Episodic Memory
Bus Window Game
Also called: picture memory game for preschool children, touch picture recognition memory task, bus window picture game
Measure how well children aged 4 and 5 recognise pictures they have seen: 32 photographs go by one at a time in a bus window, then the child touches the one seen in each of 32 pairs.
8-11 minUntimed
Free to start: 10 free participants per account, then $0.99 per participant for academic research, or any yearly plan. Try it here first, with no account.
- Papers, 2021 to 2025
- 642
- Evidence grade
- B
- Registered clinical studies
- 45
Figures for the Recognition Memory Test, the published test this game is compared with.
How these are counted
Watch the pictures go by

What it measures
A picture memory game for research with children aged 4 and 5, with an adult reading the screens aloud. The child watches pictures of everyday things go by one at a time, like the view from a bus window, and then touches the picture seen in each pair of pictures. Scores are raw research measures with no percentile.
Constructs
- Episodic memory
- Picture recognition memory
- Forced choice recognition
Research uses
- Research on picture recognition memory in young children
How the paradigm works
The Bus Window Game is a picture memory game for research with children aged 4 and 5, with an adult beside the child. A bus window shows 32 photographs of everyday things (animals, food, vehicles, things at home, things to play with and things in nature) one at a time, each for 3 seconds with half a second of passing scenery between them. The window then shows 32 pairs: a picture the child saw beside a new picture of the same kind, and the child touches the picture seen. The score is the share of scored pairs answered right, where chance is half. No answer time is scored: the test is untimed, shows no mark of right or wrong and no score on its screens, and no word is read after the instructions. The counts on its page are those of the Recognition Memory Test, the published test the game is compared with, not the game's own. The game is not that test with pictures: it has no word to read and no key, the child touches a picture, and a new picture is never a look-alike of a studied one. The validation plan also compares it with its own plain look and, at ages 4 and 5, with Day-Night.
More about the paradigm
A demonstration of 3 pictures and 1 pair with an animated hand, and a practice of 3 pictures and 3 pairs, come first, the right picture glowing for 800 ms after each practice touch. The practice runs once more when fewer than 2 of its 3 pairs are right, and a child with fewer than 2 right again ends there with no test rows: the child is not measurable, and the number of such children is reported with the study. In the test the first touch that starts on a picture answers and the touched picture is outlined for 300 ms. A touch off both pictures blocks nothing and is counted, and a second finger or a palm on a picture while the first is down, a hidden tab or a resized window leaves the pair out. A pair is scored only when its studied picture showed for at least 2,500 ms, so a hidden tab during a study picture takes its pair out too. A run with fewer than 24 scored pairs of 32 has no score. Each session takes one of 2 alternate forms of 32 pairs, and a third form of 12 pairs serves Try it now and no study.
Scores
- Pairs Answered Right%
- The share of scored pairs in which the picture seen before was touched.
- Higher is better; 0.5 is chance and 1 is perfect
- Right, First Half of the Pairs%
- The share right among the scored pairs in the first half of the test.
- Informational; shows whether attention held
- Right, Second Half of the Pairs%
- The share right among the scored pairs in the second half of the test.
- Informational; a fall may mean the child tired
- Right, Pictures Studied Early%
- The share right among the pairs whose studied picture was in the first third of the study list.
- Informational
- Right, Pictures Studied Late%
- The share right among the pairs whose studied picture was in the last third of the study list.
- Informational
- Left Picture Touched%
- The share of answers on the left picture. The studied picture is on each side equally often, so a share far from half suggests a habit of touching one side.
- Quality check; about 0.5 is expected
- Side Changescount
- How many times consecutive answers were on different sides.
- Quality check; a count near the most possible suggests alternating
- Longest Run on One Sidecount
- The most answers in a row on the same side.
- Quality check
- Usable Pairstrials
- Pairs counted in the scores. Left out: a pair never answered, one during which the screen was left, the window changed size, or a second finger or a palm touched a picture, and one whose studied picture was not shown for its full time.
- Quality check
- Practice Rounds Runcount
- How many rounds of the practice the child needed: 1 or 2. A child with fewer than 2 pairs right after the second round goes on to no test.
- Informational
- Repeat Touchescount
- Touches on a picture after the answer, which change nothing.
