HVS tracking hints
Common mistakes and quick fixes
Everyone makes these once. Here's how to make them only once.
Each card says what goes wrong, how to fix it and the fastest way to check you've fixed it.
Clicking the pool or table edges in the wrong order.
- What goes wrong
- The platform is plotted in the opposite position or paths are mirrored, so every target and quadrant measure refers to the wrong place.
- Quick fix
- Recalibrate, clicking North (top), then South (bottom), then West (left), then East (right).
- Fastest check
- Place a light object at a known position and check it plots in the same place.
Entering a ‘Platform Number’ that was never calibrated.
- What goes wrong
- The platform is not shown, appears in the top-left corner or shows as a purple marker, and target measures are not calculated for the intended position.
- Quick fix
- Remember the first calibrated position is platform 0 (usually the centre). With ‘Suggested MWM Platform Positions’, 1 to 4 are the quadrant centres, starting at NE and going clockwise. With one calibrated platform, enter 0.
- Fastest check
- Click ‘PREPARE’ and check the on-screen indicator circle lines up with the real platform.
Entering a pool diameter larger than the real water surface, or a platform diameter that does not match the platform.
- What goes wrong
- All distances and speeds are inflated, and the point where arrival is detected shifts, so the path can appear to end just before the platform.
- Quick fix
- Measure the water surface and the platform, then recalibrate with the true diameters.
- Fastest check
- Check one trial's duration against a stopwatch and its path against a printout.
Calibrating a black pool without ticking black, so white animals are not tracked.
- What goes wrong
- White animals in a black pool or dark water are not tracked until the black setting is used.
- Quick fix
- Recalibrate with the black box ticked; calibration defaults to white.
- Fastest check
- After recalibrating you should see a black mask round the pool.
Adjusting the tracking threshold before fixing the lighting.
- What goes wrong
- A threshold tuned around shadows or glare lets the cross jump away from the subject, adding false path and location data.
- Quick fix
- Run ‘CHECK LIGHTING’ first and remove shadows and glare, then set the threshold, starting at 100.
- Fastest check
- Once the dots have gone, check the cross follows the subject everywhere and run a practice trial.
Leaving auto-focus on, or changing camera settings part-way through an experiment.
- What goes wrong
- Changing camera settings mid-experiment breaks consistency with the calibration.
- Quick fix
- Focus once with the camera's own software (with HVS Image closed), turn auto-focus off, set anti-flicker to your mains frequency and note your settings.
- Fastest check
- Compare today's camera settings with the ones you noted.
Leaving the laptop's built-in webcam enabled, or the camera maker's software open.
- What goes wrong
- Only one program can use the camera, so you see no picture, a black picture or the wrong camera.
- Quick fix
- Disable the built-in webcam in Device Manager, close the camera maker's software and restart HVS Image.
- Fastest check
- Switch on the tracking lights and check the live picture shows the arena from the tracking camera.
Running a probe trial with auto-end still on.
- What goes wrong
- Auto-end decides whether a trial ends when the subject reaches the platform, so with it on the probe stops at the first arrival.
- Quick fix
- Enter the learned position as usual, tick Probe, select ‘No’ under ‘Auto-end on platform’ (‘Auto-end on escape hole’ for Barnes) and set ‘Max Seconds’, for example 60.
- Fastest check
- The probe trial should run to the full ‘Max Seconds’.
Saving the TXP file as .txt, .csv or .xls, or leaving out commas for empty fields.
- What goes wrong
- The file is not found, or subject and group appear in the wrong columns, so trials are mislabelled.
- Quick fix
- Save with the 'All files' type and a .txp ending. Each line is start, platform, subject, descriptor, for example N,2,Subject1,GroupA, and empty fields keep their commas (,,mouse1,groupA).
- Fastest check
- Open the file in Notepad and count four comma-separated fields on each line.
Clicking ‘RUN’ before or after the moment of release.
- What goes wrong
- Starting the clock at the moment of release gives accurate latencies; an early or late click shifts latency for that trial.
- Quick fix
- Click ‘PREPARE’, line up the platform or hole with the on-screen circle, then hold the remote or mouse in one hand and the subject in the other and click as you release.
- Fastest check
- Time one trial with a stopwatch and compare it with the recorded duration.
Reaching in from the top of the screen, so the tracker follows your hand or sticks to a dark platform.
- What goes wrong
- The tracker scans from the top-left and locks on to the first dark object it finds, giving a false start to the path.
- Quick fix
- Work from the south (bottom of screen) side, wear a white cap and sleeve for hard start points, and adjust lighting so the platform is no darker than the water.
- Fastest check
- Set the threshold with a subject in the south of the pool, then run a practice trial and check the cross follows the subject.
