ICA-derived cortical responses indexing rapid multi-feature auditory processing in six-month-old infants

• Published in: NeuroImage (Orlando, Fla.) Fla.), 2016-06, Vol.133, p.75-87
• Main Authors: Piazza, Caterina, Cantiani, Chiara, Akalin-Acar, Zeynep, Miyakoshi, Makoto, Benasich, April A., Reni, Gianluigi, Bianchi, Anna Maria, Makeig, Scott
• Format: Article
• Language: English
• Subjects: Auditory Cortex - physiology; Babies; Brain Mapping - methods; Data Interpretation, Statistical; EEG; Electroencephalography; Electroencephalography - methods; Event-related potential; Female; Humans; Independent component analysis; Infant; Infants; Information processing; Language; Linguistics; Male; Nerve Net - physiology; Pitch Perception - physiology; Principal Component Analysis; Rapid auditory discrimination; Reproducibility of Results; Sensitivity and Specificity; Source localization; Studies
• ISSN: 1053-8119; 1095-9572
• DOI: 10.1016/j.neuroimage.2016.02.060

The abilities of infants to perceive basic acoustic differences, essential for language development, can be studied using auditory event-related potentials (ERPs). However, scalp-channel averaged ERPs sum volume-conducted contributions from many cortical areas, reducing the functional specificity and interpretability of channel-based ERP measures. This study represents the first attempt to investigate rapid auditory processing in infancy using independent component analysis (ICA), allowing exploration of source-resolved ERP dynamics and identification of ERP cortical generators. Here, we recorded 60-channel EEG data in 34 typically developing 6-month-old infants during a passive acoustic oddball paradigm presenting ‘standard’ tones interspersed with frequency- or duration-deviant tones. ICA decomposition was applied to single-subject EEG data. The best-fitting equivalent dipole or bilaterally symmetric dipole pair was then estimated for each resulting independent component (IC) process using a four-layer infant head model. Similar brain-source ICs were clustered across subjects. Results showed ERP contributions from auditory cortex and multiple extra-auditory cortical areas (often, bilaterally paired). Different cortical source combinations contributed to the frequency- and duration-deviant ERP peak sequences. For ICs in an ERP-dominant source cluster located in or near the mid-cingulate cortex, source-resolved frequency-deviant response N2 latency and P3 amplitude at 6 months-of-age predicted vocabulary size at 20 months-of-age. The same measures for scalp channel F6 (though not for other frontal channels) showed similar but weaker correlations. These results demonstrate the significant potential of ICA analyses to facilitate a deeper understanding of the neural substrates of infant sensory processing. •Both auditory and extra-auditory cortex are involved in infant auditory processing.•Multiple cortical areas contribute to the detection of diverse acoustic features.•Source-resolved ERP measures can predict language acquisition abilities.•EEG-based imaging is a promising approach for the study of infant brain function.