Just guiding the research of face processing. They

Just by briefly looking at a face, we can identify a wide
range of information about a person. We are able to access invariant
information – such as age, sex and ethnicity. As well as changeable information
– such as emotion. Although the ease of which we are able to do this makes the
process seem uncomplicated, Face perception is arguably the most advanced
visual ability that we possess. The extensive research that has been conducted
on this skill has shown that face perception is dependent on preferential
response to faces of the network of neural mechanisms (Duchaine & Yovel, 2015).


It is long standing knowledge that face selective cells have
been found present in monkey’s inferior temporal (IT) cortex (Gross, 2005). In more
recent research using fMRI studies, it has also been found that face
selectivity is present in the same regions of the brain within monkeys (Tsao, Freiwald, Knutsen,
Mandeville & Tootell, 2003.; Tsao, Freiwald, Tootell & Livingstone,
2006) and Marmosets (Hung,
Yen, Ciuchta, Papoti, Bock, Leopold & Silva, 2015) However, findings
from these studies are not always consistent and require further research to
establish a solid relationship between these domains. In Humans however, there
are multiple regions that are responsible for the perception of faces.


and colleagues (2000) neural model, has been influential in guiding the
research of face processing. They theorised that there was a rift between the
core system that is responsible in the way that we process the visual
appearance of faces– made up of the Occipital Face Area (OFA), the Fusiform
Face Area (FFA), and the posterior Superior Temporal Sulcus (pSTS)- and a more
complex, extensive system that derives the socially relevant information from
faces, such as emotions (Visconti
di Oleggio Castello, Halchenko, Guntupalli, Gors & Gobbini, 2017).
The FFA itself has been at the centre of a debate, modularists theorise that
the FFA is specifically responsive to faces (Kanwisher & Yovel, 2009), while generalists
challenge that assumption and believe that FFA is an expert of a wide range of
visual categories, including faces but it is not specifically responsive to
them (Gauthier et al.,
2003). Beyond that is Haxby and colleagues assumption that facial
perception is taken out by a much more complex network that is beyond FFA


This essay will aim to support Haxby et al.’s (2000) model, and focus solely on
this core system and will integrate the use of
functional-Magnetic-Resonance-Imaging (fMRI) and
Transcranial-Magnetic-Stimulation (TMS) as proof for this system, as well as
bringing in some research from functional-Magnetic-Resonance-Angiography (fMR-A)
studies. These will be used to evaluate the core system that is involved in
face perception. Each






In the model
proposed by Haxby et al. (2000), the FFA is mainly responsible for processing
information such as facial identity, but is much less involved in processing
things such as facial expression, which are aspects that are changeable. From
the time when the model was introduced, this has been an area of great interest
to researchers, and has received a great deal of support from fMR-A studies.
These studies have been able to establish the degree to which the
representation of facial identity is view specific, or view invariant. They
have compared the FFA’s response to two same-identity/same-view faces to two
same-identity/different-view faces. They found a higher response to the
same-identity/different-view faces, which indicated that these faces are
characterised as different in the FFA, leading researchers to conclude that the
representation of identity is view-specifc.

Gouws & Andrews 2009; Ewbank & Andrews, 2008; Xu & Biederman, 2010;
Duchaine & Yovel, 2015).


Kanwisher et al. (1997) found reliable fMRI activation of FFA
in over 80% of their participants, the FFA favoured faces over a range of
assorted objects such as cutlery, cars and animals. Pruce et al (1996) found similar results
using fMRI techiques, finding that facial stimuli caused a significant amount
of right hemispheric activation, with characteristic patterns localised to the
FFA in comparison to when the participants were shown letter strings. These
studies both support Haxby et al.’s (2000) first FFA predication that FFA is extremely
attuned to face stimuli and lead researchers to label the FFA as the face


prediction of FFA is that it is selectively receptive to a range of invariant
facial information such as identity. Tong et al. (2000) with fMRI studies, discovered
that the FFA responded equally as strong to cartoon and human faces, as well as
showing an equal reaction to front and profile views but a very minor
activation for non-facial stimuli. This too supports the second prediction of
Haxby et al (2000).


