眼动研究：女性对美貌面孔的注视不限于偏好性别

研究做了什么

这项研究由乌得勒支大学的Johanna M. Bellue与Tom S. Roth等人完成，发表在《进化与人类行为》上。研究者把眼动仪搬到荷兰乌得勒支的Betweter科学节现场，招募了105名成人，剔除3名自认泛性恋的非二元参与者后，最终样本为65名女性和37名男性，平均年龄38岁，其中87人为异性恋，63人处于排他性关系中，平均交往约12.6年。

参与者坐在距屏幕约60厘米的隔间里，红外眼动仪每秒记录300次眼动。每人完成36个试次，每次看到两张同性别的白人面孔，一张按既有共识评分属于传统意义上的好看，一张相对普通，表情中性，停留三秒。研究者只告诉参与者这是面孔知觉研究，没有提吸引力，以免他们刻意调整。事后问卷收集年龄、性别、性取向、关系状态、交往时长和当晚饮酒量。

研究者把面孔性别与参与者自报性取向匹配，将试次分为“偏好性别”和“非偏好性别”，并在统计中控制饮酒量、照片来源库和好看面孔出现在左侧还是右侧。

自动注视与有意注视的分野

心理学把视觉注意分成两个阶段：一个是几乎不经思考的快速自动吸引，另一个是较慢的、主动选择把目光留在哪里的有意注视。这项研究分别测量了两者。

在最初的半秒内，吸引力几乎没有作用。女性在偏好性别试次中先看好看面孔的比例是50.0%，非偏好性别是50.7%，基本等于抛硬币；男性分别是53.9%和49.9%，差异在统计上不稳健。真正决定第一眼的是位置：无论面孔好不好看，约78%的试次里参与者先看左边那张。作者提到，这是大脑两半球处理视觉空间的已知倾向，也可能掩盖了任何自动的吸引力效应；热闹的节日环境同样可能干扰结果。

到了三秒的有意注视阶段，模式变得清晰。女性把55.4%的注视时间给了偏好性别中的好看面孔，54.7%给了非偏好性别中的好看面孔，两者都显著高于50%的均等线。男性在偏好性别试次中为54.9%，在非偏好性别试次中只有52.0%，后者与随机水平没有可靠差异。

男性更“对口”，但未必更强

这一结果与一些进化心理学假设不完全一致。过去有观点认为，由于外貌在男性的择偶偏好中权重更大，男性对符合自己性取向的好看面孔会表现出更强的注意力偏向。但本研究中，男性对偏好性别好看面孔的偏向并不比女性更强，直接比较两性模式的统计检验也只是略低于研究设定的显著门槛，部分原因是女性在偏好性别试次中同样看得稍久。

Roth在给PsyPost的说明中表示，男女在吸引力偏向上的差异与既有研究一致：男性更特定于自己的性取向，女性则对两性的好看面孔都有一般性偏向。他还提到，研究没有发现男性在偏好性别上的偏向更强。

作者把这一模式与此前关于性反应特异性的研究并置：男性的视觉和生理性反应往往更限定于偏好性别，女性则相对不那么限定。不过Roth特别提醒，研究测量的是目光落在哪里，而不是想和谁约会或对谁有性兴趣，女性看两性好看面孔更久，不能解读为对两性都有性兴趣。

关系状态没有带来差别

研究原本预期，处于稳定关系中的人可能会减少对符合自己性取向的好看面孔的注视，以“保护”现有关系。但结果显示，排他性关系中的参与者与单身或开放关系中的参与者，在好看面孔上的停留时间没有实质差别。

作者对此给出一种解释：样本中的伴侣关系大多已经非常稳定，平均超过十二年。在约会初期，人们或许会主动回避看有吸引力的陌生人，以维护尚不牢固的新关系、建立信任；当关系持续十多年后，伴侣可能已经足够安心，不再需要无意识地管住目光。另一种可能是，关系状态只在人们先被提醒想到浪漫或伴侣时才会起作用，而本实验没有设置这样的启动步骤。

