Attention
Face Matching Task
Also called: unfamiliar face matching task, face identity matching task, same or different face matching
Measure unfamiliar face matching: participants judge whether two photos side by side show the same person or two different people who look alike.
7-10 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
- 143
- Evidence grade
- C
- Registered clinical studies
- 8
What it measures
Measures how well people match unfamiliar faces. Each screen shows two photos side by side, and participants decide whether they show the same person or two different people who look alike, with no time limit. The main score is the percent of pairs judged correctly, with accuracy on same-person and different-person pairs, d prime and response time.
Constructs
- Unfamiliar face matching
- Face perception
- Face identity discrimination
Research uses
- Individual differences in face perception, from poor to exceptional matchers
- Face perception in aging and in neurological and developmental conditions, at the group level
- Applied research on photo identity checks
How the paradigm works
The face matching task measures how well people tell whether two photos of unfamiliar faces show the same person. Matching unfamiliar faces is harder than it looks: with simple pairs of good photos, people make 10% to 25% errors depending on the photos, while familiar faces are matched almost perfectly (Burton, White and McNeill, 2010). Accuracy on same-person pairs and accuracy on different-person pairs do not go together for unfamiliar faces, so the two are reported apart (Megreya and Burton, 2007). The skill matters wherever photo identity is checked; trained passport officers make errors too (White and colleagues, 2014).
More about the paradigm
Each screen shows two face photos side by side, and the participant answers whether they show the same person or two different people, with no time limit, by click, tap or the F and J keys. Same-person pairs show two different photos of one person, one neutral and one smiling. Different-person pairs show two people of the same gender and ethnic group who look alike, chosen by a computed similarity of their photos and their ages; one is neutral and the other smiling. The photos come from the Face Research Lab London Set (DeBruine and Jones, 2017), aligned on the eyes, in grayscale and cropped to an oval that hides hair, ears and clothing, so only the faces can be compared. A session has 80 pairs (40 in the short setting), half of each kind, in an order drawn for each session; in a full session each photo appears twice. Four practice pairs with feedback come first.
The main score is the percent of pairs judged correctly. The task also reports accuracy on same-person and on different-person pairs, d prime and the response criterion from signal detection, and median response times. Presses within 100 ms of the photos are counted as anticipations and not taken as answers. Scores support research and do not diagnose.
Scores
- Percent Correct%
- Pairs judged correctly, as a percent of the pairs answered (chance is 50%).
- Higher is better
- Same-Person Accuracy%
- Percent of same-person pairs answered Same person.
- Higher is better
- Different-Person Accuracy%
- Percent of different-person pairs answered Two different people.
- Higher is better
- d Primed'
- Sensitivity: z of the rate of Same person answers on same-person pairs minus z of that rate on different-person pairs, both log-linear corrected.
- Higher is better
- Response Criterionz
- Minus half the sum of the two z values behind d prime. Negative values mean a leaning to answer Same person.
- Informational
- Median Response Timems
- Median time from the photos appearing to the answer, over all answered pairs.
- Informational
- Median Response Time, Same Personms
- Median response time on same-person pairs.
- Informational
- Median Response Time, Different Peoplems
- Median response time on different-person pairs.
- Informational
- Pairs Correctcount
- Pairs judged correctly.
- Higher is better
- Pairs Answeredcount
- Pairs answered in the scored run.
- Quality check
- Planned Pairscount
- Pairs the session planned: 80, or 40 in the short setting.
- Informational
- Anticipationscount
- Presses within 100 ms of the photos appearing, not taken as answers.
- Quality check
Use and evidence
143 papers from 2021 to 2025 name this task in their title or abstract, up 17% compared with 2016 to 2020. That ranks it 155 of 395 test families in our census. See the counts for every test.
Show the numbers
| Year | Papers |
|---|---|
| 2025 | 25 |
| 2024 | 29 |
| 2023 | 22 |
| 2022 | 32 |
| 2021 | 35 |
| 2020 | 28 |
| 2019 | 23 |
| 2018 | 27 |
| 2017 | 24 |
| 2016 | 20 |
| 2015 | 25 |
| 2014 | 9 |
| 2013 | 17 |
| 2012 | 13 |
| 2011 | 15 |
| 2010 | 30 |
| 2009 | 10 |
| 2008 | 7 |
| 2007 | 6 |
| 2006 | 6 |
| 2005 | 3 |
| 2004 | 3 |
| 2003 | 5 |
| 2002 | 1 |
| 2001 | 0 |
| 2000 | 2 |
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
- 6
- Papers on reliability or validity
- 36
- Registered clinical studies using it as an outcome
- 8
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.
- Photos
- Grayscale front views of 42 adults from the Face Research Lab London Set, aligned on the eyes and cropped to an oval that hides hair, ears and clothing, up to 280 by 320 px each, side by side on a black panel.
- Same-person pairs
- Two different photos of one person: one neutral and one smiling. 40 people, each in one pair.
- Different-person pairs
- Two people of the same gender and ethnic group who look alike, by a computed similarity of their photos and their ages: one neutral and one smiling. Each person is paired with two others in cycles of 3 to 5, so no two people meet twice; 40 pairs.
