Facial attractiveness in the eyes of men with high arousal

Abstract Introduction Individuals differ in how they judge facial attractiveness. However, little is known about the role of arousal level and gender differences in individuals’ facial attractiveness judgments. Methods We used resting‐state electroencephalogram (EEG) to investigate this issue. A total of 48 men (aged 22.5 ± 3.03 years [mean ± SD], range: 18–30 years) and 27 women (aged 20.3 ± 2.03 years [mean ± SD], range: 18–25 years) participated in the experiment. After the EEG was collected, participants were instructed to complete a facial attractiveness judgment task. Connectome‐based predictive modeling was used to predict individual judgment of facial attractiveness. Results Men with high arousal judged female faces as more attractive (M = 3.85, SE = 0.81) than did men with low arousal (M = 3.33, SE = 0.81) and women (M = 3.24, SE = 1.02). Functional connectivity of the alpha band predicted judgment of female facial attractiveness in men but not in women. After controlling for the age and variability, the prediction effect was still significant. Conclusion Our results provide neural evidence for the enhancement of the judgment of facial attractiveness in men with high arousal levels, which supports the hypothesis that individuals’ spontaneous arousal contributes to variations in facial attractiveness preferences.


INTRODUCTION
Facial attractiveness is a key factor in facial impression (Todorov et al., 2015), which influences decisions, particularly mate choice (Buss & Schmitt, 2019;Pandey & Zayas, 2021). Highly attractive faces are associated with greater benefits. For example, individuals with attractive faces are considered healthy and as having good traits (Mengelkoch et al., 2022;Tsukiura & Cabeza, 2011). The judgment of facial attractiveness varies in different individuals (Han, Liu, et al., 2020;Little, 2014;Thiruchselvam et al., 2016). Brain functional connectivity is an objective index that can predict human behavior (Shen et al., 2017).
However, the role of individual spontaneous activation of the brain (such as arousal) in the judgment of facial attractiveness remains unclear.
Arousal plays an important role in attractiveness processing (Montoya & Horton, 2020). A previous study found that high-arousal state evoked by a swaying bridge enhanced attractiveness ratings compared with the low-arousal state evoked by a stable bridge (Dutton & Aron, 1974 ). Recently, researchers used the pheromone androstenol to evoke a high-arousal state that contributed to participants rating faces as significantly more attractive compared with the control odor (Beaton et al., 2022). The theory of misattribution of arousal proposes that a state of high arousal causes individuals to misattribute this feeling to attraction (Little, 2014;White et al., 1981).
Arousal has also been interpreted according to the context. For example, enhanced arousal was suggested to lead to polarize outcomes in the negotiation context (Brown & Curhan, 2013). In this study, the theory of misattribution of arousal was extended by focusing on the spontaneous arousal of the perceiver. This extension is meaningful because individual differences exist in the influence of context-evoked arousal on the judgment of attractiveness (Jouffre, 2015). Furthermore, the arousal level was rated subjectively in previous studies (Fayolle et al., 2015;Zhou et al., 2021). However, a resting-state electroencephalography (EEG), an objective approach, was applied in this study to measure individual spontaneous arousal. Alpha power in the resting state is the classic index of arousal, with a smaller alpha reflecting stronger arousal (Barry et al., 2020;Pivik & Harman, 1995). Therefore, the EEG method was deemed suitable to investigate the influence of individual arousal on the perception of facial attractiveness.
Gender differences exist in the processing of facial attractiveness.
