A neural marker of rapid discrimination of facial expression in 3.5- and 7-month-old infants

Abstract

Infants' ability to discriminate facial expressions has been widely explored, but little is known about the rapid and automatic ability to discriminate a given expression against many others in a single experiment. Here we investigated the development of facial expression discrimination in infancy with fast periodic visual stimulation coupled with scalp electroencephalography (EEG). EEG was recorded in eighteen 3.5- and eighteen 7-month-old infants presented with a female face expressing disgust, happiness, or a neutral emotion (in different stimulation sequences) at a base stimulation frequency of 6 Hz. Pictures of the same individual expressing other emotions (either anger, disgust, fear, happiness, sadness, or neutrality, randomly and excluding the expression presented at the base frequency) were introduced every six stimuli (at 1 Hz). Frequency-domain analysis revealed an objective (i.e., at the predefined 1-Hz frequency and harmonics) expression-change brain response in both 3.5- and 7-month-olds, indicating the visual discrimination of various expressions from disgust, happiness and neutrality from these early ages. At 3.5 months, the responses to the discrimination from disgust and happiness expressions were located mainly on medial occipital sites, whereas a more lateral topography was found for the response to the discrimination from neutrality, suggesting that expression discrimination from an emotionally neutral face relies on distinct visual cues than discrimination from a disgust or happy face. Finally, expression discrimination from happiness was associated with a reduced activity over posterior areas and an additional response over central frontal scalp regions at 7 months as compared to 3.5 months. This result suggests developmental changes in the processing of happiness expressions as compared to negative/neutral ones within this age range.

Keywords: EEG; development; facial expression of emotions; fast periodic visual stimulation; infant; visual perception.

FIGURE 1

Fast periodic visual stimulation in electroencephalography (FPVS-EEG) to isolate a neural marker of rapid discrimination of facial expression. Among the five basic emotional facial expressions used (disgust, happiness, anger, fear, sadness) and neutrality, the discrimination of three target expressions was tested (disgust, happiness and neutrality). A 2-s sequence of fast periodic stimulation is depicted with images from one individual face. From top to bottom: happy, disgust, and neutral facial expressions are respectively used in dedicated sequences and periodically displayed at the 6-Hz base rate (i.e., six pictures per second; 1 cycle ≈ 167 ms) without inter-stimulus interval, and the five other expressions are randomly displayed at the 1-Hz expression-change frequency (every 6th cycle = 1 s between each expression-change). Images were presented through ±15% randomized size variation at each cycle. This design thus isolates two dissociated responses at two different frequencies: a general visual response (6 Hz) to the rapid train of one individual face varying in size and an expression-change response (1 Hz) reflecting rapid (i.e., single-glance) discrimination of all inserted expressions from the target expression presented at the base rate. Source: KDEF database (models coded 07F, 09F, 14F, and 20F) (Lundqvist et al., 1998).

FIGURE 2

Expression-change response for the three facial expressions at 3.5 months (A) and 7 months (B). Left part. Topographical representation (superior view) of significant individual responses. For each electrode, the number of infants with a significant response (Z > 1.96, p < 0.05) is represented by circle size when at least two individual responses were significant. The smaller topographical map also indicates significant electrodes at group level (gray: Z > 1.96, p < 0.05; black: Z > 2.57, p < 0.01). Right part. 3D topographical color maps (superior view) of the expression-change response (normalized noise-corrected amplitude in arbitrary units) for each expression condition. Bottom part. Signal-to-noise ratio (SNR) of the expression-change response and surrounding frequencies (±0.2 Hz, i.e., ±6 bins) averaged across significant electrodes for each expression condition.

FIGURE 3

Expression-change response (normalized noise-corrected amplitudes, in arbitrary units) according to expressions over different sites (O1/2, P7/8, CP1/2, T7/8, and FC1/2) at 3.5 and 7 months. In 3.5-month-olds, the analysis revealed a similar topography for the two emotional expressions (i.e., disgust and happiness) that differed from neutrality with a lower contribution of O1/2 and a higher contribution of P7/8 for neutrality. In 7-month-olds, the topography of the expression-change response discriminated between disgust, neutrality, and happiness, with a lower contribution of O1/2 to the response to happiness compared to both disgust and neutrality. As for the 3.5-month-olds, P7/8 channels contributed more to the discrimination of neutrality than disgust. Note also the tendency for a higher contribution of FC1/2 to happiness, that was driven by a significant response over FC2 for happiness (see previous analyses). “*” indicated significant differences (p < 0.05).

FIGURE 4

General visual response for the three target facial expressions at 3.5 months (A) and 7 months (B). Left part. SNR was calculated on the summed harmonics values averaged between the four occipital electrodes, considering Z scores of at least one of the four occipital electrodes was significant (across harmonics Zs > 2.60, p < 0.006), showing a high SNR for each expression: around six for every expression-condition in the 3.5-month-olds, and around three in the 7-month-olds. Right part. Topographical 3D map (back view) shows a clear medial occipital cortex response of the noise-corrected amplitude (in μV) values summed until the 6th harmonic of the 6 Hz response for the three conditions.