Tilburg University
Holland
In recent years, there have been many experimental studies of autistic individuals' capacity to process faces. Although there is evidence that they are impaired on certain aspects of face perception, it is still not clear how specific this impairment is. Whereas some studies found a selective impairment in the recognition of facial expressions, other studies found a more general deficit in perceptual abilities, and some failed to find any differences between autistic people and controls. Other aspects of face perception appeared also to be affected in autistic people, such as the memory for unfamiliar faces. Furthermore, autistic individuals seemed to pay relatively more artention to the lower part of the face, they were found to make more use of verbal strategies in expression recognition, and a remarkably good performance was found on tasks with inverted photographs of faces. Until now, no serious attempt was made to place these different fmdings in a comprehensive theoretical context. The main aim of our research program was to provide such a context.
One possible explanation for the
confusing results is that autistic individuals make use of abnormal
face processing strategies. For example, autistic children seem
to be less affected when faces are presented upside-down (Langdell,
1978; Hobson et al., 1988; Tantam et al., 1989). This last finding
suggests that autistic individuals perceive faces more in terms
of their component properties alone (piecemeal processing), rather
than viewing the face as a whole (holistic processing). In this
paper a series of experiments will be discussed that investigates
perception and encoding processes in autistic adolescents and
normal children and adults, using inverted, composite and scrambled
face stimuli.
The first experiment concerned an Inversion task with relatively easy task demands. Instead of first presenting a set of photographs to be kept in memory for longer time, every trial started with the short presentation of one face in frontal view, followed by a two-alternative forced-choice recognition task two seconds later. The photographs of the faces in the recognition test were of a different view (3/4 view) than the target photograph (frontal view). This was done to encourage face-specific encoding processes in the subjects. Accuracy and reaction time data on inverted presentation were compared to performance on upright presentation of the photographs.
The results revealed that most autistic adolescents showed the normal inversion effect: upright faces were recognised faster and more accurately than inverted faces. Furtherrnore, this effect was larger for face stimuli than for object stimuli (shoes), a finding that is consistent with studies in normals. Only autistic individuals with low social IQ scores did not show the inversion effect. They were not different from the other autistics in performance on the inverted photographs, but did not, like the others, show better recognition for upright faces. This correlation of the inversion effect with social intelligence in the autistic group is interesting, since social intelligence was also found to be correlated with the ability to recognise facial expressions in a categorical perception study (Teunisse & De Gelder, submitted)
It must be noted that the presence of an inversion effect in most autistic individuals does not automaticly imply that the encoding of faces is normal. They made more errors and were slower than normal adults, even though the inspection time of the target photographs was prolonged for autistics. Their response pattern was very similar to that of the children, who are known to have smaller inversion effects. The adapted paradigm controlled for floor effects, but thereby may also have obscured developmental effects.
The results suggest that the encoding
of faces is qualitatively the same for adults, children and autistics,
but children and autistics are less efficient than adults in encoding
the configural information of the face.
Tle impairments in expression recognition and unfamiliar face recognition may also result from impaired holistic processing of faces. While configural processing concerns encoding faces on basis of the facial configuration, holistic processing is the encoding of faces as a whole, with no explicit representations of the constituent parts. Autistics perforrn extremely good on tasks where a Gestalt pattern has to be segmented into its constituent parts, as in the Embedded Figures task (Shah & Frith, 1983) and the Block Design task (Shah & Frith, 1993). Apparently, when a task requires ignoring the meaningful context, autistics do not have to resist the tendency to see the Gestalt.
Young, Hellawell and Hay (1987) constructed a task in which the upper half of a composite faces had to be recognised. They found that this was more difficult when the face halves were aligned, and thus formed a new facial Gestalt, than when they were not aligned. If autistic people are impaired in holistic processing, they are expected to show no composite effect.
This prediction was tested in an adapted version of the Composite task. The stimuli were unfamiliar faces, and the task was constructed in such a way that no learning of the faces was required. Like in the Inversion task, every trial started with presentation of the target face, immediately followed by two composite faces. Tle subject had to recognise the upper half of these composites. The orientation of the target face (3/4 view) and composite faces (frontal view) were different, to encourage face-specific encoding.
Although the autistic data were very
similar to that of the children, there was not a significant composite
effect in this group. This confirmed the hypothesis that autistic
people are impaired in holistic processing of faces. The existing
evidence suggests that this deficit is most evident in perceptual
tasks were a Gestalt figure has to be analysed for its parts.
It is less clear whether this deficit also impairs performance
on tasks that require memorising a perceptual Gestalt.
To study this suggestion, tasks were constructed that were inspired by the face superiority experiments in the seventies (Homa, Haver & Schwartz, 1976; Van Santen & Jonides, 1978; Mermelstein, Prinzmetal & Banks, 1979). In the typical face superiority task, a context stimulus (an intact face or a scrambled face) is presented on a screen for a short time, followed by a forced-choice recognition task for facial features (eyes, nose or mouth). It is found that features are better recognised when they are memorised in the context of a normal face compared to a scrambled face. Mermelstein et al. (1979) showed that the memory component is crucial for finding a face superiority effect. When the task requires a perceptual search of the context stimulus the opposite effect will be found, a face inferiority effect. In that case, the facial Gestalt inhibits finding a constituent part.
