How ‘The Dress’ Sparked a Neuroscience Breakthrough

The color debate that broke the internet raised new questions about the relationship between perception and consciousness.
A collage of images including a color wheel.
Illustration: WIRED; Getty Images

Back in 2015, before Brexit, before Trump, before Macedonian internet trolls, before QAnon and Covid conspiracy theories, before fake news and alternative facts, the disagreement over the Dress was described by one NPR affiliate as “the debate that broke the internet.” The Washington Post called it “the drama that divided the planet.”

The Dress was a meme, a viral photo that appeared all across social media for a few months. For some, when they looked at the photo, they saw a dress that appeared black and blue. For others, the dress appeared white and gold. Whatever people saw, it was impossible to see it differently. If not for the social aspect of social media, you might have never known that some people did see it differently. But since social media is social, learning the fact that millions saw a different dress than you did created a widespread, visceral response. The people who saw a different dress seemed clearly, obviously mistaken and quite possibly deranged. When the Dress started circling the internet, a tangible sense of dread about the nature of what is and is not real went as viral as the image itself.

At times, so many people were sharing this perceptual conundrum, and arguing about it, that Twitter couldn’t load on their devices. The hashtag #TheDress appeared in 11,000 tweets per minute, and the definitive article about the meme, published on WIRED’s website, received 32.8 million unique views within the first few days.

For many, the Dress was an introduction to something neuroscience has understood for a long while: the fact that reality itself, as we experience it, isn’t a perfect one-to-one account of the world around us. The world, as you experience it, is a simulation running inside your skull, a waking dream. We each live in a virtual landscape of perpetual imagination and self-generated illusion, a hallucination informed over our lifetimes by our senses and thoughts about them, updated continuously as we bring in new experiences via those senses and think new thoughts about what we have sensed. If you didn’t know this, for many the Dress demanded you either take to your keyboard to shout into the abyss or take a seat and ponder your place in the grand scheme of things.

Before the Dress, it was well understood in neuroscience that all reality is virtual; therefore consensus realities are mostly the result of geography. People who grow up in similar environments around similar people tend to have similar brains and thus similar virtual realities. If they do disagree, it’s usually over ideas, not the raw truth of their perceptions.

After the Dress, well—enter Pascal Wallisch, a neuroscientist who studies consciousness and perception at NYU. When Pascal first saw the Dress, it seemed to him that it was obviously white and gold, but when he showed it to his wife, she saw something different. She said that it was obviously black and blue. “All that night I was up, thinking what could possibly explain this.”

Thanks to years of research into photoreceptors in the retina and the neurons to which they connect, he thought he understood the roughly thirty steps in the chain of visual processing, but “all of that was blown wide open in February 2015 when the Dress surfaced on social media.” He felt like a biologist learning that doctors had just discovered a new organ in the body.

The spectrum of light we can see—the primary colors we call red, green, and blue—are specific wavelengths of electromagnetic energy, Pascal explains. These wavelengths of energy emanate from some source, like the sun, a lamp, a candle. When that light collides with, say, a lemon, the lemon absorbs some of those wavelengths and the rest bounce off. Whatever is left behind goes through a hole in our heads called the pupil and strikes the retinas at the back of the eyes, where it all gets translated into the electrochemical buzz of neurons that the brain then uses to construct the subjective experience of seeing colors. Because most natural light is red, green, and blue combined, a lemon absorbs the blue wavelengths, leaving behind the red and green to hit our retinas, which the brain then combines into the subjective experience of seeing a yellow lemon. The color, though, exists only in the mind. In consciousness, yellow is a figment of the imagination. The reason we tend to agree that lemons are yellow (and lemons) is because all our brains pretty much create the same figment of the imagination when light hits lemons and then bounces into our heads.

If we do disagree over what we see, it’s usually because the image is ambiguous in some way, and the brain of one person is disambiguating the image in a way another person is not. Pascal says that in neuroscience, the go-to examples of disambiguation are called intrapersonal bistable visual illusions—bistable because each brain settles on one interpretation at a time, and intrapersonal because every brain settles on the same two interpretations. You’ve likely seen a few of these: the duckrabbit for example, which sometimes looks like a duck and sometimes looks like a rabbit. Or the Rubin vase, which sometimes looks like a vase and sometimes looks like two people facing in silhouette.

Like all two-dimensional images, whether blobs of paint or pixels on a screen, if the lines and shapes seem similar enough to things we’ve seen in the past, we disambiguate them into the Mona Lisa, or a sailboat, or in the case of a bistable image, either a duck or a rabbit. But the Dress was something new, an interjacent bistable visual illusion—bistable because each brain settles on one interpretation at a time, but interjacent because each brain settles on only one of two possible interpretations. That’s what made the Dress so confusing to Pascal.

The same light was going into everyone’s eyes, and every brain was interpreting the lines and shapes as a dress, yet somehow all those brains weren’t converting that dress into the same colors. Something was happening between perception and consciousness, and he wanted to know what that was. So he acquired some funding and shifted the focus of his lab at NYU to tackling the mystery of the Dress while it was still going viral.

