Abstract
Face perception is believed to involve two types of processes. When the face is presented normally, separate face features are expected to integrate into a global whole (i.e., holistic/configural processing). However, when the face is presented in the unusual upside-down orientation it is decoded into elementary features (analytic processing), so its recognition is known to be severely disrupted (Farah et al., 1998, Maurer et al., 2002). In the present study we have used the network neuroscience approach to analyze brain mechanisms of holistic and analytic face processing using the face inversion effect. Stimuli were 30 grayscale photographs (15 female and 15 male faces) from WSEFEP Database (Olszanowski et al., 2015). Faces were limited by a mask highlighting only internal features. To impair holistic processing two types of facial images were created for each photograph - an inverted image and a scrambled image. All stimuli (upright, inverted and scrambled) were presented randomly for 600 ms. Forty-five participants (33F, 12M, age range 19-24) were tested. They were asked merely to look at the image (free-viewing task). During the perception EEG brain responses were recorded with 64 electrodes placed according to the international 10-10 system with a Brain Products ActiChamp amplifier (BrainProducts, Munich, Germany). Using mathematical graph theory, we calculated measures of brain integration (graph characteristic path length) and segregation (cluster coefficient of a graph) for different EEG frequency bands. We have found that the EEG cluster coefficient in theta (4-8 Hz) range varied with the types of the presented faces (F(1, 110)=7.63, p=0.01). The cluster coefficient was lowest for the inverted faces (M=0.56, SD=0.03) and highest for the upright faces (M=0.6, SD=0.02). The link between the holistic/analytic processes in face perception and the brain integration/segregation mechanisms is discussed.