STRANGERS
Two photos, one person?
Researchers found that within-person variability is often very large compared with differences between people, so unfamiliar viewers saw one person as several.
Who matches best?
WHAT THE RESEARCH SAYS
Is it the
same person?
Pair up two photos of a stranger and you'll often guess wrong.
We recognise friends from a glimpse, yet matching two pictures of someone we don't know is genuinely hard. Studies of passport staff, students and 'super-recognisers' show why, and why familiarity changes everything.
14% of fake photos accepted
Who matches best?- Strangers are hard
- Friends are easy
- Skill varies widely
- A photo is not a face
THE SHORT ANSWER
Unfamiliar faces
fool almost everyone.
Photos of one person can differ more than photos of two people.
Knowing the face is what makes matching easy.
EXPERTS
Training didn't help
Passport officers checking photos against live card holders accepted 14% of 'fraudulent' photos, and were no more accurate than students.
See the numbers
FRIENDS
Familiar faces are easy
The same photos that confused strangers were identified perfectly by people who already knew the faces.
Try it yourself
FOUR THINGS TO TRY
Test your own
matching eye.
These simple experiments follow the ideas the researchers tested. All you need is a few photos.
Sort a stranger's photos
Mix several pictures of two people you don't know and try to sort them by person; then ask someone who knows them.
Compare with a friend's face
Do the same with photos of someone you know well, and notice how much easier it feels.
Mix the cameras
The standard research test pairs photos taken with different cameras; try comparing pictures of one person taken on different devices.
Don't trust a single photo
The 2011 study suggests one photograph is not a reliable record of how someone looks.
TWO NUMBERS WORTH KNOWING
Errors at the desk,
pairs in the lab.
A real-world error rate from passport staff,
and the size of the standard test researchers use.
14%
'Fraudulent' photos accepted in live checks
Some officers were very accurate, but average performance was poor. The authors suggested security could improve by selecting staff for the skill.
PASSPORT OFFICERS · PLOS ONE 2014
168 pairs
Face pairs in the full Glasgow Face Matching Test
Viewers judge same or different for pairs photographed full-face with different cameras. A short version uses 40 pairs.
THE TEST · BEHAVIOR RESEARCH METHODS 2010
The Glasgow test correlated moderately with face memory but more strongly with matching objects, a link specific to unfamiliar faces.
WHO MATCHES BEST?
Pick a
matcher.
Choose a group of people to see how they did at deciding whether two photos show the same face. The bar shows how accurate the studies found them.
Based on studies published in Cognition (2011), PLOS ONE (2014) and Psychonomic Bulletin and Review (2009).
WHY IT MATTERS
Photo ID
has limits.
Photo ID is widely used in security settings, yet research keeps finding that people find unfamiliar faces very hard to match. The 2011 study warned that face photographs may be unsuitable as proof of identity on their own.
Large individual differences are the hopeful part: some people are far better than others, and the passport study's authors proposed that security could improve by choosing staff with that skill.
Questions
Questions people ask
Why is it hard to match two photos of a stranger?
Because photos of the same person vary a great deal. Research found this within-person variation is often very large compared with the differences between people.
Are passport officers better at it?
Not on average. In a 2014 study, passport-issuing officers accepted 14% of 'fraudulent' photos in live checks and were no more accurate than students.
Why can I recognise friends from bad photos?
Familiarity. In the 2011 study, people who knew the faces identified photos perfectly that unfamiliar viewers got wrong.
What is a super-recogniser?
Someone with exceptional face recognition. A 2009 study found four such people performed beyond the range of ordinary participants on two different tests.
How do researchers test face matching?
One standard is the Glasgow Face Matching Test: 168 pairs of full-face photos taken with different cameras, or a 40-pair short version, each judged same or different.
Sources
- White et al. (2014), Passport officers' errors in face matching, PLOS ONE
- Variability in photos of the same face (2011), Cognition
- The Glasgow Face Matching Test (2010), Behavior Research Methods
- Super-recognizers: people with extraordinary face recognition ability (2009), Psychonomic Bulletin and Review
Updated