Task 3 Flashcards
Super-recognizer
Bobak, A.K., Hancock, P.J., & Bate, S. (2016). Super‐recognisers in action: Evidence from face‐matching and face memory tasks. Applied Cognitive Psychology, 30, 81-91.
Aim
Methods
Results
Limitaions
- see whether super-recognizer are better then normal participants in matching and recognizing face
1st experiment
- 7 super-recognizer vs. 22 controls
- face matching task (either target absent or target present) 1 photo vs. 10 photos
- -> SRs were significantly better than controls when matching , especially in target absent (higher accuracy and specificity = both over 90%)
- -> SRs were also more confident
2nd
- same pps
- applied test of unfamiliar face memory –> study phase (good quality picture), filler task, recognize task (poor quality video)
- -> super-recognizer were sign. better then control, but their accuracy also was not so good
- -> super-recognizer out-preform controls in different tasks (match, recognize)
- -> the effect is more pronounce under more ideal circumstances –> SR accuracy also suffers under less ideal conditions
- small sample size cause not so many super recognizer known (to this institute)
Super-recognizer
Robertson, D.J., Noyes, E., Dowsett, A.J., Jenkins, R., & Burton, A.M. (2016). Face recognition by Metropolitan Police super-recognisers.
Aim
Methods!
Results
Limitations
- see whether super-recognizer are better then normal participants in matching and recognizing face
- 4 super recognizer (and a control group - 25) took those test:
(1) Glasgow Face Matching Test (GFMT), which is a standardised test of unfamiliar face matching.
(2) Models Face Matching Test (MFMT), an unfamiliar face matching test that was designed to be particularly difficult.
(3) Pixelated Lookalikes Test (PLT)—a new test of familiar face recognition based on very poor quality images. - SRs always sign. out-preformed the controls
- -> previous work with police officers and passport officers shows that, on average, both professional groups perform face tasks with the same levels as the general public (even with training)
- -> So, recruiting of SRs is more important because training does not do much and they are consistently better
- small sample of SR
Composites = post-event misinformation
Topp‐Manriquez, L.D., McQuiston, D., & Malpass, R.S. (2016). Facial composites and the misinformation effect: How composites distort memory.
- problems with composites
- misinformation effect def.
- types of composites
- how to improve composites?
- someone innocent might be suspected because of a resemblance to the composite (Ronald Cotton)
- a composite with a low similarity to the suspect might contaminate the memory of the eyewitness
= individuals are likely to incorporate post‐event misinformation into their own recollection of the event itself –> inaccurate face composites might be post-event misinformation!! - sketches (old-school, still used by the FBI)
- nowadays: on the computer
> PhotoFIT/Identikit: you have different versions from different features and put them together one after another (like Frankenstein’s monster) = feature-based
> Evolutionary system: response to groups of compete faces and always decide which one is more resembling and through this get closer (produce better quality composites) - combining composites of witnesses could help
- holistic is better than feature-based -> select individual features in the context of the complete face, changing size and precision of the features)
- transfer-appropriate reconstruction: witnesses should be able to make composites from a side view if they have seen the perpetrator form the side
Composites
Topp‐Manriquez, L.D., McQuiston, D., & Malpass, R.S. (2016). Facial composites and the misinformation effect: How composites distort memory.
Aim
Method
Results
Limitation
- examined how constructing and viewing facial composites affects memory accuracy
- 240 pps
- feature-based composite
- viewed face for 60 sec.
- then either distractor task or composite creation
- then one week in-between were randomly assigned saw composite or not
- after 1 week: 6-person-forced-choice test
- correct identification sign. lower for those who composed the composite or merely saw it!
- important detail! the 5 other faces were morphed versions of the composite and the face –> that would NOT be the case in real life
Composites
Charman, S.D., Gregory, A.H., & Carlucci, M. (2009). Exploring the diagnostic utility of facial composites: Beliefs of guilt can bias perceived similarity between composite and suspect.
Composites are used for 2 goals: 1) to identify the perpetrator, 2) as evidence of guilt or innocence in court
•Two studies examined how pre-existing beliefs of guilt influence similarity ratings between a suspect and a facial composite
Experiment 1
•Study 1 (n =93) demonstrated that mock-investigators’ beliefs in a suspect’s guilt inflated their subsequent similarity ratings
•If they thought they were guilty, they were likely to say there were the same
Experiment 2
•Judge similarity + guilty rating
•Study 2 (n =49) demonstrated that mock-jurors’ beliefs in a defendant’s guilt predicted their similarity ratings
•If they thought they were guilty, they were likely to say there were the same
Conclusion:
Pre-existent beliefs of guilt can alter judgement of similarity
Face matching
Megreya, A.M., Sandford, A., & Burton, A.M. (2013). Matching face images taken on the same day or months apart: The limitations of photo ID.
Jenkins, R., White, D., Van Montfort, X., & Burton, A.M. (2011). Variability in photos of the same face.
- they showed 80 pps (1:10) pictures of faces to match and they were –> say wehter one of the 10 matched
- the reference and the target picture (so same person) was either taken on the same day or several months apart
(- of course sometimes target-absent and sometimes taget-present)
–> same day: correct identified 80% of the time
–> months later: only 60%
=> shows the limitations of photo-ID
=> shows that even if taken at the same day we are not super-dooper in matching unfamiliar faces - Jenkins et al did similar experiments, but they once had participants from the UK who were unfamiliar with the Dutch celebrities and once Dutch participants, who were familiar with the celebrities
–> the Dutch participants performed well in both task (matching picture of their faces with high to the reference and with less likeness)
–> UK participants performed similarly to Megreya-study
=> variability of picture does influence matching performance very much if the face are familiar
Balsdon, T., Summersby, S., Kemp, R.I., & White, D. (2018). Improving face identification with specialist teams.
Aim
Method
Result
- evaluate real world benefits of selecting high performers based on performance in standardised tests
- pps did two sessions which were one week apart
- in the sessions they did a bunch of face matching tests (typical screening, Cambridge FMT, realistic face matching test)
- from the screening they were divided in selected and unselected (better vs. worse performance)
- selection –> modest gain in accuracy
- quite variability among the pps
- same test were good predictor for realistic face matching test
- different test were moderate predictor
Groups: - averaging scores of small groups produced substantial gains in performance –> but there was a ceiling
- benefits of grouping were the same for selected and unselected groups
- performance of groups again predictive for 2nd session
- correlation between individual and team accuracy is strong –> but good teams are not only determined by individual performance
=> face matching ability = stable
=> selection is quite important (other studies shown that training does not really improve performance), but benefits of group are equally important
Johnston, R.A., & Edmonds, A.J. (2009). Familiar and unfamiliar face recognition: A review. Memory, 17, 577-596.
see real life flash card
