An event-related brain potential study of explicit face recognition.

Abstract

To determine the time course of face recognition and its links to face-sensitive event-related potential (ERP) components, ERPs elicited by faces of famous individuals and ERPs to non-famous control faces were compared in a task that required explicit judgements of facial identity. As expected, the face-selective N170 component was unaffected by the difference between famous and non-famous faces. In contrast, the occipito-temporal N250 component was linked to face recognition, as it was selectively triggered by famous faces. Importantly, this component was present for famous faces that were judged to be definitely known relative to famous faces that just appeared familiar, demonstrating that it is associated with the explicit identification of a particular face. The N250 is likely to reflect early perceptual stages of face recognition where long-term memory traces of familiar faces in ventral visual cortex are activated by matching on-line face representations. Famous faces also triggered a broadly distributed longer-latency positivity (P600f) that showed a left-hemisphere bias and was larger for definitely known faces, suggesting links between this component and name generation. These results show that successful face recognition is predicted by ERP components over face-specific visual areas that emerge within 230 ms after stimulus onset.

Full text

Page 1

Author's personal copy
Neuropsychologia
49 (2011) 2736–
2745
Contents
lists
available
at ScienceDirect
Neuropsychologia
j ourna
l ho
me
pag
e:
ww
w.elsevier.com/locate/neuropsychologia
An
event-related
brain
potential
study
of
explicit
face
recognition
Angela
Gosling, Martin
Eimer∗
Department
of
Psychological
Sciences,
Birkbeck
College,
University
of
London,
Malet
Street,
London
WC1E
7HX,
UK
a
r
t
i
c
l
e

