Are faces really special

1
Lecture 7 Are faces really special? (Therefore
is face processing special?)
2

• Recap of Lecture 6 Faces ARE Special
• Ontogeny Faces preferred over objects. Linked to
  innate module in the brain for processing
  faces.
• Inversion Inversion cost greatest for faces.
  Evidence for configural processing of faces
  (orientation sensitive) and part-based processing
  of objects.
• Agnosia Double dissociation in agnosia.
• fMRI Evidence Fusiform gyrus specialized module
  for face processing (Kanwisher).

3

• Therefore, if one can show the following
• Ontogeny Other objects also show preferred
  processing.
• Inversion Inversion cost is also present for
  other types of image.
• Agnosia Examples of agnosia for faces and
  certain objects.
• fMRI Evidence Activation in FG for other types
  of array.
• Then, this would be evidence that faces are NOT
  special.

4
Clarifying the term Special

• The notion of special needs clarification. It
  can be
• addressed in the form of two separate questions
• Is there a part of the cortex specially activated
  by faces?
• Does face recognition involve a unique process?
• Hay Young (1982)

5
What about evidence against the notion of
face-specific processing?

• Take Ellis criteria again and re-examine the
  evidence.
• Ontogeny (Development)
• The Effect of Inversion (Cognitive)
• Different Types of Agnosia (Neural)
• Different Brain Area (Neural)

6
Developmental Mechanisms Ontogeny
7
Neonates track face-like stimuli more than
scrambled faces. Goren et al. , 1975 Johnson et
al., 1991).
Preference also evident for patterns reflecting
face-like structure. Thus faces themselves not
special (Simion et al., 2002 B.J. Dev. Psy), but
some innate structural knowledge (CONSPEC)
gt
Can replicate emotion. (Meltzoff Moore,
1977). But authors also show ability to repeat
sequences of finger movements too!
8

• Specific Problems with Ontogeny Argument
• 1. Is tracking the same as identification? Is it
  supported by cortical or subcortical mechanisms?
  Evidence that cortical mechanisms not innate but
  the product of experience (Johnson Morton,
  1991).
• 2. Replication of Emotion The Bruce Young
  Model separates emotion from identification.
• a. Neonates also show replication of finger
  movements.
• b. Newborns seem to prefer any stimulus that has
  more elements on top than on the bottom
  (Simion et al., 2002).
• c. Ellis Young (1998) argue that other
  important factors in the lives of neonates
  should also elicit greater fixation (e.g.
  breasts).

9
Evaluation of the Ontogeny Argument
Special Brain Region Not really possible to
evaluate this as early studies do not measure
brain activity. However suggestion of
subcortical mechanism (CONSPEC). CONLERN may
have cortical origins, but will discuss later if
this is special. Unique Process Since simple
top-heavy patterns also elicit tracking and
manual gestures also imitated no evidence of
unique face-specific process.
10
Cognitive Mechanisms The Effect of
Inversion (Reflects holistic processing)
11
The Issue of Stimulus Comparisons
A major problem in comparing face and object
recognition is the nature of the set of control
stimuli used. Failure to match stimuli across
several criteria can render the results
misleading.

• 1. Homogeneous vs. Heterogeneous Categories.
  (Faces vs. Houses)
• 2. Basic vs. Subordinate level categorization.
  (Bird vs. Robin)
• 3. This level of categorization becomes automatic
  for the expert viewer. Therefore necessary to
  contrast groups with similar levels of expertise
  (Dog vs. Bird vs. Face).

Expertise interacts with categorization level so
tasks examining face-specific effects need to be
contrasted with control tasks using stimuli for
which participants are similarly expert.
12
The Face Inversion Effect
Yin (1969)

• A general factor of familiarity
• (for objects in their normal orientation)
• A factor specific to faces
• Could this meet the unique process hypothesis?

13

• Diamond Carey (1986) similarly argued for a
  two-factor
• explanation for the face-inversion effect.
• Familiarity - Expertise
• Homogeneity of the Stimulus Class
• (i.e similarity of composition of the exemplars
  in a class)
• They say that if both the above criterion are met
  then you should
• Find inversion effects for non-face stimuli as
  well.

