There are various reasons for devoting a separate section to face recognition.

FACE RECOGNITION
There are various reasons for devoting a separate section to face recognition.
First, the ability to recognise faces is of great significance in our everyday lives.
Second, face recognition may differ from other forms of object recognition.
Third, we now know a considerable amount about the processes involved in face recognition.
Models of face recognition
Bruce and Young’s model
Influential models of face recognition were put forward by Bruce and Young (1986) and Burton and Bruce(1993). There are eight components in the Bruce and Young (1986) model (see Figure 3.17):
· Structural encoding: this produces various representations or descriptions of faces.
· Expression analysis: people’s emotional states can be inferred from their facial features
· Facial speech analysis: speech perception can be aided by observing a speaker’s lip movements.
· Directed visual processing: specific facial information may be processed selectively.
· Face recognition units: they contain structural information about known faces.
· Person identity nodes: they provide information about individuals (e.g., their occupation, interests).
· Name generation: a person’s name is stored separately.
· Cognitive system: this contains additional information (e.g., that actors and actresses tend to have attractive faces); and influences which other components receive attention.
The recognition of familiar faces depends mainly on structural encoding, face recognition units, person identity nodes, and name generation. In contrast, the processing of unfamiliar faces involves structural encoding, expression analysis, facial speech analysis, and directed visual processing.
Experimental evidence
Bruce and Young (1986) assumed that familiar and unfamiliar faces are processed differently, and there is much support for this assumption (see Schweinberger and Burton, 2003).For example, if we could find patients showing good recognition of familiar faces but poor recognition of unfamiliar faces, and other patients showing the opposite pattern, this double dissociation would suggest that the processes involved in the recognition of familiar and unfamiliar faces are different.
Malone et al. (1982) tested one patient with reasonable ability to recognise photographs of famous statesmen (14 out of 17 correct), but who was very impaired at matching unfamiliar faces. A second patient performed normally at matching unfamiliar faces, but had great difficulty in recognising photographs of famous people (only 5 out of 22 correct).
According to the model, the name generation component can be accessed only via the appropriate person identity node. As a result, we should never be able to put a name to a face without at the same time having available other information about that person (e.g., his or her occupation). Young, Hay, and Ellis (1985) asked participants to keep a diary record of the problems they experienced in face recognition. There were1008 incidents altogether, but participants never reported putting a name to a face while knowing nothing else about that person. In contrast, there were 190 occasions on which a participant could remember a fair amount of information about a person, but not their name.
Practically no brain-damaged patients can put names to faces without knowing anything else about the person, but several patients show the opposite pattern. For example, Flude, Ellis, and Kay (1989) studied a patient, EST, who could retrieve the occupations for 85% of very familiar people when presented with their faces, but could recall only 15% of their names.
According to the model, if the appropriate face recognition unit is activated, but the person identity node is not, a feeling of familiarity should be coupled with an inability to think of any relevant information about the person. In the incidents collected by Young et al. (1985), this was reported on 233 occasions.
Reference back to Figure 3.17 suggests further predictions. When we look at a familiar face, familiarity information from the face recognition unit should be accessed first, followed by information about that person (e.g., occupation) from the person identity node, followed by that person’s name from the name generation component. Thus, familiarity decisions about a face should be made faster than decisions based on person identity nodes. As predicted, Young et al. (1986b) found that the decision as to whether a face was familiar was made faster than the decision as to whether it was the face of a politician.
According to the model, decisions based on person identity nodes should be made faster than those based on the name generation component. Young et al. (1986a) found that participants were much faster to decide whether a face belonged to a politician than they were to produce the person’s name.
Evaluation
The model of Bruce and Young (1986) provides a coherent account of the various kinds of information about faces, and the ways in which these kinds of information are related. Another significant strength is that differences in the processing of familiar and unfamiliar faces are spelled out.
There are various limitations with the model. First, the account of the processing of unfamiliar faces is much less detailed than that of familiar faces. Second, the cognitive system is vaguely specified.Third, the theory predicts that some patients should show better recognition for familiar faces than unfamiliar ones, whereas others show the opposite pattern.