i
n
f
o
Article
Received
Received
Accepted
Available online 6 June 2011
history:
18
March
2011
in revised
form
11 May
2011
30 May
2011
Keywords:
Face
Face
Face
Event-related
Visual
processing
perception
recognition
brain
potentials
cognition
a
b
s
t
r
a
c
t
To
(ERP)
were
selective
contrast,
triggered
to
associated
stages
activated
longer-latency
faces,
face
after
determine
the
time
course
of
face
recognition
and
its
links
to
face-sensitive
event-related
potential
components,
ERPs
elicited
by
faces
of
famous
individuals
and
ERPs
to
non-famous
control
faces
compared
in
a
task
that
required
explicit
judgements
of facial
identity.
As expected,
the
face-
N170
component
was
unaffected
by
the
difference
between
famous
and
non-famous
faces.
In
the
occipito-temporal
N250
component
was
linked
to
face
recognition,
as
it
was
selectively
by
famous
faces.
Importantly,
this
component
was
present
for
famous
faces
that
were
judged
be
definitely
known
relative
to
famous
faces
that
just
appeared
familiar,
demonstrating
that
it
is
with
the
explicit
identification
of
a
particular
face.
The
N250
is likely
to
reflect
early
perceptual
of
face
recognition
where
long-term
memory
traces
of
familiar
faces
in
ventral
visual
cortex
are
by
matching
on-line
face
representations.
Famous
faces
also
triggered
a
broadly
distributed
positivity
(P600f)
that
showed
a
left-hemisphere
bias
and
was
larger
for
definitely
known
suggesting
links
between
this
component
and
name
generation.
These
results
show
that
successful
recognition
is
predicted
by
ERP
components
over
face-specific
visual
areas
that
emerge
within
230
ms
stimulus
onset.
© 2011 Elsevier Ltd. All rights reserved.
1.
Introduction
In
everyday
life,
we
encounter
familiar
and
unfamiliar
faces
in
many
tification
and
tion
by
psychological
Burton,
that
ceptual
properties
grated
recognition
successfully
of
episodic
from
ERP
the
processes
studies
different
contexts.
Even
though
the
recognition
and
iden-
of
individual
faces
seems
effortless,
many
behavioural
neuroscientific
studies
have
demonstrated
that
face
recogni-
is
in fact
a complex
cognitive
achievement
that
is mediated
a sequence
of
face-specific
brain
mechanisms.
Cognitive-
models
of
face
processing
(e.g.,
Bruce
& Young,
1986;
Bruce,
& Hancock,
1999)
have
described
different
stages
are
involved
in successful
face
recognition.
These
include
per-
processing
stages
such
as
the
initial
encoding
of
the
visual
of
a
face
and
the
subsequent
generation
of
an
inte-
representation
of
its
configural
and
holistic
features.
Face
results
when
such
perceptual
face
representations
are
matched
with
stored
memory
traces
of
visual
features
known
familiar
faces.
Following
this
match,
further
semantic
or
information
about
a particular
individual
can
be
retrieved
long-term
memory.
measures
have
been
used
in many
studies
to investigate
temporal
organisation
and
neural
basis
of
the
component
that
underlie
face
recognition.
The
majority
of
these
have
focused
on
the
face-sensitive
N170
component.
This
∗Corresponding
E-mail
author.
Tel.:
+44
020
7631
6538;
fax:
+44
020
7631
6312.
address:
m.eimer@bbk.ac.uk
(M.
Eimer).
component
triggered
between
temporal
Bötzel,
Kiss,
to
gyrus
and
the
face
recognition
It
are
eral
by
& Deouell,
tity
representations
onset
If
that
that
direct
face
investigated
order
represents
an
enhanced
negativity
that
is reliably
in response
to faces
as
compared
to non-face
stimuli
150
and
200
ms
after
stimulus
onset
over
lateral
occipito-
areas
(Bentin,
Allison,
Puce,
Perez,
&
McCarthy,
1996;
Schulze,
& Stodieck,
1995;
Eimer,
2000a,
2000b;
Eimer,
& Nicholas,
2010;
Rossion
et
al.,
2000).
The
N170
is assumed
be
generated
in occipito-temporal
cortex
and
posterior
fusiform
(Bötzel
et al.,
1995;
Rossion,
Joyce,
Cottrell,
&
Tarr,
2003),
is usually
interpreted
as
an
electrophysiological
marker
for
structural
encoding
of
faces
and
the
activation
of
perceptual
representations
that
form
the
input
for subsequent
face
processes
(Eimer,
2000b;
Sagiv
& Bentin,
2001).
is
important
to note
that
face
recognition
and
identification
unlikely
to be
directly
reflected
by
the
N170
component.
Sev-
studies
have
demonstrated
that
this
component
is
not
affected
the
difference
between
famous
and
unknown
faces
(e.g.,
Bentin
2000;
Eimer,
2000a),
although
evidence
from
face
iden-
adaptation
suggests
that
the
construction
of
individual
face
might
commence
within
170
ms
after
stimulus
(e.g.,
Caharel,
d’Arripe,
Ramon,
Jacques,
& Rossion,
2009).
the
N170
reflects
sensory-perceptual
stages
of face
processing
precede
the
recognition
of
individual
familiar
faces,
research
focuses
on
this
component
cannot
be
expected
to yield
insights
into
the
time
course
and
functional
organisation
of
recognition.
Until
now,
only
relatively
few ERP
studies
have
face-specific
ERP
components
beyond
the
N170
in
to determine
whether
and
how
these
components
are
linked
0028-3932/$
doi:10.1016/j.neuropsychologia.2011.05.025
– see
front
matter ©