14
Most classes of animal meet the criterion of
homogeneity. That is they share a common
configuration. However in general we lack
FAMILIARITY with them.
So what if this criterion is met?
15
Non-Face Inversion Effects
Diamond Carey (1986) presented upright and
inverted pictures of Dogs to dog experts
(satisfying the familiarity criterion) and
non-experts.
Significant inversion effect for dog experts.
16
Suggestion that Expertise is the result of
repeated exposure to a class of stimuli that
satisfy the criterion of Homogeneity within
class Multiple Levels of Abstraction. Can
therefore create sets of novel stimuli that meet
the above and through training can develop
expertise for these items.
Gauthier Tarr (1997) Show that Greeble
experts are susceptible to manipulations of
configural processing - better at part
recognition in whole face than in
isolation. Need to compare like with like in
contrasting face and object recognition.
17
Cognitive Effects Unique Process? Faces only
different when expertise, homogeneity and level
of abstraction are not controlled for.
Homogenous Equal Expertise Subordinate Level
Identification
VS.
Inversion does not reflect a unique process
18
But does it occur in a special brain region?
Neural Patterns (Different Patterns of Agnosia)
Is there really a dissociation between face and
object agnosia? This would be evidence for a
special brain region
Isobel Gauthier
19
Some evidence that prosopagnosia is not
face-specific. Bornstein et al. (1969) report
the case of a prosopagnosic farmer who lost the
ability to recognize his animals as well as his
family. Shuttleworth et al. (1982) also suggest
that prosopagnosia may be characterised by
difficulties in exemplar-based recognition.
Other objects possessing a degree of complexity
of form approaching that of faces (for example
various birds, automobiles, or pieces of
furniture) while usually identified by class of
object are almost always misidentified as to
individuality within that class Gauthier et
al. (1999) also report more general impairments
in subtle, subordinate level discriminations.
20
Remember Most brain damage is not nicely focal
in nature. Strokes, gunshots, head injuries etc.
are not, by their nature, very neat and tidy.
This literature is somewhat confused, but this
may be as much to do with the extent of lesions
in the patients concerned. The best we can say
is that perhaps this literature is
inconclusive. However, we can interpret it in
light of the data from fMRI in normal subjects.
21
But does it occur in a special brain region?
Neural Patterns (Activity in Fusiform Gyrus)
Is there really a differece between face and
object procesing in the ventral-temporal
cortex? This would be evidence for a special
brain region
Isobel Gauthier
22
Is the Fusiform Gyrus a face-specific region?
Look back at the Kanwisher data. Note that
activation for objects, scrambled faces, houses
and hands is above baseline. FG above baseline
for all objects.
23
Does the Fusiform Gyrus care about faces or just
the level of categorization? Gauthier et al.
(1997) Current Biology.
Participants scanned while taking part in
basic-level (BIRD) versus subordinate-level
(PELICAN) judgment about objects.
24
Region of Right and Left FG more active for
subordinate than basic level judgments. May
reflect differences in perceptual processing
across these types of judgment.
25
Can person identification in FG also be modulated
by differences in categorical processing?
Turk et al. (2005) Seeing John Malkovich.
NeuroImage
VS.
Basic Level
Subordinate
Percept is the same, but categorization differs
in specificity
26
IdentitygtOccupation Contrast Greater activation
of FG for subordinate judgment. But both tasks
activate the same regions. Not due to perceptual
differences. Responses also faster for identity
responses - automaticity of subordinate judgments.
27
Objects are automatically categorized at the
basic level (e.g. dog) whereas people tend to be
identified at the individual level (e.g. John
Malkovich). This downward shift in
identification from basic to individual/subordinat
e is characteristic of expert processing (Tanaka,
2001) and of increased brain activation in FG
(Gauthier et al., 1997 Gauthier et al., 2000
Turk et al., 2005). When objects are processed
at a subordinate level this leads to activation
of brain regions previously argued to be
face-specific.
28
Does the FG Care About Expertise with a Class of
Object? Gauthier et al. (2000)
Same Species vs. Same Model vs. Same
Person (Subordinate)
Same Species?
Same Model?
29
Does the FG Care About Expertise with a Class of
Object?
Bird Experts
Car Experts
30
Brain of a Bird Expert
FacesgtObjects
CarsgtObjects
BirdsgtObjects
31
So are Faces Special?

• Hay Young Clarification
• Ellis Criteria Special Brain Region Unique
  Process
• Development No. No.
• (Ontogeny) Sub Cortical. Hands, Patterns.
• Cognitive N/A No. Expertise
• (Inversion/Categorization) effects for dogs.
• Agnosia No? Prosopagnosic
• also has object rec.
• difficulty.
• fMRI Data No. No. Faces like
• Level of Specificity expert objects.

32
Summary Farah, Kanwisher, Yin, Duchaine
etc. Faces are special. Inversion effects,
holistic processing, different agnosia profiles
and specific activation of the fusiform
gyrus. Gauthier, Diamond Carey, Tanaka
etc. Faces are not special. We are simply face
experts through repeated exposure to faces.
Person recognition reflects differences in level
of categorization and differences in stimulus
class homogeneity. When these other factors are
controlled for face and object processing show
similar cognitive and neural profiles.
33
Reading Material Diamond Carey (1986) Journal
of Experimental Psychology General, 115,
107-117. Gauthier et al. (2000) Nature
Neuroscince, 3, 191-197. Kanwisher, McDermott
Chun (1997) Journal of Neuroscience, 17,
4302-4311. McKone, Kanwisher Duchaine (2006)
Trends in Cognitive Sciences Yin, RK (1969)
Journal of Experimental Psychology, 81,
141-145. Kanwisher N. Moscovitch M (2000) The
Cognitive Neuroscience of Face Processing,
Cognitive Neuropsychology (Special Issue), Vol
1-3.