Interpreting probe and training results

Reading probe results is part science, part detective work. These guides set out what a typical control group looks like, what impairment can look like, and which set-up problems can masquerade as a result.

Reading water maze results

Look for learning across days, a clear search of the right place in probes, and a tidy set-up. Read latency with path ratio, heading and Gallagher measures rather than on its own.

Typical control pattern: During learning, heading angles fall towards zero and well-trained subjects have a path ratio approaching 1. In probe trials, a preference for the target quadrant regardless of start point indicates spatial memory (when platforms sit at quadrant centres), and trained subjects make fewer quadrant entries, concentrated on the target. Early wall-hugging is often adaptation, and chaining is common early in learning.
Signs of impairment: Chaining in a probe after full training suggests spatial learning is impaired or absent, or spatial cues are lacking; target-quadrant time then stays near chance. Persistent thigmotaxis may reflect stress or anxiety, not understanding the task, sensorimotor impairment or low motivation. A higher Gallagher Global value means the subject searched further from the target, and this proximity measure was found to be highly sensitive to age-related impairment (Gallagher et al., 1993).
Signs of strong memory: The help content describes strong-memory signs rather than an 'enhanced' pattern: near-zero heading angles across a group in probe trials indicate strong memory, and heading is often more sensitive than whole-trial crossings or quadrant time unless a first slice is used. Be careful with whole-trial crossings, which favour subjects that keep returning (perseverating) over flexible ones that check and move on.

Set-up problems that look like results

Platform plotted in the mirrored position: the pool edges were clicked in the wrong order; recalibrate North (top), South (bottom), West (left), East (right).
Path jumps to the platform or your hand at the start: the tracker scans from the top-left and locks on to the first dark object it finds; work from the south side.
Straight lines on path plots or a frozen timer: usually background programs or uneven lighting.
Path ends just before the platform: check the calibrated pool and platform diameters first.
Swim speeds look too fast: a pool diameter larger than the real water surface inflates all distances and speeds.
Trials of about 0.1 s: a mouse or remote that double-clicks; replace or recharge it.

More detail

A non-rising platform allows non-spatial strategies, so falling latencies can be misread as spatial learning. Whole-trial quadrant time can mislead, so use average speed as a control and time slices. High inactivity can lengthen escape latency and reduce probe-trial crossings, so read it alongside the other measures. Cued trials determine whether performance factors that are unrelated to place learning are present (Vorhees and Williams, 2006).

Strategy matters as well as accuracy: before brain injury, PDAPP mice used less spatial strategies and more nonspatial systematic strategies and strategies involving repetitive looping than wild-type mice (Brody and Holtzman, 2006). Maei et al. (2009) found that the proximity measure consistently outperformed quadrant, zone and crossing measures in its ability to detect group differences in mouse probe tests.

Reading Barnes maze results

Look at errors (pokes) per hole, latency and path to the target hole and primary path length, and check the paths look clean before you interpret them.

Typical control pattern: No published pattern has been confirmed for this test yet, so compare with your own control group. HVS Image reports hole approaches counted automatically, pokes (errors) per hole, latency and path to the target hole, and primary path length (the route from the start to the first approach of the escape hole).
Signs of impairment: In the original platform study, the senescent rats were shown to exhibit poorer memory for the rewarded place (Barnes, 1979).
Signs of strong memory: No published pattern has been confirmed for this test yet, so compare with your own control group.

Set-up problems that look like results

Path jumps at the start: a temporary shadow (your hand or the start container) attracts the tracker; use a small glass beaker, a white glove and sleeve, and work from the south end.
Straight lines (short tracking drops): make lighting even and diffuse, remove black dots (shadows) in the lighting check, avoid red dots (glare) and close other programs.
Trial does not end at the escape hole: it ends by itself only if ‘Auto-end on escape hole’ is switched on; otherwise click ‘END’ when the animal fully enters the hole.

