Symposium
Chair: Catherine J. Mondloch
Discussant: Darwin Muir
This symposium will consider the origins of cortical specialization byfocussing on a system that is known to be highly specialized and localizedin adults, namely face perception. We will discuss whether a specializedsystem is present at birth, the nature of postnatal development, and therole played by postnatal experience. The goal of this symposium is tointegrate research on the development of cortical specialization for faceperception being conducted in four different labs. Together, these studiesencompass a variety of techniques (e.g., behavioral observation,event-related potentials), a wide range of ages (including newborns,children, and adults), and a variety of populations (including patientswith infantile lesions, and patients treated for congenital cataract). Theideas and data to be presented will provide a foundation for a dynamicdiscussion of various issues concerning cortical specialization.The first two papers will focus on the normal development of faceperception during infancy. The first paper tests the hypothesis that facesrepresent a special class of stimuli to newborn infants. Using apreferential looking technique the authors demonstrate preferences for arange of stimuli that satisfy the neurological constraints of the immaturevisual system (e.g., symmetrical over asymmetrical patterns). The authorsconclude that infants' preferences for faces depend on two mechanisms, onlyone of which is specific to faces. The second paper focuses on increasingcortical specificity of face processing during infancy. The authors usedevent-related potentials (ERPs) to investigate the specificity ofelectro-cortical activity elicited by human and monkey faces in adults andin 12- and 6-month-old infants. Changes in the timing and morphology of theERP responses lead to the conclusion that cortical specificity for faceprocessing increases across development.The two other papers focus on the developmental trajectories of variousaspects of face perception following developmental insult. Data frompatients suggest cortical specialization of face processing is dependentupon experience, and thus question the assumption of a very early corticalsystem that is specialized for face perception. The first of these papersstudies the effect of early visual deprivation due to bilateral congenitalcataracts on face perception both immediately following treatment andduring later childhood after patients had had several years of nearlynormal input. The second of these papers presents data from patients withinfantile brain lesions. Patients show deficits in some, but not all,aspects of face processing; it will be argued that some general learningmechanisms may be sufficient for the cortical specialization of faceperception seen in adults.Together these papers will address several key questions about thedevelopment of cortical specialization of face processing: Is therecortical specialization at birth? What brain mechanisms, present at birth,foster the development of cortical specialization? What is thedevelopmental trajectory of various aspects of face perception underdifferent developmental constraints? The arguements presented will focus onface processing, but will inform our thinking about other specializedcortical mechanisms.
Details of individual items:
paper
Although it is known that newborns prefer to orient to face-like patternsover nonface-like patterns (Goren et al., 1975; Morton & Johnson, 1991),contradictory results have been obtained with the preferential lookingtechnique. Some studies (Hersherson et al., 1967; Slater, 1993) failed toobtain a preference, whereas others were able to demonstrate it (Kleiner,1987; Valenza, et al., 1996). This apparent contradiction may be explainedby making reference to the two hypotheses that have been proposed toexplain face preferences at birth: the sensory hypothesis and thestructural hypothesis.The sensory hypothesis (Banks & Salapatek, 1981; Kleiner & Banks, 1987)states that at birth preferences for any class of stimuli, including faces,are due to the amount of energy in the stimuli. Only if the sensorycharacteristics of two stimuli are identical, is the structure deemed toplay a role in producing the preference (Kleiner, 1990, 1993). Faces arenot special in either case. The structural hypothesis maintains that facesare special because of the structural organization of their internalfeatures. At birth the visual system is thought to possess a subcorticalmechanism (Conspec) that allows the system to detect and to orient to faces(Johnson & Morton, 1991).In the present study, the hypothesis that faces represent a special classof stimuli to newborns was indirectly examined in three experiments using apreferential looking technique that measured number of orienting responsesand total fixation time. The idea was that, if the structure of a face isnot special, a preference should be found for all stimuli that share withfaces those structural properties that satisfy the neurological constraintsof the immature visual system. Such characteristics may include symmetryand number or position of high-contrast areas.Experiments 1 and 3 investigated the role of symmetry in the upper part ofthe stimulus. In Experiment 1 the stimuli were five blobs in the correctlocations to form two Ts-one upright and the other upside-down. InExperiment 3 the same five blobs formed two Us-one upright and the otherupside-down. Experiment 2 investigated the role of high-contrast areas inthe upper part of the stimulus.Results demonstrate that in every experiment newborns looked longer atstimuli that either were symmetrical or had more numerous high-contrastareas in the upper part. However, they did not orient more frequently tothose stimuli they fixated longer. Only in one experiment number oforientations showed the same trend as total fixation time. These findingscan be interpreted by assuming that face preferences at birth depend on twomechanisms. One is not specific to faces and induces the newborn to looklonger at every stimulus that possess certain structural properties. Theother is specific to faces and induces the newborn to preferentially orientto and/or track facelike stimuli that are presented in the periphery of thevisual field.
