Episodic Memory
Hidden Path Learning
Also called: hidden maze learning, hidden pathway maze, maze learning test, invisible maze task, stylus maze
Measure spatial learning: participants find a path hidden in a 10 by 10 grid one square at a time from right and wrong feedback, over repeated rounds and once more after a break.
8-12 minUntimed
Free to start: 10 free participants per account, then $0.99 per participant for academic research, or any yearly plan. Preview any test free after you sign up.
- Papers, 2021 to 2025
- 23
- Evidence grade
- C
- Registered clinical studies
- 66
What it measures
Measures spatial learning and memory with a path hidden in a 10 by 10 grid of squares. Participants find the path one square at a time from right and wrong feedback, over several rounds on the same path and once more after a short break. The main score is the number of wrong squares over the learning rounds.
Constructs
- Spatial learning
- Visuospatial working memory
- Error monitoring
Research uses
- Memory and learning research
- Repeated testing in aging and drug studies
- Executive-function research
- Within-study group comparison
How the paradigm works
Hidden Path Learning measures spatial learning and memory: participants find a path hidden in a grid of squares one step at a time, from right and wrong feedback, and learn it over repeated rounds. It follows the hidden maze of Milner (1965), in the 10 by 10 computer form that Papp, Snyder, Maruff, Bartkowiak and Pietrzak (2011) describe. No published maze or path is used: every path is grown by our code from the session's seed by stated rules.
More about the paradigm
The path runs from the top-left square to the bottom-right square, which has a flag, with 28 moves and 12 turns. It moves one square at a time, up, down, left or right, never comes back to a square and never runs beside itself, so every square next to the participant that is not the next step is off the path. The participant starts on the top-left square, marked with a blue dot, and clicks or taps the next square. A square on the path turns green and the dot moves onto it; a square off the path turns red and the dot stays. Moving diagonally, skipping a square, tapping the same square or going back along the path breaks the rules and counts as an error too. A round ends at the flag, or at a limit of 4 minutes.
The standard session has 5 learning rounds on the same path (or 3), then 120 seconds of true-or-false arithmetic (60 to 600 in the settings) and one more round on the same path. One practice round on a 5 by 5 grid, the same for everyone, comes first. Every tap is kept with its time and input, and the scorer reads the taps again on the saved path.
The main score is the total number of errors over the learning rounds: taps on any square that is not the next square of the path. The errors of each learning round and of the round after the break, the wrong moves and rule breaks, repeated errors (a wrong square tapped again from the same square), steps along the path per second and the time taken are also reported. Each score is a count or a total for the session; none compares one round with another. Taps within 100 ms of the grid appearing or of the last tap taken are counted as anticipations and not taken. Scores support research and do not diagnose.
Scores
- Total Errorscount
- Taps on a square that is not the next square of the path, over the learning rounds.
- Lower is better
- Errors, Round 1count
- Errors in the first learning round.
- Lower is better
- Errors, Round 2count
- Errors in the second learning round.
- Lower is better
- Errors, Round 3count
- Errors in the third learning round.
- Lower is better
- Errors, Round 4count
- Errors in the fourth learning round; blank when the session ran 3 rounds.
- Lower is better
- Errors, Round 5count
- Errors in the fifth learning round; blank when the session ran 3 rounds.
- Lower is better
- Errors After the Breakcount
- Errors in the round on the same path after the break.
- Lower is better
- Wrong Movescount
- Taps on a square next to the participant's square that is off the path, over the learning rounds.
- Lower is better
- Rule Breakscount
- Taps on the same square, on the square just left, or on a square not next to the participant's square, over the learning rounds.
- Lower is better
- Repeated Errorscount
- Errors on a square already tapped wrongly from the same square in the same round, over the learning rounds.
- Lower is better
- Steps per Secondscore
- Steps along the path per second over the learning rounds.
- Informational
- Rounds Completedcount
- Learning rounds that reached the end of the path.
- Higher is better; an unfinished round is flagged
- Unfinished Roundscount
- Rounds, learning or after the break, that reached the time limit before the end of the path.
- Above 0 is flagged for review
- Total Timeseconds
- Time from each learning round's onset to the end of the path, summed; a round not finished counts its limit.
- Informational
- Anticipationscount
- Taps within 100 ms of the grid appearing or of the last tap taken, not taken.
