Attention
Composite Face Task
Also called: composite face paradigm, composite face illusion, face composite effect, holistic face processing task
Measure holistic face processing: two faces in a row, each built from two people, and a same or different judgment of the top halves only, aligned or shifted.
8-10 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
- 214
- Evidence grade
- C
- Registered clinical studies
- 3
What it measures
Measures holistic face processing. Participants judge whether the top halves of two faces match, ignoring the bottom halves, when aligned or shifted apart.
Constructs
- Holistic face processing
- Face perception
- Selective attention to face parts
Research uses
- Face perception research
- Individual differences in holistic processing
- Within-study comparison of composite effects
How the paradigm works
The composite face task measures holistic face processing: how strongly people see a face as a whole even when asked to attend to one part of it. Each trial shows two faces one after the other, each made of the top half of one person and the bottom half of another, with a thin light line between them. Participants decide whether the top halves show the same person and ignore the bottom halves, pressing F for same and J for different. The first face stays for 250 ms; after a 250 ms blank, the second stays until the answer or 2 seconds. The first face is always aligned. The second is aligned, or its halves are shifted apart by half a face width. The bottom halves are the same or different regardless of the tops, so on congruent trials they give the same answer as the tops and on incongruent trials the opposite one. Practice gives feedback; the test gives none.
More about the paradigm
The design follows Young, Hellawell and Hay (1987) and the complete design of Richler, Cheung and Gauthier (2011), with 160 trials across 8 conditions and the timing of Gao et al. (2011). The photos come from the Face Research Lab London Set (CC BY 4.0): grayscale neutral front views of 20 white adults, 10 women and 10 men, cropped to an oval that hides hair and clothing, with both halves from the same gender. Settings can run the partial design, which shows only different bottoms (Rossion, 2013), and add upside-down trials, on which the target is still the half with the eyes.
The main score is the composite effect: the congruency effect in d prime on aligned trials minus the same effect on misaligned trials (Richler and Gauthier, 2014). A larger effect means the bottom halves were harder to ignore when aligned; it is not better or worse performance. Scores support research and do not diagnose.
Scores
- Composite Effectd'
- Congruency effect in d prime on aligned trials minus the same effect on misaligned trials. Blank in the partial design.
- Larger means the bottom halves were harder to ignore when aligned; not better or worse
- Congruency Effect, Alignedd'
- d prime on congruent trials minus d prime on incongruent trials, aligned second face.
- Larger means more influence of the bottom halves
- Congruency Effect, Misalignedd'
- d prime on congruent trials minus d prime on incongruent trials, misaligned second face.
- Larger means more influence of the bottom halves
- Partial Composite Effect%
- On trials with the same top and a different bottom, the share answered same when misaligned minus when aligned.
- Larger means aligned bottoms made the same top look different more often
- d Prime, Alignedd'
- Sensitivity to the top half over all aligned trials.
- Higher means better judgment of the top halves
- d Prime, Misalignedd'
- Sensitivity to the top half over all misaligned trials.
- Higher means better judgment of the top halves
- Accuracy%
- Share of all trials answered correctly; a missed answer counts as wrong.
- Higher is better; chance is 0.5
- Mean Correct Timems
- Mean time of correct answers from the onset of the second face; answers under 100 ms are left out.
- Context for accuracy
- Composite Effect in Timems
- Slowing on incongruent over congruent trials when aligned minus the same slowing when misaligned.
- Larger means more slowing from aligned bottom halves; not better or worse
- Inversion Effectd'
- Composite effect on upright trials minus the composite effect on upside-down trials. Blank unless upside-down trials are on.
- Larger means the composite effect depends more on upright faces
Use and evidence
214 papers from 2021 to 2025 name this task in their title or abstract, down 26% compared with 2016 to 2020. That ranks it 119 of 395 test families in our census. See the counts for every test.
Show the numbers
| Year | Papers |
|---|---|
| 2025 | 42 |
| 2024 | 34 |
| 2023 | 41 |
| 2022 | 40 |
| 2021 | 57 |
| 2020 | 49 |
| 2019 | 56 |
| 2018 | 63 |
| 2017 | 50 |
| 2016 | 71 |
| 2015 | 56 |
| 2014 | 51 |
| 2013 | 62 |
| 2012 | 70 |
| 2011 | 69 |
| 2010 | 101 |
| 2009 | 45 |
| 2008 | 27 |
| 2007 | 19 |
| 2006 | 31 |
| 2005 | 19 |
| 2004 | 19 |
| 2003 | 15 |
| 2002 | 9 |
| 2001 | 4 |
| 2000 | 7 |
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
- 1
- Papers on reliability or validity
- 40
- Registered clinical studies using it as an outcome
- 3
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.
- Faces
- Grayscale neutral front views of 20 adults, 10 women and 10 men, all white, from the Face Research Lab London Set, cropped to an oval that hides hair and clothing.
- Composites
- The top half of one person above the bottom half of another of the same gender, cut halfway between the pupils and the base of the nose, with a 3-pixel light line between them.
- Trial
- Fixation 500 ms, first face 250 ms (standard; 100 to 1,000), blank 250 ms, then the second face until a response or 2,000 ms, with 700 to 900 ms between trials.
- Alignment
- The first face is always aligned. The second is aligned, or misaligned by half a face width in all (105 pixels), the top half to the left and the bottom half to the right.