A version of the standard memory search paradigm with intact and scrambled faces (Mermelstein et al., 1979) was used to examine face superiority effects in children, adults and autistic adolescents. The stimuli were chimeric faces of a male person, in which the facial features from four different expressions were combined, both in a facial configuration and in a scrambled configuration. This allowed not only the study of the face superiority effect by comparing performance on normal faces with scrambled faces, but also the possible existence of an expression superiority effect, by comparing the configuration of a congruent expression (e.g. happy eyes with a happy mouth) with that of an incongruent expression (e.g. happy eyes with a sad mouth). The results showed that the three groups performed better in the face condition than in the scrambled condition, although this face superiority effect was smaller for autistic people and children than for adults. Furthermore, facial features were recognised somewhat better when they were encoded in a face with a congruent expression than in an incongruent expression, which is the first empirical evidence for an expression superiority effect.
The fact that the superiority effects were smaller for children and autistic individuals, might have been due to the longer exposure times of the context stimuli (1 sec) for these groups. Tle longer exposure times were necessary, because a pilot study revealed that these subjects performed at chance level with shorter presentation of the stimuli (150 msec). The longer presentations might have allowed the children and autistics to use post-perceptual piecemeal encoding strategies that are by definition less sensitive to context effects.
The poor performance of children
and autistic individuals in the pilot study may be related to
their inefficiency in processing the configural aspects of a face,
which was already suggested by the smaller inversion effects and
the poor performance on unfamiliar face recognition tasks. Configural
encoding, which is thought to be face-specific (Carey & Diamond,
1994), may contribute to a great extend to the face superiority
effect. In contrast to holistic encoding, in which the constituent
parts of a face are not explicitly represented, the individual
features are encoded in relation to each other when a face is
encoded configurally, and this may facilitate later recognition
of the isolated features. The less efficient configural encoding
ability in autistics and children would thus yield smaller face
superiority effects. Developmental findings suggest that this
ability improves with age in normal children, whereas autistic
individuals do not seem to develop their configural processing
ability significantly.
The findings of our studies suggest
that in some aspects of face processing autististic individuals
are very similar to 10-year-old children. Like children, autistic
individuals use global processing strategies that are based on
overall similarity. Furthermore, autistic individuals and normal
children are less able than adults to use the second-order relational
features (the configuration) of a face for encoding a face in
memory. Therefore, they must rely more on post-perceptual strategies
based on piecemeal information. On the other hand, autistic individuals
differ from children in their perception of Gestalt patterns.
Perception in autistic people seems less concept-driven, they
are less inclined to automatically attach meaning to a stimulus,
even when they potentially have the relevant knowledge to understand
the meaning.
These findings seem to support the
theory of Frith (1989) of weak central coherence in autism. Our
studies contribute to a further specification of the theory in
that it states that the deficit is primary perceptual. Meaningful
and familiar stimuli are not perceived as Gestalts, because top-down
processes that use long-term memory representations for automatic
perception and recognition are impaired. The fundamental proposal
is that the mental representations in autistic people may be relatively
intact, but the use of these representations for perception is
impaired. T'his interpretation is in line with the executive
function approach (e.g. Ozonoff et al., 1991), which assumes that
organised search and working memory is impaired in autistic individuals.
Such a deficit would not only impair perception and attention
processes, but also mental operations that require active use
of incoming information. It may also explain why autistic people
pay attention to other parts of the face than normals (Langdell,
1978). Their scanning of the face seems to be less guided by meaning,
and therefore they will not pay more attention to parts of the
face that contain important social information, such as the eye
region.
REFERENCES
Carey, S. & Diamond, R. (1994). Are faces perceived as configurations more by adults than by children? Visual Cognition, 1, 253-274.
Frith, U. (1989). Autism: explaining the enigma. Basil Blackwell.
Hobson, R.P., Ouston, J. & Lee, A. (1988). What's in a face? The case of autism.
British Journal of Psychology, 79, 441-453.
Homa, D. Haver, B. & Schwartz, T. (1976). Perceptibility of schematic face stimuli:
Evidence for a perceptual Gestalt. Memory and Cognition, 4(2), 176-185.
Langdell, T. (1978). Recognition of faces: an approach to the study of autism. Journal of Child Psychology and Psychiatry, 19, 255-268.
Mermelstein, R., Banks, W. & Prinzmetal, W. (1979). Figural goodness effects in
perception and memory. Perception and Psychphysics, 26(6), 472-480.
Ozonoff, S., Pennington, B.F. & Rogers, S.J. (1991). Executive function deficits in highfunctioning autistic individuals: Relationship to theory of mind. Journal of Child Psychology and Psychiatry, 32, 1081-1105.
Shah, A. & Frith, U. (1983). An islet of ability in autistic children: a research note.
Journal of Child Psychology and Psychiatry, 24, 613-620.
Shah,A. & Frith, U. (1993). Why do autistic individuals show superior performance on the block design task? Journal of Child Psychology and Psychiatry, 8, 1351-1364.
Tantam, D., Monagham, L., Nicholson, H. & Stirling, J. (1989). Autistic children's ability to interpret faces: a research note. Journal of Child Psychology and Psychiatry, 30, 623-630.
Teunisse, J.P. & de Gelder, B. (submitted). Impaired categorical perception of facial expressions in high-functioning autistics.
Van Santen, J.P.H. & Jonides, J. (1978). A replication of the face-superiority effect.
Bulletin of the Psychonomic Society, 12, 378-380.
Young, A.W., Hellawell, D. & Hay, D.C. (1987). Configurational information in face perception. Perception, 16, 747-759.