for each one, including old and new positions in reversal learning. See zones around the platform.
Calibration: Calibration tells the software the size of your pool or arena and where each target is. All distances use the calibrated pool diameter.
Chaining: Chaining is circling at a fixed distance from the wall, roughly the platform's distance. It is a non-spatial, procedural strategy that can coexist with good latencies.
Chaining annulus: The chaining annulus is a ring-shaped zone at the platform's distance from the wall, used to spot chaining. Animals that circle inside that ring can cut latency without knowing where the platform is. See the chaining annulus.
Circular zones: Circular zones split the pool into rings. Click the explanatory diagram button on the analysis screen to see which letter is A, B or C in your version.
Close encounter (counter): A close encounter is an entry into a circle (the counter) around a target position, which is less strict than an exact crossing.
Cognitive flexibility: Cognitive flexibility in the water maze is the ability to abandon a previously learned place or cue when the rule changes, as in reversal learning or a dual-strategy trial. Perseveration on the old solution is its opposite.
Cone test: The cone test asks whether the subject's path stayed inside a cone aimed from the start point toward the goal. It is a directional accuracy check that, like the corridor test, works for any start-to-goal pairing. Related to heading angle and path efficiency ratio.
Corrected integrated path length (CIPL): Corrected integrated path length adjusts path length for how far the start was from the goal, so trials from near and far starts can be compared fairly. It is often used alongside raw path length in ageing and strategy studies.
Corridor test: The corridor test asks whether the subject swam inside a corridor of set width along the ideal line from start to goal. It normalises for different start distances and is a clean readout of directional accuracy. See also heading error and cone test.
CST (custom results) file: A CST file is your chosen set of results, one line per trial and one column per variable, ready to open in Excel.
Cued platform: A cued (visible) platform trial marks the platform with a local cue so the animal can navigate by sight rather than by a room map. It checks vision, swimming and motivation before or after hidden-platform training. Contrast with a hidden platform.
Cumulative search error: Cumulative search error (Gallagher cumulative) sums the average distance from the goal in successive one-second bins. Higher means more search error; it is often used across training. It is the training counterpart of average proximity. See Gallagher proximity.

D

Detection Area: The Detection Area is the % of each arm that counts as an entry: at 100% any entry counts; at 25% only entries to the far quarter.
Direct Path Measures: Direct Path Measures classify whether a trial began with an accurate approach. There are six classes in two groups of three: to Target and to Close Encounter.
Direct path to close encounter: Direct path to close encounter is a Direct Path class in which the first accurate approach reaches the close-encounter zone around the target rather than the exact spot. See Direct Path Measures and close encounter (counter).
Direct path to target: Direct path to target is a Direct Path class in which the subject sets off on an accurate approach that reaches the platform itself. It is strong evidence of knowing where to go from the start. See Direct Path Measures and direct swim.
Direct swim: A direct swim is a nearly straight path from the start point to the platform (or learned place). It is the classic spatial solution and the opposite of thigmotaxis, chaining or random search. See search strategies.
Directed search: Directed search is swimming that heads toward the goal area but is not yet a perfect direct swim. The path may curve or overshoot, yet the bias toward the learned place is clear. Related to focal search and initial spatial intent.
Distal cues: Distal cues are large, three-dimensional cues placed far from the pool on room walls or fixed structures, visible to the animals.
Dual-strategy trial: A dual-strategy trial pits a proximal cue against the remembered place (for example by moving a local marker while the platform stays put, or the reverse). Close-encounter measures at each calibrated position show whether the subject followed the cue or the place. See calibrated positions.

