Over the past 48 hours, a controversy has erupted over social media: does a photograph depict a blue dress with black lace? Or is it a white dress with gold lace? A minority have responded that they see some combination of the two positions. Those who report the dress as gold and white understand the front of the dress to have been photographed in shadow; these shadows cause the dress to take on a bluish hue and the gold lace a dark brown, or so they say. Those who report that the dress is blue and black (or blue and brown) see the dress as front-lit under yellowish light, causing the black lace to appear brown.
The debate involves interesting philosophical issues concerning language that I will comment on in this post; as far as I know, I will be the first person to comment on the particular aspects of the debate that I identify here. I do not plan to resolve the debate; instead, I will use the debate to illustrate some of the interesting reasons that communication can break down and some of the perplexities of color terms. (As far as I am concerned, the debate was resolved when my friend, Mallorie Nasrallah, who is a professional photographer, determined that white balancing the image reveals that the dress is blue and black. Meanwhile, numerous faulty explanations emerged, including multiple pseudo-psychological explanations. Her analysis was confirmed when the dress was identified on Amazon.)
At first pass, when we talk about “the color of an object”, what we typically mean is the way that a neurophysiologically typical individual will perceive that object when white light is shined on that object (I say “at first past” because this definition is not perfect; reflections from white light can alter the color of an object, for example). However, under different lighting conditions, the color that we literally see will differ, even though the color we interpret and report might not. To use an example from one of my professors in undergrad, the color of a barn might appear different during midday than at dusk, even though we might not change what color we say that the barn is. Furthermore, the color of an object is not the same as the color of a photograph depicting the object. Photographs of a barn at midday and at dusk might differ in color, even though many people will report, based on the photos, that the barn is the same color. This reveals an important ambiguity related to color-terms.
The following sentence is ambiguous, in that it has two meanings: “The dress is blue.” Here is one possible interpretation:
- The actual object — the dress, itself — is blue.
Here is a second interpretation:
- The area of the picture occupied by the dress is blue.
Notice that those who support the conclusion that the dress is gold-white deny (1). This does not mean that they deny (2). This explains some of the frustration that I have seen from individuals involved in the debate: all parties see that there are blue pixels in the photo, but not everyone is straightforward in reporting this observation. Instead, people are straightforward in reporting what color they interpret the dress to be, which is not the same color as its representation (I am reminded Rene Margritte’s Treachery of Images: Ceci n’est pas blanche!). In other words, everyone is seeing the same thing (no one is looking at the photo and seeing pure white) but their reports differ because their interpretations differ. Importantly: interpretations of the debate which depend upon people literally seeing differently (e.g. receiving different stimuli) are incorrect because everyone sees blue in the image, but not everyone reports that blue as representing a blue dress. This means that the debate has much more to do with the exegesis of the image than with neurophysiology or psychology.
A second problem has to do with what referents color terms have. Some individuals who have told me that they see the dress as gold and white have gone on to explain that they see the dress as bluish white. But what exactly is the distinction between bluish white and whitish blue? I have some idea of what this distinction might amount to — bluish white is closer to white than to blue and whitish blue is closer to blue than it is to white — but, as I read it, the color of the photo is ambiguous between the two. Those who read the dress as bluish white might report the dress as white; those who read the dress as whitish blue might report the dress as blue; yet both see the dress as the same color. In some sense, they even interpret the dress as the same color but use different terms to report that interpretation.
There is a second problem with reference: what color is gold? When I hear the term ‘gold’, what immediately comes to mind is an image of Scrooge McDuck diving into a vault of gold coins (as depicted in the opening to Duck Tales). The animators chose to depict McDuck’s coins as a vibrant yellowish color because gold, when in direct light, has a rich yellow tone. However, in shadow, gold has a dark brown color, similar to bronze. When individuals reported to me that they saw gold, I was perplexed because I expected to see yellow in the image. What they meant was that the dress has gold lace which appears dark when in shadow.
This exchange brings to the fore the importance of doing philosophical analysis. Conversation breaks down when we fail to make the proper sort of semantic distinctions because we end up talking past each other; by carefully unraveling and unpacking sentences, we can (often) determine that we do not disagree with others at all.
And people wonder what philosophy is useful for…