Elsevier

NeuroImage

Volume 19, Issue 3, July 2003, Pages 1180-1193
NeuroImage

Regular article
Cortical specialisation for face processing: face-sensitive event-related potential components in 3- and 12-month-old infants

https://doi.org/10.1016/S1053-8119(03)00076-4Get rights and content

Abstract

The adult N170 is considered to be an electrophysiological marker of specialised mechanisms for face processing, but little is known about its developmental origin. Previous work has identified two face-sensitive infant ERP components (N290 and P400) (J. Cog. Neurosci. 14 (2002), 199). In the present study, we assessed the specificity (to upright human faces) of these infant components at 3 and 12 months. At 12 months the degree of specificity observed in both components was similar to that seen in the adult N170. In contrast, at 3 months of age the N290 and P400 did not show the same level of specificity for human faces as that observed at 12 months. Our findings suggest that (1) both face-sensitive components increase in their specificity for upright human faces during development, and (2) the adult N170 is not preceded by a single developmental precursor, but rather emerges as a consequence of the integration of two functionally and morphologically distinct components (N290 and P400).

Introduction

Evidence from single-cell recordings in monkeys, human neuropsychology, and functional neuroimaging converge to suggest that faces are processed by specialised neural and cognitive mechanisms that differ from those involved in the processing of other objects. For example, functional magnetic resonance imaging studies show that regions within the fusiform gyrus are more active when adults passively view faces than when they view other objects such as hands or chairs Puce et al 1995, Puce et al 1996, Kanwisher et al 1997 and the pattern of activation over the ventral temporal cortex differs for faces compared to other objects (Ishai et al., 1999). In addition, electrophysiological studies which measure scalp-recorded brain electrical potentials (event-related potential, ERP) have consistently shown that faces elicit a negative potential that occurs most prominently over occipito–temporal scalp regions and peaks between 140 and 170 ms after stimulus onset (Bentin et al., 1996). This ERP component, the N170, is sensitive to faces in that it is of larger amplitude and shorter latency to faces compared with most other stimuli, including hands, feet, trees, cars, letters, or words Botzel and Grusser 1989, Botzel et al 1989, Bentin et al 1996, George et al 1996. Moreover, the amplitude and/or latency of the N170 also differs between upright human faces and closely related stimuli such as monkey or ape faces and inverted human faces Eimer 2000c, Rebai et al 2001, Carmel and Bentin 2002, de Haan et al 2002. Whilst the functional significance of the N170 in adult face processing has yet to be fully elucidated, it is thought to reflect processing related to the structural encoding of faces as opposed to face recognition or face identification Eimer 2000a, Eimer 2000b. This conclusion is based on findings demonstrating that the N170 is sensitive to experimental manipulations that alter the basic structural properties of faces, e.g., inversion, feature scrambling George et al 1996, Eimer 2000a, Eimer 2000b but is unaffected by the familiarity Bentin et al 1996, Eimer 2000b or emotional expression of faces (Eimer and Homes, 2002; but see Pizzagalli et al., 2002 for evidence of affect-modulated brain electrical activity in fusiform gyrus 160 ms after stimulus onset).

Despite previous efforts to establish the nature and degree of cortical specialisation for face processing, there is still considerable debate regarding how differences between face and object processing originate. There are currently several viewpoints that differ from each other in two main ways (reviewed in de Haan and Halit, 2001): (1) the degree to which cortical specialisation for face processing is determined by genetic or environmental factors and (2) the specificity of cortical processing mechanisms for faces. According to one account, the neural mechanisms underlying face processing are genetically determined and are specific to faces Farah 2000, Farah et al 2000. In this view, a cortical module specifically dedicated to the processing of faces processing exists from birth. Evidence to support this view comes from a case study of a patient (Adam) who, as a result of a viral infection, suffered brain damage at only 1 day old. Tests carried out when he was 16 years of age revealed a disproportionate impairment in face recognition compared to object recognition (Farah et al., 2000). Farah et al. (2000) interpret these results as evidence that the distinction between face and object recognition, and the anatomical localisation of face recognition, are “explicitly specified in the genome” (p. 122).