There are
also claims that the FFA is involved in expression processing (Bernstein & Yovel 2015).
Ganel, Valyear,
Goshen-Gottstein & Goodale (2005) found that the FFA responds very
strongly to facial expression when it is both attended and unattended and
following this, its response is influenced by how intense the facial expression
is. fMR-A studies similarly found that the FFA is attuned to variations in
facial expressions across faces (Xu & Biederman, 2010). 
Furthermore, when studying a patient who suffered brain damage which
resulted in the destruction of the right FFA, but no damage to the right OFA or
right pSTS who had poor expression recognition, it was suggested that the FFA
must contribute to expression recognition for static faces at the very least (Dalrymple et al. 2011).
With this being said, it is indicated that the FFA and the pSTS are responsible
for expression processing, however they may extract different types of
information from an expression. Said, Haxby & Todorov (2011) presented participants with computer generated faces, these faces
differed from average faces in different ways. In one condition the faces
varied in expression, in the other they varied in face shape regarding typical
face shape, but not in expression. The FFA was found to show a reaction to
deviations from the average face in both conditions, where the pSTS was only
sensitive to those that differed in expression. Suggesting that the response of
the FFA could reflect a broad sensitivity to shape information, where the pSTS
may only be responsive to facial stimuli that give emotional information.






The functions
that the Occipital Face Area carries out are complimentary to that of
FFA. Haxby et al. (2000) theorise that it processes invariant information prior
to FFA processing it, and focuses specifically on facial features. These claims
were tested by Pitcher et
al. (2009) who used TMS studies to substantiate this. Using stimuli
consisting of faces, bodies and novel objects, they demonstrated a triple
dissociation in OFA using perceptual discrimination tasks that involved stimuli
consisting of faces, bodies as well as novel objects. Alongside the OFA they
tested the extrastriate body area (EBA) and the Lateral Occipital complex
(LOC). Firstly they used fMRI to confine the appropriate locations with the
region of interest (RoI). RoI allows regions of the brain to be studied
objectively without an individual’s anatomical differences having an effect (Kanwisher & Yovel, 2009).  Hereafter the TMS was applied and delivered
discerning effects, where OFA disruption did influence the participant’s
decisions on facial stimuli, it did not affect the participant’s decisions on
any of the other stimuli types. EBA was found to influence perception on bodies
and LOC influenced perception of novel objects, disruptions to these areas also
did not show any effect on participant’s decisions of other stimulus types. Showing
that there must be distributed modality, the discerning processing of different
perceptual categories must take place in distinct regions and therefore
supporting the Haxby et al. (2000) model.



The third key
prediction of Haxby et al’s (2000) model is that STS reacts to the unpredictable
and changeable characteristics of faces, such as gaze movement, facial
expression and movement of the mouth. fMRI and TMS studies have been used in
order to corroborate the claims of Haxby et al.’s (2000) model.


A group who
have been of high interest in this area are individuals with Autistic Spectrum
Disorder (ASD). This is due to the association with abnormal gaze and emotion
perception (Golarai et al.
2006). Pelphrey et
al. (2005) discovered, when using an event related fMRI, that in
individuals with ASD, STS was not prone to intention violation, as were the
control group. In the study an actors gaze would differ when a new object
entered the scene, the gaze would be moved to the expected – the object, or the
unexpected- an open space. Both groups equally noticed the change in gaze, but
STS activation differed significantly between them. It was also found that STS
was more active when the individual was attending more to gaze than facial
identity, this proposes the uniqueness from FFA.


TMS has also
been used to localise and disrupt STS. Grossman et al. (2005) delivered a 10-minute
succession of repetitive low frequency stimulations to individuals. They were
delivered at 1Hz as this is used to produce lengthier effects than if it was
delivered at 10Hz. When placed over the right posterior STS these stimulations compromised
biological motion perception, this was measured by point light animations. This
observation shows the STS’s fundamental role in understanding dynamic
information, which is extremely important following facial perception.