- Quality check
- Planned Roundstrials
- Learning rounds and the round after the break that the session planned.
- Informational
- Total Roundstrials
- Rounds in the scored run.
- Fewer than planned is flagged for review
Use and evidence
23 papers from 2021 to 2025 name this task in their title or abstract, down 57% compared with 2016 to 2020. That ranks it 323 of 395 test families in our census. See the counts for every test.
Show the numbers
| Year | Papers |
|---|---|
| 2025 | 0 |
| 2024 | 6 |
| 2023 | 6 |
| 2022 | 6 |
| 2021 | 5 |
| 2020 | 7 |
| 2019 | 15 |
| 2018 | 7 |
| 2017 | 11 |
| 2016 | 14 |
| 2015 | 12 |
| 2014 | 7 |
| 2013 | 4 |
| 2012 | 9 |
| 2011 | 6 |
| 2010 | 2 |
| 2009 | 8 |
| 2008 | 9 |
| 2007 | 6 |
| 2006 | 1 |
| 2005 | 3 |
| 2004 | 1 |
| 2003 | 1 |
| 2002 | 0 |
| 2001 | 1 |
| 2000 | 1 |
Use grade E
Fewer than 30 papers from 2021 to 2025.
Evidence grade C
At least 2 test-retest papers, 30 reliability or validity papers or 3 registered clinical studies.
- Papers on test-retest reliability
- 6
- Papers on reliability or validity
- 18
- Registered clinical studies using it as an outcome
- 66
How the grades work
Use: A is 1,000 or more papers from 2021 to 2025, B 300 to 999, C 100 to 299, D 30 to 99 and E fewer than 30. Evidence: A means at least 25 papers report its test-retest reliability and at least 25 registered clinical studies use it as an outcome; B, at least 8 test-retest papers, and either 5 registered clinical studies or 150 reliability or validity papers; C, at least 2 test-retest papers, 30 reliability or validity papers or 3 registered clinical studies; D, little published measurement evidence found. The counts measure use. They include every paper that names the task in its title or abstract. Grades describe the published research on this task, not ConductCognition's version of it.
Source: ConductCognition test census 0.14.0, counted 31 October 2026 from OpenAlex and ClinicalTrials.gov.
Task design
The trial sequence as the test runs in ConductCognition.
- Grid
- 10 by 10 squares. The path runs from the top-left square to a flag in the bottom-right square, with 28 moves and 12 turns, and never runs beside itself.
- Paths
- Grown from the session's seed by a random walk that keeps the finish in reach, drawn again until it has the set moves and turns. No published maze or path is used.
- Moves
- Click or tap the next square up, down, left or right of the blue dot. A square on the path turns green for 300 ms and the dot moves onto it; a square off the path turns red. Diagonal moves, skipped squares, the same square and going back count as errors too.
- Rounds
- 5 learning rounds on the same path (standard; or 3), each up to 4 minutes, then 120 seconds of true-or-false arithmetic (standard; 60 to 600) and one more round on the same path.
- Practice
- One round on a 5 by 5 grid with a path of 10 moves, the same for everyone, followed by its number of wrong squares.
- Data
- Every tap with its time and input. The scorer reads the taps again on the saved path. Taps within 100 ms of the grid or of the last tap are counted as anticipations and not taken.
Papp and colleagues (2011) describe the 10 by 10 computer form with a path of 28 moves and 11 turns, in which the participant taps the last right square again after an error. Here the paths are our own, with 12 turns, and the blue dot stays on the last right square, so no step back is needed after an error.
Quality checks
Each session is marked valid, needs review or invalid, with the reason, before you see its scores.
Scored rows: hidden_path_learning_trial from the scored run, one per learning round and one for the round after the break. A round has no chance level, so accuracy is not judged against chance. A round with no tap before its time limit is a missed response: more than max(3, 25%) invalidate and any is flagged. A round that ends before the end of the path, or a short run, is flagged and withholds the percentile.
Why these rules
There is no correct option to compare with chance, so accuracy is not judged against chance. Errors can only be counted on the way to the end of the path, so a round left unfinished gives an incomplete count; a round with no tap at all is not an attempt and signals non-engagement.