Pascal’s hunch was that different people saw different dresses because when we aren’t sure what we are seeing, when we are in unfamiliar and ambiguous territory, we disambiguate using our priors, short for prior probabilities—the layers of pattern recognition generated by neural pathways, burned in by experiences with regularities in the external world. The term comes from statistics and has come to mean any assumption the brain carries about how the world outside should appear given how it has appeared in the past. But the brain goes further than this: In situations of what Pascal and his colleague Michael Karlovich call “substantial uncertainty,” the brain will use its experience to create illusions of what ought to be there but isn’t. In other words, in novel situations the brain usually sees what it expects to see.

Pascal says this was well understood in color vision. We can tell a sweater is green when our closet is very dark, or a car is blue under a cloudy night sky, because the brain does a little Photoshopping to help us in situations where differing lighting conditions alter the appearance of familiar objects. We each possess a correction mechanism that recalibrates our visual systems to “discount the illuminant and achieve color constancy to preserve object identity in the face of changing illuminations.” It does that by altering what we experience to match what we’ve experienced before. There’s a great example of this in an illusion created by vision researcher Akiyoshi Kitaoka.

It looks like a bowl of red strawberries, but the image contains zero red pixels. When you look at the photo, no red light enters your eye. Instead, the brain assumes the image is overexposed by blue light. It turns down the contrast a bit and adds a little color where it was just removed, which means the red you experience when you look at those strawberries isn’t coming from the image. If you’ve grown up eating strawberries and spent a lifetime seeing strawberries as red, when you see the familiar shape of a strawberry, your brain assumes they should be red. The red you see in Kitaoka’s illusion is generated internally, an assumption made after the fact and without your knowledge, a lie told to you by your visual system to provide you with what ought to be the truth.

Pascal figured the photo of the Dress must have been a rare, naturally occurring version of the same phenomenon. The image must have been overexposed, which made the truth ambiguous; people’s brains were disambiguating it by “discounting the illuminant” they assumed was present, all without their knowledge.

The photo had been taken on a dreary day. It was taken with a cheap phone. One portion of the image was bright, and the other was dim. The lighting was ambiguous. Pascal explained that the color that appeared in each brain was different depending on how each brain disambiguated the lighting conditions. For some, it disambiguated the ambiguous as black and blue; for others, as white and gold. As with the strawberries, people’s brains accomplished this by lying to them, by creating a lighting condition that wasn’t there. What made this image different, he says, was that different brains told different lies, dividing people into two camps with incompatible subjective realities.

Chasing that hypothesis, Pascal thought he had an explanation for that. After two years of research with more than 10,000 participants, Pascal discovered a clear pattern among his subjects. The more time a person had spent exposed to artificial light (which is predominantly yellow)—typically a person who works indoors or at night—the more likely they were to say the Dress was black and blue. That was because they assumed, unconsciously, at the level of visual processing, that it was artificially lit, and thus their brains subtracted the yellow, leaving behind the darker, bluish shades. However, the more time a person had spent exposed to natural light—someone who works during the day, outside, or near windows—the more likely they were to subtract blue and see it as white and gold. Either way, the ambiguity never registered.

Whatever colors people saw subjectively, the image never seemed ambiguous because consciously people experienced only the output of their processes, and the output differed depending on a person’s prior experiences with light. The result was a lie told to them by their brains that felt obviously true.

His lab came up with a term for this: SURFPAD. When you combine Substantial Uncertainty with Ramified (which means branching) or Forked Priors or Assumptions, you will get Disagreement.

In other words, when the truth is uncertain, our brains resolve that uncertainty without our knowledge by creating the most likely reality they can imagine based on our prior experiences. People whose brains remove that uncertainty in similar ways will find themselves in agreement, like those who saw the dress as black and blue. Others whose brains resolve that uncertainty in a different way will also find themselves in agreement, like those who saw the dress as white and gold. The essence of SURFPAD is that both groups feel certain, and among the like-minded, it seems that those who disagree, no matter their numbers, must be mistaken. In each group, people then begin searching for reasons why those in the other group can’t see the truth—without entertaining the possibility that they aren’t seeing the truth themselves.

Another example of SURFPAD in action was the different reactions to the Covid-19 vaccines as they rolled out to the public in 2020. Most people weren’t experts on vaccines or epidemiology, so the information on how it worked and what to do was both novel and ambiguous. To resolve that uncertainty, people used their prior experiences with vaccines and doctors, their existing levels of trust in scientific institutions, and their current attitudes toward the government to make sense of it all. For some, that led to the conclusion that vaccines were probably safe and effective. For others, it led to a hesitancy that matured into suspicions of conspiracy. For both, the people who saw things differently seemed blind to the truth.

When we encounter novel information that seems ambiguous, we unknowingly disambiguate it based on what we’ve experienced in the past. But starting at the level of perception, different life experiences can lead to very different disambiguations, and thus very different subjective realities. When that happens in the presence of substantial uncertainty, we may vehemently disagree over reality itself—but since no one on either side is aware of the brain processes leading up to that disagreement, it makes the people who see things differently seem, in a word, wrong.


Adapted from How Minds Change: The Surprising Science of Belief, Opinion, and Persuasion by David McRaney, published by Portfolio, an imprint of the Penguin Publishing Group, a division of Penguin Random House, LLC. Copyright © 2022 by David McRaney.