To cope with sex-specific adaptive problems, in the mate choice, men place greater value on female attractiveness: an attractive female face signals a potential mating opportunity for men (Buss & Schmitt, 2019).
Men also show stronger activation of the area of the brain associated with reward when judging facial attractiveness than women (Cloutier et al., 2008). Men are able to quickly process female faces, and the effect was not influenced by the difficulty of task, which suggests that male process female faces automatically (Klümper et al., 2020). Conversely, an attractive female face poses a potential threat to other women's reproductive success (Buss & Schmitt, 2019). For this reason, women prefer fewer female applicants because of intrasexual com-petition (Luxen & Vijver, 2006). Furthermore, one study found that women's arousal was unrelated to their male facial attractiveness judgment (Hagerman et al., 2017), and others determined that when using music to evoke arousal, women but not men judged male faces as more attractive than in the silent control condition (Marin & Rathgeber, 2022;Marin et al., 2017). Women rate male faces as significantly less attractive than men rate female faces (Fisher, 2004). Additionally, previous studies have found some meaningful results concerning women rating male facial attractiveness. Bech-Sørensen and Pollet (2016) showed that women emphasized facial attractiveness less than men, valued a high earning potential more than men did, and had a greater preference for marrying someone older. The sexual strategy theory also suggests that men and women have evolved to pursue different mating strategies, with men being more attentive to facial attractiveness, while women value characteristics signaling resources (e.g., power, status; Buss & Schmitt, 2019), which are rarely reflected in faces. Therefore, the female face was considered the target in this study.
To further determine the association between the brain and judgment of facial attractiveness, we employed connectomebased predictive modeling (CPM), which has been widely applied to the prediction of individual behavior (Cai et al., 2020). CPM aims to identify functional connections throughout the brain that can predict a specific behavioral measure based on individual differences in connectivity strength. One important feature of CPM is that a set of functional connections, recognized as a predictor of a behavioral variable in one sample, can effectively generalize the same behavioral variables in independent samples.
By using CPM, researchers have successfully predicted individual childhood aggression (Ibrahim et al., 2022), processing speed in older adults (Gao, 2020), and loneliness (Feng et al., 2019), identifying related biomarkers. However, an EEG biomarker that can predict individual facial attractiveness preference remains unknown.
In summary, we aimed to explore the influence of spontaneous arousal on individual judgment of facial attractiveness and gender differences. Since women value male resources more than facial attractiveness in mate selection (Buss & Schmitt, 2019), only female faces were used as the target faces in the current study. Men (mate choice) and women (intrasexual competition) were instructed to judge the attractiveness of female faces. Resting-state EEGs were also collected.
We expected that men with high arousal levels would judge female faces as more attractive than would those with low arousal levels (H1).
Because of gender differences, an attractive female face signals mate choice in men but intrasexual competition in women (Luxen & Vijver, 2006). Thus, we expected that men would give higher scores to female faces than women would (H2). CPM was used to identify the individual preference in the judgment of facial attractiveness. As previously mentioned, arousal is strongly associated with attractiveness, particularly for men in terms of judging women's faces (Little, 2014). Thus, we expected that the functional connectivity of the alpha band could predict men's perception of female attractiveness (H3).

