Neural categorization of voices at four months of age
(2022) The International Congress of Infant Studies (ICIS) — Location: Ottawa, Canada (7.July.2022)
Files
No attached file found for this publication.
Details
- Authors
Calce, Roberta Pia 
- et. al.
- Abstract
- Human voices are among the most relevant sounds in our daily life. Neuroimaging studies in adults demonstrated the existence of regions in the superior temporal sulcus- the “temporal voice areas” (TVAs)- that respond preferentially to voices vs other sounds (Belin et al., 2000). Further behavioral and electrophysiological studies also suggested that this voice sensitivity is not accounted for by low-level auditory features, but rather by higher-level categorization of vocal sounds (Barbero et al., 2021; Agus et al., 2017). How early in the developing brain such selective categorical response occurs remains however poorly understood. Auditory developmental research has mainly focused on language development (Sharifian et al., 2019) investigating the ability of infants to discriminate between native and foreign languages, or differentiate speech from non-speech sounds, phonemes, and even syllables (Gasparini et al., 2021; Dehaene-Lambertz et al., 2010; Karzon et al., 1989.), without directly addressing whether voices are processed differently than other sounds. When investigating voice perception, studies could not disentangle the role of low-level acoustic properties of voice sounds, leaving open the question of whether infants can process voices as one homogeneous perceptual category (Grossmann et al., 2010). In the present study, we investigate whether a voice-selective response occurs already in the infant brain, relying on electroencephalography (EEG) frequency-tagging. This approach provides an objective marker of the brain’s automatic ability to categorize vocal information with a high signal-to-noise ratio within a short testing duration, making the technique optimal for testing infants. Twenty-three 4-month-old infants listened to streams of heterogeneous sounds presented periodically at 3.33 Hz to elicit a brain response at the same frequency in the EEG amplitude spectrum. Importantly variable voice stimuli were inserted as each third sound (1.11 Hz) to elicit an additional response in the EEG spectrum if the infant brain discriminates voices from other sounds and generalizes across heterogeneous vocal exemplars (standard condition). Moreover, infants were also presented with a scrambled version of the same stimuli that disrupts intelligibility but preserves spectral content. Here, a response at the voice rate would be interpreted as merely driven by low-level features. The analysis is conducted in the frequency domain at sounds (3.33 Hz) and voice rates (1.11 Hz) in both conditions. Results show no amplitude difference between standard and scrambled conditions at the general sound frequency (3.33 Hz), suggesting that intelligible sounds are processed similarly to their scrambled version. Importantly, we observe a neural response over temporal channels at the voice presentation rate (1.11 Hz), that is significantly larger in the standard condition, this pattern being consistent across individual infants. Taken together, these results indicate that the infant brain successfully recognizes variable voice excerpts as items of the same category. The comparison with their scrambled version suggests that voice selectivity does not merely rely on acoustic frequency content, bringing evidence of high-level voice categorization as early as 4 months of age.
- Affiliations
Citations
Calce, R. P., Rekow, D., Barbero, F., Kiseleva, A., Talwar, S., Leleu, A., Collignon, O., & et al. (2022). Neural categorization of voices at four months of age. The International Congress of Infant Studies (ICIS), Ottawa, Canada. https://hdl.handle.net/2078.5/238379