Leaving antivirus scans, updates or other programs running during tracking.
- What goes wrong
- Background programs cause straight lines on path plots, frame drops or a frozen trial timer.
- Quick fix
- Close all other programs, schedule antivirus or security scans outside tracking, and use a High Performance power plan with USB suspend off and sleep set to Never.
- Fastest check
- Look for straight lines on the path plot or a frozen trial timer in a practice trial.
Using an optical mouse or a remote that does not click reliably once.
- What goes wrong
- A double or missed click ends a trial at once or fails to start it.
- Quick fix
- Use a long-range wireless presenter or track-ball mouse with about 10 m range that keeps the cursor steady while you walk.
- Fastest check
- If trials end at once or fail to start, try another mouse.
Letting the HVS folder fill with 10,000 or more .hvs files.
- What goes wrong
- Files numbered 10000 onwards do not show, so your newest trials are missing from the analysis list.
- Quick fix
- Back up the HVS folder, create a subfolder and move older files into it, then click ‘ANALYZE’ again.
- Fastest check
- Count the .hvs files in the HVS folder: it must hold fewer than 10,000.
Relying on target-quadrant time when the platform is not at a quadrant centre.
- What goes wrong
- Quadrant measures are only meaningful when platforms sit at quadrant centres.
- Quick fix
- Use close passes and Gallagher measures instead, or place platforms at quadrant centres in future studies.
- Fastest check
- Check on the plot where the platform sits: quadrants start at NE and go clockwise, with North at the top of the image.
Changing ‘Advanced’ values part-way through a study, or not recording them.
- What goes wrong
- These settings change the Direct Path results, which is why they should be kept constant and reported.
- Quick fix
- Keep Counter diam x, ThigRatio and Heading distance constant across a study and report them.
- Fastest check
- Check the exported Heading range rather than assuming a value.
Running HVS Image with Windows number formats that use a comma as the decimal mark.
- What goes wrong
- Results can show negative paths, or calibration will not save.
- Quick fix
- In Control Panel > Region choose UK or USA formats ('.' decimal, ',' list separator) and restart HVS Image; do the same on analysis PCs.
- Fastest check
- Check a results file for negative paths.
Accepting a slice latency that equals the whole trial duration.
- What goes wrong
- An out-of-date build reports the wrong latency for slices in which the platform is not reached.
- Quick fix
- Check the build on the main control panel and install the current version: slice latency should be the slice duration if the platform is not reached in it. Name each results file after its slice.
- Fastest check
- Check the build on the main control panel.
Analysing copies of .hvs files that were damaged in transfer, for example via USB.
- What goes wrong
- Corrupted copies lose data, giving short latencies, points outside the pool or no plot.
- Quick fix
- Use the original files or backups made at recording, and always keep the originals.
- Fastest check
- Look for short latencies, points outside the pool or a missing plot.
Deleting .sum or .hvs files after exporting results.
- What goes wrong
- Without them you cannot make a different results file later.
- Quick fix
- Keep .sum files to make new variable selections and .hvs files to make new .sum files; for VR also keep the .BOND files.
- Fastest check
- Check your backup holds the .hvs (and .BOND) files for every trial in the results.
Opening a CST file without choosing comma as the separator.
- What goes wrong
- CST files are in CSV format, so the comma separator is what splits the variables into columns.
- Quick fix
- Open the CST in Excel or Google Sheets and, if asked, choose comma as the separator.
- Fastest check
- Check the first row shows your chosen variables in separate columns.
Calibrating Barnes holes on a screen that was not opened from the Barnes maze button.
- What goes wrong
- If the holes are not masked, check the Barnes box is ticked, which shows you opened the Barnes maze.
- Quick fix
- Close the screen and open the Barnes maze; its Barnes box is ticked for you. Then click Calibrate, click the table edges (top, bottom, left, right), then click each hole.
- Fastest check
- After ‘Save and Return’ the holes should be masked by white discs.
Waiting for a Barnes trial to end by itself when ‘Auto-end on escape hole’ is set to ‘No’.
- What goes wrong
- Ending late or early changes escape latency.
- Quick fix
- On the Barnes maze Run screen set ‘Auto-end on escape hole’ to ‘Yes’ or ‘Delay in Secs’, or leave it on ‘No’ and click ‘END’ (or the remote) when the animal fully enters the hole.
- Fastest check
- Check the ‘Auto-end on escape hole’ setting on the Barnes maze Run screen.
Letting your hand or the start container cast a shadow at the start of the trial.
- What goes wrong
- A temporary shadow attracts the tracker, so the path jumps at the start or shows straight lines.
- Quick fix
- Use a small glass beaker as the start container, wear a white glove and sleeve, stand at the south end, or click start just after your hand has left the area.