to these views, some speculate that STS is activated from any generic stimuli
that transmits intention, not just face-specific stimuli, as it was shown that
STS was activated when basic geometric shapes transmitted intention (Golarai et al., 2006). Contrary
to this, other research done using fMRI indicated that STS is receptive to both
moving and motionless eyes, as well as mouths, but did not respond to moving
shapes (Puce et al.,
1998).  Nevertheless, both of
these perspectives suggest that STS activation is imperative in reacting to
signals that are important to prospective behaviours that occur after facial


is a type of visual agnosia that is specific to faces, however it is very rare
that an individual just has prosopagnosia, it is extremely common for them to
have other recognition deficits along side it (Kanwisher, 2000). Anatomical
differences between patients have made it extremely difficult to generalise
information in support or opposition of claims made in regards to the
condition, however TMS has made in possible for researchers to recreate the
effects in healthy patients to demonstrate modularity within the face
perception network. The lack of redundancy within the FFA, OFA and pSTS have
been theorized to diminish the capability for reorganisation amongst the intact
regions, and therefore damage to any individual region can be more disastrous
(DeGutis, Chiu, Grosso & Cohan, 2014). fMRI studies have shown that the FFA
is responsive to both face configuration as well as parts of the face, whereas
OFA and pSTS are responsive to the parts of the face, but not the correct
configuration of these (Liu
et al., 2010). Studies have shown the functional autonomy that is
present in the face processing network. Barton (2008) discovered that patients who had
lesions to their right occipital-temporal regions had much more specific
discrepancies when perceiving facial structure and configuration, in
particularly the eye area, while those who had anterior temporal damage had
much more difficulty in accessing memories of faces.



There are many studies that back up the claims of Haxby et
al’s (2000) model and exhibit the dissociation amid the mechanisms of the
system. In particularly, Hoffman
& Haxby (2000) used fMRI and repetition-detection tasks, firstly
participants were instructed whether they were required to concentrate on the
identity of the face, or the direction of the gaze. Identity was predominantly
related to FFA and OFA, whereas gaze was linked with STS. By using this block
design, they determined invariant and changeable processing systems, one that
processes identity, and the other that processes expression and gaze, which is
consistent with the forecasts of Haxby et al’s (2000) model.

In Summary, there is a wealth of evidence from fMRI, TMS and fMR-A studies
to support three apparent regions that are involved in the perception of faces,
in turn supporting the Haxby et al. (2000) model. FFA is extremely responsive
to the generic facial stimuli and it also processes identity holistically,
additionally it is thought to be related to face memory (Kanwisher & Yovel, 2009; Golorai et
al., 2006). OFA
is an area that is much more responsive to facial features (Pitcher at al., 2009)
and STS is receptive to the unpredictable and changeable information that a
face can give us (Golorai
et al., 2006).  Although these
regions are all extremely valuable and important in the perception of faces,
Haxby et al. (2000) clarified that face perception is also reliant on external
processes. Using fMRI studies, Rossion et al. (2012) discovered multiple regions that are
involved in the process that go beyond the core system, these regions were
discovered by using multi-dimensional stimuli. They concluded that connectivity
between systems is just as important as modular functioning in the perception
of faces, and Haxby et al. (2000) proposed a link between and within the core
and extended networks.

Haxby et al. (2010) supported this view by proposing a further discrepancy
between maximal and significant activation within the facial perception
network. The FFA, OFA and STS maximally react to invariant and changeable
information respectively, but extended systems, for example the amygdala are proposed
to respond significantly to this information. Haxby et al (2010) suggested that
this is because they are more vulnerable to the more generic traits that
stimuli possess. This discrepancy is valuable in supporting the idea of
distributed-modularity. Turati
(2004) proposed that this allows general expertise effects to be
supportable as non-facial stimuli may actually resemble a face in generic ways.
For example the front of a building may comprise of symmetry similar to that of
a face, in turn activating FFA. In disagreement to this claim Kanwisher & Yovel (2009)
found that facial stimuli reliably activated FFA compared to non-facial

An interesting claim was made by Tsao & Livingstone (2008), they proposed that all visual
stimuli have the same issues. These issues are image changes regarding
position, illumination and occlusion. Futhermore, Turati (2004) suggested that faces are unique and
special because they have noticeable perceptual configuration, they proposed
that this is due to us innately having non-specific perceptual boundaries.