Is there a general skill for telling if two images match?
People who are good at judging whether two faces match also tend to be good at matching fingerprints, bullet casings and even novel prints, suggesting a general 'matching' ability.
Source
Match me if you can: Evidence for a domain-general visual comparison ability
Study at a glance
- Design
- Cross-sectional — Two preregistered online correlational studies in which every participant completed four same/different visual comparison tasks plus discriminant-validity measures, analysed with correlations, Bayes factors and principal component analysis.
- N
- N=124 · 124 Prolific participants in Experiment 1 and a separate 124 in Experiment 2 (US residents); no exclusions in either.
- Population
- Adult online participants from the general population with no forensic training.
- Outcome
- Signal-detection sensitivity (d') on face, fingerprint, firearms (cartridge case) and artificial-print comparison; intrinsic motivation (Exp. 1); visual search reaction time and visual statistical learning (Exp. 2).
Structured fields used in claim comparison tables when every cited study has a complete layer.
What they did
In two preregistered online studies, 124 adults each judged whether pairs of faces, fingerprints, cartridge cases from firearms, and artificial prints stamped from carved potatoes came from the same source. Performance was scored as sensitivity (d'), which separates accuracy from a tendency to say 'same'. Experiment 1 also measured intrinsic motivation, and Experiment 2 added a visual search task and a visual statistical learning task to test whether any shared ability was just general visual skill.
What they found
In both studies, sensitivity on all four comparison tasks correlated positively, and one principal component on which all four loaded explained 41.99% (Experiment 1) and 34.92% (Experiment 2) of the variance. Intrinsic motivation, visual search and statistical learning did not load on this component and were largely unrelated to comparison accuracy, with Bayes factors mostly supporting no correlation. Correlations between tasks were only moderate, and face matching had the most task-specific variance, suggesting a mix of general and stimulus-specific skill.
The limits
What it doesn't show
Participants were untrained novices, so the authors note it is unclear whether a general matching ability still matters for professional examiners, whose expertise is usually domain-specific. The fingerprint tasks had below-standard reliability (alpha around .61-.62), which can weaken correlations involving them. The design is correlational, so it identifies shared variance but not the mechanism behind it (e.g. holistic vs featural processing). Samples were online US or Prolific volunteers, and the applied suggestion of selecting examiners by this ability was not tested directly.
Key terms
- Visual comparison (matching)
- Judging whether two images, such as two faces or two fingerprints, come from the same source.
- Sensitivity (d')
- A signal-detection measure of how well someone discriminates matches from non-matches, independent of their bias to answer 'same' or 'different'.
- Domain-general ability
- A skill that applies across many kinds of material rather than being tied to one type of stimulus or expertise.
- Principal component analysis (PCA)
- A statistical technique that summarises many correlated measures into a smaller number of components capturing shared variance.
- Discriminant validity
- Evidence that a measure is not related to measures of theoretically different constructs, here motivation, visual search and statistical learning.
Flashcards
0 of 10 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
What was the main evidence for a domain-general matching ability?
Common questions
Why include artificial prints made from potatoes?
Nobody has prior experience with them, so good performance on that task cannot be explained by familiarity or training.
Couldn't the shared factor just mean some people try harder?
The authors tested this: intrinsic motivation did not correlate with matching accuracy and loaded on a separate component.
Does this mean fingerprint examiners should practise face matching?
No. The authors stress there were also stimulus-specific skills, and they suggest the finding is most useful for selecting trainees rather than cross-training experts.
More on Perception