The area that does holistic processing comes
The fusiform face area (FFA) is an area of the brain in the inferiortemporal lobe that characteristically responds preferentially to face-typestimuli compared to other stimuli (BAARS& GAGE, 2013; Kanwisher, McDermott, & Chun, 1997). However, there is controversywithin the field as to whether this increased activity is due to a specificresponse to faces as opposed to a general response to things with which one hasexpertise. Because humans are extremely social creatures, and a large amount ofsocial interaction depends on properly interpreting facial expressions,everyone needs to build a certain amount of expertise with faces in order toparticipate normally in society (Haxby,Hoffman, & Gobbini, 2002). This paper will evaluate theevidence for and against the specialization of the FFA as a face processingarea.
Beforewe look at the studies that provide evidence for and against the fusiform facearea being face specific, we can look at different theories that could be usedto argue for or against the likelihood of a brain area like it existing,specifically a brain area that processes faces holistically instead of featureby feature. The most interesting debate in theory about the use of a brain areathat does holistic processing comes from the evolutionary perspective. The mainquestion evolutionary psychologists ask about the fusiform face area is whatthe effect of having a separate brain area for holistic processing would be onfitness. One side of the debate argues that the evidence for holistic faceprocessing in non-humans faces is insufficient to conclude that there is clear holisticprocessing occurring (Burke& Sulikowski, 2013). For example, though changing facialmarkings on cows makes them less likely to be approached by herd mates (Coulon,Baudoin, Heyman, & Deputte, 2011), suggesting that they are no longerrecognized, it is not conclusive whether or not this is due to a speciallyevolved holistic processing of faces or the simple cooption of preexistingvisual discrimination structures. However, researchers on the other side arguethat because both humans and primates have difficulties learning inverted(upside down) or split faces (faces which are made by aligning the top half ofone face with the bottom half of another) this suggests that the evolvedprocess for processing faces is both holistic and configuration based (McKone,Kanwisher, & Duchaine, 2007), and FFA activation is correlated tothis type behavioral process suggesting that it is an area evolved specificallyfor this type of processing (Kanwisheret al., 1997).
Theoriginal study that identified the Fusiform Face Area was conducted byKanwisher et al. in 1997. It argued found that there was a specific area in theinferior temporal lobe that activates preferentially to faces – specificallyintact (not visually scrambled) faces – than it does to any other stimulus. Theyused this to argue that the region is selectively involved in facialprocessing. Since then there has been debate in the field about whether thisconclusion is accurate. We will start by looking at studies that disagree withthe conclusion that Kanwisher et al.
came to, and then look at the response tothese critiques. Thereare many different critiques of the face specificity argument, most (if notall) of which argue that though the FFA is more active when viewing faces, thiseffect is due to expertise, because faces are something we have a lot of reasonto build expertise in identifying. The first study we will look at that makesthis argument is about general processing of faces of different races. In thisstudy, they asked participants to recognize both individual features and whole imagesof White and Asian faces. What they found was that participants were muchbetter at holistic recognition of same race faces than of different race faces.Because the White participant sample was taken in Australia and the Asianparticipant sample was taken in Hong Kong, this difficulty mirrored therelative amounts of experience the participants had with people of other races (Hayward,Rhodes, & Schwaninger, 2008). This finding suggests that, on thebehavioral level, face processing is at least partially experience based.
However this study looks at general facial processing and not specifically atactivation in the FFA. Perhaps activation in the FFA does not relate toexperience in the way behavioral responses do. A similar study by a differentlab found that differential memory for same race and different race faces wascorrelated with FFA activation (Golby,Gabrieli, Chiao, & Eberhardt, 2001), showing that it is not just generalfacial processing, but specifically FFA facial processing that plays a role inexperience based learning. Other studies look at whether faceand object processing in the FFA can really be distinguished. One study foundthat when bird and car experts are shown pictures of birds and cars their FFAis more active than when they are shown pictures of objects they do not haveexpertise with (Gauthier,Skudlarski, Gore, & Anderson, 2000).