Participants
Eighty participants were recruited from a university. Two men and three women were excluded due to excessive EEG artifacts (an EEG amplitude that was greater than 100 μV for more than half of the total time

Procedure
Resting-state EEG data for 200 s were recorded for each participant.
The participants sat in a chair and were instructed to maintain a comfortable posture, close their eyes but remain awake, and refrain from moving.
After resting-state EEG acquisition, participants performed the experimental tasks. The procedure of this experiment was presented using E-Prime software (version 3.0; Figure 1). Each trial was preceded by a fixation cross in the center of the screen for 500 ms, followed by a 2000-ms female face image at random. A schematic of a 7-point Likert-type scale (1 = very unattractive and 7 = very attractive) was then displayed, and participants were instructed to judge facial attractiveness by pressing a number from 1 to 7 on a keyboard.
The schematic scale disappeared when the participants did not press the key within 5000 ms. A total of 80 trials were conducted in the formal experiment.

EEG data recording and processing
EEG data were recorded at a sample rate of 500 Hz, with electrode caps containing 64 Ag/AgCl electrodes arranged in a 10−20 system and referenced to the midline frontal-central electrode (FCz). Electrode impedance was maintained below 5 kΩ.
EEG data were imported into EEGLAB (Delorme & Makeig, 2004) in MATLAB (Mathworks, v2016a). The processing analysis steps of resting-state EEG data were as follows: (1) Through visual inspection, bad channels that contained excessive noise or artifacts were marked and interpolated. (2) The EEG data were re-referenced to an average reference. (3) The EEG data were filtered offline using a bandpass filter (0.1−30 Hz after a notch filter for 50 Hz line noise). (4) An independent component analysis was used to remove the components related to eye movements, blinks, head movement, electrocardiogram, and other types of artifacts (Delorme et al., 2007). (5) Data were deleted from the analysis if artifacts at the signal channel accounted for more than 50% of the recording time. After preprocessing, behavioral and EEG data of five participants (two men and three women) were excluded from further analysis. (6) Finally, the data were subscribed to the time-frequency analysis and CPM analysis.

Data analysis
A time-frequency analysis was used to quantify neural activity using the FieldTrip toolbox ). An EEG power spectrum was calculated using the fast Fourier method applied on Hanning-

Control and variability analysis
Age was controlled to rule out its potential influence on the predictive model. LOOCV and the premutation test were performed 1000 times and were calculated. To verify the stability of the results, we used thresholds of p < .05 and p < .01.

Behavioral results
The one-way ANOVA revealed a

EEG results
Alpha

F I G U R E 2
The results of attractiveness judgment and alpha power among high-and low-arousal men and women. *p < .5; **p < .01.

F I G U R E 3
The alpha power of high-and low-arousal men and women evoked during the resting state.
lower than that of women (p = .04; Figure 3). This indicated that the groups had been successfully divided according to arousal.

CPM results
For men, the CPM results revealed that the combined network successfully predicted judgment of facial attractiveness using alpha functional connectivity (r = .37, p < .01), which remained unchanged after the permutation test (p perm = .01). However, the positive (r = .20, p = .18, p perm = .42) and negative networks (r = .15, p = .32, p perm = .48) failed to predict judgment of facial attractiveness. No significant results were found in the prediction models constructed using the other bands in men (see Table 1).
For women, the combined network of the delta band significantly predicted judgment of facial attractiveness, which was also successfully predicted after the premutation test (r = .42, p = .03, p perm = .03).
The other bands of the network did not predict the judgment of facial attractiveness (see Table 1; Figures 4 and 5).

F I G U R E 4 (A)
The combined network of the alpha that predicts the behavior performance in men and (B) the combined network of the delta that predicts the behavior performance in women.

Age control
After controlling for age, the alpha functional connectivity in men remained significant for the combined network (r = .33, p = .02,  Note: Dash (-) indicates that no functional connectivity can predict the behavior performance. *p < .5; **p < .01.

F I G U R E 5
The p values without (the top line) and with (the bottom line) permutation test that were calculated by the connectome-based predictive modeling.
p perm = .44). No other significant results were observed (see Tables 2 and 3).

Variability results
When the threshold was set at .01, the combined network of the alpha could predict judgment of attractiveness in men (r = .37, p < .01, p perm = .03). No other significant differences were observed (see Tables 2 and 3).

DISCUSSION
This study examined the influence of individual arousal on judgment of facial attractiveness and gender differences. We found that men with high arousal judged faces as more attractive than did men with low arousal and women. The arousal index of alpha predicted a judgment of facial attractiveness in men. Overall, these results provide evidence that men with high arousal are associated with higher facial attractiveness judgments. Furthermore, alpha oscillations formed the association between arousal and attractive facial preference. Our results TA B L E 2 Results of control and validation analyses for the prediction of the judgement of facial attractiveness in men. Note: The negative r value means failure to predict the behavior performance.