In contrast to this view, two alternative accounts emphasise the critical role of experience viewing faces. One such account argues that neural specialisation for faces arises during development due to processes that parallel perceptual skill learning in adults Diamond and Carey 1986, Gauthier and Nelson 2001. According to this view, development of specialised mechanisms occurs as a result of experience in making fine-grained discriminations between numerous visually similar category members, and is not unique to faces per se. The development of specialised mechanisms responsible for the processing of upright faces thus occurs over time as a result of our extensive exposure to a large number of individual faces. In this view, mechanisms that are involved in the processing of faces are no different from those involved in tasks that require visual expertise for discriminating other categories with complex, visually similar members. These domain-general learning mechanisms may appear to be specific to faces only because faces are a stimulus category for which humans consistently develop a high level of expertise. Evidence to support this view has been obtained from behavioral and neuroimaging studies showing that effects previously thought to be specific to faces can also be obtained with other categories for which participants show expertise Gauthier et al 1999, Gauthier et al 2000, Tarr and Gauthier 2000, Rossion et al 2002.

A third perspective explains data on the development of face processing in terms of developmental mechanisms that generate increasingly specialised processing within cortical areas Nelson 1993, Nelson 2001, Johnson 2000, Johnson 2001. From this perspective, early in development a given region of cortical tissue may be activated by a broad range of stimuli. However, over time the selectivity of the cortical tissue increases so that it responds to only certain kinds of stimuli, e.g., human upright faces. Thus, in this view an initially general mechanism becomes domain-specific. One possibility is that this process occurs via mechanisms that are analogous to those responsible for cortical specialisation of speech Nelson 1993, Nelson 2001. In the first months of life infants are able to discriminate between a broad range of speech sounds, but with increasing age and experience, the range of speech sounds that infants respond to narrows. For example, 6-month-old infants can discriminate speech sounds from both native and nonnative languages, whereas 12-month-olds and adults can only discriminate speech sounds that are part of their native language Kuhl et al 1992, Cheour et al 1998. In a similar way, the face processing system may develop from a broadly tuned, nonspecific, complex-figure recognition system to a system tuned in to the type of faces seen in the natural environment, i.e., upright human faces Nelson 1993, Nelson 2001. This view is supported by a recent study showing a decrease in discrimination abilities for nonhuman faces with age: 6-month-olds can discriminate between individual monkeys and humans, while 9-month-old infants and adults tested with the same procedure discriminate only between members of their own species (Pascalis et al., 2002).

One way to examine in more detail this third hypothesis is to investigate whether face-specific cortical areas are present early in life. Ethical and practical difficulties prohibit the use of most neuroimaging methods with healthy human infants; however, ERPs are one method that is suitable for use with this population. Information regarding the existence and stimulus-specificity of face-sensitive ERP components early in life is useful for at least two reasons. First, it can provide one source of evidence relevant to developmental theories of face processing, and second, it can provide an important tool for investigating developmental disorders of social information processing, e.g., autism (Dawson et al., 2002). The first study to investigate face-sensitive components during infancy showed that at 6 months of age a positive component peaking at 400 ms over occipital electrodes was of shorter latency for faces compared to objects (de Haan and Nelson, 1999). However, the results of this study are difficult to interpret because there was no adult comparison group and because some of the objects were toys that had schematic faces.