- Informational
- Touches Off the Picturescount
- Touches that began away from both pictures, which block nothing.
- Informational
- Pairs With an Extra Contacttrials
- Pairs in which a second finger touched a picture while the first was down.
- Quality check
- Answers by Touch%
- The share of answers given by touch rather than by mouse or pen.
- Informational
Use and evidence
642 papers from 2021 to 2025 name this task in their title or abstract, down 3% compared with 2016 to 2020. That ranks it 40 of 711 test families in our census. These counts belong to the Recognition Memory Test, the published test this game is compared with. This game is a new measure, so its own reliability and norms come from use. See the counts for every test.
Show the numbers
| Year | Papers |
|---|---|
| 2025 | 147 |
| 2024 | 155 |
| 2023 | 106 |
| 2022 | 109 |
| 2021 | 125 |
| 2020 | 125 |
| 2019 | 149 |
| 2018 | 126 |
| 2017 | 132 |
| 2016 | 133 |
| 2015 | 130 |
| 2014 | 124 |
| 2013 | 132 |
| 2012 | 138 |
| 2011 | 109 |
| 2010 | 134 |
| 2009 | 107 |
| 2008 | 109 |
| 2007 | 97 |
| 2006 | 89 |
| 2005 | 69 |
| 2004 | 76 |
| 2003 | 52 |
| 2002 | 55 |
| 2001 | 53 |
| 2000 | 59 |
Use grade B
300 to 999 papers from 2021 to 2025.
Evidence grade B
At least 8 test-retest papers, and either 5 registered clinical studies or 150 reliability or validity papers.
- Papers on test-retest reliability
- 17
- Papers on reliability or validity
- 186
- Registered clinical studies using it as an outcome
- 45
Measured on ConductCognition
Pooled across the studies that feed our platform norms, by device group. A figure needs 50 or more people.
No pooled figure yet.
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 the Recognition Memory Test, the published test this game is compared with, not the game's own.
Source: ConductCognition test census 0.58.0, counted 31 October 2026 from OpenAlex and ClinicalTrials.gov.
Task design
The trial sequence as the test runs in ConductCognition.
- Pictures
- 169 photographs of everyday things in six groups: animals (43), food (34), vehicles (20), things at home (22), things to play with (20) and nature (30). Each shows one thing, whole and by itself, cropped square at 512 px, with no person, no scene and no text or brand a child could read. The photographs are credited on the Terms page. A pair is two pictures of one group, so the group gives no clue, and the studied and the new pictures are balanced over six features (brightness, colour strength, amount of edge, size of the thing, warmth and distance from the centre) and over where the photograph came from. No picture shows a thing that the Recognition Memory Test or another of our tests for children aged 4 to 12 shows or names, so a study can run them with the same child.
- Forms
- 2 study forms of 32 pairs each and a third form of 12 pairs for Try it now, each a committed list drawn from a stated seed, which a test draws again. A study form holds 3 to 6 pairs of each group and no picture twice. The studied picture is on the left in 16 pairs and on the right in 16, never more than 3 in a row, so a chooser that always takes one side, or follows a repeating pattern of sides, is right on 14 to 18 of 32. In each form the studied and the new pictures differ by at most a quarter of a standard deviation in each feature, and a chooser that takes the larger or the smaller value of one feature, or goes by the photograph's source, is right on .375 to .625 of the pairs. Each participant gets the study forms in an order drawn once for them, and each session takes the next form. A study never serves the third form.
- Study
- The 32 studied pictures come one at a time in a bus window, each for 3,000 ms on the frame clock, with 500 ms of passing scenery between them, or a soft fade when the device asks for less motion. No group comes more than twice in a row, and every group is in each third of the list. The child only watches: nothing is touched and no picture comes twice.
- Test
- After a screen that says the bus stops, each of the 32 pairs shows a studied picture beside a new one of the same group, in an order that does not follow the study order (a rank correlation of .25 or less). The child touches the picture seen. There is no time limit and no feedback: the touched picture is outlined for 300 ms and the next pair comes 500 ms later. The first touch that starts on a picture answers. A touch off both pictures blocks nothing and is counted, a later touch on a picture changes nothing and is counted, and a second finger or a palm on a picture while the first is down leaves the pair out. There is no keyboard route.