E

Egocentric strategy: An egocentric strategy uses body-centred rules such as "swim left" or "keep a fixed distance from the wall", rather than a map of the room. Chaining and some forms of thigmotaxis are egocentric. Contrast with an allocentric strategy.
Elevated plus maze: The elevated plus maze is a widely used test of anxiety-related behaviour: subjects are placed at the junction of the four arms, facing an open arm.
Entropy (strategy entropy): Strategy entropy summarises how mixed or unpredictable a subject's strategy labels are across trials. High entropy means the subject keeps switching approaches; low entropy means one strategy dominates. It is useful in ageing studies that compare flexible and stereotyped search.
Equal-area circular zones: Equal-area circular zones divide the pool into rings of equal area rather than equal width. Outer rings are thinner; inner rings are thicker. They are an alternative to equally spaced rings when you want each zone to have the same chance occupancy. See circular zones.
Escape latency: Latency is the time to reach the target. It doesn't show strategy and is affected by speed and start distance, so read it with other measures.

F

First bearing: First bearing is the direction of the subject's earliest movement relative to the ideal line to the goal. It is closely related to heading angle and is often taken in the first second or two of the trial.
Focal search: Focal search (also called target-focused search) is searching tightly around the learned platform location. In a probe it is a strong spatial-memory signature and can succeed as a measure even when exact platform crossings are few. See search strategies.
Forced swim measures: Forced swim results give % time climbing, diving, swimming and floating, plus total path, duration, average speed and % time moving.
Forced swim tracking band: For forced swim, the tracking area is a band around the water line, sized so the animal's highest point sits in the middle third when swimming or floating.
Freezing: Freezing is staying still for a set time. You decide how many seconds the subject must stay still, and within what distance, for it to count.
'From Poolside' (Direct Path class): "From Poolside" describes where the direct approach began (near the wall), not continuous wall-following. Sustained wall-following is reported separately as Thigmotaxis.

G

Gallagher by segment: Gallagher by segment applies Gallagher proximity measures to successive portions of a trial (for example successive time slices). It shows whether search tightened or wandered as the trial went on. See time slice and Gallagher proximity.
Gallagher cumulative: Gallagher cumulative is another name for cumulative search error: the sum of one-second average distances to the goal. Higher values mean more error and are typically used across training days.
Gallagher global: Gallagher global is another name for average proximity: the mean distance to the goal over the trial. Lower is better and is especially informative in probe trials.
Gallagher proximity: Gallagher proximity measures how close the subject stayed to the target. Global is the average distance (lower = closer; used for probes); Cumulative sums one-second average distances (higher = more search error; used for training).

H

Heading angle: Heading angle shows how directly the subject set off towards the target. Angles near zero mean it headed straight for it.
Heading error: Heading error is the angular difference between the subject's initial direction and the ideal line to the goal. Angles near zero mean it set off the right way. It is the same family of measure as heading angle. See heading angle.
Hidden platform: A hidden platform sits just under opaque water so it cannot be seen. Finding it depends on distal cues and place learning. Contrast with a cued platform.
HVS 6D VR: HVS 6D VR runs virtual experiments such as the VR water maze in a Quest 3 headset; participants can walk on a treadmill or navigate in other ways, such as with the arrow keys.
.hvs file: The .hvs file is the raw data file saved for each trial. Keep every one: your SUM and CST results can be recreated from them.

I

Inactivity (floating): Inactivity is the percentage of trial time spent below a speed threshold, such as floating in rodents or standing still in VR.
Initial spatial intent: Initial spatial intent labels an early portion of the trial that is already biased toward the learned place, even if the rest of the swim is messy. It captures knowing where to go at the outset. Related to direct path to target and heading angle.
Inter-trial interval: The inter-trial interval is the wait between successive trials for the same subject. It affects fatigue, stress and how much is remembered from the previous swim. Protocol pages discuss typical choices for rats and mice.

L

Latency to close encounter: Latency to close encounter is how quickly the subject first enters the close-encounter zone around a calibrated position. It credits a near miss that exact platform crossings would ignore. See close encounter (counter).

M

Morris water maze: In the Morris water maze, subjects learn to escape from opaque water onto a hidden platform, using cues around the room.

O

Omnidrive treadmill and treadplate: The Omnidrive treadmill or treadplate lets participants walk in VR. Assemble it first, in its final position, then set up the headset around it.
Opaque water: Opaque water (made cloudy with a non-toxic agent) hides the platform so subjects must use room cues rather than sight of the goal. Clear water is used for cued platform checks. Depth and temperature matter as much as opacity.
Open field: In the open field the subject explores an arena while the tracker records where it goes. HVS Image gives % time, % path, entries and latency for grid cells, plus speed and freezing.

