Abstract:

The facial width-to-height ratio (fWHR) is a stable perceptual structure of all faces. It is calculated by dividing the face width (the distance between the left and right zygion) by the face height (the distance between the eyebrow and the upper lip). Previous studies have demonstrated that men's facial width-to-height ratio is a reliable clue to noticing aggressive tendencies and behavior. Individuals with higher fWHR were considered by observers as more aggressive than those with lower fWHR. The researchers proposed that this may be related to facial expression. Observers more readily saw anger in faces with a relatively high fWHR and more readily saw fear in faces with a relatively low fWHR. However, it is unclear what the neural mechanism of fWHR is, particularly in the absence of attention. The present study investigated this issue by recording visual mismatch negativity (vMMN), which indicates automatic processing of visual information under unattended conditions. We hypothesized that faces with high fWHR would elicit a larger vMMN compared to faces with low fWHR. If the above result is related to the fact that high fWHR faces appear angrier and low fWHR faces appear more fearful, then high fWHR faces displaying an angry expression would evoke vMMN and low fWHR faces displaying a fearful expression would evoke vMMN.

Participants performed a size-change-detection task on a central cross, while random sequences of faces were presented in the background using a deviant-standard-reverse oddball paradigm. High fWHR faces (deviant stimuli) were presented less frequently among low fWHR faces (standard stimuli), or vice versa. This paradigm allows us to investigate the vMMN induced by the same physical stimulus, as the same stimulus is utilized as both the deviant and the standard stimulus in different blocks, thus reducing the influence of lower-level physical stimulus attributes on ERP components. 41 (19 females, 21.05 ± 1.70 years) and 25 (13 females, 20.56 ± 1.635 years) Chinese participated in Experiment 1 and 2, respectively. In Experiment 1, faces with neutral expressions were used. We employed 2 (fWHR: high vs. low) × 2 (stimuli: deviant vs. standard) within-subject design. The occipital-temporal vMMN (the deviant stimuli elicited more negative responses than the standard stimuli) emerged in the latency range of 200~500 ms for faces with high fWHR (200~250 ms: 4.117 ± 0.591 vs. 4.685 ± 0.582 μV, p < 0.001, 95% CI = [-0.804, -0.331]; 250~300 ms: 3.273 ± 0.562 vs. 4.869 ± 0.553 μV, p < 0.001, 95% CI = [-2.043, -1.150]; 300~350 ms: 2.026 ± 0.532 vs. 3.725 ± 0.510 μV, p < 0.001, 95% CI = [-2.114, -1.284]; 350~400 ms: 2.104 ± 0.483 vs. 3.692 ± 0.443 μV, p < 0.001, 95% CI = [-2.064, -1.113]; 400~450 ms: 1.163 ± 0.463 vs. 2.936 ± 0.431 μV, p < 0.001, 95% CI = [-2.231, -1.316]; 450~500 ms: 0.331 ± 0.449 vs. 2.231 ± 0.434 μV, p < 0.001, 95% CI = [-1.889, -0.752]) and in the latency range of 200~250 ms (4.117 ± 0.591 vs. 4.685 ± 0.582 μV, p < 0.001, 95% CI = [-0.804, -0.331]) and 300~350 ms (2.563 ± 0.648 vs. 3.256 ± 0.588 μV, p = 0.009, 95% CI = [-1.207, -0.179]) for faces with low fWHR (Figure 1). More importantly, faces with high fWHR elicited a higher vMMN than those with low fWHR faces in the 300~350 ms latency range (-1.728 ± 0.242 vs. -0.693 ± 0.254 μV, p = 0.010, 95% CI = [-1.804, -0.266]).

In Experiment 2, faces with expressions of fear and anger were used. We employed 2 (fWHR: high vs. low) × 2 (stimuli: deviant vs. standard) × 2 (face expression: angry vs. fearful) within-subject design. Results showed that high-fWHR faces displaying an angry expression elicited a vMMN in the 200~250 ms at P4/PO8 electrode sites (P4: 2.291 ± 0.547 vs. 2.694 ± 0.542 μV, p = 0.039, 95% CI = [-0.784, -0.022]; PO8: 1.298 ± 0.669 vs. 1.966 ± 0.664 μV, p = 0.011, 95% CI = [-1.166, -0.169]) and 300~400 ms latency ranges (300~350 ms: P3: 1.068 ± 0.361 vs. 1.492 ± 0.291 μV, p = 0.009, 95% CI = [-0.731, -0.116]; PO5: 0.689 ± 0.580 vs. 1.097 ± 0.525 μV, p = 0.044, 95% CI = [-0.804, -0.012]; PO8: 0.775 ± 0.636 vs. 1.348 ± 0.702 μV, p = 0.049, 95% CI = [-1.143, -0.002]. 350~400 ms: P3: 0.613 ± 0.307 vs. 0.979 ± 0.229 μV, p = 0.031, 95% CI = [-0.696, -0.036]; PO8: 0.730 ± 0.553 vs. 1.343 ± 0.587 μV, p = 0.035, 95% CI = [-1.180, -0.047]), while low-fWHR faces displaying a fearful expression elicited a vMMN in the 250~400 ms latency range (250~300 ms: 1.484 ± 0.600 vs. 1.911 ± 0.551 μV, p = 0.026, 95% CI = [-0.797, -0.056]; 300~350 ms: 0.239 ± 0.538 vs. 0.820 ± 0.510 μV, p = 0.022, 95% CI = [-1.069, -0.092]; 350~400 ms: 0.657 ± 0.435 vs. 1.109 ± 0.390 μV, p = 0.035, 95% CI = [-0.870, -0.035]), especially in the left hemisphere (Figure 2).

To gain a better understanding of the effect of facial expression on the degree of automatic processing in high and low fWHR, we compared vMMN responses to faces with high fWHR presenting neutral and angry expressions, and faces with low fWHR showing neutral and fear expressions (Table 1 and 2). The results revealed that faces with high fWHR displaying an angry expression elicited smaller vMMN than those displaying a neutral expression (300~350 ms at PO5 site: t(64) = -3.654, p = 0.001, Cohen’s d = 0.272, 95% CI = [-2.180, -0.639]; 300~350 ms at PO8 site: t(64) = -3.455, p = 0.001, Cohen’s d = 0.289, 95% CI = [-2.581, -0.690]; 350~400 ms at PO8 site: t(64) = -3.279, p = 0.002, Cohen’s d = 0.305, 95% CI = [-2.538, -0.617]).

In conclusion, the present findings suggest that the facial width-to-height ratio is associated with automatic processing and provide new electrophysiological evidence for the different mechanisms underlying high and low fWHR faces under unattended conditions. The automatic processing of high fWHR exhibits greater neural activity than that of low fWHR, which might be related to facial expressions representing facial aggression. Consistent with previous studies, the current finding demonstrates that automatic processing of high and low fWHR is promoted by expressions of anger and fear, respectively. At the same time, due to the automatic processing of facial expressions, the automatic processing of faces with high fWHR is weakened by angry faces relative to neutral faces.

Key words: face, facial width-to-height ratio, emotion, vMMN, automatic processing