On 26 February 2015, a photograph of a striped dress did something no photograph had done before: it split the internet into two camps that could not comprehend each other. Roughly half the world saw white and gold. The other half saw blue and black. Both sides were certain. Both sides thought the other was joking.
They weren’t. And a decade later, vision scientists are still writing papers about why.
A wedding, a phone, and a bad exposure
The photo began as an ordinary family moment. Cecilia Bleasdale photographed a dress she planned to wear to her daughter’s wedding in Scotland and sent the picture to the family. They immediately disagreed about its colours. A friend of the couple, musician Caitlin McNeill, posted the picture to Tumblr asking for help — and within hours it had escaped into the wider internet. BuzzFeed’s poll on the image drew tens of millions of views in a day. Celebrities weighed in. Taylor Swift saw blue and black and declared herself “confused and scared.”
For the record: the dress, sold by British retailer Roman Originals, is blue and black. The company confirmed it, then watched the garment sell out.
Why your brain picked a side
The explanation sits in a quirk of vision called colour constancy. Your brain doesn’t just read the light coming off an object — it quietly subtracts the light it believes is falling on the object. That’s why a white shirt still looks white at sunset, even though the light reflecting off it is orange.
The dress photo is badly overexposed and gives almost no clues about its lighting. So your brain guesses. If it assumes the dress is in cool, bluish daylight shadow, it subtracts blue — and you see white and gold. If it assumes warm artificial light, it subtracts yellow — and you see blue and black. Same pixels, two different corrections.
Research that followed made it stranger. Studies published in the Journal of Vision in 2015 found the split was real and measurable across large populations rather than a joke or a screen-settings issue. Later work by NYU neuroscientist Pascal Wallisch suggested a person’s assumptions may track their lived experience of light: people who spend more waking hours in daylight leaned white-and-gold, night owls leaned blue-and-black.
Why it still matters
Before the dress, most vision scientists assumed that two healthy people looking at the same image would broadly agree on its colours. The dress proved that assumption wrong at planetary scale, and it handed researchers a rare natural experiment: a single stimulus that reliably divides observers. It has since appeared in hundreds of academic papers on perception.
It also previewed something about the internet itself: that millions of people can look at the same thing, in good faith, and genuinely see something different. That lesson aged rather well.