P

Passes close to the platform: Passes close to the platform count entries into a small zone around the goal (or another calibrated position). They are less strict than exact crossings and pair well with percentage of time close and Gallagher proximity.
Path efficiency: Path efficiency is another name for the path efficiency ratio: actual path divided by the shortest direct path. A value of 1 is perfect; higher means a longer detour. (Some authors invert the ratio; report which form you used.)
Path efficiency ratio: The path ratio is actual path / shortest direct path. A ratio of 1 is perfectly direct and lower is better.
Path length: Path length is the distance travelled, in metres. It depends on start distance.
Percentage of time close: Percentage of time close is the share of the trial spent inside a close-encounter zone around the goal. It rewards lingering near the right place and is sensitive in probes when crossings are sparse.
Perseveration: Perseveration is repeating an old solution after the rule has changed, such as searching the former platform quadrant after a reversal. It is a useful index of cognitive inflexibility and shows up in quadrant time, close encounters at the old place, and strategy labels.
Place learning: Place learning is learning where something is in the environment, using a map of distal cues. The classic hidden-platform water maze is a place-learning task. Contrast with cue learning on a cued platform.
Platform crossing: A crossing is a pass over the exact platform location. Whole-trial crossings favour subjects that keep returning over flexible ones that check and move on.
Platform Number (Hole Number): The ‘Platform Number’ tells each trial which calibrated position is the target. On the Barnes screen it is the ‘Hole Number’.
Pool circling: Pool circling counts complete loops enclosing the pool centre; doubling back does not affect the count.
Probe trial: A probe trial has no platform to find; it shows whether the subject remembers where the platform was. Enter the learned position, tick Probe and turn Auto-end off.
Procedural strategy: A procedural strategy is a habit-like solution such as chaining or persistent pool circling. It can improve latency without hippocampal place learning. Separating procedural from spatial solutions is a main reason to classify search strategies.
Proximal cues: Proximal cues are landmarks at or near the pool (a marker on the platform, a pattern on the wall of the pool). They support cue navigation rather than a room map. Distal cues sit farther away on the room walls. See distal cues and dual-strategy trial.

Q

Quadrant entry sequence: The quadrant entry sequence is the order in which the subject visited the quadrants, which helps reveal its strategy.
Quadrants and target quadrant time: The pool is divided into four quadrants, starting at NE and going clockwise. Time in the target quadrant in a probe shows spatial memory when platforms sit at quadrant centres.

R

Random search: Random search is swimming without a clear bias toward the goal, the wall, or a fixed radius. Paths look unstructured. Early in training it is common; late in training it suggests the subject has not settled on a spatial or procedural solution.
Reference memory: Reference memory in the water maze is knowledge that stays true across trials, such as where the platform is when it is fixed for a block of days. Contrast with working memory designs where the place changes often.
Reversal learning: In reversal learning the platform moves to a new place after acquisition. Subjects must stop searching the old location and learn the new one. Close encounters and Gallagher measures at both calibrated positions show perseveration versus updating. See platform zones.