This double dissociation was obtained by Malone et al. (1982), but has proved difficult to replicate. For example, Young et al. (1993) studied 34brain-damaged men, and assessed their familiar face identification, unfamiliar face matching, and expression analysis. Five patients had a selective impairment of expression analysis, but there was much weaker evidence of selective impairment of familiar or unfamiliar face recognition.
Interactive activation and competition model
Burton and Bruce (1993) developed the Bruce and Young (1986) model and the theory of Valentine et al. (1991), and this was developed further by Burton, Bruce, and Hancock (1999). Burton and Bruce’s interactive activation and competition model adopted a connectionist approach (see Figure 3.18). The face recognition units (FRUs) and the name recognition units (NRUs) contain stored information about specific faces and names, respectively. Person identity nodes (PINs) are gateways into semantic information, and can be activated by verbal input about people’s names or by facial input. Thus, they provide information about the familiarity of individuals based on either verbal or facial information. Finally, the semantic information units (SIUs) contain name and other information about individuals (e.g., occupation, nationality).
Experimental evidence
The model has been applied to associative priming effects found with faces. For example, the time taken to decide whether a face is familiar is reduced when the face of a related person is shown immediately beforehand (e.g., Bruce & Valentine, 1986). According to the model, the first face activates SIUs, which feed back activation to the PIN of that face and related faces. This then speeds up the familiarity decision for the second face. Since PINs can be activated by both names and faces, it follows that associative priming for familiarity decisions on faces should be found when the name of a person (e.g.,Prince Philip) is followed by the face of a related person (e.g., Queen Elizabeth). Precisely this has been found (e.g., Bruce & Valentine, 1986)
One difference between the interactive activation and competition model and Bruce and Young’s(1986) model concerns the storage of name and autobiographical information. These kinds of information are both stored in SIUs in the Burton and Bruce (1993) model, whereas name information can only be accessed after autobiographical information in the Bruce and Young (1986) model. The fact that the amnesic patient, ME could match names to faces in spite of not accessing autobiographical information is more consistent with the Burton and Bruce (1993) model (de Haan, Young, and Newcombe, 1991). In similar fashion,Cohen (1990) found that faces produced better recall of names than of occupations when the names were meaningful and the occupations were meaningless. This could not happen according to the Bruce and Young (1986) model, but poses no problems for the Burton and Bruce (1993) model.
Covert face recognition
We turn now to patients suffering from pro- sopagnosia, a condition in which familiar faces cannot be recognised consciously but common objects are recognised. In spite of the lack of con- scious recognition of faces, prosopagnosics often show evidence of covert recognition which can be assessed in various ways. For example, Young, Hellawell, and de Haan (1988) gave prosopag- nosics the task of deciding rapidly whether names were familiar or unfamiliar. They performed this task more rapidly when presented with a related priming face immediately before the target name, even though they could not recognise the face overtly. That is an example of covert recognition assessed by a behavioural measure. Covert recog- nition can also be assessed physiologically. For example, when people with prosopagnosia are presented with familiar and unfamiliar faces, there are larger skin conductance responses (reflecting a state of physiological arousal) to the familiar faces even when overt recognition is at chance level (e.g., Tranel Damasio, 1988).
Schweinberger and Burton (2003) developed the previous models of Burton and Bruce (1993) and Burton et al. (1999). (see Figure 3.19). One of their key assumptions was that the same func- tional system is used in overt recognition (involv- ing conscious recognition) and covert recognition as assessed by behavioural measures (as in the priming study of Young et al., I 988). In essence, the brain damage suffered by prosopagnosies means that the links between face recognition units and person identity nodes are weakened at loca- tion A (see Figure 3.19). Covert face recognition is less affected by this damage than is overt face recognition, because measures of covert recogni- tion are more sensitive.
Evidence Three kinds of evidence support the notion that the same system is used in overt recognition and covert recognition assessed behaviourally. First, patients who have reasonably good overt face per- ception typically show more covert recognition than patients with very poor overt face perception (see Schweinberger & Burton, 2003). Second, there are no cases in which patients show intact overt face recognition but impaired covert recog- nition as assessed by behavioural measures. This is exactly what would be expected from the theory. Third, if prosopagnosics have a weakly function- ing face recognition system, they might be able to recognise faces overtly if the task were very easy. As predicted, overt face recognition can be produced in prosopagnosics when several faces are presented and they are informed that all of them belong to the same category (e.g., Morrison, Bruce, & Burton, 2003).