2011 Elsevier Ltd. All rights reserved.

Page 2

Author's personal copy
A. Gosling,
M.
Eimer
/ Neuropsychologia
49 (2011) 2736–
2745
2737
to face
face
face
that
enhanced
N250r
and
hemisphere
Huddy,
The
N170,
that
suggested
is,
tation
(Schweinberger
are
& Taylor,
for
Jentzsch,
are
pre-existing
that
Results
(2006)
aspects
a face
tion
et
otherwise
presented
the
get
N250
tion
(1995,
in
representation
ually
between
and
N250
with
an
ably
no difference
faces
N170
Taken
ponents
recognition.
memory
vated
interpretation
not
experiments
uals
studies
2000a),
faces
faces
400
eral
negativity
with
(2000),
recognition
and
identification.
In most
of
these
studies,
repetition
paradigms
have
been
used
as
a
tool
to manipulate
familiarity.
What
was
consistently
found
in these
studies
was
relative
to previously
unseen
faces,
repeated
faces
trigger
an
negativity
at inferior
occipito-temporal
electrodes.
This
component
typically
reaches
its
maximum
between
230
ms
280
ms
after
stimulus
onset,
and
is often
larger
over
the
right
(e.g.,
Begleiter,
Porjesz,
& Wang,
1995;
Schweinberger,
& Burton,
2004;
Schweinberger,
Pfutze,
& Sommer,
1995).
repetition-sensitivity
of
the
N250r
suggests
that
unlike
the
it reflects
processes
sensitive
to the
familiarity
of
faces
take
place
after
their
initial
perceptual
analysis.
It has
been
that
the
N250r
is associated
with
face
recognition,
that
with
a successful
match
between
a
perceptual
face
represen-
and
an episodic
memory
trace
of
a previously
seen
face
& Burton,
2003).
Even
though
N250r
components
triggered
reliably
by repetitions
of unfamiliar
faces
(e.g.,
Itier
2004;
Schweinberger
et
al.,
1995),
they
are
usually
larger
familiar
faces
(e.g.,
Herzmann,
Schweinberger,
Sommer,
&
2004),
which
indicates
that
face-specific
memory
traces
activated
more
strongly
by
repetitions
of faces
that
have
long-term
representations
than
by
repeated
faces
are
pre-experimentally
unknown.
from
a study
by
Tanaka,
Curran,
Porterfield,
and
Collins
suggest
that
N250/N250r
components
reflect
two
different
of
face
recognition
– the
acquired
short-term
familiarity
of
that
is
repeated
in an experimental
context,
and
the
activa-
of
long-term
representations
of
known
familiar
faces.
Tanaka
al.
(2006)
asked
participants
to identify
a
previously
studied
but
unknown
target
face
that
could
appear
in a sequentially
series
of
distractor
faces.
One
of
these
distractors
was
participants’
own
face,
the
others
were
unfamiliar
faces.
Tar-
faces
triggered
a right-hemisphere
dominant
occipito-temporal
that
was
similar
in terms
of its
latency
and
scalp
distribu-
to the
N250r
component
described
by
Schweinberger
et al.
2004).
However,
this
N250
to target
faces
only
emerged
the
second
half
of
the
experiment,
suggesting
that
an
episodic
of
a previously
unfamiliar
target
face
builds
up
grad-
(see
also
Kaufmann,
Schweinberger,
& Burton,
2009,
for
links
the
N250
and
the
acquisition
of
new
face
representations,
Krigolson,
Pierce,
Holroyd,
& Tanaka,
2009,
for
evidence
that
the
also
develops
during
the
acquisition
of perceptual
expertise
non-face
objects).
In contrast,
participants’
own
faces
elicited
N250
component
throughout
the
experiment,
which
presum-
reflects
their
long-term
familiarity.
Interestingly,
there
was
in N170
components
in response
to participants’
own
and
unfamiliar
faces,
which
underlines
the
conclusion
that
the
is
not
sensitive
to face
recognition.
together,
these
findings
suggest
that
N250/N250r
com-
are
linked
to early
visual–perceptual
stages
of
face
These
components
may
be
generated
when
episodic
traces
in face-specific
occipito-temporal
cortex
are
acti-
by
matching
on-line
perceptual
face
representations.
If
this
was
correct,
N250
components
should
be
observed
only
in the
context
of
face
repetition
paradigms,
but
also
in
where
ERPs
elicited
by
well-known
famous
individ-
and
unfamiliar
faces
are
directly
compared.
In two
previous
that
used
this
procedure
(Bentin
& Deouell,
2000;
Eimer,
consistent
ERP
modulations
sensitive
to the
familiarity
of
were
indeed
obtained.
Relative
to unfamiliar
faces,
famous
triggered
an enhanced
negativity
that
was
maximal
around
ms
after
stimulus
onset
(N400f).
In contrast
to the
distinct
lat-
occipito-temporal
focus
of
the
N250/N250r
component,
this
to famous
faces
was
much
more
broadly
distributed,
a frontocentral
maximum
