Attention
Change Blindness Task
Also called: flicker paradigm, flicker task, flicker change detection task, change blindness flicker task
Measure how long it takes to notice a change: a scene and a changed version take turns with brief blanks, and participants click the change once they find it.
5-8 minTimed: desktop or laptop with a mouse or trackpad
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
- 168
- Evidence grade
- C
- Registered clinical studies
- 1
What it measures
A measure of how long it takes to notice a change in a scene. A scene of 30 small pictures and a second version, in which one picture disappears, changes color or moves, take turns with a short blank between them, so the change does not stand out. Participants click the change as soon as they find it. The score is the median time to find it.
Constructs
- Visual attention
- Change detection
- Scene perception
- Visual short-term memory
Research uses
- Attention research
- Aging research
- Within-study group comparison
How the paradigm works
The change blindness task measures how long it takes to notice a change in a scene when nothing draws the eye to it. It uses the flicker paradigm of Rensink, O'Regan and Clark (1997): a scene and a changed version take turns, each shown for 240 ms and followed by an 80 ms blank, and each version is shown twice before the other. The blanks hide the local motion that would otherwise point to the change, so the change is found only when attention reaches the changed object. Participants click the change as soon as they find it.
More about the paradigm
In place of the photographs of the original studies, each scene holds 30 small pictures on a grid of 8 by 5 places, arranged by ConductCognition's code from Noto Emoji pictures, as Rensink (2000) used arrays of items on a jittered grid. The blank is the panel's own light gray, so only the pictures go off and on. In each scene one picture disappears, changes color or moves to the next place, Rensink and colleagues' three kinds of change. The standard session has 3 practice scenes, after which the change is circled, then 24 test scenes, 8 of each kind, in an order drawn from the session's seed, with the changed picture spread evenly over the four quarters of the panel. A scene runs for up to 60 seconds (30 to 60 as a setting). The scenes are the same in every session.
The main score is the median time to find the change, counting a scene not found, by a click elsewhere or by the time limit, at the time limit. The median over scenes found, the share found, the share of clicks on the change, scenes timed out, clicks away from the change and the median for each kind of change are also reported. Clicks before the first change, or within 100 ms of it, are counted as anticipations and not taken. Displays are timed to screen frames, and the share of scenes whose blanks all ran their full length is reported, since a blank cut short can show the change. Scores support research and do not diagnose.
Scores
- Median Detection Timems
- Median time from the first version's onset to a click on the change, over all test scenes; a scene not found counts at the time limit.
- Lower is better
- Median Time, Changes Foundms
- Median time to click the change, over scenes in which it was found.
- Lower is better
- Changes Found%
- Share of test scenes in which the change was clicked within the time limit.
- Higher is better
- Location Accuracy%
- Share of clicks within 40 px of the changed picture's center, either place for a move.
- Higher is better; under 0.75 is flagged for review
- Scenes Timed Outcount
- Test scenes with no click before the time limit.
- Lower is better
- Clicks Away from the Changecount
- Test scenes ended by a click away from the change.
- Lower is better
- Median Detection Time, Presencems
- The main score over scenes in which a picture disappears and comes back.
- Lower is better
- Median Detection Time, Colorms
- The main score over scenes in which a picture changes color.
- Lower is better
- Median Detection Time, Locationms
- The main score over scenes in which a picture moves to the next place.
- Lower is better
- Anticipationscount
- Clicks before the first change or under 100 ms after it. They are not taken as answers.
- Lower is better
- Frame Intervalms
- Median time between screen frames on the participant's device.
- Informational
- Displays on Time%
- Share of test scenes whose blanks each lasted at least 80 ms less one frame.
- Below 0.9 is flagged for review
- Planned Scenestrials
- Test scenes the settings planned.
- Informational
- Total Scenestrials
- Test scenes shown in the scored run.
- Fewer than planned is flagged for review
Use and evidence
168 papers from 2021 to 2025 name this task in their title or abstract, down 12% compared with 2016 to 2020. That ranks it 141 of 395 test families in our census. See the counts for every test.
Show the numbers
| Year | Papers |
|---|---|
| 2025 | 32 |
| 2024 | 40 |
| 2023 | 25 |
| 2022 | 32 |
| 2021 | 39 |
| 2020 | 22 |
| 2019 | 35 |
| 2018 | 43 |
| 2017 | 49 |
| 2016 | 41 |
| 2015 | 46 |
| 2014 | 47 |
| 2013 | 33 |
| 2012 | 50 |
| 2011 | 48 |
| 2010 | 117 |
| 2009 | 49 |
| 2008 | 45 |
| 2007 | 46 |
| 2006 | 42 |
| 2005 | 42 |
| 2004 | 39 |
| 2003 | 36 |
| 2002 | 30 |
| 2001 | 28 |
| 2000 | 39 |
Use grade C
100 to 299 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
- 2
- Papers on reliability or validity
- 28
- Registered clinical studies using it as an outcome
- 1
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.
- Display
- 30 pictures 56 px wide on a light gray panel of 720 by 450 px, one to a place on a grid of 8 by 5 places 90 px apart, each up to 12 px off the center of its place.
- Flicker
- After a 1,000 ms empty panel, each version for 240 ms, then the empty panel for 80 ms; each version shows twice before the other, until a click or the time limit. Displays switch on screen frames.
- Changes
- One picture disappears, changes color (its hues turned by 120, 180 or 240 degrees) or moves to the next place: 8 scenes of each kind, with the changed picture twice in each quarter of the panel for each kind.