- Fastest check
- Check the first part of the path plot for jumps or straight lines.
Leaving the subject size too small on a dry maze, so droppings are tracked.
- What goes wrong
- Setting the size carefully means the tracker follows the animal and ignores droppings.
- Quick fix
- With no faeces on the maze, move the slider right until the cross appears, then keep going until the graticules are larger than any droppings and slightly smaller than the animal.
- Fastest check
- Watch the cross stay on the animal when it passes a dropping.
Analysing Barnes trials without setting the thigmotaxis band and hole counter size.
- What goes wrong
- These settings turn your paths into hole visits, errors and search measures.
- Quick fix
- In ‘Advanced’, set the thigmotaxis band so the holes fall in it (for example 0.65) and the counter size for hole investigation (for example 2.0), and ignore water-maze-only measures.
- Fastest check
- Double-click one trial and check the holes fall within the band on the plot.
Clicking the bottom-right corner before the top-left when calibrating a rectangular arena.
- What goes wrong
- Clicking in the wrong order mirrors the plot.
- Quick fix
- Click the top-left corner first, then the bottom-right, then ‘Save (and return)’.
- Fastest check
- Test with a light object at a known position.
Using a background or objects that contrast with the floor the way the animal does.
- What goes wrong
- The aim is that only the animal is tracked, not objects or gaps between arenas.
- Quick fix
- Make the whole background one contrasting colour (white for dark or hooded animals, black for white animals, including gaps between arenas) and colour or mask objects to match the floor.
- Fastest check
- With an animal in place, check the cross follows only the animal everywhere in the arena.
Ticking black for a dark animal on a light floor.
- What goes wrong
- The black setting is meant only for light animals on a dark floor or maze.
- Quick fix
- Tick black only for light animals on a dark floor.
- Fastest check
- Check the cross follows the animal during threshold set-up.
Choosing a grid size that puts two objects in one cell, or splits an object across cells.
- What goes wrong
- Object exploration cannot be separated if objects share or straddle cells. You can re-analyse with other grid sizes at any time.
- Quick fix
- Pick a grid size that puts each object in its own cell, and place objects so they fall inside grid cells.
- Fastest check
- Remember cells are numbered down the left column first, then the next column, and check each object's cell number.
Entering subject IDs for several arenas in the wrong order.
- What goes wrong
- Data are attributed to the wrong animals.
- Quick fix
- Separate subject IDs, and descriptors, with pipes (|). For ‘2x2 Fields’ the order is top-left, top-right, bottom-left, bottom-right.
- Fastest check
- Before running, read the IDs back against the arena positions.
Treating a zone latency of -1 as a number in statistics.
- What goes wrong
- Averaging -1 values would wrongly shorten mean latencies.
- Quick fix
- Read -1 as 'zone not visited' and handle it as such in your analysis.
- Fastest check
- Filter the latency columns for -1 before averaging.
Not checking or reporting the freezing criteria when freezing or inactivity matter.
- What goes wrong
- Freezing and inactivity values depend entirely on these criteria.
- Quick fix
- Set the time window and movement for freezing, for example less than 0.1 cm in 2 s for freezing and less than 5 cm in 10 s for inactivity, and report them.
- Fastest check
- Check the criteria shown on the analysis screen before batch analysis.
Calibrating the whole container instead of a band around the water line.
- What goes wrong
- This band is what lets the software tell climbing, swimming, floating and diving apart.
- Quick fix
- Calibrate a band around the water line, sized so the animal's highest point sits in the middle third when swimming or floating, the top third when climbing and the bottom third when diving.
- Fastest check
- Watch one animal climb and dive, and check its highest point moves into the top and bottom thirds.
Tracking the tail or water droplets instead of the body.
- What goes wrong
- Tracking the body rather than the tail keeps % climbing, swimming and floating accurate.
- Quick fix
- Move ‘Set X-Y Size’ to the right until the cross stays on the body.
- Fastest check
- Watch the cross during a practice trial: it should not jump to the tail or droplets.
Analysing the dummy files after running three containers on a 3 x 3 grid.
- What goes wrong
- Dummy files add empty trials to your results.
- Quick fix
- Exit HVS Image and move the .hvs files ending 4 to 9 to another folder before analysis; delete them only after a successful analysis.
- Fastest check
- Count the files per set: nine are made, and those ending 4 to 9 belong to the six dummy names.
Clicking the maze points out of order, or not right-clicking the final point.
- What goes wrong
- The maze outline is what lets the software count every arm entry.
- Quick fix
- Click the points shown in the small example image, in that order, using a right-click on the final point.
- Fastest check
- Check the maze outline appears before ‘Save (and return)’.
Misreading which arm is which in the results.