年龄的探索性分析只显示微弱而不确定的倾向：男性的吸引力偏向随年龄略有上升，女性则略有下降。

个体差异与解释边界

研究者还做了统计模拟，估计新招募一个人会怎样表现，而不只看组平均。结果是一名新的女性参与者约有80%的概率对两性的好看面孔都看得更久；一名新的男性参与者对偏好性别好看面孔的这一概率约为76%，对非偏好性别只有约62%。这说明组层面的趋势虽然可靠，个体之间的观看习惯差异相当大。Roth也提到，多数人确实存在某种吸引力偏向，但也有人没有，这至少说明它不是对所有人都成立的固定特质。

解读结果时还有几项限制。实验在热闹的公共科学节进行，流程必须简短，好看与否依赖照片库既有的平均评分，而个人口味各异，参与者未必觉得被标为好看的那张真的好看，这可能稀释测得的注视模式。现场招募也导致样本中女性、有伴侣者和高学历城市居民偏多，异性恋占多数。面孔材料只有白人，且多为二三十岁的年轻人，可能影响年长参与者的反应。作者建议未来用更大、更均衡的样本，加入不同年龄的面孔，并让参与者自己评分，同时询问关系满意度和嫉妒等细节。

When viewing pairs of faces, men tend to linger selectively on attractive faces that match their sexual orientation, whereas women spend more time looking at attractive faces regardless of the person’s sex. A new eye-tracking study published in Evolution and Human Behavior suggests that this preference occurs during deliberate, sustained viewing rather than in the first split-second of automatic attention. The findings also indicate that being in a committed relationship does not necessarily stop people from looking longer at attractive faces.

In psychology, attentional bias refers to the tendency for our visual system to prioritize certain types of information over others in our environment. Facial attractiveness, which is the subjective judgment of how aesthetically pleasing a face is, acts as a strong magnet for human gaze. From an evolutionary perspective, physical beauty is often thought to serve as a signal of health and fertility, which may explain why human cognition is tuned to notice desirable potential mates.

How this visual pull operates depends heavily on both who is looking and who is being looked at. In a 2003 study of young adults, researchers outlined several competing possibilities for how gender and mating motives guide our eyes. One possibility is that people focus mainly on attractive members of their preferred sex to find mates, while another suggests that both men and women look longer at attractive women: men to evaluate potential partners, and women to assess potential romantic rivals. A third proposed that men in particular would show a strong pull toward attractive faces of their preferred sex, because physical attractiveness tends to weigh more heavily in men’s mate preferences.

Subsequent research has complicated this picture by examining how relationship status and different stages of mental processing shape our gaze. For instance, a 2018 eye-tracking study indicated that single people exhibit a stronger visual pull toward attractive faces than people in committed relationships, who may dial down their attention to protect their existing bond. Similarly, a study covered by PsyPost in 2023 found that when viewing faces matching their sexual preference, people’s brains initially react strongly to both very attractive and very unattractive faces before settling into sustained neural activity specifically for the attractive ones.

Psychologists separate visual attention into two distinct stages: immediate attention and volitional attention. Immediate attention is a rapid, automatic reflex that pulls our eyes toward a stimulus before we consciously think about it. Volitional attention is the slower, deliberate process of choosing where to keep looking to gather more information.

Separating these two stages matters for attractiveness research. A 2023 study of speed-daters, led by Tom S. Roth, the senior author of the new study, found that people’s personal ratings of facial attractiveness predicted where both men and women directed their deliberate gaze, while results for quicker, more automatic glances were mixed across the sexes. However, almost all prior eye-tracking studies on facial attractiveness have relied on college undergraduates, whose casual dating habits may not reflect how older adults in long-term relationships view attractive strangers.

To address this, lead author Johanna M. Bellue and her colleagues, including Roth, an assistant professor at Utrecht University, set out to test how immediate and deliberate attention to attractive faces varies by sex, sexual preference, age, and relationship status in the general public.