- Conditions
- 20 trials (standard; 10 to 40) in each of 8 conditions: same or different top, same or different bottom, aligned or misaligned. That makes 160 trials, half with women and half with men, with no more than 4 of the same right answer in a row and a rest after every 80.
- Keys
- F for the same top half and J for a different one.
- Settings
- The partial design keeps only different bottoms (80 trials at the standard count). Upside-down trials add the same number again, mixed in, with the target still the half with the eyes.
- Practice
- 8 trials with feedback, with faces the test does not use.
The design follows Young, Hellawell and Hay (1987) and the complete design of Richler, Cheung and Gauthier (2011), with the display times of Gao et al. (2011). The photos come from the Face Research Lab London Set (CC BY 4.0).
Quality checks
Each session is marked valid, needs review or invalid, with the reason, before you see its scores.
Scored rows: composite_face_test from the scored run, F for the same top half and J for a different one (chance 50%, half the trials have the same top) within 2,000 ms. Below chance invalidates and not above chance is flagged. One key on 90% of answers invalidates when accuracy is not above chance. Missed responses and responses under 100 ms follow the timed-test rule. A mean first-face timing error above 20 ms 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). Missed responses and responses under 100 ms keep the platform's timed-test rule: more than max(3, 25%) missed responses or more than max(2, 20%) responses under 100 ms invalidate, and any is flagged. A response under 100 ms cannot be a reaction to the stimulus. The composite effect compares judgments of faces seen for a moment, so a first face that could not hold its planned time changes the measure.
No percentile. Scores and quality flags only.
Running this test
Settings you can change
| Setting | Standard | Allowed |
|---|---|---|
| Trials per condition | 20 | 10 to 40 |
| First face (ms) | 250 | 100 to 1,000 |
| Design | Complete (standard) | Complete (standard), Partial |
| Add upside-down trials | Off | On or off |
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.
- 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 Composite Face Task in a web browser (ConductCognition task version 1.0.0, scoring version 1.0.0; Young, Hellawell and Hay, 1987; Richler, Cheung and Gauthier, 2011), with the standard settings (trials per condition: 20; first face (ms): 250; design: Complete; add upside-down trials: Off). The main score was composite effect (congruency effect in d prime on aligned trials minus the same effect on misaligned trials).
ConductCognition versions: task version 1.0.0, scoring version 1.0.0. Every export records the versions each session used, for a methods section.
- Young, A. W., Hellawell, D., & Hay, D. C. (1987). Configurational information in face perception. Perception, 16(6), 747-759. https://doi.org/10.1068/p160747
- Richler, J. J., Cheung, O. S., & Gauthier, I. (2011). Holistic processing predicts face recognition. Psychological Science, 22(4), 464-471. https://doi.org/10.1177/0956797611401753
- Gao, Z., Flevaris, A. V., Robertson, L. C., & Bentin, S. (2011). Priming global and local processing of composite faces: Revisiting the processing-bias effect on face perception. Attention, Perception, & Psychophysics, 73(5), 1477-1486. https://doi.org/10.3758/s13414-011-0109-7
- Richler, J. J., & Gauthier, I. (2014). A meta-analysis and review of holistic face processing. Psychological Bulletin, 140(5), 1281-1302. https://doi.org/10.1037/a0037004
- Rossion, B. (2013). The composite face illusion: A whole window into our understanding of holistic face perception. Visual Cognition, 21(2), 139-253. https://doi.org/10.1080/13506285.2013.772929
- DeBruine, L., & Jones, B. (2017). Face Research Lab London Set [Data set]. figshare. https://doi.org/10.6084/m9.figshare.5047666.v5
Source papers
- Young AW, Hellawell D, Hay DC (1987). Configurational information in face perception. Perception, 16(6):747-759.Paradigm source, Adults
- Richler JJ, Cheung OS, Gauthier I (2011). Holistic processing predicts face recognition. Psychological Science, 22(4):464-471.Paradigm source, Adults
- Gao Z, Flevaris AV, Robertson LC, Bentin S (2011). Priming global and local processing of composite faces: revisiting the processing-bias effect on face perception. Attention, Perception, & Psychophysics, 73(5):1477-1486.Paradigm source, Adults
- Richler JJ, Gauthier I (2014). A meta-analysis and review of holistic face processing. Psychological Bulletin, 140(5):1281-1302.Paradigm source, Adults
- Rossion B (2013). The composite face illusion: A whole window into our understanding of holistic face perception. Visual Cognition, 21(2):139-253.Paradigm source, Adults
- 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 Composite Face Task take?
About 8 to 10 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?
Composite Effect, Congruency Effect, Aligned, Congruency Effect, Misaligned, Partial Composite Effect, d Prime, Aligned, d Prime, Misaligned, Accuracy, Mean Correct Time, Composite Effect in Time and Inversion Effect. Each session is also marked valid, needs review or invalid, with the reason.
How widely is the Composite Face Task used in research?
214 papers from 2021 to 2025 name it in their title or abstract, down 26% compared with 2016 to 2020. That ranks it 119 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 Composite Face Task have?
In the published research, 1 paper reports its test-retest reliability and 3 registered clinical studies use it as an outcome (evidence grade C in the ConductCognition test census).
Does it report a percentile?
No. Scores and quality flags only.
Which settings can a researcher change?
Trials per condition (10 to 40), first face (ms) (100 to 1,000), design (Complete (standard), Partial) and add upside-down trials (On or off). 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.