The picture is somewhat different when covert recognition is assessed physiologically. We might expect that no patients would have intact overt face recognition combined with impaired covert recognition assessed by skin conductance re- sponses reflecting increased arousal. In fact, that expectation has been disproved in patients with Capgras delusion, who believe that very familiar people have been replaced by impostors, aliens, or doubles. Patients with Capgras delusion show overt recognition of familiar faces, but do not produce enhanced skin conductance responses to such faces (Ellis, Lewis, Moselhy, & Young, 2000; Hirstein & Ramachandran, 1997). However, they do show covert face recognition assessed by behavioural measures (Ellis et al., 2000).
What is going on here? According to Schweinberger and Burton (2003), skin conductance responses to familiar faces are produced via an arousal response in the amygdala, and patients with Capgras delusion have damage at location B (see Figure 3.19). In contrast, covert recognition assessed behaviourally and overt recognition both involve person identity nodes. This two-pathway assumption neatly fits the data.
Evaluation
Schweinberger and Burton's (2003) model provides a satisfactory account of covert face recognition in prosopagnosics. There is reasonable evidence that covert recognition assessed behaviourally involves the same functional system as overt recognition, whereas covert recognition assessed physiologically does not. These findings are as predicted by the model. In future research, it may be worth investigating the assumption that the two pathways from face recognition units are totally separate from each other.
Are faces special? Yes!
It has often been assumed there is something special about face recognition. Three main reasons have been put forward to support this assumption. First, it is argued that faces are processed differently from objects. For example, Farah (1990, 1994a) put forward a two-process model of object recognition in which two processes or forms of analysis were distinguished:
(1) Holistic analysis, in which the configuration or overall structure of an object is processed. (2) Analysis by parts, in which processing focuses on the constituent parts of an object.
Farah (1990, 1994a) argued that holistic analysis and analysis by parts are involved in the recognition of most objects, and reading words or text mostly involves analytic processing. However, face recognition depends mainly on holistic processing. According to Farah et al. (1998, p. 484), "[Holistic processing] involves relatively little part decomposition", meaning that explicit representations of parts of the face (e.g., face, mouth) are of minor importance.
Second, there are numerous brain-imaging studies investigating whether face recognition involves different brain areas from object recognition. It has been found in many studies that there is an area of the brain (mainly the fusiform gyrus) apparently specialised for face processing (see below). This area is often referred to as the fusiform face area.
Third, there is research on patients suffering from prosopagnosia, a condition in which patients cannot recognise familiar faces but can recognise familiar objects. This inability to recognise faces occurs even though prosopagnosic patients can still recognise familiar people from their voices and names. These findings suggest that the brain area damaged in patients with prosopagnosia is involved in processing faces but not familiar objects.
Evidence
Farah (1994a) studied holistic or configural pro- cessing of faces and objects. Participants were presented with drawings of faces or houses, and were told to associate a name with each face and each house. Then they were presented with whole faces and houses or with only a single feature (e.g., mouth, front door). Their task was to decide whether a given feature belonged to the individual whose name they had been given previously. Recognition performance for facial features was much better when the whole face was presented than when only a single feature was presented (see Figure 3.20). In contrast, recognition for house features was very similar in whole and single-feature conditions. These findings suggest that holistic analysis is more important for face recognition than for object recognition Farah et al. (1998) pointed out that the findings of Farah (1994a) indicated that faces are stored in Memnon) in a holistic form, but did not show that faces are perceived holistically. They filled this gap in a series of studies. Participants were presented with a face, followed by a mask, Followed by a second face. The task was to decide whether the second face was the same as the first. The key manipulation was the nature of the mask, which consisted either of a face arranged randomly or of a whole face. The crucial prediction was as follows: "If faces are recognised as a whole and part representation plays a relatively small role in face recognition, then a mask made up of face parts should be less detrimental than a mask consisting of a whole face" (Farah et al 1998, p. 485).