More detail

Paths are not artificially smoothed, so jagged paths reflect jerky movement. An animal may investigate a hole without entering, which is why many users leave ‘Auto-end on escape hole’ on ‘No’ and end the trial themselves. Set the thigmotaxis band so the holes fall in it before you read approaches and errors.

Barnes (1979) tested rats on a circular platform which favoured the use of spatial cues for its solution, and found the amount of synaptic enhancement was statistically correlated with the ability to perform the circular platform task.

Reading open field and object test results

In the open field, wall-hugging (thigmotaxis) and time in the outer zones are the usual anxiety-related measures; read them with path length and speed so that activity changes are not mistaken for anxiety.

Typical control pattern: When mice are introduced into an open field they explore mainly the peripheral zone, and this tendency to remain close to the walls decreases gradually during the first minutes of exploration (Simon et al., 1994).
Signs of impairment: Anxiogenic-like profile: reference anxiogenic drugs increased thigmotaxis in mice (Simon et al., 1994). Higher inactivity or freezing may also reflect fatigue, stress or anxiety, pharmacological effects or motor impairment, so check speed and path length too.
Signs of strong memory: Anxiolytic-like profile: anxiolytic drugs reduced thigmotaxis in mice (Simon et al., 1994), and in rats thigmotaxis was suppressed by anxiolytic agents (Treit and Fundytus, 1988).

Set-up problems that look like results

Cross jumps in one area: adjust the threshold with the animal in that area.
Cross follows the tail or droppings: move ‘Set X-Y Size’ to the right for large animals in small boxes.
Wrong colour of cross: a green cross tracks dark animals on a light background and a red cross white animals on a dark background; recalibrate with the correct background colour setting.
Objects or gaps between arenas are tracked: the whole background must be one contrasting colour, and objects should match the floor or be masked.

More detail

Whole-trial averages can hide effects such as early interest in a novel object, so use time bins. For object tests, body-centre tracking scores time near objects, while sniffing and paw contacts are scored manually. For inner versus outer totals, sum the relevant zone columns in Excel. The open field is a very popular animal model of anxiety-like behaviour (Prut and Belzung, 2003).

Prut and Belzung (2003) note that some compounds with a different spectrum of therapeutic efficacy were poorly effective as anxiolytics in the open field test, suggesting that this paradigm may not model features of anxiety disorders. Treit and Fundytus (1988) factored out general activity using analysis of covariance. No preference pattern for novel object recognition is given here; interpret object preference with the method and criteria of the protocol you followed.

Reading multi-arm maze results (EPM, radial, T and Y mazes)

For the elevated plus maze, compare open-arm with closed-arm % time and entries; for radial, T and Y mazes, look at the entry sequence and repeat entries.

Typical control pattern: Elevated plus maze: rats made significantly fewer entries into the open arms than into the closed arms, and spent significantly less time in open arms (Pellow et al., 1985). Radial arm maze: rats performed efficiently, choosing an average of more than seven different arms within the first eight choices (Olton and Samuelson, 1976). T-maze: on the second of two quick trials the rodent tends to choose the arm not visited before (Deacon and Rawlins, 2006).
Signs of impairment: Elevated plus maze, anxiogenic-like profile: compounds that cause anxiety in man significantly reduced the percentage of entries into, and time spent on, the open arms (Pellow et al., 1985). T-maze: both spontaneous and rewarded alternation are very sensitive to dysfunction of the hippocampus (Deacon and Rawlins, 2006).
Signs of strong memory: Elevated plus maze, anxiolytic-like profile: an increase in open arm activity (duration and/or entries) reflects anti-anxiety behaviour (Walf and Frye, 2007).

Set-up problems that look like results

No tracking cross: check the background colour setting in calibration, then set Size just above zero and sweep the threshold slowly.
A trial lasting 0.1 s points to a mouse that double-clicks; try another.
Entry counts look wrong: check the Detection Area; at 100% any entry counts, at 25% only entries to the far 25% of the arm.

