Resumen
Representation of self-face is vulnerable to cognitive bias, and consequently, people often possess a distorted image of self-face. The present study sought to investigate the neural mechanism underlying distortion of self-face representation by measuring event-related potentials (ERPs) elicited by actual, aesthetically enhanced, and degraded images of self-face. In addition to conventional analysis of ERP amplitude and global field power, multivariate analysis based on machine learning of single trial data were integrated into the ERP analysis. The multivariate analysis revealed differential pattern of scalp ERPs at a long latency range to self and other familiar faces when they were original or aesthetically degraded. The analyses of ERP amplitude and global field power failed to find any effects of experimental manipulation during long-latency range. The present results indicate the susceptibility of neural correlates of self-face representation to aesthetical manipulation and the usefulness of the machine learning approach in clarifying the neural mechanism underlying self-face processing.