Figure 3
From: Deepfake smiles matter less—the psychological and neural impact of presumed AI-generated faces
Differences in activity of the P1, N170, EPN, and LPP components depending on Information (Real / Fake) and Emotion (Angry / Neutral / Happy) during the rating task. Left: Grand-average amplitudes following face onset, pooled across the respective region of interest (ROI) for each component. The gray shading marks the time range taken into account for the analyses for each component. Middle: For each component, topographies show the differences in activity between Angry and Neutral, and Happy and Neutral faces for each Information condition. Asterisks indicate a significant difference between the respective conditions observed in the analyses. Right: Amplitude differences between Angry and Neutral, and Happy and Neutral faces for both Information conditions. The colored dots represent by-participant differences of the respective conditions; the black dot represents the mean difference; error bars display 95% confidence intervals around the mean. Asterisks indicate significant differences between the respective conditions observed in the analyses. *** p < .001, ** p < .01, * p < .05; n.s. = non-significant.