S

Scanning: Scanning (whole-area scanning) is searching widely across the pool, often with looping paths that cover many regions. Target scanning is the tighter variant focused near the goal. Scanning can look energetic while still being non-spatial. See focal search and search strategies.
Search error: Search error is how far the subject's path was from the goal, summarised over time. Cumulative search error and average proximity are the common forms. Lower error means a more accurate search.
Serial search: Serial search is checking locations in a fixed order (for example successive Barnes holes clockwise). It can solve a maze quickly without a spatial map. In the pool, sequential visits around the quadrants can play a similar role. See quadrant entry sequence.
Spatial learning: Spatial learning is learning about places and the relations between them. In the water maze it usually means forming a map from distal cues so the platform can be found from new start points. See place learning and allocentric strategy.
Start position: The start position is where the subject is released. Varying starts (for example N, S, E, W) forces use of a room map rather than a fixed turning response. Start distance also affects raw latency and path length, which is why corrected measures exist.
Strategy classification: Strategy classification assigns labels such as direct swim, chaining, thigmotaxis or scanning to a trial (or to parts of it) from the path. Criteria based on time-stamped positions and angular span reduce false positives from paths that only briefly cross a zone. See behaviour identification.
Subject Size: Subject Size sets how big the tracked object must be. For water maze leave the slider at the left; for dry mazes set it so the tracking box is larger than any droppings but smaller than the animal.
SUM (summary) file: A SUM file holds the analysis of a batch of trials. Highlight the trials, type a name in ‘Name your Summary (SUM) file’ and click ‘Analyze Batch or Series’.
Swim speed: Swim speed is how fast the subject moves through the water. Average speed includes pauses; active speed excludes slow floating. Speed is a control for motor and motivational differences when you read escape latency. See average and active speed and swim speed.

T

Target-focused search: Target-focused search is another name for focal search: concentrated searching around the learned place rather than the whole pool.
TARGET_RADIUS and TARGET_POSITION: In the VR water maze, the trial ends when the participant reaches the target edge defined by TARGET_RADIUS (metres) and TARGET_POSITION.
Target scanning: Target scanning is searching in the immediate neighbourhood of the learned platform location, typically with small loops or passes close to the spot. It is a spatial probe signature closely related to focal search.
Thigmotaxis: Thigmotaxis is the tendency of an animal to stay close to the walls of an arena, often called wall-hugging. Early wall-hugging is often adaptation; if it persists it may reflect stress or anxiety.
Thigmotaxis band: The thigmotaxis band is the outer ring next to the wall used to measure wall-hugging. Percentage of time and path in the band, and angular span of wall-following episodes, quantify thigmotaxis. See thigmotaxis.
Thigmotaxis with exploration: Thigmotaxis with exploration is wall-following combined with wider excursions into the open water. It often marks a subject that is beginning to leave the wall. See thigmotaxis.
Thigmotaxis with incursions: Thigmotaxis with incursions is wall-following broken by short trips into the pool before returning to the edge. It sits between pure wall-hugging and freer search. See thigmotaxis.
Time bins (open field and NOR): Time bins split an open field or NOR trial into periods so you can see how behaviour changes during the trial.
Time slice: A time slice analyses only part of a trial. Slice start 0 and Slice time 30 give the first 30 s; start 30 and time 30 give the next 30 s.
TRACKING_DIAMETER and PLAYER_SCALE: TRACKING_DIAMETER sets the size of the tracked area in a vrScript. For the VR water maze, a diameter of 200 (cm) is recommended so results are comparable with the rodent maze.
Tracking threshold: The threshold sets how the tracker separates the subject from the background. Start at 100 and check the cross follows the subject.
TRIAL_START_POINT (VR positions): VR positions use six numbers: east, height, north, then three angles. In the water maze, east and north are proportions of pool radius.
TXP (Master Experiment) file: A TXP file is a simple list of trials, easily made in Excel: start point, platform number, subject and descriptor, one row per trial in run order.

V

Visible platform: Visible platform is another name for a cued platform trial: the goal can be seen, so success checks sensorimotor ability rather than place memory.
vrScript file: Each VR experiment is a plain-text .vrScript file, run by pressing its function key after ‘Run HVS 6D VR’. Scripting is text editing, not programming.

W

Whishaw index: The Whishaw index (Whishaw's directionality or error index) scores how directly the subject headed toward the goal early in the trial, based on the path's deviation from the ideal line. It belongs with heading error, path efficiency and the corridor and cone tests as a directional accuracy measure.
Working memory (water maze): Working-memory water maze designs change the platform place frequently (for example each day), so the subject must remember the current location within the session rather than a long-term fixed place. Contrast with reference memory.

Many of these terms have their own explainer on hvsimage.com, starting with what the Morris water maze is.

Spotted a mistake, or know a better source? Tell us and we'll fix it and say so on our corrections page.