observed
by
Bentin
and
Deouell
and
a centroparietal
maximum
by Eimer
(2000a). This
marked
N400f
processes.
visual
N400f
of
is
tic
Hillyard,
(2000a)
tivity
The
not
about
tion
to
(1986),
In
yet
ERP
identifying
stimulus
has
component
that
mation.
N250/N250r
far only
paradigms,
to
ferences
non-repeated
could
viously
familiarity
recognition,
of
The aim of the
tions,
correlates
with
faces.
personalities,
faces
tified
were
non-famous
visual
components
was
2000a), or only
Bentin
to categorise
point
familiar,
ERPs
arately
contrasted
to be
famous
to identify
findings,
the
question
ponent
by
topographic
difference
suggests
that
the
N250/N250r
and
components
reflect
temporally
and
functionally
separate
While
the
N250
is likely
to reflect
the
activation
of
face
memory
in domain-specific
ventral
visual
areas,
the
is more
plausibly
associated
with
the
subsequent
activation
face-related
episodic
or semantic
memory.
This
interpretation
supported
by
the
fact
that
the
semantic
processing
of linguis-
material
is
associated
with
N400
components
(e.g.,
Kutas
&
1980).
In addition,
Bentin
and
Deouell
(2000)
and
Eimer
found
that
the
N400f
was
followed
by an enhanced
posi-
to famous
faces
(P600f),
which
was
again
broadly
distributed.
functional
interpretation
of
this
longer-latency
component
is
yet
clear.
It may
reflect
the
generic
reduction
of
uncertainty
facial
identity
that
is associated
with
the
explicit
recogni-
of
a particular
individual,
or could
be linked
more
specifically
later
stages
of
face
recognition
postulated
by
Bruce
and
Young
such
as
the
retrieval
of
a person’s
name.
summary,
ERP
investigations
of face
recognition
have
not
provided
a clear
picture
of
the
links
between
face-specific
components
and
brain
processes
involved
in recognizing
and
familiar
faces.
For
example,
the
question
at what
post-
latency
recognition-specific
ERP
components
first
emerge
not
been
conclusively
answered.
It seems
clear
that
the
N170
is associated
with
perceptual
stages
of
face
processing
occur
prior
to the
activation
of
identity-related
infor-
However,
evidence
that
face-specific
occipito-temporal
components
are
linked
to face
recognition
has
so
been
obtained
in experiments
that
used
face
repetition
but
not
in studies
that
directly
compared
ERP
responses
famous
and
unfamiliar
faces.
The
interpretation
of
ERP
dif-
that
emerge
from
the
contrast
between
repeated
and
faces
is complicated
by
the
fact
that
these
differences
reflect
the
activation
of an episodic
memory
trace
of
a
pre-
encountered
face
that
is unrelated
to its
identity
(a
mere
effect),
or could
be
linked
more
directly
to explicit
face
that
is,
to the
activation
of
long-term
representations
known
faces.
present
experiment
was
to resolve
these
ques-
and
to provide
new
insights
into
the
electrophysiological
of
explicit
face
recognition.
Participants
were
presented
a
sequence
of
photographic
images
of
famous
and
non-famous
Famous
faces
showed
actors,
politicians,
musicians,
sports
and
other
celebrities
well-known
in the
UK.
These
were
selected
from
a
larger
sample
because
they
were
iden-
most
consistently
by participants
in a
pilot
study.
All
faces
cropped
and
placed
in oval
frames
(see
Fig.
1).
Famous
and
faces
were
matched
with
respect
to their
low-level
properties.
In contrast
to previous
studies
that
compared
ERP
to famous
and
non-famous
faces,
where
facial
identity
either
task-irrelevant
(Bentin
&
Deouell,
2000, Exp.
1;
Eimer,
relevant
for
a
minority
of target
faces
(politicians;
& Deouell,
2000,
Exp.
2 and
3),
participants
were
instructed
each
individual
face
in terms
of
its
identity
on
a
four-
scale.
Faces
could
be
classified
as
definitely
known,
merely
unfamiliar,
or definitely
unknown.
were
computed
for
famous
and
non-famous
faces,
sep-
for
each
of
the
four
response
categories.
One
analysis
ERPs
to correctly
classified
famous
faces
(those
judged
definitely
known
or familiar)
and
correctly
classified
non-
faces
(those
judged
to be
unfamiliar
or
unknown),
in order
ERP
correlates
of
face
recognition.
In line
with
earlier
the
N170
component
was
expected
to be unaffected
by
difference
between
famous
and
non-famous
faces.
The
critical
was
whether
famous
faces
would
elicit
an
early
N250
com-
with
a distinct
occipito-temporal
distribution
(as
described
Schweinberger
et al.,
1995,
2004,
and
Tanaka
et
al.,
2006,
in