- Practice
- A demonstration with an animated hand shows 3 pictures and 1 pair. A practice of 3 pictures and 3 pairs follows, the right picture glowing for 800 ms after each touch, and its first pair holds pictures of two groups. The practice runs once more when fewer than 2 of its 3 pairs are right, and a child with fewer than 2 right again ends there with no test rows, and the session is marked invalid.
- Pictures and touch
- A test card is at least 144 px across and up to 288 px, two side by side with 24 px between, and the photograph on it is 72% of the card, 104 to 207 px. A study card is at least 200 px and up to 288 px, its photograph 144 to 207 px. Each screen is laid out from the stage before it shows. The look and older-adult mode change no picture: older-adult mode makes the words 1.2 times larger and the screen buttons 77 px tall instead of 64 px. The game look adds hills and a road behind the window and the plain look shows the same pictures on white. A stage (the window less the progress bar) under 360 by 292 px holds the game with a line asking for a larger window and takes no answer: a study picture needs a stage of 248 by 292 px and a pair of pictures one of 360 by 236 px.
- Schedule
- Run it as one visit, or add a retest as a visit of the same study 7 to 28 days later: each session takes the next of the 2 study forms by itself. Keep each child on one device class and one input at every visit. Children's forms are never pooled across studies: each study's split-half and checks are counted within that study.
- Scores
- The share of scored pairs answered right (the main score: chance is half and higher is better). The share right in each half of the test and for pictures studied in the first and in the last third come beside it, with the share of answers on the left, the changes of side, the longest run on one side, the usable pairs, the practice rounds, repeat touches, touches off the pictures, pairs with an extra contact and the share of answers by touch. A pair is scored when it was answered with no extra contact, no hidden tab and no resize, and its studied picture showed for at least 2,500 ms. A run with fewer than 24 scored pairs of 32 (9 of 12 on the Try it now form) has no score, and no answer time is scored.
- Devices
- Tablets and computers. A finger, a mouse or a pen answers, so use a tablet or a touch-screen computer for a child. Phones do not run the game, and the plain Recognition Memory Test serves people who use a keyboard only. No answer is timed, so no timing check applies. An adult sits beside the child, reads each screen aloud and neither points nor answers.
The design is the forced-choice recognition test in which a picture seen before is paired with a new one, a choice that leaves out the child's own criterion for saying yes (Rollins and colleagues 2023), with each picture shown for 3 seconds as in the picture recognition test given to children of 4 and 6 (Ngo and colleagues 2018). A new picture is plainly different from the studied one and never a look-alike, so the test asks for recognition and not for telling similar pictures apart. The photographs, the pairs, the groups, the forms, the bus window, the practice, the words and the looks are our own. Older children and adults are likely to score near the top of this test, so the game is made for ages 4 and 5, and age 5 is provisional, expected near the top. A score near the top also lowers the split-half, so read a child's share right beside the other measures. This is a new measure for research with children: its reliability and norms come from use, the figures published in those studies are not its own, and it computes no percentiles. The game keeps no score on its own screens; Try it now shows the share of its 12 pairs answered right when the run ends. For a retest 7 to 28 days later, add a second visit to the same study.
Quality checks
Each session is marked valid, needs review or invalid, with the reason, before you see its scores.
Scored rows: bus_window_game test pairs that were answered, with no second finger or palm on a picture, the tab not hidden, the window not resized, and whose studied picture showed for at least 2,500 ms. A child with no test row after a failed practice (fewer than 2 of 3 pairs right in each of two rounds) cannot be measured and the session is invalid. Fewer scored pairs than 75% of the form's own pairs, rounded up (24 of 32, 9 of 12), invalidates. One side on 81% or more of the answers (26 of 32), a change of side on 90% or more of the steps (28 of 31), 62.5% or fewer right (20 of 32) and a second finger or palm on more than 10% of the pairs are flagged and the score stands. No answer time is judged. A hidden tab withholds the percentile.
