Background. Face recognition and face-name memory are important aspects of interpersonal communication. Event-related potential (ERP) methods and their source localization provide opportunities to investigate the spatiotemporal characteristics of brain activity in this domain. However, in current cognitive models of face recognition, the source localization of the N400 component associated with personal identity nodes (PINs) in semantic information remains insufficiently studied.

Objective. This study aimed to identify differences in brain activity when recognizing long-familiar celebrity faces versus recently learned faces, in particular — to test whether faces containing more semantic information evoke an N400 component with different amplitude and latency, and to determine the source localization of these differences.

Study Participants. The study included 66 university students (mean age = 21.83 years, SD = 3.57): 33 females (17 Chinese and 16 Russian) and 33 males (18 Chinese and 15 Russian). All were right-handed.

Methods. The study was carried out using electroencephalography (EEG) and implemented the methods of evoked potential and source localization. To process and analyze the obtained EEG data, a brain activity localization method, the "Virtually Implanted Electrode," developed by A.V. Vartanov (Russian Patent No. 2785268), was used.

Results. Source localization analysis showed that the N400 component differed significantly between long-familiar celebrity faces (OLD) and newly learned faces (LRN) in the medial globus pallidus, thalamus, left hippocampus, ventral striatum, and right VO1 and VO2 areas, with OLD faces eliciting greater N400 amplitude. Latency differences were observed in the range of 300–480 ms. Analysis of the directed connectivity graph indicated that both face categories activated similar neural networks, with VO1 R, VO1 L, and VO2 L showing strong activation, and the ventral striatum, left medial globus pallidus, and left thalamus contributing to the activation of the right medial globus pallidus with high correlation coefficients (r > 0.93).

Conclusions. This study demonstrates that face-name memory engages a broad neural network encompassing cortical and subcortical structures. Differences in N400 between long-familiar and newly learned faces likely reflect information contained in personal identity nodes (PINs). Key components of this network include the medial globus pallidus, thalamus, left hippocampus, ventral striatum, and VO1 and VO2 areas in the right hemisphere.