Page 3

Author's personal copy
2738
A.
Gosling,
M.
Eimer
/ Neuropsychologia
49 (2011) 2736–
2745
Fig.
positions.
1. Examples
of
famous
faces
(top)
and
non-famous
faces
(bottom)
used
in the
present
experiment.
Matching
famous
and
non-famous
faces
are
shown
in corresponding

Page 4

Author's personal copy
A. Gosling,
M.
Eimer
/ Neuropsychologia
49 (2011) 2736–
2745
2739
the
broadly
Bentin
component
rapid
In
distinguish
effects
of
tion,
explicitly
were
famous
primarily
ilar
as
ferences
there
and
correctly
incorrectly
included
tion.
a
encoding
conditions
facial
tion
(see
determine
nition
unimpaired
context
of
face
repetition
procedures),
or only
later
and
more
distributed
N400f
and
P600f
components
(as
observed
by
& Deouell,
2000, and
Eimer,
2000a).
The
presence
of
an N250
to famous
faces
would
provide
new
evidence
for
the
activation
of
long-term
memory
traces
in ventral
visual
areas.
previous
face
repetition
experiments,
it was
difficult
to
ERP
correlates
of
explicit
face
identification
from
associated
with
the
identity-unspecific
generic
familiarity
repeated
faces.
To dissociate
these
two
aspects
of
face
recogni-
a second
analysis
contrasted
ERPs
to famous
faces
that
were
recognized
(“definitely
known”)
and
famous
faces
that
judged
to be
merely
familiar.
If the
ERP
differences
between
and
non-famous
faces
identified
in the
first
analysis
were
driven
by the
explicit
recognition
of
individual
faces,
sim-
effects
should
emerge
in the
contrast
between
ERPs
to known
compared
to just
familiar
famous
faces.
In contrast,
if these
dif-
reflected
mere
familiarity
rather
than
explicit
recognition,
should
be
no
systematic
differences
between
ERPs
to known
just
familiar
famous
faces.
A final
analysis
contrasted
ERPs
to
classified
non-famous
faces
and
to famous
faces
that
were
categorised
as
unfamiliar
or unknown.
This
analysis
was
to reveal
possible
ERP
evidence
for
covert
face
recogni-
An
N250r
to repeated
famous
faces
has
been
observed
when
competing
high-load
perceptual
task
was
present
during
face
(e.g.,
Neumann
& Schweinberger,
2008),
that
is,
under
that
have
previously
been
linked
to covert
processing
of
identity
(Jenkins,
Burton,
& Ellis,
2002).
Covert
face
recogni-
also
remains
a controversial
issue
in the
study
of
prosopagnosia
Schweinberger
& Burton,
2003),
which
makes
it important
to
whether
electrophysiological
evidence
for
covert
recog-
might
even
be
present
in neurologically
and
functionally
participants.
2. Methods
2.1.
Participants
Nineteen
paid
volunteers
were
tested,
and
informed
consent
was
obtained
from
all
number
aged
Fifteen
of
them.
Three
participants
were
excluded
from
analyses
due
to an insufficient
of
trials
after
EEG
artefact
rejection.
The
remaining
16
participants
(8 males;
21–37
years;
mean
age
27.3
years)
had
normal
or corrected-to-normal
vision.
were
right-handed,
one
was
left-handed.
2.2.
Stimuli
and
procedure
Stimuli
were
presented
on
a CRT
monitor
at a viewing
distance
of
100
cm.
E-
Prime
presentation
famous
sample
where
face
for
by
the
dians,
musicians.
larger
sion
examples
to
contours.
All
ahead,
of
The
famous
order,
throughout
by
each
tives
Response
software
(Psychology
Software
Tools,
Pittsburgh,
PA)
was
used
for
stimulus
and
behavioural
response
collection.
The