“Previous studies on attractiveness bias have mostly been performed with heterosexual undergraduate students as a sample,” Roth told PsyPost. “However, to understand this phenomenon it is more interesting to see whether similar effects are present in a broader sample, containing people of different ages, relationship statuses, and sexual orientations.”

The researchers recruited 105 adults attending the Betweter Festival, an annual public science festival held at a music venue in Utrecht, the Netherlands. After excluding three non-binary participants, who all reported being pansexual and so had no “non-preferred” faces to compare, the final sample included 65 women and 37 men. The participants had an average age of 38, and most (87 of the 102) were heterosexual, while the 63 participants in exclusive relationships had been with their partners for an average of about 12.6 years.

Each participant sat in a booth about 60 centimeters away from a computer monitor equipped with an infrared eye-tracker that recorded their eye movements 300 times per second. To prevent participants from consciously altering their behavior, the researchers told them only that the study examined general face perception, without mentioning attractiveness. Across 36 main trials, participants viewed pairs of same-sex portrait photos, either two men or two women, with neutral expressions.

Each pair featured one conventionally attractive face and one less attractive face, selected from established academic photo databases based on pre-existing consensus ratings. All of the faces were of white adults, mostly in their twenties and thirties. The face pairs remained on the screen for three seconds while participants looked wherever they wanted. Afterward, participants completed a paper questionnaire detailing their age, sex, sexual orientation, relationship status, relationship length, and how many alcoholic drinks they had consumed at the festival that evening.

The authors matched each participant’s stated sexual orientation with the sex of the faces shown on screen to categorize every trial as either “preferred” or “non-preferred.” For example, female faces were coded as preferred for heterosexual men and homosexual women, while all faces were coded as preferred for bisexual participants. In their statistical models, the team controlled for alcohol consumption, which photo database the images came from, and whether the attractive face appeared on the left or right side of the screen.

When looking at immediate attention, measured by which of the two faces a participant’s eyes landed on first within half a second of the images appearing, facial attractiveness had no noticeable effect. Women looked at the attractive face first about 50.0% of the time for their preferred sex and 50.7% of the time for their non-preferred sex, which is essentially a coin toss. Men looked at the attractive face first in 53.9% of preferred-sex trials and 49.9% of non-preferred-sex trials, a difference that was not statistically robust.

Instead of beauty, the first automatic glance was heavily dictated by screen position. Participants looked at the left image first in roughly 78% of the trials, regardless of how attractive the person was. This strong leftward habit is a well-known quirk of how the human brain’s hemispheres process visual space, and it may have masked any immediate pull toward attractive faces, as could the noisy festival setting. The null result also contrasts with some earlier reaction-time studies that found a rapid pull toward attractive faces, which the authors suggest may reflect differences in the type of attention each method captures.

A different pattern emerged when the researchers examined volitional attention, which was measured by the proportion of total viewing time spent looking at the attractive face over the full three seconds. Women demonstrated a general appreciation for facial beauty, spending 55.4% of their viewing time on the attractive face when looking at their preferred sex and 54.7% when looking at their non-preferred sex. Both figures were statistically above the 50% mark that would indicate equal viewing.

Men, by contrast, were more selective with their deliberate gaze. They spent 54.9% of their viewing time on the attractive face when the pair matched their sexual preference, but only 52.0% when viewing their non-preferred sex, a figure that did not reliably differ from chance. Men’s bias toward attractive faces of their preferred sex was no stronger than women’s, and a statistical test comparing the two sexes’ patterns directly fell just short of the study’s threshold, partly because women also looked slightly longer at attractive faces of their preferred sex.

The authors note that the pattern parallels earlier research suggesting that men’s visual and physical sexual responses tend to be specific to their preferred sex, whereas women’s tend to be less specific.

