Seminar 5: divided visual field and ERP

Download Report

Transcript Seminar 5: divided visual field and ERP

The anatomical basis of face recognition:
evidence from studies of intact individuals:
What is the anatomical basis of face recognition in
humans?:
Right hemisphere seems particularly important for
face-recognition.
Farah (1990): 65% of 81 prosopagnosics had
bilateral damage, 29% RH only, 6% LH only.
RH important for configural processing, LH for
featural?
Which hemisphere is most
important for face
recognition?
Divided-field studies with normal
people (Bourne, Vladeanu and
Hole, 2008):
Stimuli in extreme left visual field
go first to right hemisphere, and
vice versa.
Both hemispheres can recognise
faces.
LH: featural processing.
RH: configural processing.
RH faster than LH with complete faces.
Blurring affects LH more than RH.
Features-only affects RH more than LH.
Mean RT (+ 1 SE) to normal and blurred
faces as a function of visual field of
presentation.
Mean RT (+ 1 SE) to complete faces
and the eyes only as a function of
visual field of presentation.
950
1200
1100
LVF-RH
RVF-LH
900
1000
850
900
800
700
Normal
Face type
Face type
Blurred
Mean RT (ms) + 1 SE
Mean RT (ms) + 1 SE
800
LVF-RH
RVF-LH
750
700
Complete
Eyes only
Face type
Bourne and Hole (2003): hemispheric differences in
processing incomplete faces:
Complete
Eyes
missing
Nose
missing
Mouth
missing
Bourne and Hole (2003): hemispheric differences in
processing incomplete faces:
Mean RT difference (ms) + 1 SE
300
Eyes missing
Nose missing
Mouth missing
p = .002
p = .012
200
p = .088
100
0
LVF-RH
RVF-LH
VF of presentation
Bars represent
the difference
between the
complete face
condition and
each
experimental
condition. The
greater the
difference the
more detrimental
the effect of the
manipulation.
LH (featural) copes worse with missing features than RH (configural).
Tong, Nakayama, Moskowitz, Weinrib and Kanwisher (2000):
fMRI study of fusiform responses to
face-like stimuli, eyes, houses and
non-face objects.
FFA response similar for cat, cartoon
and human faces (with/without eyes);
weaker for schematic faces and eyes
alone;
equal for front and profile views, but
declining as face rotated away from
view;
weakest for non-face objects and
houses.
Conclusion: fusiform gyrus responds
best to facial configurations plus
features - involved in "face"
perception/detection.
Schiltz, Dricot, Goebel and Rossion (2010):
fMRI adaptation study of neural responses to composite faces.
Right middle fusiform ("FFA") sensitive to composites - treats
them as "new" faces.
Right FFA involved in "holistic" processing.
Lee, Anaki, Grady and Moscovitch (2012):
fMRI study of responses to face halves separated in time or space.
Behavioural data: ISI 0 and ISI 200 similar; ISI 800 and Misaligned similar to
each other, and worse than ISI 0 and ISI 200.
ISI 800: activated face processing regions (more bilaterally) plus areas
involved in attention and working memory (strategic processing?)
ISI 0 and ISI 200: better identification correlated with increased activity in
“configural processing” network (R fusiform, middle occipital, bilateral
superior temporal, inferior/middle cingulate and frontal cortex).
ISI 800 and Misaligned: better identification correlated with less activity in
these regions.
Suggest configural and analytic processing regions oppose each other.
Gorno-Tempini and Price (2001) PET/MRI study:
Four visual tasks:
(a) Famous face matching.
(b) Non-famous face matching.
(c) Famous building matching.
(d) Non-famous building matching.
Category-specific perceptual processing:
Faces (famous and non-famous) activate fusiform gyrus.
Buildings (famous and non-famous) activate parahippocampal gyrus.
Shared analysis of semantic processing:
Fame (faces or buildings) activates left anterior middle temporal gyrus.
Event-related potential (ERP) studies:
(a) P1: (100 ms after presentation of a visual stimulus) occipital
response to low-level stimulus characteristics (e.g. luminance and
contrast).
(b) N170) occipito-temporal response, larger for faces than other
objects: structural encoding or detection of a face-like pattern
ALSO VPP (vertex positive potential, a fronto-central positive
counterpart to N170)
(c) P2 (200 ms after presentation) more fine-grained analysis of featural
distances or facial distinctiveness
(d) N250: related to processing of facial identity – more negative for
repetitions of familiar faces compared to non-repeated familiar faces
(N250r effect) – facilitated access to representations of familiar faces.
N250r also increases during face learning.
(e) LPC (“late positive component”, 400-700 ms at central-parietal sites:
related to retrieval from episodic memory (old/new effect)