Only one study has used the same procedure to compare face-sensitive ERP components in infants (6 months) and adults (de Haan et al., 2002). In that study, the specificity of infant and adult ERP responses was determined by comparing their responses to upright and inverted faces and to human and monkey faces. Inverted faces were used because they are identical to upright faces in their basic psychophysical properties, but adults are less familiar with them, and are worse at recognising them (Yin, 1969). Since inversion typically has less or no influence on recognition of other objects (Yin, 1969), the “inversion effect” is considered to be an index of specialisation for face processing. Indeed, it has been argued that the inversion effect on the adult N170 is a better measure of its face-sensitivity even than difference between faces and objects (Rossion et al., 2000). Monkey faces were used because they share the basic structure of human faces (e.g., two eyes, above a nose, above a mouth), but adults are less familiar with them and are also worse at recognising them Pascalis and Bachevalier 1998, Pascalis et al 1998, Pascalis et al 1999, Pascalis et al 2001. The adult N170 showed specificity to upright human faces in that: (1) it was of smaller amplitude and shorter latency for upright human faces compared with all other face conditions and (2) face-inversion increased the amplitude and latency for human faces only (de Haan et al., 2002). The results obtained with 6-month-olds were consistent with the de Haan and Nelson (1999) study and showed that faces elicited a P400 occurring over posterior electrodes. Like the adult N170, the P400 was influenced by inversion: its amplitude was more negative for inverted than upright faces. However, unlike the adult N170, this effect was not specific to human faces and also occurred for monkey faces. A negative component preceding the P400 was also observed over posterior regions and peaked at 290 ms after stimulus onset (N290). Like the adult N170, the N290 showed sensitivity to the species of faces in that it differed in amplitude between monkey and human faces. In contrast to the adult N170, the N290 showed no sensitivity to the orientation of faces. Thus, while the N290 and P400 are possible candidates for an “infant N170”, neither component exhibited the same degree of specificity for upright human faces observed at the adult N170 (de Haan et al., 2002).

In summary, it remains uncertain which, if any, of the two candidate infant components corresponds to the adult N170. One purpose of this study was to address this question by examining whether either component goes on to show adult-like specificity in response at an older age. Thus, in Experiment 1 we recorded high-density ERPs in 12-month-olds using the same stimuli as de Haan et al. (2002). A second purpose of this study was to investigate whether any degree of face-sensitivity could be observed at a younger age than has previously been tested. Thus, in Experiment 2 we used the same procedure with 3-month-old infants.

Section snippets

Experiment 1: introduction

As described above, the ERP responses of 6-month-old infants do not share the same degree of specificity to upright human faces as the adult N170 (de Haan et al., 2002). Whilst the adult N170 shows sensitivity to both the species and orientation of faces, an equivalent component has not yet been identified within the infant ERP. Instead, the effects of species and orientation occur at two separate time points within 6-month-old’s ERP. Experiment 1 investigated whether either of the two infant

Participants

The final sample consisted of 26 12-month-old infants (11 males) with an average ag of 12 months (range = 11.2–12.2 months). All were born full-term and were of normal birthweight. An additional 59 infants were tested but were excluded from further analysis due to eye and/or body movements that resulted in recording artefacts (n = 58) or due to a procedural error (n = 1).

Stimuli

The stimuli were 20 colour images of human female faces and 20 colour images of macaque monkey faces. The faces subtended a

Experiment 1: results

Both human and monkey faces elicited a negative component that peaked approximately 290 ms after stimulus onset (N290), which was followed by a positive component that peaked at 400 ms (P400) after stimulus onset. The amplitude and latency of both ERP components were analysed to test for effects of species and orientation. Fig. 1 shows the grand-averaged waveforms from two posterior temporal recording sites for upright and inverted human faces and upright and inverted monkey faces. Means (M)

Experiment 1: discussion

The purpose of Experiment 1 was to identify and characterise an ERP component in 12-month-olds whose response properties most closely resemble the adult N170. We focused on two ERP components (the N290 and P400) which at 6 months of age were found to be sensitive to either the species or orientation of faces, but not both. We aimed to see whether by 12 months of age either component would, like the adult N170, show an interaction of species by orientation. The results showed that both

Experiment 2: introduction

The aim of Experiment 2 was to investigate the development of cortical specialisation for faces by examining the functional characteristics of the N290 and P400 in 3-month-olds. As in Experiment 1, infants passively viewed images of upright and inverted human faces and upright and inverted monkey faces. Since 3-month-olds have had even less experience of faces than 6-month-olds we hypothesised that their ERP components would either exhibit the same or less specificity for human faces than