- Answer
- A click within 40 px of the changed picture's center, in either place for a move, finds the change. The first click ends the scene; clicks before the first change or within 100 ms of it are counted and not taken.
- Length
- 24 scenes (12 to 24) in a seeded order, each running up to 60 seconds (30 to 60).
- Practice
- 3 scenes, one of each kind, each followed by the change circled for 2,500 ms.
- Scores
- The median time to find the change, with a scene not found counted at the time limit, then the median over changes found, the share found, the share of clicks on the change and the median for each kind of change.
Rensink, O'Regan and Clark (1997) used photographs of real scenes; this version uses scenes of small pictures arranged by code, so its times are not comparable with theirs. The blank is the panel's own light gray, so only the small pictures go off and on, about three times a second.
Quality checks
Each session is marked valid, needs review or invalid, with the reason, before you see its scores.
Scored rows: change_blindness_test from the scored run, one per scene. A click is on the change when it lands within 40 px of the changed picture's center (either place for a move); a click at random does so about 1.6% of the time (3.1% for a move), so clicks on the change no more often than that are flagged and fewer than that invalidate. Clicks in one quarter of the panel on 90% of scenes invalidate when they are not on the change more often than chance; the changes are spread evenly over the four quarters. Not finding a change is what the test measures, so scenes that run out of time never invalidate on their own, a session with no clicks is invalid, and an unbroken run of timed-out scenes covering half the scenes (at least 8) is flagged and its percentile withheld. Clicks before the first change, or within 100 ms of it, are not taken as answers. Under 75% of at least 4 clicks on the change, under 90% of scenes with every blank on time, or a run short of its planned scenes is flagged and its percentile withheld.
Why these rules
Accuracy significantly below chance (one-sided exact test, p < .05) cannot come from guessing; it means the response mapping was reversed or the task was not followed (Slick, Sherman and Iverson 1999 treat below-chance forced-choice performance as definite response bias). Chance-level accuracy can be a real result in impaired participants, so it is only flagged. One option on at least 90% of answered trials, when the correct options are balanced and accuracy is not above chance, means the answers did not depend on the items (the task counterpart of long-string responding in surveys; Curran 2016). Missed targets are what this test measures, so missed responses never invalidate on their own. A session with no responses is invalid. An unbroken run of misses covering at least half the response trials (minimum 8) may mean the participant stopped, so the session is flagged and its percentile withheld. The 80 ms blank is what hides the change: a blank cut short by a slow screen shows the change as motion, so scenes whose blanks ran short change what was measured.
Running this test
Settings you can change
| Setting | Standard | Allowed |
|---|---|---|
| Scenes | 24 | 12 to 24 |
| Time limit per scene (seconds) | 60 | 30 to 60 |
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
- Timed test: participants need a desktop or laptop with a mouse or trackpad. 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 Change Blindness Task in a web browser (ConductCognition task version 1.0.0, scoring version 1.0.0; Rensink, O'Regan and Clark, 1997), with the standard settings (scenes: 24; time limit per scene (seconds): 60). The main score was median detection time (median time from the first version's onset to a click on the change, over all test scenes; a scene not found counts at the time limit).
ConductCognition versions: task version 1.0.0, scoring version 1.0.0. Every export records the versions each session used, for a methods section.
- Rensink, R. A., O'Regan, J. K., & Clark, J. J. (1997). To see or not to see: The need for attention to perceive changes in scenes. Psychological Science, 8(5), 368-373. https://doi.org/10.1111/j.1467-9280.1997.tb00427.x
- Rensink, R. A. (2000). Visual search for change: A probe into the nature of attentional processing. Visual Cognition, 7(1-3), 345-376. https://doi.org/10.1080/135062800394847Flicker displays of arrays on a jittered grid, with blanks of the background color.
- Andermane, N., Bosten, J. M., Seth, A. K., & Ward, J. (2019). Individual differences in change blindness are predicted by the strength and stability of visual representations. Neuroscience of Consciousness, 2019(1), niy010. https://doi.org/10.1093/nc/niy010A click on the changed object as the answer, and test-retest reliability.
Source papers
- Rensink, O'Regan and Clark (1997): the flicker paradigm and its three kinds of change. Rensink (2000): flicker displays of arrays on a jittered grid. Andermane, Bosten, Seth and Ward (2019): a flicker task answered by clicking the changed object, with its test-retest reliability.Paradigm source, Adults
Questions
How long does the Change Blindness Task take?
About 5 to 8 minutes, as listed in the test library.
What device does a participant need?
Timed test: participants need a desktop or laptop with a mouse or trackpad. The browser must be Chrome, Edge, Firefox or Safari.
What scores does it produce?
Median Detection Time, Median Time, Changes Found, Changes Found, Location Accuracy, Scenes Timed Out, Clicks Away from the Change, Median Detection Time, Presence, Median Detection Time, Color, Median Detection Time, Location, Anticipations, Frame Interval, Displays on Time, Planned Scenes and Total Scenes. Each session is also marked valid, needs review or invalid, with the reason.
How widely is the Change Blindness Task used in research?
168 papers from 2021 to 2025 name it in their title or abstract, down 12% compared with 2016 to 2020. That ranks it 141 of 395 test families in the ConductCognition test census (use grade C). Source: ConductCognition test census 0.14.0, counted 31 October 2026 from OpenAlex and ClinicalTrials.gov.
How much reliability evidence does the Change Blindness Task have?
In the published research, 2 papers report its test-retest reliability and 1 registered clinical study uses it as an outcome (evidence grade C in the ConductCognition test census).
Which settings can a researcher change?
Scenes (12 to 24) and time limit per scene (seconds) (30 to 60). 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.