“The Dress” and Uncovering the Hidden Uncertainties in Visual Perception

NYU Center for Data Science
3 min readApr 16, 2024
Photograph: Cecilia Bleasdale

Our brains constantly make smart guesses to fill in the gaps when interpreting the world around us. This process happens automatically, behind the scenes, making us largely unaware of the extent to which our visual perceptions are based on assumptions. In a recent episode of Scientific American’s new podcast series “Uncertain,” CDS Clinical Associate Professor of Data Science and Psychology Pascal Wallisch discussed his research probing how the brain handles the hidden uncertainties involved in perceiving the world.

The episode, titled “Think Seeing is Believing? Think Again,” features Wallisch’s work on the famous “dress” photo that went viral in 2015. While some people saw the dress as white and gold, others adamantly perceived it as blue and black. Wallisch, who has discussed the dress before, explained that the dress itself was actually blue and black, but the ambiguous lighting in the photograph introduced uncertainty. People’s brains made different assumptions to resolve this ambiguity based on their past visual experiences, leading to the stark disagreement.

“Your brain doesn’t say ‘I have absolutely no idea’,” Wallisch said in the podcast. “Your brain tells you ‘I think it’s white.’ That’s it. It doesn’t flag to you ‘Oh, and by the way, I made a guess here’.”

From a data science perspective, said Wallisch, the brain is effectively doing “missing data handling,” a key challenge for any data scientist. And it is doing it in a smart way. The brain fills in the gaps in the available visual information based on prior knowledge and experiences, similar to how data scientists use various techniques to handle missing data in datasets.

To further investigate how prior assumptions shape visual perception, Wallisch conducted experiments using images of pink Crocs shoes photographed under green lighting, which made them appear gray. Surprisingly, people who recognized that the socks were white (based on their experiences with those types of socks always being white) perceived the Crocs as pink, mentally subtracting the green light. In contrast, those who assumed the socks could actually be green saw the Crocs as gray.

“Just because you’re sure does not mean you’re right,” Wallisch emphasized. “This uncertainty is hidden from you.”

Wallisch’s research highlights the importance of intellectual humility and being open to the idea that our perceptions might not always align with reality. By studying how the brain fills in uncertainties, his work at the intersection of data science, psychology, and neuroscience aims to shed light on the complex processes underlying the construction of our subjective realities.

CDS’ Fox Lab, directed by Wallisch, conducts studies with high statistical power and epistemic integrity to empirically answer questions about personality and cognition. The lab’s interdisciplinary approach reflects the far-reaching implications of understanding the hidden uncertainties that shape our perceptions and decisions.

As Wallisch noted in the podcast, “I think we need a culture of awareness and intellectual modesty that counteracts this evolutionary tendency to jump to conclusions and then commit to them.” By uncovering the hidden uncertainties in visual perception, his research reminds us to approach our own perceptions with a healthy dose of humility and openness to alternative perspectives.

By Stephen Thomas



NYU Center for Data Science

Official account of the Center for Data Science at NYU, home of the Undergraduate, Master’s, and Ph.D. programs in Data Science.