Experiment 2: methods

The stimuli, general methods for data collection, and data analysis as well as statistical analysis were the same as in Experiment 1. The only difference was the two time windows that were chosen to capture the face-sensitive components. These were: (1) N290 (120–440 ms) and (2) P400 (440–690 ms). The time windows differed from those selected for the 3-month-olds because as predicted their ERP components occurred at a later time point within the ERP waveform than those observed in 12-month-olds

Experiment 2: discussion

The aim of Experiment 2 was to investigate the development of cortical specialisation for faces by examining the functional characteristics of the N290 and P400 in 3-month-olds. The N290 peaked at around 350 ms after stimulus onset and was most prominent at medial recording sites. The delayed latency of the N290 and its medial scalp distribution are consistent with results obtained in 12-month-olds (Experiment 1). The N290 showed sensitivity to the species of a face in that it was of larger

General discussion

The general aim of the present study was to characterise the developmental trajectory of face-sensitive ERP components over the first year of life. Experiment 1 aimed to investigate the degree of face specificity of two ERP components (N290 and P400) previously identified in 6-month-olds (de Haan et al., 2002) by comparing 12-month-old infants’ ERP responses to upright and inverted human faces or upright and inverted monkey faces. The results showed that both the species and orientation of

Acknowledgements

This work was supported by the Medical Research Council Program Grant G97 155 87 to M.H.J. We thank the parents and infants who participated in the study; Jane Singer, Leslie Tucker, Agnes Volein, and Eileen Mansfield for their help with recruiting and testing participants; and Gergo Csibra and Haralambos Hatzakis for the help with computer programming. In addition, we thank Daphne Maurer and Sybil Geldart (McMaster University) and Olivier Pascalis (University of Sheffield) for kindly providing

References (52)

  • M.J. Taylor et al.

    ERP evidence of developmental changes in processing of faces

    Clin. Neurophysiol.

    (1999)
  • D.M. Tucker

    Spatial sampling of head electrical fieldsthe geodesic sensor net

    Electroenceph. Clin. Neurophysiol.

    (1993)
  • N. Tzourio-Mazoyer et al.

    Neural correlates of woman face processing by 2-month-old infants

    NeuroImage

    (2002)
  • S. Bentin et al.

    Electrophysiological studies of face perception in humans

    J. Cog. Neurosci.

    (1996)
  • K. Botzel et al.

    Electric brain potentials evoked by pictures of faces and non-facesa search for “face-specific” EEG-potentials

    Exp. Brain Res.

    (1989)
  • K. Botzel et al.

    The search for face-specific evoked potentials

  • M. Cheour et al.

    Development of language-specific phoneme representations in the infant brain

    Nature Neurosci.

    (1998)
  • G. Dawson et al.

    Neural correlates of face and object recognition in young children with autism spectrum disorder, developmental delay, and typical development

    Child Dev.

    (2002)
  • M. de Haan et al.

    Neural bases and development of face recognition during infancy

  • M. de Haan et al.

    Brain activity differentiates face and object processing in 6-month-old infants

    Dev. Psychol.

    (1999)
  • M. de Haan et al.

    Specialization of neural mechanisms underlying face recognition in human infants

    J. Cog. Neurosci.

    (2002)
  • R. Diamond et al.

    Why faces are and are not specialan effect of expertise

    J. Exp. Psychol. General

    (1986)
  • M. Eimer

    The face-specific N170 component reflects late stages in the structural encoding of faces

    Neuroreport

    (2000)
  • M. Eimer et al.

    An ERP study on the time course of emotional face processing

    Neuroreport

    (2002)
  • M. Eimer et al.

    Prosopagnosia and structural encoding of facesevidence from event-related potentials

    Neuroreport

    (1999)
  • M.J. Farah

    The Cognitive Neuroscience of Vision

    (2000)
  • Cited by (0)

    View full text