This study was originally critiquedbecause from certain angles bird faces and cars can resemble human faces, but asubsequent study that used images of these objects specifically chosen to notresemble human faces got the same result (Xu,2005). This activation to objects forwhich a character has expertise can also be created: Gauthier et al. conducteda study in which participants were trained to identify and distinguish novel semiface-like objects called ‘greebles’. When the researchers compared activationof face-specific areas of the FFA before and after expertise training, theyfound that activation in the FFA increased for greebles, but not for faces (Gauthier,Tarr, Anderson, Skudlarski, & Gore, 1999). Gaining expertise in anidentification task increases activation in “face” areas of the fusiform gyrus,which suggests that these areas are more expertise based than face based. Several studies were used to respondto Gauthier et al.’s bird and car expertise paper that aimed to distinguish theway faces and objects of expertise are processed. In a paper that inspiredthese studies, researchers had people try to identify components of compositefaces made by combining the top half of one face with the bottom half ofanother.
They found that people have much more trouble identifying the halveswhen they are aligned to form a single composite face than when they are notaligned. However, this specific difficulty identifying face components incomposites disappears when the faces are inverted (flipped upside down) (Young,Hellawell, & Hay, 2013). These findings suggest that face processing is specific to theconfiguration and direction of face elements. McKone et al. repeated thisexperiment (composites and inversion), but added non-face images which someparticipants had expertise with.
Here they found that the identificationdifficulty effects of composites and inversion do not affect identification,which suggests that face processing is done differently than processing ofobjects with expertise. This is relevant when it comes to fMRI data, because ithas been found that activation in the FFA depends on having most or all of theelements of a face visible in the correct configuration (Tong,Nakayama, Moscovitch, Weinrib, & Kanwisher, 2000). Another defense of the specificity ofthe FFA is that in the case of increased activation to items of expertise, theincrease in signal in the FFA to images of items of expertise is due to theparticipant paying more attention to these items, and the increase in signalfor items of expertise is higher in other areas of the cortex than it is in theFFA (McKoneet al., 2007). A critique of the ‘greeble’ studiesis that the training that teaches participants how to categorize greebles doesnot actually confer expertise, and during identification participants are usingordinary object recognition mechanisms to complete the task (Duchaine,Dingle, Butterworth, & Nakayama, 2004). This claim is supported by two casestudies of people with acquired prosopagnosia: both were able to learn andperform well on greeble identification tasks, but unable to do the same forface recognition tasks, and in one subject this learning occurred even thoughthey did not have a fully functioning FFA (Rezlescu,Barton, Pitcher, & Duchaine, 2014), which strongly suggests that theFFA is not necessary for greeble learning. The other common critique of certainpapers that are arguing for the expertise model is that the region they definedas the FFA extends past what is typically defined as the FFA, and isincorporating non FFA signal (McKoneet al.
, 2007; Rezlescu et al., 2014). Inreading the arguments presented by both sides of this debate I have beensomewhat convinced away from the pure face-processing model of the FFA.However, I do still believe that the FFA plays an important role in theholistic processing and identification of faces. There is clearly evidence forboth expertise based processing and face specific configuration basedprocessing in the FFA. Looking at the two arguments I am entirely willing tobelieve that both things can be true simultaneously and occur in the same area.
In individual cell recordings of an analogous region to the FFA in monkeys, researchersfound that though the vast majority of cells were extremely reactive to faces,there were also some cells that were reactive to other things (Tsao,2006). There are many different cells inthe FFA, and it is entirely possible that different cells can be accomplishingdifferent tasks, or that the reaction FFA cells have to faces is a combinationof a reaction to an item of expertise and a reaction to a specific configurationof features – we have seen cells that are tuned in similar ways in other areasof the brain in class. I am willing to grant that as far as we know, only facesare processed in the holistic, configuration based way that has been describedby the studies I explained earlier, but that does not mean that nothing elsecan be processed by the same brain area.