Temporal dynamics of face selection mechanism in the context of similar and dissimilar faces: ERP evidence for biased competition within the ventral occipito-temporal cortex using ICA.

This study was designed to investigate the spatio-temporal characteristic of face selection in the context of the Biased Competition (BC) model using electroencephalography (EEG). Pairs of similar (SIM) or dissimilar (DISS) faces were presented simultaneously. Subjects had to attend to one face (ATT) or ignore both faces (IGN). According to the BC account, simultaneously presented faces compete for representation. Spatial attention biases these competitive interactions towards neural processing of the attended face alone. A preceding functional magnetic resonance imaging (fMRI) study confirmed the validity of BC as selection principle in the occipital face area (OFA) and in the fusiform face area (FFA) (IGN SIM<IGN DISS; ATT SIM=ATT DISS) (Gentile and Jansma, 2010). The directionality of the similarity modulation within the IGN condition was interpreted in line with the assumption that within FFA similar faces are more likely to be encoded by largely overlapping population of neurons, hence leading to a stronger competition. In the present EEG experiment we investigated when neural competition and the bias effect take place. Using independent component analysis (ICA), we focused on face-processing-related components N170 and N250. Whereas we could not find any context-dependent modulation of the N170, we observed a context x task interaction within the N250 time-window. Together, the fMRI and EEG results suggested that faces competed for representation (IGN SIM<IGN DISS) in the N250 time-window within FFA. In addition, at the same latency (250 ms post-stimulus onset) and region of interest (FFA) the data supported the idea that attention solved neural competition in favor of the relevant face (ATT SIM=ATT DISS). We also observed a main effect of attention frontally within the N2b time-window, i.e. about 50 ms later than the attentional bias that solved competitive interactions. Based on these findings, it is plausible to conclude that this frontal activity was not directly involved in the bias control during selection in FFA.