The neural dynamics of familiarity‐dependent face identity representation

Abstract

Recognizing a face as belonging to a given identity is essential in our everyday life. Clearly, the correct identification of a face is only possible for familiar people, but ‘familiarity’ covers a wide range—from people we see every day to those we barely know. Although several studies have shown that the processing of familiar and unfamiliar faces is substantially different, little is known about how the degree of familiarity affects the neural dynamics of face identity processing. Here, we report the results of a multivariate EEG analysis, examining the representational dynamics of face identity across several familiarity levels. Participants viewed highly variable face images of 20 identities, including the participants' own face, personally familiar (PF), celebrity and unfamiliar faces. Linear discriminant classifiers were trained and tested on EEG patterns to discriminate pairs of identities of the same familiarity level. Time‐resolved classification revealed that the neural representations of identity discrimination emerge around 100 ms post‐stimulus onset, relatively independently of familiarity level. In contrast, identity decoding between 200 and 400 ms is determined to a large extent by familiarity: it can be recovered with higher accuracy and for a longer duration in the case of more familiar faces. In addition, we found no increased discriminability for faces of PF persons compared to those of highly familiar celebrities. One's own face benefits from processing advantages only in a relatively late time‐window. Our findings provide new insights into how the brain represents face identity with various degrees of familiarity and show that the degree of familiarity modulates the available identity‐specific information at a relatively early time window.

Your place or mine? The neural dynamics of personally familiar scene recognition suggests category independent familiarity encoding

“…More recently, Kovács et al (2023) found that the that familiarity enhances identity decoding in the time range of 200-400 ms, while earlier neural representations appear to be relatively insensitive to the level of familiarity. Our study builds upon and extends these findings by demonstrating that familiarity enhances not only face representations but also those of non-face stimulus categories.…”

“…Our results are further based on only highly personally familiar stimuli. Different degrees of familiarity (Li et al, 2022;Kovács et al, 2023), like perceptual and contextual familiarity (Kovács, 2020), should be related to the same processing mechanism but evoke slightly differential activations (Ambrus et al, 2021;Dalski et al, 2022a). Indeed, Beldzik and colleagues (2021)have shown that personally familiar scenes and known, but personally unfamiliar scenes, engage different brain networks related to an egocentric or more allocentric worldview, respectively.…”

“…Indeed, Beldzik and colleagues (2021)have shown that personally familiar scenes and known, but personally unfamiliar scenes, engage different brain networks related to an egocentric or more allocentric worldview, respectively. As the level of familiarity is closely related to and enhances face identity processing gradually (Dobs et al, 2019;Kovács et al (2023), it would be interesting to test if a similar identity effect exists for scenes as well, e.g., a differentiation between ones' very well known, own bathroom and the bathroom at a railway station which is visited only irregularly. This, however, should only be a byproduct of scene processing, since identification is not the primary goal of scene perception and neural mechanisms should not be tuned explicitly towards it (Steel et al, 2021).…”

“…This is however unlikely to explain the results of the main, group-level analysis where decoding was performed on a cross-participant level. Nonetheless, the encoding of familiarity degree for scenes is an interesting field and should be tested in the future, similarly to that of faces (Li et al, 2022;Kovács et al, 2023).…”