Abstract
Human adults are extremely efficient at detecting faces in complex visual scenes, being able to categorize a wide range of visual stimuli as faces and discriminate those from other visual categories accurately and rapidly (Crouzet et al., 2010; Rousselet et al., 2003). Yet the developmental course and the neural basis of this remarkable ability remain unknown. To clarify these issues, we performed EEG recording (32-channel) of 15 4- to 6-month-old infants viewing 20-second sequences of images flickering at 6 Hz (i.e., 6 images/second). A stimulation sequence contained repetitive series of 4 unsegmented pictures of objects (O) followed, every fifth stimulus, by a picture of a face (F) (Figure 1). Objects and faces were centred but varied substantially in size, colour, lighting and viewpoint, with faces varying also in gender, age, ethnical origin and expression. Infants performed between 5 to 12 trials (1 minute and 40 seconds to 4 minutes of experimentation). EEG trials were averaged for each infant separately and Fourier transformed into the frequency domain. A sharp response was found over the medial occipital lobe exactly at the base stimulation frequency (6 Hz) (averaged signal-to-noise ratio (SNR) at Oz: 8.87; Figure 2, left). More importantly, there was a distinct response at the oddball (face) frequency (1.2 Hz), particularly salient at the right occipito-temporal electrode P8 (averaged SNR at P8: 2.56; Figure 2, right). These findings have three implications: first, they show that 4- to 6-month-old infants discriminate faces from other complex non-face objects and generalize across faces despite their high physical variance. Second, they indicate that the right hemisphere advantage for face perception emerges early in infancy and is therefore independent of left lateralized letter/word representations emerging later during development. Finally, they highlight the power of fast periodic visual stimulation to characterize infants high-level visual processes rapidly and objectively.
Meeting abstract presented at VSS 2014