stimulus
set
contained
80
and
80 non-famous
faces.
The
famous
faces
were
selected
from
a larger
of
129
faces.
This
selection
was
based
on
the
results
of a pilot
experiment
each
face
was
shown
for
400
ms,
and
eight
participants
had
to identify
each
by
naming
the
person
and
stating
their
profession.
The
80
famous
faces
selected
inclusion
in the
main
experiment
were
those
faces
that
were
explicitly
identified
at
least
six
of
the
eight
pilot
participants.
They
were
celebrities
widely
known
to
general
public
in the
UK,
and
included
actors/actresses,
politicians,
chefs,
come-
entrepreneurs,
models,
members
of
the
royal
family,
sports
personalities,
and
For
each
of
the
80
famous
faces,
one
non-famous
face
was
selected
from
a
sample
to provide
a match
in terms
of
gender,
approximate
age,
facial
expres-
and-low
level
visual
attributes
such
as contrast
and
brightness
(see
Fig.
1 for
of famous
and
matched
non-famous
faces).
All
face
stimuli
were
converted
greyscale,
resized,
and
cropped
into
an
oval
shape,
thereby
removing
their
outer
These
image
transformations
were
performed
in Adobe
Photoshop
CS3.
face
stimuli
were
presented
at fixation
in a full
front
view,
with
eye
gaze
straight
against
a light
grey
background
(17.6
cd/m2). They
subtended
a
visual
angle
7.4◦× 4.9◦, and
their
average
luminance
was
21.9
cd/m2.
experiment
consisted
of
eight
blocks
of
80
trials
per
block.
In all
blocks,
and
non-famous
faces
were
presented
with
equal
probability
and
in random
so that
each
individual
famous
and
non-famous
face
was
shown
four
times
the
experiment.
Each
face
was
presented
at fixation
for
400
ms,
followed
a blank
interstimulus
interval
of
1300
ms.
Participants’
task
was
to report
on
trial
whether
they
recognized
a particular
face
by
choosing
one
of
four
alterna-
(definitely
known
– seems
familiar
– seems
unfamiliar
– definitely
unknown).
alternatives
were
mapped
to four
horizontally
arranged
response
keys.
Participants
the
felt
their
mapped
of
block
delivered
were
instructed
to classify
a face
as definitely
known
only
if
they
knew
person’s
name
and
profession,
and
to use
the
‘seems
familiar’
category
if they
that
a particular
face
was
familiar
without
being
able
to name
the
person
or state
profession.
Each
response
key
was
labelled
with
its
response
category,
and
was
to the
index
and
middle
fingers
of the
left
and
right
hand.
The
assignment
keys
and
response
categories
was
kept
constant
across
all
participants.
A training
containing
20
different
famous
faces
and
20
different
non-famous
faces
was
prior
to the
first
experimental
block.
2.3.
EEG
recording
and
data
analysis
EEG
was
DC-recorded
with
a BrainAmps
DC
amplifier
(upper
cut-off
frequency
40
from
P7,
system).
canthi
for
and
line
after
(reflecting
vertical
exceeding
Following
of
unfamiliar,
was
trodes
was
for
measured
(Fz,
tudes
and
performed
ducted.
that
correctly
famous
merely
Table
famous
ses,
third
trials
Greenhouse–Geisser
appropriate.
Hz,
500
Hz
sampling
rate)
and
Ag–AgCl
electrodes
mounted
on
an
elastic
cap
23
scalp
sites
(Fpz,
F7,
F3,
Fz,
F4,
F8,
FC5,
FC6,
T7,
C3,
Cz,
C4,
T8,
CP5,
CP6,
P3,
Pz,
P4,
P8,
PO7,
PO8,
and
Oz,
according
to the
extended
international
10–20
Horizontal
electrooculogram
(HEOG)
was
recorded
bipolarly
from
the
outer
of
both