- Answer
- Same person or Two different people, by click, tap, or F and J, with no time limit. Presses within 100 ms of the photos are counted and not taken as answers. A 500 ms blank follows each pair.
- Pairs
- 80 (standard; 40), half of each kind, in an order drawn for each session with no person in two pairs in a row and no more than 4 of one answer in a row. Each half of the order holds 20 pairs of each kind, and the 40 setting runs the first half. A rest after 40 pairs.
- Practice
- 4 pairs of two people the test does not use, with feedback for 1,500 ms.
The format follows the unfamiliar face matching literature (Bruce et al., 1999; Megreya and Burton, 2007; Burton, White and McNeill, 2010), with photos and pairs of our own: every photo appears twice in a full session, and no screen names or labels a face. The photos come from the Face Research Lab London Set (CC BY 4.0) and hide hair, so only the faces can be compared. They include 21 of the Face Memory Test's 24 distractor faces and 17 of the Face-name Memory Task's 24 foils, so in a study with either, run face matching after them or in a separate session.
Quality checks
Each session is marked valid, needs review or invalid, with the reason, before you see its scores.
Scored rows: face_matching_test from the scored run, one same-or-different answer per pair of photos with no time limit (chance 1 in 2). Accuracy significantly below chance invalidates and not above chance is flagged. One answer on 90% of pairs invalidates when accuracy is not above chance; half the pairs show one person and half two people. Answers under 200 ms on at least half the pairs invalidate. A run short of its planned pairs is flagged for review.
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). An answer under 200 ms is faster than a simple visual reaction, so it cannot reflect the item. Following response-time effort screening for untimed tests (Wise and Kong 2005), the session is invalid only when at least half the answers are that fast.
Running this test
Settings you can change
| Setting | Standard | Allowed |
|---|---|---|
| Number of pairs | 80 (standard) | 80 (standard), 40 |
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 Face Matching Task in a web browser (ConductCognition task version 1.0.0, scoring version 1.0.0; Bruce et al., 1999; Megreya & Burton, 2007; Burton, White & McNeill, 2010), with the standard settings (number of pairs: 80). The main score was percent correct (pairs judged correctly, as a percent of the pairs answered (chance is 50%)).
ConductCognition versions: task version 1.0.0, scoring version 1.0.0. Every export records the versions each session used, for a methods section.
- Burton, A. M., White, D., & McNeill, A. (2010). The Glasgow Face Matching Test. Behavior Research Methods, 42(1), 286-291. https://doi.org/10.3758/BRM.42.1.286
- Megreya, A. M., & Burton, A. M. (2007). Hits and false positives in face matching: A familiarity-based dissociation. Perception & Psychophysics, 69(7), 1175-1184. https://doi.org/10.3758/BF03193954Reporting accuracy on same-person and different-person pairs apart; 84 pairs shown until a response.
- Bruce, V., Henderson, Z., Greenwood, K., Hancock, P. J. B., Burton, A. M., & Miller, P. (1999). Verification of face identities from images captured on video. Journal of Experimental Psychology: Applied, 5(4), 339-360. https://doi.org/10.1037/1076-898X.5.4.339
- White, D., Kemp, R. I., Jenkins, R., Matheson, M., & Burton, A. M. (2014). Passport officers' errors in face matching. PLoS ONE, 9(8), e103510. https://doi.org/10.1371/journal.pone.0103510Errors in photo identity checks by trained officers.
- Hautus, M. J. (1995). Corrections for extreme proportions and their biasing effects on estimated values of d'. Behavior Research Methods, Instruments, & Computers, 27(1), 46-51. https://doi.org/10.3758/BF03203619The log-linear correction behind d prime.
- Stanislaw, H., & Todorov, N. (1999). Calculation of signal detection theory measures. Behavior Research Methods, Instruments, & Computers, 31(1), 137-149. https://doi.org/10.3758/BF03207704d prime and the response criterion.
- DeBruine, L., & Jones, B. (2017). Face Research Lab London Set [Data set]. figshare. https://doi.org/10.6084/m9.figshare.5047666.v5
Source papers
- Burton AM, White D, McNeill A (2010). The Glasgow Face Matching Test. Behavior Research Methods, 42(1):286-291.N 300, 18-80
- DeBruine L, Jones B (2017). Face Research Lab London Set. figshare. Dataset. doi:10.6084/m9.figshare.5047666.v5Paradigm source, Adults
Questions
How long does the Face Matching Task take?
About 7 to 10 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?
Percent Correct, Same-Person Accuracy, Different-Person Accuracy, d Prime, Response Criterion, Median Response Time, Median Response Time, Same Person, Median Response Time, Different People, Pairs Correct, Pairs Answered, Planned Pairs and Anticipations. Each session is also marked valid, needs review or invalid, with the reason.
How widely is the Face Matching Task used in research?
143 papers from 2021 to 2025 name it in their title or abstract, up 17% compared with 2016 to 2020. That ranks it 155 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 Face Matching Task have?
In the published research, 6 papers report its test-retest reliability and 8 registered clinical studies use it as an outcome (evidence grade C in the ConductCognition test census).
Which settings can a researcher change?
Number of pairs (80 (standard), 40). 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.