What did Farah et al. (1998) find? As predicted, face-recognition performance was better when part masks were used than when whole masks were used. This finding suggests that faces were processed holistically. In other conditions, the effects of part and whole masks on word and house recognition were assessed. The beneficial effects of part masks over whole masks were less with house stimuli than with faces, and there were no beneficial effects at all with word stimuli. Thus, there seemed to be less holistic processing of object (house) and word stimuli than of faces.
Farah and Aguirre (1999) carried out a metaanalysis of PET and fMRI studies designed to see whether separate brain regions are associated with face and object recognition. The findings were somewhat inconsistent. However, parts of the right fusiform gyrus were more likely to be active during face recognition than object recognition. For example, Kanwisher, McDermott, an Chun (1997) used fMRI to compare brain activity to faces, scrambled faces, houses, and hands. There was face-specific activation in parts of the right fusiform gyrus. Similar findings have been reported since Farah and Aguirre's (1999) review. For example, Pelphrey et al. (2003, p. 959) used fMRI while participants viewed faces and common objects, finding that "Face activated areas were localised to the fusiform and inferior temporal gyri and adjacent cortex."
As we have seen, patients with prosopagnosia cannot recognise familiar faces even though they can recognise familiar objects. This may occur because there are specific processing mechanisms used only for face recognition. An alternative possibility is that more precise discriminations are required to distinguish between two specific faces than to distinguish between familiar objects (e.g., a chair and a table). Farah (1994a) obtained evidence that prosopagnosic patients can be good at making precise discriminations for objects. LH (a patient with prosopagnosia) and control participants were presented with various faces and pairs of spectacles, and were then given a recognition-memory test. LH was at a great disadvantage to the controls on face recognition, but performed comparably on spectacle recognition.
Evidence that the fusiform face area is damaged in prosopagnosics was reported by Hadjikhani and de Gelder (2002). They found with fMRI that prosopagnosics showed similar activation of the mid-fusiform gyrus and the inferior occipital gyrus to faces and to objects. These findings suggest that damage to these areas played a role in their problems with face processing.
Earlier we mentioned Farah's (1990) theory, according to which face recognition involves primarily holistic processing, reading involves primarily analytic processing, and object recognition involves holistic and analytic processing. We can test this theory by considering brain-damaged patients falling into the following categories: prosopagnosia; visual agnosia (deficient object recognition); and alexia (problems with reading in spite of good ability to comprehend spoken language and good object recognition). Farah (1990) studied the co-occurrence of these conditions in 87 patients. What would we expect from her theory? First, patients with visual agnosia (having impaired holistic and analytic processing) should also suffer from prosopagnosia and/or alexia. This prediction was confirmed. There were 21 patients with all three conditions, 15 patients with visual agnosia and alexia, 14 patients with visual agnosia and prosopagnosia, but only 1 patient who may have had visual agnosia on its own. However, a few cases of visual agnosia without prosopagnosia or alexia have been reported subsequently, For example, Humphreys and Rumiati (1998) tested a 72-year-old woman with visual agnosia. Her face recognition and visual word recognition were both within normal limits, which is contrary to Fatah's theory.
Second, and most importantly, there was a double dissociation between prosopagnosia and alexia. There were 35 patients who had prosopagnosia but not alexia, and there are numerous reports of patients with alexia without prosopagnosia. These findings suggest that the brain mechanisms underlying, face recognition differ from those underlying word recognition.
Third, it is assumed that reading and object recognition both involve analytic processing. Thus, patients with alexia (who have problems with analytic processing) should be impaired in their object recognition. This contrast with the conventional view that patients with "pure" alexia have impairments only to reading abilities. Behrmann, Nelson, and Sekuter (1998) studied six patients who seemed to have 'pure" alexia. Five of these patients were significantly slower than normal participants to name visually complex pictures.
Fourth, patients with severe deficits in both holistic and analytic processing should also have greatly impaired object recognition. In Farah's (1990) data, there were no patients who had prosopagnosia and alexia without visual agnosia. However, this pattern was observed by Buxbaum, Glosser, and Coslett (1999). Their patient, WB, had severe prosopagnosia and alexia, but nevertheless performed reasonably well on tests of object recognition. This is inconsistent with Farah's theory.

Дата добавления: 2015-11-04