- What goes wrong
- Swapped arms reverse open- and closed-arm results.
- Quick fix
- Arms are listed from the left arm clockwise (plus maze: left, top, right, bottom).
- Fastest check
- Compare one trial's arm entries with its trace.
Not stating, or changing, the Detection Area between groups.
- What goes wrong
- Arm entries depend on the Detection Area; you can re-analyse with other percentages any time.
- Quick fix
- Choose one Detection Area (the % of each arm that counts as an entry), for example 100% for any entry or 25% for the far quarter only, and report it.
- Fastest check
- Check the Detection Area selected before batch analysis.
Plugging the treadmill adapter into a different USB port, so its Windows input name no longer matches OMNIDRIVE_INPUT.
- What goes wrong
- The participant walks on the treadmill but does not move in VR.
- Quick fix
- Plug the adapter into the usual USB port, rename the input 'Omnidrive' in Windows Sound Settings if it shows as 'Microphone', and check OMNIDRIVE_INPUT in your vrScript matches exactly.
- Fastest check
- Ask the participant to step and watch whether they move in VR.
Leaving OMNI_SENS and SMOOTH_MOTION untuned, so movement runs on or steps are missed.
- What goes wrong
- Unintended movement adds path the participant did not walk.
- Quick fix
- Restore the values HVS supplied, then lower OMNI_SENS a little if small vibrations move the participant, raise it if light steps are missed, and shorten SMOOTH_MOTION (for example 0.5 to 0.3) if movement runs on after stopping.
- Fastest check
- Ask the participant to take light steps and then stop, and watch the view.
Pressing Enter while the participant is still moving.
- What goes wrong
- Stopping first means the participant lands safely in the pool, ready to begin.
- Quick fix
- Ask the participant to stop moving, then press Enter to place them at the trial start point.
- Fastest check
- Before each trial, confirm the participant is standing still.
Typing a SET_SCENE name that does not exactly match a supplied scene name.
- What goes wrong
- The scene name must match exactly, letter for letter.
- Quick fix
- Copy the scene name exactly from the supplied example files.
- Fastest check
- If the vrScript has an error, scroll to the bottom of the error log to see which line to fix.
Changing TRACKING_DIAMETER or PLAYER_SCALE from the water maze values.
- What goes wrong
- These settings make VR results directly comparable with the rodent water maze.
- Quick fix
- For the water maze use TRACKING_DIAMETER 200 and PLAYER_SCALE 0.16; TRACKING_AREA_NORTH and WEST are 0, SOUTH and EAST equal the diameter.
- Fastest check
- Read the general lines at the top of each vrScript before the session.
Leaving another OpenXR runtime active, or SteamVR out of date.
- What goes wrong
- The headset freezes, drops the Link connection or shows the SteamVR home instead of HVS VR.
- Quick fix
- In SteamVR settings set OpenXR Runtime to SteamVR, set SteamVR to immediate automatic updates (stable, not beta), then restart Steam, SteamVR and the PC.
- Fastest check
- Start a session and check the headset shows HVS VR, not the SteamVR home.
An mp3 file name in MAIN_MP3 or END_WITH_MP3 that does not match the file exactly.
- What goes wrong
- The session moves on when the mp3 ends, so the audio file also controls session timing.
- Quick fix
- Check the file exists and the name matches exactly; audio starts after you press Enter.
- Fastest check
- If the scene runs on after the audio, check END_WITH_MP3.
Starting a heart-rate session before the Polar H10 is connected.
- What goes wrong
- Heart rate is not recorded for that session.
- Quick fix
- Moisten the sensor; if the laptop view says the device is not connected or found, click refresh, and turn off Bluetooth on nearby phones with the Polar app.
- Fastest check
- Check the bottom-right of the laptop view before pressing Enter.
Installing or running HVS Image without administrator rights.
- What goes wrong
- You can see 'File not found' (error 53), 'Path not found' (error 76) or a registration error.
- Quick fix
- Back up the HVS Image folder, rerun the setup file with Run as administrator, and set the desktop shortcut to always run as administrator.
- Fastest check
- Right-click the shortcut > Properties > Shortcut > Advanced and check Run as administrator is ticked.
Pasting the activation code into the green box at the top instead of the red box at the bottom.
- What goes wrong
- The module stays locked until the code is accepted.
- Quick fix
- Delete any old code, remove spaces before or after, and paste the new code into the red box; OK then becomes clickable. Use the code for the right module.
- Fastest check
- The OK button becomes clickable once the code is in the right box.
Many of these mistakes start at calibration; HVS Image has a page on calibrating a Morris water maze for tracking.
Spotted a mistake, or know a better source? Tell us and we'll fix it and say so on our corrections page.