paper
Evidence from a variety of sources suggests that face processing by adultsis mediated by specialized and localized cortical mechanisms. For example,functional magnetic resonance imaging (fMRI) studies show that ventral andinferior occipito-temporal cortex, particularly specific areas of thefusiform gyrus, is more active while viewing faces than while viewingobjects (Kanwisher et al., 1997; Sergent, Ohta, & MacDonald, 1992), hands(Kanwisher et al., 1997) scrambled faces (Puce et al., 1995), or texturesPuce et al., 1996). This type of specialisation may in part emerge from ourextensive experience with human faces over the course of development. Byinvestigating when and how the cortical processing system becomesadult-like, it is possible to put some constraints on hypotheses about howthe initial organisation and functioning of the cortex and the inputarising form the environment interact to create the pattern of specificityseen in adults.We used high event-related potentials (ERPs) to investigate the specificityof electro-cortical activity elicited by human faces in adults and 12- and6-month-old infants. Activity of cortical face processing areas is alsodetectable in scalp recorded ERPs. In adults the N170 component in the ERP(peaking between 120-200 ms after stimulus onset) is sensitive to faces inthat: a) it is of shorter latency for upright faces compared to otherstimuli such as inverted or scrambled faces, hands, feet, the whole body,trees, cars, buildings, shapes, letters or words (e,g,m Bentin, Allison,Puce, Perez & McCarthy, 1996; Seeck & Grusser, 1992), and b) its amplitudeis smaller or absent for objects compared to upright faces (Bentin et al.,1996), but larger for deviant faces (e.g., upside down or with the featuresscrambled) than for normal faces (Bentin et al., 1996; George, et al.,1996). In this paper we report a series of experiments designed todetermine the development and specificty of such face-sensitive potentialsduring infancy. ERPs wer recorded to upright and inverted human faces in6-month-olds, 12-month-olds and adults. At all ages inversion influencedface processing. For adults the N170 was of larger amplitude and longerlatency for inverted than upright faces, and for 12-month-olds a similarpattern was seen at a slightly longer latency. A striking difference intiming and morphology was seen in 6-month-olds: at this age upright faceselicited a greater positivity over occipital electrodes peaking atapproximately 350 ms. Analysis indicates that face specific processingoccurs at longer latencies after stimulus onset in infants than in adults.46urther experiments comparing responses elicited by upright and invertedmonkey faces showed that these effects of inversion were specific tohuman faces in adults but not in 6-month-olds. These results are consistentwith the hypothesis that face processing is less specialized in earlyinfancy than in adulthood, and that specific mechanisms for face processingdevelop with experience. We interpret the differences in timing offace-specific processing with age in terms of specialisation shift forwardwith age to an earlier stage of cortical processing.
paper
Recent theories of face perception emphasize the importance of early visualexperience: Johnson and Morton (1991) argue that biases for faces in thefirst few weeks of life constrain input to the developing cortex and,hence, influence its later specialization for face processing. de Schonenand Mathivet (1989) suggest further that such early experience mayespecially affect cortical mechanisms in the right hemisphere (e.g.,configural-global processing) because some cortical networks mature morequickly in the right hemisphere than in the left hemisphere during thefirst months of life, a time when babies only process low spatialfrequencies. We examined the influence of early visual experience bycomparing patients treated for bilateral congenital cataracts that haddeprived them of patterned visual input during early infancy to subjectswith normal visual history. We tested face-processing abilities bothimmediately after treatment for cataracts and during later childhood afterpatients had had several years of nearly normal visual input.Infants were presented with three different pairs of face and nonfacestimuli using a method based on the Teller acuity card procedure. Normalnewborns preferred a head outline with three blobs located so as torepresent eyes and a mouth (Config) over an outline with those featuresinverted. This preference disappeared by 6 weeks of age, suggestingsubcortical mediation. Preferences for the two other face stimuli emergedat 6 and 12 weeks of age, suggesting increasing cortical influence overface perception (Mondloch et al., 1999). To assess the role of early visualexperience, we are testing infants treated for bilateral congenitalcataract within 1 hour of their first visual experience. All patientstested to date (n3D5) were at least 6 weeks old at the time of treatment.Unlike normal infants at this age, 3 of the 5 patients preferred Configover its inverted version. Four of the patients were at least 12 weeks ofage at the time of treatment, an age at which normal infants prefer aschematic face over its photo-negative. Unlike normal infants, not onepatient showed this preference. The results to date suggest that theincreased cortical control of face perception during early infancy dependsupon early visual experience.Previous testing with older children who had been treated for congenitalcataract suggested deficits in configural processing (i.e., processing ofthe spatial relations among features) (Geldart et. al., 1998). Patients'(n3D17) performed normally or nearly normally except on a task requiringthe recognition of identity despite changes in head orientation--a taskthat likely requires configural processing. To probe the nature ofpatients' deficits, we showed subjects pairs of faces that (1) wereidentical, (2) differed only in the configuration of facial features, or(3) differed only in the shape of individual features. The 8 patientstested to date performed poorly only on face pairs differing in theconfiguration of features. Data from patients treated for unilateralcongenital cataract show that these deficits cannot be attributed to pooracuity. Together the data support the hypothesis that at least some aspectsof cortical specialization of face perception are influenced by earlyvisual input.
paper
Face processing has been taken as one of the best example of the existenceof a system specialized in the processing of a specific category ofobjects. The existence of a dedicated cortical system in adults howeverdoes not tell anything on how it becomes specialized in this particularability during development. The question here is to understand what are thefactors that contribute to this specialization and how plastic is thecortex for developing this particular ability. The experimental argumentssupporting the assumption of a very early cortical specific system will bedescribed (behavioral studies and brain imagery studies). However, we havesuggested several years ago that assuming the existence of a specificsystem is not necessary for understanding these data. Rather, some pre- andpost-natal general mechanisms (including perceptual learning mechanisms)converge towards building a system that becomes progressively specific. Weshall show that each of the arguments in favour of an early specific systemhave to be modulated by other data, in particular by data from brainlesions in early infancy. We shall conclude that there is no reason tobelieve that there exists a specific face processing system early inpost-natal life, but rather a convergence of several mechanisms which guidethe learning about faces and which work towards the formation of alocalized and specialized system. When only some of these constraintsguiding learning about faces are present, several aspects of faceprocessing might develop along a different trajectory, according to whichconstraint is preserved and at what time these E5constraints emerge.