Running this test
Settings you can change
| Setting | Standard | Allowed |
|---|---|---|
| Learning rounds | 5 (standard) | 5 (standard), 3 |
| Break before the delayed round (seconds) | 120 | 60 to 600 |
Yearly plans can change settings within these limits. Free and pay-as-you-go studies use the standard settings.
Need a setting this test does not have?
Devices and languages
- Untimed test: runs on desktops, laptops and tablets. Participant devices
- Languages: English, Spanish, French, German, Italian, Portuguese (Brazil), Chinese (Simplified) and Japanese.
- Available on: Every plan, including the free plan.
- Recruiting online? See the setup guides for Prolific, Sona Systems, CloudResearch Connect and Qualtrics.
Cite and describe
Participants completed the Hidden Path Learning task in a web browser (ConductCognition task version 1.0.0, scoring version 1.0.0; Milner, 1965; Papp et al., 2011), with the standard settings (learning rounds: 5; break before the delayed round (seconds): 120). The main score was total errors (taps on a square that is not the next square of the path, over the learning rounds).
ConductCognition versions: task version 1.0.0, scoring version 1.0.0. Every export records the versions each session used, for a methods section.
- Milner, B. (1965). Visually-guided maze learning in man: Effects of bilateral hippocampal, bilateral frontal, and unilateral cerebral lesions. Neuropsychologia, 3(4), 317-338. https://doi.org/10.1016/0028-3932(65)90005-9The hidden maze: a path found step by step from feedback and learned over repeated runs.
- Papp, K. V., Snyder, P. J., Maruff, P., Bartkowiak, J., & Pietrzak, R. H. (2011). Detecting subtle changes in visuospatial executive function and learning in the amnestic variant of mild cognitive impairment. PLoS ONE, 6(7), e21688. https://doi.org/10.1371/journal.pone.0021688The 10 by 10 computer form: a hidden path of 28 moves, five learning rounds and a delayed round, with errors counted against the movement rules.
Source papers
- Milner B (1965). Visually-guided maze learning in man: effects of bilateral hippocampal, bilateral frontal, and unilateral cerebral lesions. Neuropsychologia, 3(4):317-338.Paradigm source, Adults
- Papp KV, Snyder PJ, Maruff P, Bartkowiak J, Pietrzak RH (2011). Detecting subtle changes in visuospatial executive function and learning in the amnestic variant of mild cognitive impairment. PLoS ONE, 6(7):e21688.Paradigm source, Older adults in the source
Questions
How long does the Hidden Path Learning take?
About 8 to 12 minutes, as listed in the test library.
What device does a participant need?
Untimed test: runs on desktops, laptops and tablets. The browser must be Chrome, Edge, Firefox or Safari.
What scores does it produce?
Total Errors, Errors, Round 1, Errors, Round 2, Errors, Round 3, Errors, Round 4, Errors, Round 5, Errors After the Break, Wrong Moves, Rule Breaks, Repeated Errors, Steps per Second, Rounds Completed, Unfinished Rounds, Total Time, Anticipations, Planned Rounds and Total Rounds. Each session is also marked valid, needs review or invalid, with the reason.
How widely is the Hidden Path Learning used in research?
23 papers from 2021 to 2025 name it in their title or abstract, down 57% compared with 2016 to 2020. That ranks it 323 of 395 test families in the ConductCognition test census (use grade E). Source: ConductCognition test census 0.14.0, counted 31 October 2026 from OpenAlex and ClinicalTrials.gov.
How much reliability evidence does the Hidden Path Learning have?
In the published research, 6 papers report its test-retest reliability and 66 registered clinical studies use it as an outcome (evidence grade C in the ConductCognition test census).
Which settings can a researcher change?
Learning rounds (5 (standard), 3) and break before the delayed round (seconds) (60 to 600). Settings can be changed on yearly plans; free and pay-as-you-go studies use the standard settings.
Which languages can participants use?
Participants can take this test in English, Spanish, French, German, Italian, Portuguese (Brazil), Chinese (Simplified) and Japanese. English is on every plan, and the other languages come with yearly plans. Have a native speaker read the machine-translated text before launch.
Is this a clinical test?
No. ConductCognition is for research use only. It runs, scores and exports the test; clinical interpretation is outside the platform.
Every test: the full test list.
Run this test in your next study
Start free with 10 free participants on any test, then use participant credits or a yearly plan.