eyes.
An electrode
placed
on
the
left
earlobe
served
as reference
online
recording,
and
EEG
was
re-referenced
off-line
to the
average
of
the
left
right
earlobe.
Electrode
impedances
were
kept
below
5 k?. No additional
off-
filters
were
applied.
EEG
was
epoched
offline
from
100
ms
before
to 700
ms
stimulus
onset.
Epochs
with
activity
exceeding
±30
?V in the
HEOG
channel
horizontal
eye
movements)
or ±60
?V at Fpz
(indicating
eye
blinks
or
eye
movements)
were
excluded
from
analysis,
as were
epochs
with
voltages
±80
?V at any
other
electrode.
artefact
rejection,
EEG
waveforms
were
averaged
for
all
combinations
face
type
(famous
versus
non-famous)
and
response
category
(known,
familiar,
unknown),
relative
to a 100
ms
pre-stimulus
baseline.
N170
amplitude
quantified
as ERP
mean
amplitude
obtained
at
lateral
occipito-temporal
elec-
P7/P8
in the
160–200
ms
time
interval
after
stimulus
onset.
N250
amplitude
measured
as mean
amplitudes
in a
230–400
ms
post-stimulus
time
window,
lateral
posterior
electrode
pairs
P7/8
and
PO7/8.
P600f
mean
amplitudes
were
during
the
subsequent
400–700
ms
time
window,
at midline
electrodes
Cz,
Pz),
and
at lateral
electrode
pairs
(F3/4,
FC5/6,
C3/4,
P3/4).
ERP
mean
ampli-
were
also
quantified
during
the
230–400
ms
interval
for the
same
midline
lateral
electrodes.
Repeated
measures
analyses
of
variance
(ANOVAs)
were
on
these
mean
amplitude
values.
Three
sets
of
ERP
analyses
were
con-
The
first
set
compared
ERPs
to correctly
categorised
famous
faces
(faces
were
classified
as known
or familiar)
and
ERPs
to non-famous
faces
that
were
categorised
as unfamiliar
or unknown.
The
second
set
compared
ERPs
to
faces
as a function
of
whether
they
were
classified
as definitely
known
or
familiar.
‘Familiar’
responses
to famous
faces
were
relatively
infrequent
(see
1).
After
artefact
rejection,
four
participants
retained
ten
trials
or less
where
faces
were
judged
as familiar.
These
were
excluded
from
this
set
of
analy-
which
therefore
only
included
data
from
the
remaining
twelve
participants.
The
set
of
analyses
compared
ERPs
to famous
and
non-famous
faces
obtained
in
where
they
were
both
categorised
as unfamiliar
or unknown.
For
all
analyses,
corrections
to the
degrees
of
freedom
were
performed
where
3.
Results
3.1.
Behavioural
results
Table
1 shows
the
frequency
with
which
participants
chose
one
of
and
good,
nitely
unfamiliar
for
1.67.
to
tered
to
the
faces
the
four
alternative
response
categories,
separately
for
famous
non-famous
faces.
Overall,
face
recognition
performance
was
with
more
than
70%
of
all
famous
faces
classified
as
defi-
known,
and
nearly
75%
of
all
non-famous
faces
categorised
as
or
definitely
not
known,
resulting
in
an
overall
d?value
the
discrimination
between
famous
and
non-famous
faces
of
To test
whether
individual
non-famous
faces
were
more
likely
be
judged
as
known
or familiar
when
they
had
been
encoun-
before,
accuracy
data
were
quantified
separately
for
the
first
fourth
presentation
of each
face.
Ordinal
position
did
not
affect
probability
of
known
or familiar
classifications
for
non-famous
(F < 1).
Table
Mean
in
1
frequency
(in
percent)
of
each
of
the
four
alternative
classification
judgements
response
to famous
or
non-famous
faces.
Definitely
known
Familiar