More detail

HVS Image lists per-arm data from the left-most arm clockwise, and for the elevated plus maze reports open versus closed % time and entries. You can re-analyse with other Detection Area percentages at any time, so report the one you used. Neither novelty nor illumination was a significant contributor to the behaviour of the rats on the plus maze (Pellow et al., 1985).

In the radial maze, there was a small but reliable recency effect, with the likelihood of a repetition error increasing with the number of choices since the initial instance (Olton and Samuelson, 1976). Pellow et al. (1985) found that neither antidepressants nor major tranquilisers had a specific effect on the plus maze.

Reading forced swim results

HVS Image gives % time climbing, diving, swimming and floating, plus speed and path. Check the tracking band and tail tracking before you interpret the percentages.

Typical control pattern: No published pattern has been confirmed for this test yet, so compare with your own control group.
Signs of impairment: No published pattern has been confirmed for this test yet, so compare with your own control group.
Signs of strong memory: No published pattern has been confirmed for this test yet, so compare with your own control group.

Set-up problems that look like results

Cross jumps to a raised tail or water droplets on the container wall: move ‘Set X-Y Size’ to the right.
Climbing, swimming and diving look misclassified: check the tracking band height, so the animal's highest point is in the top third when climbing, the middle third when swimming or floating and the bottom third when diving.
Cross will not stay with the animal at any slider position: improve lighting and camera settings for an even background, good contrast and no glare.

More detail

The software uses the highest part of the animal within the tracking area to classify climbing, swimming/floating and diving, and separates swimming from floating by speed. With three containers run as a 3 x 3 grid, only the files for the real containers are your data.

No forced swim paper is yet in the verified publications database (a classic reference has been proposed and awaits approval), so no control, impaired or enhanced pattern is given here.

Reading HVS 6D VR water maze results

VR trials are analysed just like the water maze, so the same measures and cautions apply. Rodent and human VR values should not be compared directly.

Typical control pattern: No VR-specific control pattern is verified yet, so read results as for the water maze: quadrant measures are only meaningful when targets sit at quadrant centres; otherwise use close encounters and Gallagher proximity.
Signs of impairment: When humans are required to use spatial cues to navigate to a hidden escape platform, patients with hippocampus resections display severe impairments in spatial navigation relative to age-matched controls (Astur et al., 2002).
Signs of strong memory: No published pattern has been confirmed for this test yet, so compare with your own control group.

Set-up problems that look like results

Movement continues after the participant stops: shorten SMOOTH_MOTION; if small vibrations or sounds move the participant, lower OMNI_SENS a little.
Light steps are missed: raise OMNI_SENS.

More detail

VR analysis shows only the platform position used for the trial. Probe trials run to the maximum time even when the target is reached, so keep the learned position in TARGET_POSITION so it is shown on plots and analysed. HVS 6D VR does not record experimenter assistance, so note any help you gave.

Astur et al. (1998) consistently found that males navigate to the hidden platform better than females across a variety of measures. Thornberry et al. (2021) report an inconsistency of use of the virtual water maze, particularly with respect to the environmental setup and procedures, so report your environment and cues in full.

One way to look at how memory changes during a probe is the Devan-Baker time bin analysis.

Sources

Maei (2009). What is the most sensitive measure of water maze probe test performance?. Frontiers in Integrative Neuroscience 3. doi:10.3389/neuro.07.004.2009
Blokland et al. (2004). A detailed analysis of rats’ spatial memory in a probe trial of a Morris task. Behavioural Brain Research 154(1):71-75. doi:10.1016/j.bbr.2004.01.022
Gallagher et al. (1993). Severity of spatial learning impairment in aging: Development of a learning index for performance in the Morris water maze. Behavioral Neuroscience 107(4):618-626. doi:10.1037/0735-7044.107.4.618

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