Social Cognition
Face Impressions Task
Also called: face trait judgments, facial first impressions, trustworthiness and dominance ratings
Measure first impressions of faces: participants rate how trustworthy and dominant each face looks, then rate it again. Compare agreement and consistency between repeated ratings.
6-9 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
- 226
- Evidence grade
- C
- Registered clinical studies
- 2
What it measures
Measures first impressions of faces. Participants rate how trustworthy and dominant each face looks, then rate it again. Scores show agreement and consistency between repeated ratings.
Constructs
- Face trait impressions
- Repeat-rating agreement
- Impression consistency
Research uses
- Social perception research
- First impression research
How the paradigm works
People form shared first impressions of faces along trustworthiness and dominance. This task measures those impressions and how stable they are when the same faces appear again. It uses 24 London Set neutral front photographs, or a fixed short form of 14, rated on a 0 to 100 bar with no initial choice and no deadline. Each trait has its own block in each of two passes. The session seed sets trait order and shuffles each block. Two separate practice faces precede the test.
More about the paradigm
Repeat agreement is 100 minus the mean absolute rating change. Trait-specific Pearson correlations measure consistency across faces. First-pass means describe the impressions given. Ratings chosen under 100 ms are counted as anticipations and excluded. There is no correct answer: these judgments describe first impressions, not the people pictured. For agreement between people, compare first-pass ratings of the same faces across participants. Trial exports identify the face, trait and pass for these comparisons. The session score is repeat agreement, not agreement with a reference population.
The source used 66 photographs, three presentations and a 1 to 9 scale. This shorter adaptation uses London Set photographs, two passes and the library’s rating bar. Norms are off. Scores support research and do not diagnose.
Scores
- Repeat agreementpoints
- 100 minus the mean absolute change between repeated ratings of the same face and trait.
- Informational
- Trustworthiness consistencyscore
- Pearson correlation between the two trustworthiness ratings across faces. Null with fewer than three pairs or no variation.
- Informational
- Dominance consistencyscore
- Pearson correlation between the two dominance ratings across faces. Null with fewer than three pairs or no variation.
- Informational
- Mean trustworthiness impressionpoints
- Mean first-pass trustworthiness rating, from 0 to 100.
- Informational
- Mean dominance impressionpoints
- Mean first-pass dominance rating, from 0 to 100.
- Informational
- Trustworthiness repeat agreementpoints
- 100 minus the mean absolute change in trustworthiness ratings.
- Informational
- Dominance repeat agreementpoints
- 100 minus the mean absolute change in dominance ratings.
- Informational
- Complete repeat pairscount
- Face and trait pairs with two usable ratings.
- Informational
- Usable ratingscount
- Ratings in the scored run, excluding choices under 100 ms and missing or out-of-range answers.
- Informational
- Planned ratingscount
- Number of ratings planned for this run.
- Informational
- Median rating timems
- Median time from picture onset to confirmation, for usable ratings.
- Informational
- Anticipationscount
- Ratings first chosen under 100 ms, excluded from the scores.
- Informational
Use and evidence
226 papers from 2021 to 2025 name this task in their title or abstract, up 32% compared with 2016 to 2020. That ranks it 135 of 582 test families in our census. See the counts for every test.
Show the numbers
| Year | Papers |
|---|---|
| 2025 | 41 |
| 2024 | 55 |
| 2023 | 47 |
| 2022 | 44 |
| 2021 | 39 |
| 2020 | 45 |
| 2019 | 37 |
| 2018 | 29 |
| 2017 | 29 |
| 2016 | 31 |
| 2015 | 14 |
| 2014 | 22 |
| 2013 | 21 |
| 2012 | 20 |
| 2011 | 17 |
| 2010 | 11 |
| 2009 | 8 |
| 2008 | 8 |
| 2007 | 2 |
| 2006 | 0 |
| 2005 | 2 |
| 2004 | 1 |
| 2003 | 2 |
| 2002 | 2 |
| 2001 | 1 |
| 2000 | 0 |
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
- 3
- Papers on reliability or validity
- 31
- Registered clinical studies using it as an outcome
- 2
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.
Note on the counts: includes economics studies of trust in faces.
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 0.39.0, counted 8 November 2026 from OpenAlex and ClinicalTrials.gov.
Task design
The trial sequence as the test runs in ConductCognition.
- Faces
- 24 neutral front photographs from the London Set, or a fixed subset of 14. All are grayscale, cropped to an oval and matched in luminance and contrast. One face is shown at a time.
- Ratings
- 96 ratings in the standard form, or 56 in the short form: each face twice for trustworthiness and twice for dominance.