Unfamiliar

Definitely
unknown
Famous
Non-famous

71.77
12.54

9.80
12.91

9.08
38.57

9.32
35.96

Page 5

Author's personal copy
2740
A.
Gosling,
M.
Eimer
/ Neuropsychologia
49 (2011) 2736–
2745
Fig.
to
constructed
(230–400
for
of
2.
Grand-averaged
ERPs
based
on
sixteen
participants,
obtained
in the
700
ms
interval
after
stimulus
onset
(relative
to a 100
ms
pre-stimulus
baseline)
in response
famous
faces
classified
as
definitely
known
or
familiar
(solid
lines)
and
non-famous
faces
classified
as unfamiliar
or
unknown
(dashed
lines).
The
topographic
map
was
by spherical
spline
interpolation
(see
Perrin,
Pernier,
Bertrand,
& Echallier,
1989)
and
shows
difference
amplitudes
obtained
during
the
N250
time
interval
ms
post-stimulus)
by subtracting
ERP
mean
amplitudes
in response
to non-famous
faces
from
mean
amplitudes
to famous
faces.
Enhanced
negative
amplitudes
famous
faces
are
shown
in blue,
an enhanced
positivity
in red.
(For
interpretation
of
the
references
to colour
in this
figure
legend,
the
reader
is referred
to the
web
version
the
article.)
Mean
RT
across
all
face
and
response
categories
was
912
ms.
Par-
ticipants
(889
famous
egorisation
943
famous
definitely
gory)
unfamiliar
were
faster
to respond
to famous
than
non-famous
faces
ms
versus
936
ms;
t(15)
= 3.23;
p < .01).
The
classification
of
faces
as
definitely
known
was
faster
(742
ms)
than
their
cat-
as
familiar,
unfamiliar
or definitely
unknown
(1011
ms,
ms,
and
859
ms,
respectively;
all
t(15)
> 3.7;
all
p < .02).
For
non-
faces,
responses
were
faster
when
they
were
classified
as
known
or definitely
unknown
(866
ms
for
either
cate-
relative
to trials
were
they
were
judged
to be
familiar
or
(1040
and
972
ms;
all
t(15)
> 4.3;
all
p < .001).
3.2.
ERP
results
3.2.1.
ERPs
Fig.
lapsed
familiar)
miliar
modulation
linked
ulus
to correctly
classified
famous
and
non-famous
faces
2 shows
ERPs
triggered
in response
to famous
faces
(col-
across
faces
classified
as
definitely
known
or
as
merely
and
ERPs
to non-famous
faces
that
were
classified
as
unfa-
or unknown.
As
expected,
there
was
no familiarity-related
of
the
N170
component.
Differential
ERP
modulations
to facial
identity
started
approximately
230
ms
after
stim-
onset.
At
lateral
occipito-temporal
electrodes,
an
enhanced