Lecture 3- face recognition/perception Flashcards
briefly describe the challenge for the system we use to recognise faces?
face recognition has to deal with all these challenges that are over and above the identity, different images of same faces that can look different < our recognition system has to cope with these different elements
what do we use as the basis for our representations?
we try to extract some inner variant about each individuals face and use this as our basis.
identify 5 processing stages in a possible model of face recognition…
1) image of face, description of face, stored descriptions of faces ( face recognition units, matching of faces) , person identities (other routes to identy, voice/name ect), Names
briefly describe the basic idea of face recognition?
you need to create some sort of internal description about the face and match it with what you have stored in your memory, some stored descriptions of the visual properties of faces you know and you access some knowledge you have about a person > eg their name.
identify 3 key pieces of evidence for the different stages?
diary study ( qualitive feel to it) , patterns of breakdpwn after brain injury (neuropsychological evidence), laboratory experiments.
true or false, patients suffering brain-damage are unable to do certain tasks?
true
what is aperceptive prosopagnosia?
defective perceptions of faces < issue with description of face stage of model < unable to form coherent descriptions of faces < unable to access their intact knowledge of about what familiar faces look like
what is associative prosopagnosia?
impaired face memory (issue with stored descriptions of faces model) < faces look okay but patients because their stored knowledge is disrupted, the faces just don’t appear to be familiar to them.
semantic impairments?
damage to person identities stage of model.
anomia?
patients have a problem retrieving names
identify two pieces of experimental evidence for stages?
decision speeds, repetition priming.
It is faster to recognise a face if you have been primed to it before hand, true or false?
true
what is within domain repetition priming >
faces will prime faces
brennan, baugley, bright and bruce (1990) induced what state, briefly explain?
the tip of the tongue (TOT) state, they asked subs to name people from semantic descriptions, subs given various cues and they recorded the percentage they resolved
what is the TOT state?
when you know that thing/that person but you are just unable to access that information
identify ways of resolving TOT states?
1) someones initals will help resolve TOT states, showing a face wont help as their is no direct link between FACE RECOGNITION UNIT and NAMES in model.
what percent of people resolved TOT state using initals compared to faces
initals - 47% , faces- 15%
face/person recognition involves……
perceptual and memory processes, which are at least partially, seperable.
identify key types of socially important meanings from faces?
identity, age, gender, mood, speech, gaze, attention, attractiveness, dominance, trustworthiness
according to Bruce and young 1986, identify the sequence of stages involved in naming of familiar faces?
structural encoding, face recognition units, person identity nodes, name generation.
expression tasks are …..
unaffected by familiarity
identify a common double dissociation found In neurological patients suffering from prosopagnosia?
prosopagnosics can often do expressions ( can recognize expression judgements) , opposite pattern also found ( people unable to make expressions judegments) but can tell you the identity of the person.
identity is often associated with what region of the brain?
FUSIFORM FACE AREA, FUSIFORM GYRUS
expression tasks are often associated with what brain region?
the superior temporal sulcus ( not typically the fusiform face area)
different cells in primate temporal lobe are different for?
identity and expressions
functional separation of processes involved in extracting identity, expression, gaze and facial speech information from faces, true or false?
true
what is acquired prosopagnosia?
face blindness that occurs following some brain injury.
face-specific defecits =
prosopagnosia
different representations are involved in face and object recognition…. explain?
objects = parts based, faces= configural/holistic objects = edges , faces = surface properties
according to YIN, 1969, face recognition dissproporitionately impaired by?
inversion
face recognition may involve?
configuration of features aswell as features themselves.
it is significantly slower to name composite than non-composite stimuli…..TRUE OR FALSE
TRUE
with reference to composite effects, what interfered with the identification of constituent parts?
perception of a novel face configuration
trying to recognise a composite face forms what?
a new configuration of features < makes it difficult to extract the identity of either the top or bottom half.
evidence suggests that upright, normal faces are represented…
configurally
what impairs configural processing?
when faces are inverted
configural processing is not as important when?
in object recognition
Biedermans geons specificed by edges can be obtained from where?
primal sketch
what plays no role in defining geons?
surface characteristics
in what way do faces preserve surface characteristics?
pigmentation, shading
identify a way of disrupting peoples ability to identify the faces?
simply negating faces and inverting ( similar effects)
presenting a face in which form is really disruptive?
in photographic negative
identify the 2 sources of information adversely affected by negation?
pigmentation > skin and hair colour and variations in these
2) pattern of shading and shadows which may help specify 3D structures of the face
line drawings are difficult to recognize unless …
information about pigmentation and/or shading is preserved
representations mediating face recognition preserve…
surface information
surfaces (pigmentation and shading ) is more important for?
faces than objects
faces are….
more than the sum of their parts, whereas, object recognition is part based.
what is the expertise hypothesis?
face recognition is special only because it involves a task of within-category recognition at which we are all experts.
diamond and carey, 1986, hypothesised that inversion can…
affect recognition of dogs by dog show judges
changing the location of face features of the features on greebles does what?
makes them harder to recognise
‘experts with greebles’ are more susceptible to?
effects of inversion and negation than novices ( gauther et all 1998) CHECK PRINTED NOTES NOW
many prosopagnosias are disrupted at?
other within-category discriminations but not all, others can do some within-category discriminations.
prosopagnosia may not just be a face-specific disorder < explain?
it affects ability to make within-category discriminations.
faces are special?
true
devlin 1976 reported?
no conviction should be based on eyewitness ID alone, if conviction rests substantially on ID, jury should be cautioned to risk of error.
what is the innocence project?
individuals who were wrongfully convicted and DNA evidence exonerated them
what if DNA is contaminated or not available?
eye witness testimony carries a great deal of weight with jurors, particiualry where witness is confident.
give a reason as to why people are so confident with eyewitness testimonys>
impressive picture memory
eyewitness memory for unfamiliar faces is said to be
poor
processes/representations involved in familiar/unfamiliar face recognition….
are different
what are estimator variables and identify some?
these affect accuracy of identifications but can NOT be controlled by the justice system, tend to revolve around other factors such as….individual differences, other race effect, in group biases, delay, confidence, stress, duration of encounter, quality of witness description.
estimator variables contrast with
system variables
identify and explain system variables
justice system can control and these are often the influence of the retrieval stage of information : linup composition, prior exposure to mugshots, target absent warnings, sequential vs simultaneous lineups, moving vs static lineups, double blind procedure
difficulty with matching faces from CCTV/PHOTOS/VIDEOS ect…
difficulties matching other race faces, check notes
how is a composite created?
from memory of an unfamiliar face. it is produced following a traumatic experience and will contain inaccuracies.
what are photofit facial composites…
using witness accounts to create a photofit for most like criminal
why are photofit so bad?
featural facial composities, produces poor likeliness of targets, constructing photofit from memory is difficult and produces poor results.
the evofit system?
evolving a likeliness via various combination rounds of faces < current system used by many police forces, check notes.
identify ways of improving photofit/composite/id methods
use holistic systems and exploit resemblance (evofit ect), better ways to tap memory (cognitive interview, holistic interview or both), morph composites from separate witnesses. blur external features, split composites to disrupt configural processing.
why are composites so poor?
verbal mediation, feature-based construction, created from memory so inaccuracies, differential focus on external and internal features of unfamiliar witness and familiar composite “recogniser”
explain ‘blurring features’
witnesses are unfamiliar with face they are trying to recall, external features are important in unfamiliar face recognition, those trying to name composites will be familiar with the faces and internal features relatively more important for familiar faces
what is split composites?
producing a composite from memory so it will contain inaccuracies. CHECK PRINTED NOTES
