Do masks hurt face recognition as much as sunglasses?
Masks and sunglasses made it equally harder to tell that two photos showed the same stranger, and a year of living with masks did not make people any better at it.
Source
Face masks versus sunglasses: limited effects of time and individual differences in the ability to judge facial identity and social traits
Study at a glance
- Design
- Human experiment — Within-subjects manipulation of face covering (none, mask, sunglasses, both) on face matching and trait ratings; timepoint (June 2020, Feb 2021, Aug 2021) between subjects; Experiment 2 compares DP and SR groups with age-matched controls
- N
- N=135 · Experiment 1: 135 analysed typical adults (44, 44 and 47 per timepoint) after exclusions from 150 recruited. Experiment 2: 12 people with developmental prosopagnosia and 10 super-recognisers, each compared with 25 control participants drawn from Experiment 1
- Population
- UK adults aged 18-60 recruited online, plus lab-screened individuals with developmental prosopagnosia and super-recognisers
- Outcome
- Face matching sensitivity (A) and response bias (b), accuracy and response time; 7-point ratings of trustworthiness, competence and attractiveness
Structured fields used in claim comparison tables when every cited study has a complete layer.
What they did
Online participants judged whether two photos showed the same unfamiliar person, where one photo was edited to add a medical mask, sunglasses, both, or nothing; they also rated the faces for trustworthiness, competence and attractiveness. Separate groups were tested in June 2020, February 2021 and August 2021 to see whether the effects shrank as people got used to masks. A second experiment gave the same matching task to people with developmental prosopagnosia (very poor face recognition) and super-recognisers (exceptional face recognition), and related the mask effect to self-reported face ability.
What they found
Masks and sunglasses each lowered matching sensitivity by a similar amount, and wearing both was worse than either alone, but performance stayed well above chance. The drop came from failing to recognise matching pairs, not from mistaking different people as the same; masks also made people more likely to answer 'different'. None of these effects changed across the three timepoints. Sunglasses lowered all three trait ratings, whereas masks alone did not reduce trustworthiness or attractiveness and actually raised competence ratings. Prosopagnosic and super-recogniser groups differed in overall accuracy but were not disproportionately affected by coverings, apart from super-recognisers showing a smaller mask effect on matched pairs.
The limits
What it doesn't show
Timepoint was compared across different groups of people rather than following the same people over time, so it cannot show how any one person adapted. Masks and sunglasses were digitally edited onto database photos taken about 15 minutes apart, and participants had already seen the faces in the rating tasks, so accuracy was near ceiling and may overstate real-world performance. The occluded areas were not matched in size, so the study cannot say whether location or amount of coverage matters. The prosopagnosia and super-recogniser groups were small (12 and 10 people) and self-referred, limiting power to detect group differences.
Key terms
- Unfamiliar face matching
- Deciding whether two simultaneously shown photos depict the same person when the viewer does not know that person.
- Signal detection sensitivity (A)
- A bias-free measure of how well someone distinguishes same-person from different-person pairs, from 0 (chance) to 1 (perfect).
- Response bias (b)
- A tendency to favour one answer; here, higher values mean a conservative tendency to say 'different'.
- Developmental prosopagnosia
- A lifelong, severe difficulty recognising faces despite normal vision and intelligence.
- Super-recogniser
- A person whose face recognition ability is far above population norms.
- PI20
- A 20-item self-report questionnaire of everyday face recognition difficulty; higher scores mean poorer self-reported ability.
Flashcards
0 of 11 answers reviewed
Research intelligence for this paper
See its role on concept claims, tensions it is part of, placement history, and related discoveries.
Quiz yourself
In Experiment 1, how did the effect of masks on face matching change from June 2020 to August 2021?
Common questions
If masks cover more of the face than sunglasses, why was the damage similar?
The authors suggest that which features are hidden matters more than how much area is hidden, and that viewers flexibly use whatever cues remain across the face, at least when they have unlimited time.
Why did coverings hurt 'same' trials but not 'different' trials?
Rejecting a mismatch can rely on a single visible difference, but deciding two photos are the same person needs confidence that the whole face matches, which occlusion undermines.
Does this mean people never adapt to masks?
Not quite: the study found no change across groups tested over about a year, but it did not train anyone or track the same individuals, so targeted practice might still help.
More on Face processing