Why these rules
The game is a forced choice between two pictures, so a child who does not look is right half the time. Chance is not tested against, because a low share can be a real result in a child with memory difficulty and the answers of a child of 4 or 5 are noisy; instead the pattern of sides is read. The studied picture is on each side equally often and its side never repeats more than 3 times in a row, so one side on 81% or more of the answers (26 of 32) means the pictures were not compared, and a child who touches the sides in turn, changing side on 90% or more of the steps (28 of 31), shows the same. A share right at 62.5% or below (20 of 32 or fewer) is close to chance (a fair guesser reaches 20 or fewer of 32 in about 94% of runs), so it is flagged and the score stands. A share needs most of the form's own pairs, so fewer than 75% of them scored (24 of 32, 9 of 12) is invalid, whatever the form. A child who cannot get 2 of 3 practice pairs right in two rounds has not understood the game, so the run is not a measurement of memory; the number of such children is reported with the study. No time is scored, so no answer-speed rule applies. Thresholds are ConductScience choices, not published ones. A hidden tab lets the study, delay or response timers run without the participant, so a percentile is withheld after one.
No percentile. Scores and quality flags only.
Running this test
Settings you can change
| Setting | Standard | Allowed |
|---|---|---|
| Look | Game | Game, Plain |
Every plan can change the look, which never changes how a result is scored.
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.
- No percentile. Scores and quality flags only. How norms work
- Recruiting online? See the setup guides for Prolific, Sona Systems, CloudResearch Connect and Qualtrics.
Cite and describe
Participants completed the Bus Window Game in a web browser (ConductCognition task version 1.0.0, scoring version 1.0.0; Ngo and colleagues, 2018), with the standard settings (look: Game). The main score was pairs answered right (the share of scored pairs in which the picture seen before was touched).
ConductCognition versions: task version 1.0.0, scoring version 1.0.0. Every export records the versions each session used, for a methods section.
- Ngo, C. T., Newcombe, N. S., & Olson, I. R. (2018). The ontogeny of relational memory and pattern separation. Developmental Science, 21(2), e12556. https://doi.org/10.1111/desc.12556Children aged 4 recognised pictures they had seen for 3 seconds less well than children aged 6 and adults, who did about alike (full text read). The game takes the 3 seconds a picture and the ages 4 and 5, with a plain choice of two pictures in place of that study's judgment task and three buttons.
- Rollins, L., Huffman, D. J., Walters, L. A., & Bennett, K. (2023). Prolonged development of forced-choice recognition when targets are paired with non-corresponding lures. Journal of Experimental Child Psychology, 236, 105742. https://doi.org/10.1016/j.jecp.2023.105742In a forced-choice recognition task whose format pairs a studied picture with a new one, children aged 4 did less well than children aged 6 and adults, who did not differ (the authors' preprint was read, and the published article was not). The game takes that pairing, which leaves out the child's own criterion.
Source papers
Questions
How long does the Bus Window Game take?
About 8 to 11 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?
Pairs Answered Right, Right, First Half of the Pairs, Right, Second Half of the Pairs, Right, Pictures Studied Early, Right, Pictures Studied Late, Left Picture Touched, Side Changes, Longest Run on One Side, Usable Pairs, Practice Rounds Run, Repeat Touches, Touches Off the Pictures, Pairs With an Extra Contact and Answers by Touch. Each session is also marked valid, needs review or invalid, with the reason.
How widely is the Recognition Memory Test used in research?
These counts belong to the Recognition Memory Test, the published test this game is compared with. This game is a new measure, so its own reliability and norms come from use. 642 papers from 2021 to 2025 name the task in their title or abstract, down 3% compared with 2016 to 2020. That ranks it 40 of 711 test families in the ConductCognition test census (use grade B). Source: ConductCognition test census 0.58.0, counted 31 October 2026 from OpenAlex and ClinicalTrials.gov.
How much reliability evidence does the Recognition Memory Test have?
These counts belong to the Recognition Memory Test, the published test this game is compared with. This game is a new measure, so its own reliability and norms come from use. In the published research, 17 papers report the task's test-retest reliability and 45 registered clinical studies use it as an outcome (evidence grade B in the ConductCognition test census).
Does it report a percentile?
No. Scores and quality flags only.
Which settings can a researcher change?
Look (Game, Plain). Every plan can change the look, which never changes how a result is scored.
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.
Related tests