The internet phenomenon known as “the dress” swept across screens globally in late February 2015, igniting a fervent debate that tested even the most fundamental aspects of human perception. While psychologist Robert Zajonc’s “mere exposure effect” suggests repeated exposure breeds liking, the ubiquity of “the dress” quickly pushed this theory to its limits. Many swiftly grew weary of the endless discussions and images, yet for vision scientists, including myself, the phenomenon remained—and remains—a source of profound fascination.
Why? Because “the dress” did something remarkable: it challenged the very foundations of our understanding of color vision. Prior to 2015, the scientific community largely believed that color vision was a well-charted territory. “The dress” ripped up that map. Suddenly, we were confronted with a stimulus that revealed a stark divide in perception, with no immediate explanation. It was akin to stumbling upon an uncharted continent, an entirely new frontier in visual science. Adding to its intrigue, “the dress” emerged organically, captured in a photograph in England, unlike the carefully controlled stimuli typically used in vision labs.
Slate
Beyond the realm of scientific inquiry, “the dress” illuminated a common, often unspoken assumption: that everyone perceives the world in fundamentally the same way. Disagreements about something as seemingly objective as color can be unsettling, hinting at ignorance, malice, or even irrationality. Our visual experiences are deeply ingrained as truth, which helps explain the intensity of the disputes that erupted when “the dress” first surfaced, leaving science scrambling for answers.
However, in the years since, our understanding of this visual puzzle has sharpened considerably. A key factor in the varied interpretations of “the dress” lies in how our brains perceive lighting conditions. The photograph, taken with a cellphone, presented considerable ambiguity regarding illumination. Was the dress photographed indoors under artificial light or outdoors in natural light? Was it front-lit or backlit, casting it into shadow?
Our brains, far from being passive receivers of information, are active interpreters. In situations of uncertainty, like the ambiguous lighting in the dress photo, the brain confidently fills in the gaps by making assumptions. These assumptions aren’t arbitrary; they are rooted in our past visual experiences. For example, in uncertain visual conditions, we tend to perceive objects as moving slower than they are, likely because slow-moving objects are statistically more common in our environment. Color perception, it turns out, is no different.
Because the lighting conditions in the dress image are unclear, individuals make different assumptions about them. These varying assumptions are the root cause of the divergent color perceptions. This conclusion is supported by my research, which involved over 13,000 participants, including many Slate readers, who completed surveys detailing their perception of the dress and other aspects of their visual experience.
The dress is, in reality, blue and black. Yet, initially, the majority perceived it as white and gold. Our research revealed a significant correlation: those who assumed the dress was in shadow were far more likely to see it as white and gold. Why? Because shadows are enriched in blue light. Our visual system, in an attempt to perceive the “true” color of the dress, might subconsciously subtract the bluish cast of the shadow, resulting in a yellowish or gold perception. Similarly, those who believed the dress was illuminated by natural daylight were also more inclined to see white and gold. Daylight, like shadows, contains a higher proportion of short-wavelength (blue) light compared to artificial incandescent light, which is richer in longer wavelengths (yellow to red).
This raises another intriguing question: why do individuals make different assumptions about lighting in the first place? The answer likely lies in our individual histories of light exposure. While directly measuring someone’s lifetime exposure to different wavelengths of light is impractical, we can consider proxies.
One such proxy is chronotype—whether someone is a “lark” (early riser) or an “owl” (late riser). Larks, who are awake during daylight hours, are naturally exposed to more daylight. Owls, conversely, spend more time awake under artificial, often incandescent, light. Therefore, we hypothesized that larks would be more likely to assume daylight illumination and perceive the dress as white and gold, while owls, accustomed to artificial light, might be less likely to make this assumption.
Indeed, our findings supported this hypothesis. The effect was subtle but statistically significant: the more a person identified as a lark, the greater the likelihood they perceived the dress as white and gold. Furthermore, owls were more inclined to assume artificial lighting conditions in the photo.
It’s crucial to remember that this is a subtle, average effect. Individual light exposure histories are complex and influenced by numerous factors beyond chronotype. Even a self-proclaimed owl might work a day job under bright daylight. However, on average, owls are exposed to more incandescent light than larks.
Interestingly, I, a self-confessed owl, initially saw the dress as white and gold. However, I also instinctively assumed it was photographed in shadow, which likely overrode my owl-typical lighting assumption. Furthermore, our visual perception isn’t static. My own perception of the dress shifted to blue and black after a few days of intense online exposure and, crucially, learning its actual colors. This suggests that knowledge and subsequent visual input can recalibrate our assumptions. Some individuals may have such finely tuned perceptual systems that their perception can flip rapidly with new information, but these cases appear to be rare.
The “dress” phenomenon, while seemingly trivial, underscores a crucial point about scientific rigor. Despite the viral explosion of “the dress” in 2015, truly rigorous scientific studies, including the paper I co-authored and another published concurrently, took considerable time to emerge. Good science demands thoroughness and replication. I prioritized validating my findings through internal replication before seeking publication, delaying the process by over a year, even though 97% of the initial data was collected within a month of the dress going viral.
This commitment to robust methodology is increasingly vital in today’s scientific landscape. The current academic environment often incentivizes rapid publication and prioritizes quantity over quality. This pressure contributes to the troubling “replication crisis” across various scientific fields, where a significant portion of published findings fail to stand the test of time. Examples abound, from social psychology to cancer biology and genomics.
The lesson from “the dress” extends beyond visual perception; it serves as a microcosm for the broader need for a shift in scientific practice. For civilization to progress, we require reliable, high-quality science. Paradoxically, research on a seemingly frivolous internet meme like “the dress” might just illuminate a path forward.
In the spirit of #citizenscience, I encourage you to participate in a brief (five-minute) survey. While I cannot disclose the specifics to avoid biasing results, the survey aims to address remaining questions about the “dress” effect and related visual stimuli. And I appreciate your continued patience as my colleagues and I delve deeper into the fascinating world of color perception, sparked by a dress that broke the internet.