Anatomical location of processes involved in face
recognition (Schweinberger and Burton (2003):
Fusiform gyrus
Structural encoding N170
(superior temporal sulcus)
RH
LH
Lingual gyrus
Parahippocampal
gyrus
Face recognition
(Fusiform gyrus) N250R
Name
PIN
(Left temporal
lobe)
(Anterior
temporal lobe)
Semantic
information
(Anterior
medial
temporal
lobe)
N400
Integrative
device
Arousal to
familiar face
(Amygdala)
Attribution
processes
Skin
conductance
response
Event-related Potential (ERP) studies of face processing
(Schweinberger 2003):
N170:
Generated from posterior lateral occipitotemporal
cortex (superior temporal sulcus).
Larger for faces than most other visual stimuli. Not
human face-specific: also produced by car "faces",
ape faces, schematic faces and inverted faces.
Unaffected by face familiarity or face priming. i.e., not
related to face recognition.
Correlate of structural encoding, identification of facelike configurations?
N250R:
Strongly right hemisphere. Affected by familiarity of
faces, and larger for personally-familiar faces than
famous faces. Activity modulated in response to
repeated faces (even if diffferent views each time,
though strongest with identical images). Probably
generated from fusiform gyrus.
Most response from human faces; then ape faces;
no response to inverted faces or car "faces".
Correlate of "face recognition units"?
N400:
Anterior medial temporal cortex.
Correlate of "person identity nodes" (postperceptual response to individuals)?
ERP and experience:
Wiese, Wolff, Steffens and Schweinberger (2013):
Effects of experience on own-age bias.
Young experts (geriatric nurses) and novices.
Recognition task with old and young faces.
OAB in novices but not experts.
Larger N170 and P2 for young than old faces in both groups (i.e. similar
for early perceptual processing).
Novices: N250 more anterior repetition effects for own- than other-age
faces. Experts: no differences.
Novices : larger LPC for old than young faces. Experts: no differences.
Conclusion: experience with other-age faces does not affect early
perceptual processing; affects later stages related to memory retrieval.
ERP and experience:
Balas and Saville (2015):
Undergraduates from small (500-1000)
or large (30,000-100,000) towns.
CFMT.
Classifcation task: respond “chair” or
“face” to upright or inverted stimuli.
Small-town: poorer face memory and
an N170 that was less specific to faces
(a smaller difference between face and
chair N170s for small-town subjects).
Conclusion: the number of faces
encountered during early experience
affects adult face processing/brain
structure.
ERP and experience:
Dundas, Plaut and Behrmann (2014):
Adults: larger LH N170 for words, RH N170 for faces.
7-12 year olds: adult pattern for words, but not faces: similar-sized
N170s over both hemispheres.
Hypothesis:
Hemispheric development of face and word recognition ae
interdependent: both compete for fine-grain visual analysis.
Word lateralisation precedes and drives later face lateralisation.
(As reading ability increases, decreased LH/increased RH FG
actviation to faces).
Dundas,Plaut and Behrmann (2013): DVF study in children,
adolescents and adults. Reading scores predict face selectivity
(hemifield superiority for matching faces or words in the two visual
fields).
Barbeau, Taylor, Regis, Marquis, Chauvel and Liegeois-Chauvel (2008):
Intra-cranial ERP study of time-course of famous face recognition.
Massively distributed processing from 110 -600 msec post-stimulus - at least
seven structures involved.
Processing is not "one-way" frontal areas influence "earlier"
stages.
FG - invariant aspects of faces;
STS - changeable aspects.
Perirhinal cortex -signals
"familarity".
Temporal structures recognition.
(Dark blue = periods when recognition
effects were found).
Jonas et al. (2012):
Intra-cranial electrode stimulation study in
epileptic woman, K.V.
Epilepsy due to right inferior occipital
cortical dysplasia: normal face recognition.
Stimulation of right inferior occipital gyrus
(Occipital face Area) produced transient
prosopagnosia (no naming or semantic
information).
‘‘the facial elements were in disarray’’
Face-sensitive intra-cranial N170 from OFA
(contacts 07 and 09) .
Right OFA is necessary for normal face
perception: possibly has critical role in
configural processing.
Outstanding questions:
How do the hemispheres cooperate during normal
face processing?
Are the RH and LH really specialised for configural
and featural processing, or are these merely
reflections of generalised differences in processing
modes? (RH – global, LH – local).
In particular, is featural processing really a mode of
face processing, or merely a strategy to cope with
odd-looking faces in psychology experiments?