Combining Real-Time Neuroimaging With Machine Learning to Study Attention to Familiar Faces During Infancy: A Proof...

Area of Science:

• Developmental neuroscience
• Cognitive neuroscience
• Infant social cognition

Background:

• Early social development relies on processing familiar and novel faces.
• Neural attention to familiar faces increases in the first 6 months.
• By 12 months, group analyses show no difference in brain responses to familiar versus unfamiliar faces.

Purpose of the Study:

• To investigate individual infant neural responses to faces varying in familiarity.
• To determine if apparent null findings in group analyses mask heterogeneous individual preferences.
• To apply Neuroadaptive Bayesian Optimisation (NBO) for real-time analysis of infant brain responses.

Main Methods:

• Sixty-one infants (5-12 months) viewed morphed familiar-to-unfamiliar faces.
• Electroencephalography (EEG) recorded brain activity.
• A Bayesian Optimisation algorithm identified stimuli eliciting maximal Negative central (Nc) event-related potential (ERP) amplitude for each infant.

Main Results:

• Neuroadaptive Bayesian Optimisation (NBO) predicted an optimal stimulus in 85% of infants, indicating individual preferences.
• No group-level difference in Nc amplitude was found between familiar and unfamiliar faces.
• Lower attrition rates (15%) were observed compared to traditional methods (22%).

Conclusions:

• The lack of a group-level preference for familiar over unfamiliar faces in middle infancy is due to heterogeneous individual preferences.
• NBO is effective in identifying subgroups with distinct brain activation patterns.
• Individual attentional preferences were not correlated with age or social behavior.