FaceID Flashcards
Basics
- fundamental for social functioning
- research focus on understanding cognitive processes
Memory for familiar and unfamiliar people
- familiar very good even in poor condition
- unfamiliar very poor even in good conditions
- feature based or holistic?
Tanaka and Farah (1993)
- evidence for holistic processing
- test accuracy on normal and scrambled faces
- location important effects on performance
- effect disappear when inverted
Feature Based Facial ID
- may be incorrect
- poor progress in auto recognition for photographic effects (negation, inversion, caricature)
Identifying Face vs Objects
- two different processes
- patient with Face ID impairment can recognise objects (Farah, 1996)
- neuropsychology support for different areas of the brain responsible for each (Kanwisher et al, 1997, Haxby et al, 2000)
Yin (1969)
- worse at ID of inverted faces than inverted objects
- suggests different processes involved
Recognising Inverted Faces
- less accurate than upright recognition
- slower than upright
- may be special mechanisms for recognising upright faces
Diamond and Carey (1986)
- face recognition for faces and dogs
- dog experts impaired for inverted faces
- suggests exposure results in the inversion effect
McNeil and Warrington (1993)
- farmer with prosopagnosia
- can remember sheeps faces
Facial ID in Forensic Settings
- unfamiliar hard to reconstruct, recognising and describe
- eye witness testimony compelling
- failures of ID ascribed to memory system problems ( encoding, storage, retrieval)
- 3 basic process of acquisition, storage and retrieval influence eyewitness memory
Bruce and Young (1986)
- recognising familiar vs unfamiliar
- familiar:
- view centred
- expression independent
- multiple exposure, different contexts and viewpoints
- unfamiliar
- view centred only
- on exposure
- hair, face, lighting
Eyewitness ID Procedures
- show ups
- simultaneous line up
- sequential line up
- field ID
- mug shots
Steblay et al (2001)
- sequential line up superiority effect
- meta analysis
- more careful attention and decision making
- most accurate
McQuiston-Surrett et al (2006)
- questions sequential superiority effect
- meta analysis
- target absent - sequential better
- target present - simultaneous better
Differences Between Countries
- line up: 6 in US, 9 in UK, 20 in Australia
- legislation: controlled in UK, non-binding guideline in UK
- live parades in the past in UK but replaced by video due to cost and time
Bruce and Burton (1999, 2001)
- 1 in 10 task
- video and photos
- good conditions
- many misses, incorrect and even more false positives
Cognitive Model of Face Recognition
Bruce and Young (1986)
- Structural encoding
- perceive face and create visual description - Face recognition units
- match description to stores descriptions - Person identity nodes
- access semantic info in LTM (occupation, personality) - Name retrieval units
- access name from speech output lexicon
Cognitive Model Critique
- decisions about familiarity, then person knowledge, then name (Young et al, 1986)
- can’t recall name without other info (Flude et al, 1989)
- accounts for difference in familiar and unfamiliar face processing
- explains feelings of familiarity without knowing anything more about the face
- explains why names no produced without other info
Model for Distributed System for Face Perception
Haxby and Gobbing (2010)
Core systems
-central in visual analysis
- invariant features and changeable features represented differently
- e.g. posterior STS: dynamic features for gestures
Extended systems
- match visual description to stored
- 3 sets of brain areas involved in person knowledge, motor stimulation and emotions
Burton et al (1999)
Experiment one
- police vs public
- 3 conditions; familiar, unfamiliar, police
- unfamiliar and police same accuracy but familiar much better
Experiment two
- 4 conditions; normal, hide gait, body, face
- all familiar with target
- worst performance when faces covered
- faces important to recognition
White et al (2014)
Experiment one - person to photo test - passport officers - many errors, wrongly rejected and more fraudulent wrongly accepted Experiment two - photo to photo test - passport officers and students - poor for match, better for mismatch - no difference between passport officers and students
Composites
- better at unlocking phone that individual photo (Robertson et al, 2015)
- averages of famous improve ID (Burton et al, 2005)
- more accurate when averages used, even in unfamiliar (White et al, 2014)
