Holistic face processing is mature at 4 years of age: Evidence from the composite face effect

Abstract

Although it is acknowledged that adults integrate features into a representation of the whole face, there is still some disagreement about the onset and developmental course of holistic face processing. We tested adults and children from 4 to 6 years of age with the same paradigm measuring holistic face processing through an adaptation of the composite face effect [Young, A. W., Hellawell, D., & Hay, D. C. (1987). Configurational information in face perception. Perception, 16, 747–759]. In Experiment 1, only 6-year-old children and adults tended to perceive the two identical top parts as different, suggesting that holistic face processing emerged at 6 years of age. However, Experiment 2 suggested that these results could be due to a response bias in children that was cancelled out by always presenting two faces in the same format on each trial. In this condition, all age groups present strong composite face effects, suggesting that holistic face processing is mature as early as after 4 years of experience with faces.

Introduction

An important paradox characterizes the development of humans’ face processing abilities. On the one hand, neonates tested several hours after birth already show face processing abilities, preferring to orient their attention toward face-like patterns as compared to scrambled faces (Goren et al., 1975, Morton and Johnson, 1991) or being able to differentiate their mother’s face from a stranger’s face (Bushnell, 2001, Bushnell et al., 1989, Pascalis et al., 1995). On the other hand, developmental studies have shown that face processing abilities develop rather slowly and progressively (Geldart, Mondloch, Maurer, de Schonen, & Brent, 2002). For instance, children’s performance in identity and facial expression processing improves tremendously between 4 and 11 years of age (Bruce et al., 2000) and reaches maturity only after puberty (Carey et al., 1980, Chung and Thomson, 1995, Mondloch et al., 2002).

It is yet unclear whether children simply process faces less efficiently than adults (i.e., a quantitative difference) or whether qualitatively different processes are used by adults and children. For instance, it is widely acknowledged that adults’ face recognition relies not only on the process of individual facial features but also on the relations between these features, the so-called configuration of faces (for a review, see Mondloch et al., 2002). The ability of children to process faces configurally has been debated frequently in the literature (e.g., Baenninger, 1994, Brace et al., 2001, Freire and Lee, 2001, Mondloch et al., 2004, Mondloch et al., 2003, Mondloch et al., 2002). The current view is that adult expertise in configural processing is especially slow to develop (Mondloch et al., 2002) even if it already emerges during infancy (Turati, Sangrigoli, Ruel, & de Schonen, 2004) and early childhood (Cohen and Cashon, 2001, Deruelle and de Schonen, 1998). To complicate matters further, the definition of face configuration varies considerably between authors and may appear to be somewhat confusing in the face literature. There are at least two types of configuration that have been conceptually differentiated (Goffaux and Rossion, 2006, Maurer et al., 2002, Rossion and Gauthier, 2002). First, configural information may refer to metric distances between facial features such as the interocular or eye–mouth distance. These distances between facial features can be measured and manipulated on the stimulus, and the sensitivity of the face processing system to perceive and encode this information can be tested in discrimination or recognition tasks (e.g., Barton et al., 2001, Freire et al., 2000, Haig, 1984, Leder et al., 2001). The second type of configuration is referred to as holistic processing. It is more difficult to grasp because it refers to a way of handling a face stimulus rather than information that can be manipulated independently of the observer. The concept was probably first introduced by Francis Galton, who noticed that facial features were not perceived and analyzed separately; that is, the face stimulus was processed as a whole unit or as a Gestalt (Galton, 1883). Numerous phenomena exemplify this holistic processing of faces in real-life situations or in the laboratory (e.g., Davidoff and Donnelly, 1990, Farah et al., 1998, Goffaux and Rossion, 2006, Hole, 1994, Homa et al., 1976, Sergent, 1984, Tanaka and Farah, 1993, Young et al., 1987).

Two experimental paradigms have been widely used to provide evidence for face holistic processing: the composite face paradigm (Young et al., 1987) and the whole–part paradigm (Davidoff and Donnelly, 1990, Tanaka and Farah, 1993). In the whole–part paradigm, participants are trained to name a series of faces, and they recognize face features (eyes, nose, or mouth) better when these features are embedded in the whole face stimulus than when they are presented in isolation (Tanaka & Farah, 1993). In the initial composite face paradigm, a composite stimulus was made by joining the top half of a familiar face (cut below the eyes) with the bottom half of another familiar face. Observers were slower to name the top half of such a composite face when the top and bottom parts were vertically aligned, creating a new face stimulus, than when the same top and bottom parts were offset laterally (i.e., misaligned). Both effects have been found with unfamiliar faces in matching tasks (e.g., Endo et al., 1989, Farah et al., 1998, Hole, 1994, Hole et al., 1999, Goffaux and Rossion, 2006, Le Grand et al., 2004, Michel et al., 2006).

Although a number of developmental studies have addressed the question of the ability to perceive metric distances between facial features (e.g., Baenninger, 1994, Brace et al., 2001, Freire and Lee, 2001, Mondloch et al., 2002, Mondloch et al., 2003, Mondloch et al., 2004), few studies have directly tested holistic face processing in children. To our knowledge, only four studies have been conducted using different paradigms explicitly measuring holistic face processing: the composite face paradigm (Carey & Diamond, 1994), the whole–part advantage paradigm (Pellicano and Rhodes, 2003, Tanaka et al., 1998), and a categorization task (Schwarzer, 2002). Furthermore, there is still some disagreement about the onset and the developmental pattern of holistic face processing. On the one hand, studies using the whole–part advantage paradigm (Pellicano and Rhodes, 2003, Tanaka et al., 1998) have suggested that 4- and 6-year-olds process faces as holistically as adults (see also Carey & Diamond, 1994); on the other hand, Schwarzer (2002) attested that 2- to 5-year-olds prefer to categorize faces on the basis of their constituent parts (by focusing on a single attribute) more than holistically (in terms of overall similarity), suggesting that young children rely less on holistic processing than do adults. Thus, the question remains as to whether young children process faces holistically and, if so, whether there is sudden onset of holistic processing around a given age, such as between 4 and 6 years of age, or a gradual developmental pattern.

In the current study, we aimed to clarify the question of the emergence and development of holistic face processing by testing adults and 4- to 6-year-olds with the exact same paradigm. To this end, two behavioral experiments were conducted with adults and children using the composite face paradigm. This paradigm is considered as providing the most compelling evidence of holistic face processing (Maurer et al., 2002) and does not present the limits associated with the whole–part paradigm such as the lack of specific instructions about encoding strategy (Goffaux and Rossion, 2006, Michel et al., 2006). In the current study, as compared with the initial study of Young and colleagues (1987) and subsequent experiments, different parameters were modified to accommodate young participants. First, faces were presented to the participants with no time limit. Second, faces were presented simultaneously so that there was no memory component involved in the task. Third, the upper parts of all faces were slightly colorized in red to help the youngest children performing the task adequately (Fig. 1). We reasoned that if the “quantitative” developmental view of holistic processing (Carey and Diamond, 1994, Pellicano and Rhodes, 2003, Tanaka et al., 1998) was correct, all tested children and adults should present a composite face effect. Moreover, the younger children’s recognition accuracy should be poorer than that of adults (Carey et al., 1980, Chung and Thomson, 1995, Geldart et al., 2002, Mondloch et al., 2002). In contrast, according to a “qualitative” viewpoint extended from the switch hypothesis (Carey and Diamond, 1977, Schwarzer, 2002), the composite face effect should emerge at a certain age, testifying to the emergence of holistic face processing abilities.