![[Adobe Stock]](https://sp-ao.shortpixel.ai/client/to_webp,q_glossy,ret_img,w_750,h_375/https://www.psypost.org/wp-content/uploads/2024/08/online-dating-750x375.jpg)
When users swipe through profiles on dating applications, their brains make split-second decisions based on limited visual information. A new study suggests that these rapid judgments rely on two distinct cognitive processes: one that assesses facial beauty and another that interprets the “vibe” or social context of a photo. Published in Computers in Human Behavior, the research reveals that while a beautiful face is a powerful asset, combining it with intense displays of wealth or social status can actually backfire.
The researchers refer to this phenomenon as the “Two-Pathway Heuristic Model” of impression formation. This model proposes that the human brain utilizes efficiency-oriented mental shortcuts to handle the high volume of profiles found on digital platforms. One pathway triggers an immediate emotional response to physical aesthetics. The second pathway involves a rapid inference of a person’s lifestyle and values based on the background and context of their photographs.
Junkyu Jang, a researcher at the Korea Advanced Institute of Science and Technology (KAIST), served as the first author of the paper. Jang and his colleagues sought to understand how these two pathways operate in a real-world setting. Previous research often relied on small laboratory experiments or surveys that did not reflect the high-speed environment of actual dating apps. The team aimed to fill this gap by analyzing a massive dataset of user interactions.
The study utilized data from a major heterosexual dating platform in South Korea. The dataset included information from 10,619 users who had received evaluations from at least 15 other people. This provided a robust sample for statistical analysis. The researchers did not rely on human volunteers to rate the tens of thousands of photos in the dataset. Instead, they employed a cutting-edge artificial intelligence tool known as a Large Multimodal Model (LMM).
The specific model used, InternVL-G, is designed to interpret images in a way that mimics human perception. The team provided the AI with specific prompts to evaluate different aspects of each profile picture. This allowed for the consistent quantification of subjective qualities that are typically difficult to measure in large-scale studies. The AI assessed facial attractiveness based on symmetry and features.
Beyond simple beauty, the researchers tasked the AI with quantifying “social attractiveness,” which they colloquially termed “vibe.” To make this abstract concept measurable, the team drew upon the theories of French sociologist Pierre Bourdieu regarding forms of capital. They programmed the model to look for three specific types of visual cues: social capital, economic capital, and cultural capital.
Social capital was defined by visual signals suggesting a rich social life. The AI looked for images containing groups of friends or scenes depicting enjoyable social events. These cues imply that a user is likable and possesses a strong network of peers. Economic capital involved signals of financial stability. The model scanned for luxury items, high-end clothing, or upscale environments.
Cultural capital was the third component of the “vibe” assessment. This involved looking for signs of sophisticated tastes or intellectual curiosity. The AI identified cues such as engagement in artistic activities, travel to cultural landmarks, or the presence of books and musical instruments. By breaking down the vague concept of “vibe” into these constituent parts, the researchers could analyze exactly which elements contributed to a successful match.
The results of the study highlighted the overwhelming power of the first pathway: facial attractiveness. For both men and women, having a beautiful face was a primary predictor of matching success. The effect was particularly strong for male profiles. The researchers found that moving from the lowest to the highest level of facial beauty could double a man’s matching rate.
This finding supports the “gatekeeper” hypothesis of online dating. Women tend to be more selective than men on these platforms. Consequently, a man’s physical appearance often acts as a strict initial filter. If a male user does not meet a certain threshold of attractiveness, he is often screened out before other qualities can be considered.
Once the initial hurdle of physical appearance is cleared, the second pathway comes into play. The study showed that a positive “vibe” substantially increased matching chances for both genders. Users who displayed high levels of social and cultural capital received more matches. This suggests that people look for cues of warmth, sociability, and shared interests to infer a potential partner’s personality.
The impact of economic capital showed a distinct gender divide. Visual displays of wealth and luxury positively influenced matching rates for men. However, the same signals had a much weaker effect for female profiles. This aligns with evolutionary psychology theories suggesting that women may prioritize resource acquisition potential in partners more than men do.
The most nuanced finding of the study emerged when the researchers analyzed how beauty and “vibe” interact. Common intuition might suggest that maximizing every positive trait would lead to the best results. The data proved otherwise. The analysis revealed a pattern of diminishing returns when high facial attractiveness was combined with strong signals of social or cultural capital.
For users who already possessed a high degree of facial beauty, adding intense signals of social status or cultural sophistication resulted in fewer matches than the model predicted for an additive effect. This negative interaction was statistically clear. When a user is extremely good-looking, an overly curated background may seem redundant or even suspicious.
The authors propose several psychological mechanisms to explain this counterintuitive result. One possibility is that “too perfect” profiles trigger negative stereotypes. A highly attractive man who also displays an excessive number of party photos might be perceived as a “player” or lacking in commitment. Similarly, a highly attractive woman displaying intense highbrow cultural capital might be viewed as “snobbish” or unapproachable.
Another explanation involves the concept of authenticity. In the digital realm, users are wary of deception and catfishing. A profile that scores top marks in every visual category can appear manufactured or inauthentic. The cognitive load required to process such an abundance of positive signals may also lead evaluators to be more skeptical.
These findings challenge the “more is better” strategy often employed by dating app users. A balanced profile appears to be more effective than one that attempts to signal superiority in every category. If a user is naturally physically attractive, they might benefit from a more modest presentation of their lifestyle. Conversely, users with average facial attractiveness can significantly boost their appeal by showcasing a fun social life or interesting hobbies.
The use of the Large Multimodal Model to conduct this analysis represents a step forward for computational social science. Validating the AI’s performance was a key part of the study. The researchers compared the AI’s ratings with human ratings on a benchmark dataset. The AI achieved a high level of accuracy, demonstrating that modern machine learning can reliably quantify abstract human judgments like “beauty” and “vibe.”
There are limitations to the study that provide context for the findings. The data came exclusively from South Korea, a country with specific cultural norms regarding beauty and dating. The results might vary in Western contexts or other parts of Asia. Additionally, the study focused solely on heterosexual matching patterns. The dynamics of impression formation in LGBTQ+ dating communities could operate differently.
The researchers also noted that they only analyzed the visual components of the profiles. Textual biographies and prompts also play a role in decision-making. However, the study prioritized images because visual processing is the dominant mechanism in the rapid-fire environment of swipe-based apps. Future research could investigate how text and images interact to confirm or contradict one another.
This research offers a detailed look at the split-second decisions that define modern romance. It confirms that while humans are visually driven creatures, we are also constantly scanning for context. We look for clues about who a person is and how they live. But as the study shows, there is a fine line between an impressive profile and one that tries too hard.
The study, “Beauty vs. Vibe: Deconstructing visual appeal in online dating with large multimodal models,” was authored by Junkyu Jang, Soonjae Kwon, and Sung-Hyuk Park.