Earlier this year a dress nearly broke the internet. A photo of the striped frock (which is actually blue and black) was posted on Tumblr and it quickly became apparent that it looked very different to different people, spawning furious arguments and lively scientific commentary.
Specifically, people disagreed vehemently over whether it was white and gold (that’s my perception) or blue and black. Now, writing in the journal Cortex, researchers in Germany have published the first study to scan people’s brains while they look at the dress, and the neural findings appear to support earlier, psychological explanations of the phenomenon.
When the dress story went viral, psychologists were quick to explain that this dress provided a striking example of how our perception of the world arises from a combination of incoming sensory information and our interpretation of that information. In the case of colour perception, when light bounces off an object and hits your retina, its mix of wavelengths is determined by the colour of the object and the nature of the light source illuminating it. Your brain has to disentangle the two. Usually it does this very well allowing for something called “colour constancy” – the way that objects of the same colour are perceived the same even under different illumination conditions. However, the mental processing involved in colour perception does leave room for interpretation and ambiguity, especially when the nature of the background illumination is unclear as is the case with the photo of the dress (another illusion that hacks the limitations of this aspect of our visual system is the checker shadow illusion).
For the new study, Lara Schlaffke and her colleagues scanned the brains of 28 people with normal vision while they looked at the photo of the dress. Fourteen of the participants see the dress as white and gold and 14 see it as blue and black. The key finding is that the people who see the dress as white and gold showed extra activation in a raft of brain areas, including in frontal, parietal (near the crown of the head) and temporal (near the ears) regions. Yet, no group differences emerged in a control condition when the participants simply looked at large coloured squares that matched two of the colours that feature in the dress, but without any contextual information also visible (see figure, above).
These results are broadly consistent with the idea that the white/gold perceivers were engaged in more interpretative mental processing when looking at the dress. To oversimplify, their perceptual experience of the dress is based less purely on the “bottom up”, raw sensory information arriving at their eyes, and is distorted more by their own assumptions and expectations about the background illumination. The extra activity in their brains during the dress viewing is likely, at least in part, a neural correlate of all this interpretative, “top down” processing.
What the new study can’t answer is whether this extra neural processing (or which aspects of it) in the white/gold group is the cause of their perceptual experience of the dress, or the consequence. However, the researchers describe some future approaches that could help address this quasi-philosophical conundrum: for example, by using transcranial magnetic stimulation (TMS) to temporarily disrupt the extra localised neural activity seen in the people who experience the dress as white and gold, we could ask: will they still experience the illusion?
Meanwhile, as someone who’s firmly in the white/gold camp, I take satisfaction from this study: I might see the dress as the “wrong” colours, but at least this isn’t due to simple-mindedness, but rather it’s because my brain’s working overtime, doing clever tricks in the background. I’m pretty sure that must be an advantage in at least some situations.
Schlaffke, L., Golisch, A., Haag, L., Lenz, M., Heba, S., Lissek, S., Schmidt-Wilcke, T., Eysel, U., & Tegenthoff, M. (2015). The brain’s dress code: How The Dress allows to decode the neuronal pathway of an optical illusion Cortex, 73, 271-275 DOI: 10.1016/j.cortex.2015.08.017