Combined network
contribute to the understanding of how a positive first impression is formed in facial attractiveness.
The enhancement in facial attractiveness is due to familiarity (Carr et al., 2017), emotion (Han et al., 2022;Han, Liu, et al., 2020), good traits , and generalization to similar faces (Han, Hu, et al., 2020). We found the influence of arousal on the judgment of facial attractiveness, namely, men with high arousal rate female faces as more attractive, thereby supporting the misattribution of arousal hypothesis, which suggests individual misattributed arousal due to attractive features (White et al., 1981). The result adds to evidence that individual arousal contributes to improving facial attractiveness.
Moreover, the novel finding of the current study is that spontaneous arousal, rather than evoked arousal, is involved in the judgment of facial attractiveness.
Individual differences in attraction preferences also exist among the genders. Men value physical attractiveness of potential partners more than women (Schwarz & Hassebrauck, 2012); thus, they do not require attentional resources when processing female faces but not when processing same-sex faces. Whether the target face is male or female, women utilize cognitive resources for processing (Klümper et al., 2020). These results are consistent with the evolutionary perspective on facial attractiveness-that is, men easily capture a potential mate's facial attractiveness, and cues of fertility and healthiness, which could increase reproductive success (Buss & Schmitt, 2019). Furthermore, we found that men with high arousal gave higher attractiveness ratings to female faces. Because we used the resting-state EEG before the participants viewed attractive faces, the enhancement of facial attractiveness may be associated with the general arousal state rather than sexual arousal specifically (Jozifkova & Konvicka, 2009), thus supporting the idea that general arousal improves judgment of attractiveness in men.
Women gave lower scores to female attractiveness. The same pattern has been observed in previous studies, which is interpreted as intrasexual competition leading women to judge same-sex targets as less attractive (Luxen & Vijver, 2006). In addition, women's arousal is not related to their judgment of facial attractiveness in men (Hagerman et al., 2017). Our results support the sexual strategy theory in which men and women have different mating preferences. As such, women prefer indirect cues (e.g., social status), whereas men prefer direct cues (Buss, 1998;Buss & Schmitt, 2019). Moreover, the alpha power of lowarousal men was significantly lower than that of women. One reason for this is that men have higher levels of testosterone than women, which diminish the alpha power (Vogel et al., 1971). Another reason may be that woman are more vulnerable to stress-induced hyperarousal than men (Bangasser et al., 2019). Furthermore, women with arousal levels higher than those of men did not have significant differences in the scores of facial attractiveness. Only men with high arousal levels gave higher attractiveness scores to target faces, indicating that arousal enhances facial attractiveness specifically in men.
Functional connectivity of alpha oscillations predicted individual preferences for facial attractiveness. Traditionally, researchers have focused on explaining observed behavior by manipulating independent variables, which is helpful in elucidating the nature of cognitive processing and behavioral responses Thiruchselvam et al., 2016). In this study, we further used CPM to predict individual attractive preference, which revealed that alpha is a biomarker to predict judgment of facial attractiveness in men, supporting the idea that the misattribution of arousal contributes to an individual's facial attractiveness preference (Marin et al., 2017). The results also expanded the theory of misattribution of arousal in that individual spontaneous arousal, not evoked arousal, was associated with facial attractiveness judgment. Additionally, a combined network of delta bands predicted facial attractiveness ratings in women, which may be due to the motivation based on the intrasexual competition, because delta oscillations are related to the motivation process (Knyazev, 2012). These results suggested that the prediction mechanism is sex specific for observers.
Nevertheless, this study had some limitations. First, we only used faces with middle-level attractiveness. However, high-and low-attractive faces have a different processing mechanism . Researchers can further investigate whether the results in this study are accurate when using high-and low-attractive faces. Second, male faces were not presented to women in this study. Thus, future studies can explore whether male faces are rated as more attractive when presented to women with high arousal.
In conclusion, an individual's general arousal level plays an important role in the judgment of facial attractiveness, particularly for men.
Men with high arousal were more attracted to female faces than men with low arousal or women. Our results suggest that individual biological characteristics (i.e., arousal) contribute to the diversity of social functions.

CONFLICT OF INTEREST STATEMENT
The authors declare no conflicts of interest.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available on request from the corresponding author upon reasonable request.