- Order
- Four blocks: one per trait in each of two passes. The session seed sets which trait comes first and shuffles the faces separately in every block.
- Answer
- A 0 to 100 bar from Not at all to Extremely, with no numbers shown and no initial choice. Click, tap or use the arrow keys, then confirm. The picture stays visible; there is no deadline.
- Timing
- A 250 ms confirmation pause and a 500 ms blank between pictures. Response time is descriptive.
- Practice
- 2 separate faces, one rating for each trait, with feedback that the rating was recorded.
- Scoring
- Repeat agreement is 100 minus the mean absolute change between repeat ratings. Consistency is Pearson r for each trait. Mean ratings use the first pass. Choices under 100 ms are excluded.
- Agreement between people
- Compare first-pass ratings of the same faces across participants in your study. Trial exports identify the face, trait and pass for these comparisons. Session results report repeat agreement.
These judgments describe first impressions, not the people pictured. No answer is right or wrong. The task uses London Set photographs and a 0 to 100 bar instead of the source’s photographs and 1 to 9 scale. Norms are off. Scores support research and do not diagnose. Counting the 24 test and 2 practice people, this task shares people with Face Memory Test (12), Face-Name Memory Task (10), Face Matching Task (26), Chimeric Faces Task (7), Part-Whole Face Task (13), Thatcher Illusion Task (8), Two-Tone Pictures Task (11), Verbal Overshadowing Task (17). Counts include the other tasks' practice pictures and foils. Use separate sessions or counterbalance task order when combining these tests. The Reverse Correlation Task uses an average face, not an individual photograph. Including practice pictures and foils, this task shares people with Real or AI Face Task (26), Uncanny Valley Task (7).
Quality checks
Each session is marked valid, needs review or invalid, with the reason, before you see its scores.
Scored rows: face_impressions_rating from the run with the most ratings, with duplicate stimulus occurrences counted once. Ratings have no correct answer. Repeated answers are accepted. Choosing a point under 200 ms on at least half the trials invalidates; incomplete runs are flagged. Ratings chosen under 100 ms are counted as anticipations and excluded from the scores.
Why these rules
There is no correct option to compare with chance, so accuracy is not judged against chance. Ratings describe the participant's impressions, so neither low nor repeated ratings prove an error. 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.
No percentile. Scores and quality flags only.
Running this test
Settings you can change
| Setting | Standard | Allowed |
|---|---|---|
| Number of faces | 24 faces, 96 ratings (standard) | 24 faces, 96 ratings (standard), 14 faces, 56 ratings |
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, tablets and phones. 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 Face Impressions Task in a web browser (ConductCognition task version 1.0.0, scoring version 1.0.0; Oosterhof and Todorov, 2008), with the standard settings (number of faces: 24 faces, 96 ratings). The main score was repeat agreement (100 minus the mean absolute change between repeated ratings of the same face and trait).
ConductCognition versions: task version 1.0.0, scoring version 1.0.0. Every export records the versions each session used, for a methods section.
- Oosterhof, N. N., & Todorov, A. (2008). The functional basis of face evaluation. Proceedings of the National Academy of Sciences, 105(32), 11087-11092. https://doi.org/10.1073/pnas.0805664105Trustworthiness and dominance, repeated self-paced trait ratings, and the distinction between impressions and actual traits.
Source papers
- Oosterhof, N. N., & Todorov, A. (2008). The functional basis of face evaluation. Proceedings of the National Academy of Sciences, 105(32), 11087-11092.Paradigm source
Questions
How long does the Face Impressions Task take?
About 6 to 9 minutes, as listed in the test library.
What device does a participant need?
Untimed test: runs on desktops, laptops, tablets and phones. The browser must be Chrome, Edge, Firefox or Safari.
What scores does it produce?
Repeat agreement, Trustworthiness consistency, Dominance consistency, Mean trustworthiness impression, Mean dominance impression, Trustworthiness repeat agreement, Dominance repeat agreement, Complete repeat pairs, Usable ratings, Planned ratings, Median rating time and Anticipations. Each session is also marked valid, needs review or invalid, with the reason.
How widely is the Face Impressions Task used in research?
226 papers from 2021 to 2025 name it in their title or abstract, up 32% compared with 2016 to 2020. That ranks it 135 of 582 test families in the ConductCognition test census (use grade C). Source: ConductCognition test census 0.39.0, counted 8 November 2026 from OpenAlex and ClinicalTrials.gov.
How much reliability evidence does the Face Impressions Task have?
In the published research, 3 papers report its test-retest reliability and 2 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?
Number of faces (24 faces, 96 ratings (standard), 14 faces, 56 ratings). 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.