“We find that men and women differ in their attractiveness bias,” Roth explained. “Men are specifically interested in attractive faces that align with their sexual orientation, while women have a general bias toward attractive faces of both sexes. This is in line with previous research. We don’t find that men have a stronger bias toward attractiveness in their preferred sex, as has been suggested by some evolutionary hypotheses.”

The findings are in line with research covered by PsyPost earlier this year, which found that more attractive faces drew more direct eye movements but did not speed reactions on a task measuring attention shifts that happen without moving the eyes. Although the two studies used different tasks, both suggest that attractiveness influences where people choose to look more than it captures attention automatically.

The researchers also ran statistical simulations to predict how a newly recruited individual would behave, rather than just looking at group averages. They found that a new female participant would have about an 80% probability of looking longer at attractive faces of either sex, while a new male participant would have a 76% probability for attractive faces of his preferred sex, but only about 62% for his non-preferred sex. This wider uncertainty at the individual level suggests that while the group trend is reliable, personal viewing habits vary considerably from one person to the next.

“While humans do seem to have an attractiveness bias of some kind, we also found that there is quite some variation between participants,” Roth noted. “While the majority of individuals indeed have a bias, others do not. This shows that it is at least not a fixed trait that generalizes to everyone.”

Contrary to expectations and some prior university-based studies, relationship status did not meaningfully alter how long people looked at attractive faces of their preferred sex. Participants in exclusive relationships spent just as much time looking at attractive faces as participants who were single or in open relationships. In addition, an exploratory check of participant age showed only weak, inconclusive tendencies, with attractiveness bias slightly increasing with age in men and slightly decreasing in women.

“We were also interested in effects of relationship status, but we did not find any,” Roth said. “We had expected that people might look less at attractive faces that aligned with their preference when they were in a relationship, as they want to ‘protect’ their relationship. However, our sample mostly contained data from participants who were already in very established relationships with a long duration. Maybe our expected effect would apply more to new relationships specifically.”

The authors noted that this difference in relationship maturity may help explain why their festival sample behaved differently from student populations. In the early stages of dating, which are common among college students, people may actively avoid looking at attractive strangers to protect a fragile new bond and build trust. Once a relationship has lasted over a decade, partners may feel secure enough that they no longer need to unconsciously police where their eyes wander during a casual viewing task. It is also possible that relationship status only dampens attention when people are first primed to think about romance or their partner, a step used in several earlier studies that this experiment omitted.

There are a few things to keep in mind when interpreting these results. “One important caveat is that we measured where people looked, not whom they wanted to date or felt sexually attracted to,” Roth added. “Women looking longer at attractive faces of both sexes should therefore not be interpreted as evidence of sexual interest in both sexes.”

Because the experiment took place at a bustling public festival, the researchers had to keep the procedure brief and relied on pre-existing average ratings to define which faces were attractive. People have diverse personal tastes, meaning a participant might not have personally found the designated “attractive” face appealing, which could dilute the measured gaze patterns.

The festival setting also meant that the researchers recruited whoever volunteered rather than sampling to balance groups, resulting in noticeably more women and coupled people than men and singles. The photo databases featured only white faces, mostly of young adults in their twenties and thirties, which may have skewed how older participants responded to the images. Future studies could test larger, evenly balanced groups using faces of many different ages, while also asking participants to rate the faces themselves and answer detailed questions about their relationship satisfaction and jealousy.

“Although our science-festival sample covered a broader age range than a typical student sample, it still overrepresents highly educated people from urban settings, and most participants were heterosexual,” Roth said. “Recruiting people from a wider range of backgrounds and sexual orientations would help us establish how broadly these patterns apply. We would also like to account more directly for individual preferences: the faces that are attractive on average are not necessarily the ones that every individual finds attractive.”

The study, “Sexual preference modulates attentional bias towards attractive faces in men, but not women,” was authored by Johanna M. Bellue, David Terburg, Mirre E. Balke, Tessa W. Coster, Jorg J.M. Massen, and Tom S. Roth.