Pierre-Louis Audette¹ (), Justin Duncan¹, Caroline Blais¹, Daniel Fiset¹; ¹Université du Québec en Outaouais

In our presentation at VSS2023, we explored the correlation between N170 face sensitivity and diagnostic information processing during a 10-identities recognition task (Audette et al., 2023). We utilized sparse facial stimuli generated with Bubbles, and previously published classification images (Royer et al., 2018) were employed to quantify the available diagnostic information on a stimulus basis (0-100%). Remarkably, we observed a linear increase in N170 amplitude as diagnostic information increased (r=-0.98), indicating thorough processing of diagnostic information during face identification. However, the revealed amount of diagnostic information strongly correlated with the amount of facial surface (r=0.99), suggesting that this effect may be influenced not only by diagnostic information but also by the overall amount of facial information. Consequently, the second phase systematically examined whether N170 and N250 reflect the quantity of diagnostic information processed by the brain while keeping the amount of facial information constant. Sparse facial stimuli were generated using a method similar to the first phase. To standardize the facial surface across stimuli, we applied inverse bubbles to an average face of the 10 identities, replacing face regions hidden by bubbles with non-diagnostic facial information. The stimuli were categorized into 12 bins ranging from 0.001% to 100%, with an additional 0% condition (average face). EEG data were collected from 10 participants during 1,300 trials of a 10-identities recognition task. We analyzed N170 and N250 peak amplitudes at PO8, along with behavioral responses. Results indicated that only N250 peak amplitude varied: as diagnostic information increased, N250 amplitude linearly increased (r=-0.90). Moreover, N250 amplitudes across bins strongly correlated with recognition accuracies (r=-0.86). Thus, N250 seems to reflect in-depth processing of diagnostic information during face identification, while N170 responds to the ease of categorizing the stimulus as a face.

Acknowledgements: This work was supported by grants from the Natural Sciences and Engineering Research Council of Canada (RGPIN-2022-04350).