Late positive event-related potentials in electroencephalography can distinguish acquaintances from strangers when both groups recognize faces

Abstract

Recognition of a familiar face can be important for not only basic cognitive neuroscience but also advanced personal authentication schemes. When a person looks at faces, social and emotional aspects as well as appearance are processed in the brain, which can induce different cognitive processes between genuine users and imposters. For the development of such an authentication scheme, we proposed a new authentication paradigm using familiar faces together with electroencephalography (EEG) and compared neural activity between acquaintances and strangers while they searched for known faces via an event-related potential (ERP) analysis. 28 pictures of human faces were prepared among which 4 were designated as target faces. 29 participants belonged to one of 2 groups: the acquaintance group of 14 participants indeed knew 4 target faces for long time (> 1 years) whereas the stranger group of 15 participants had not known a priori any face but were informed of 4 target faces before the task. In each trial, participants were shown 8 faces arranged in the form of a 3x3 matrix with the center slot empty. The face matrix was presented for different durations of 1, 1.5, 2 or 2.5 s over four blocks. Half of trials in each block included a target face (i.e. familiar condition) and the rest did not (i.e. unfamiliar condition). Participants searched for a target face and located it in the matrix using a numerical keypad (3x3 matrix for of 9 keys). If they could not find the target, they pressed the center key on the keypad. EEG and eye tracking data were recorded simultaneously. The epochs for ERP analysis were time-locked to the onset of the longest fixation on correct answer in the familiar condition or the average response time of the familiar condition in the unfamiliar condition. Error rates were similar between two groups (p > 0.05). The block with 1s stimulus duration yielded a significantly higher error rate than others (p < 0.001). The ERP analysis revealed that late positive potentials (LPPs) at FC1 and F3 were significantly larger in the acquaintance group than in the stranger group (p < 0.05) even when the both groups successfully identified the target faces. This is in line with preceding studies showing that LPP reflects more elaborate processing such as long-term memory traces. The current study suggests plausibility of our new paradigm as a novel authenticating scheme using EEG and face recognition.