Wednesday, March 25, 2015

"A" Is Red, "I" Is White, "X" Is Black -- Um, Why?

This is just the kind of dorky thing I think is cool. Check out this graph of the color associations for different letters for people with grapheme-color synesthesia.
[click on the picture for full size, if it's not showing properly]

This is from a sample of 6588 synesthetes in the US, reported in Witthoft, Winawer, and Eagleman 2015. Presumably, they're not talking to each other. But there's a pretty good agreement that "A" is red, "X" is black, and "Y" is yellow. But you knew that already, right?

Now some of these results seem partly explicable: "Y" is yellow, maybe, because of the word "yellow" starts with "Y". That might also work for "R" red, "B" blue, and "G" green. For "A" I think of the big red apple with the "A is for apple" posters that ubiquitously decorate kindergarten classrooms. But "O" is not particularly associated with orange in this chart, nor "W" with white. And why are "X" and "Z" black? Because we're tired because it's near the end of the alphabet and our eyelids are starting to droop doesn't seem like a good answer. (Does it?)

You might wonder whether it's only synesthetes who have this consensus of associations, and how stable such associations are over time or between countries.

You're in luck, then, because here's another cool chart, from Australia in 2005!

[again, click for clearer view]

The colored bars are synesthetic respondents and the hatched bars are non-synesthetic respondents. The patterns are similar between synesthetes and non-synesthetes, but maybe with the non-synesthetes tending toward stronger associations between the color and the initial letter of the color word. Furthermore, again "A" is red, "I" is white, and "X" and "Z" are black. US and Australian synesthetes seems to agree that "O" is white, but the Australian non-synesthetes like their "O" orange. For some reason, "D" is now brown (47%!).

There are some older US data from the underappreciated early introspective psychologist Mary Whiton Calkins in her classic 1893 paper on synesthesia. [Pop quiz: Who are the only three people to have been president of both the American Psychological Association and the American Philosophical Association? Answer: William James, John Dewey, and Mary Whiton Calkins.] She reports that synesthetes tend to associate "I" with black and "O" with white. "O" being white matches the synesthete reports from the US and Australia in 2015 and 2005, but Calkins's black "I" is different. Calkins reports this possible explanation for the whiteness of "O", from one of her participants, seeming to find it plausible: O "= cipher = blank = sheet of white paper".

Witthoft et al. 2015 found that almost a sixth of their participants born in the US in the late 1970s (but not those born before 1967) seem to have letter-color associations that match much better than chance with the colors of the letters of this then-popular magnet toy:

[image source]

Neat finding. Of course, the darned toy has "X" purple and "Z" orange, so it's all wrong!

Brang, Rouw, Ramachandran and Coulson 2011 find a weak tendency for similarly-shaped letters to associate to similar colors in US sample. Irish-based Barnett et al. 2008 and British-based Simner et al. 2015 find broadly similar patterns to the other recent English-language populations.

Spector and Maurer 2011 find that even pre-literate English-speaking Canadian toddlers associate "O" and "I" with white and "X" and "Z" with black, though they do not share older participants' associations of "A" with red, "B" with blue, "G" with green, and "Y" with yellow. They hypothesize that jagged shapes ("X" and "Z") might be more likely to have shaded portions in a natural environment than non-jagged shapes ("O" and "I"), and that other, later associations might be language based. However, color maps of Swiss research on German-language synesthetes (Beeli, Esslen, and Jaencke 2007) shows no such relationship (see the chart on p. 790) -- for example with more participants associating "X" with white or light gray than with black or dark gray (though Simner et al. have a German subset which do show black associations with "X" and "Z"). Beeli et al. find a weak tendency for higher frequency letters to be associated with higher saturation colors in a German-language sample. Rouw et al. 2014 found that Dutch and English-speaking non-synesthetic participants had similar associations for "A" (red), "B" (blue), "D" (brown), "E" (yellow), "I" (white), and "N" (brown). Hindi participants, with their different alphabet, had a rather different set of associations -- though the first letter of the Hindi alphabet was also associated with red. They speculate that the first letter in each alphabet gets a "signal" color.

Okay, so now you know!

Let me leave you then, with this highly unnatural thought:



  1. I can think of a few stories to make sense of at least some of these findings. A is the first letter and red is the first colour. Black is always at the end of colour lists. Almost all the vowels are pale colours, because vowels are liquids.

    But actually, the ease with which these explanations spring to mind worries me. There's no reason obvious to me why a neurotypical should be able to figure out the connections of someone who is literally wired differently. Which leads me to the null hypothesis: what are the chances that all this is just chance? The paper says "most frequently chosen color match for each letter hereafter as the modal choices," i.e. all a letter has to do is have about 10% of people name a particular colour (assuming that the colours are drawn from the 11 reputed main colour words in English). Certain of the letters - F, for example - really don't show much consistency to me at all.

    So I'm dubious about the stats. I'm also dubious about the concepts used. I know that the letter-colour links for synaesthetes is widely reported, but I struggle with how it could be one of their primary experiences. Colour is a primary, physical category for people; letters are a very secondary category. Running this analysis seems like... I dunno, asking people what their favourite letter of the alphabet is, then asking who their favourite author is, and looking for correlations. If you look, you'll find 'em, but the concepts just don't seem to be on equivalent levels.

  2. chinaphil: The stats seem pretty solid to me in at least some of the cases. Eyeball check the first figure, for example. It seems pretty clear that the A, Y, and Z are not chance arrangements. Some of the other stuff could be chance without sufficient correction for multiple comparisons, or file-drawer issues, I agree. How primary synesthesia experience is -- that's a good question. I inclined toward skepticism early in my reading about this years ago, and I still have some skepticism, but a closer look at the research both on the possible neuroanatomy and on the behavioral tests has quieted my skepticism somewhat.

  3. Yes, that wasn't quite what I was getting at. I agree that for some letters the correlations are very striking; but the authors used a method which derives a "consensus colour" for all letters, whether there is a consensus or not. That's a bit dodgy.

    And I wasn't trying to suggest that synaesthesia is not a primary type of experience, rather that if it is a primary experience, we should surely get more relevant data by examining how it works with other primary experiences - e.g. colour and sound, shape and temperature - rather than with highly cultural artifacts like letters.

    Thinking further about it, I wonder what this kind of research can tell us about qualia in neurotypical people. For example, the old question, how do I know if my red is the same as your red... well, synaesthetes are good examples of people for whom my red is not your red. What do colourblind synaesthetes perceive? If they are RG colourblind, do they get the same synaesthetic associations with both red and green, or are their associations culturally conditioned, so they are separate, even though they cannot see the two colours?

    I still can't get my head around this idea of different levels, though. If a synaesthete can perceive C as a colour, do they also perceive the colour as C? Does the linking always involve one primary(?) sense datum? Or can synaesthesia blend together two higher level concepts, so that the letter C also means the number 4, or the concept of an apartment?

    Odours might be an interesting test case, because they are much less culturally defined - do synaesthetes who experience olfactory blending show more/less similarity than those with visual blending, and is it more/less stable? That might tell us something about the extent to which their sense data are being scrambled by (a) some very early wiring, or (b) are subject to cultural conditioning, presumably via cortical processing.

    1. I have synesthesia, and for me the letter 'C' had always been a very specific shade of light green. But no, if I encounter that color elsewhere, it does not make me think of the letter 'C' specifically. That's just my personal experience, however. Letters and numbers have always had their own specific color, for as long as I can remember.

  4. Yes, these are all pretty interesting questions! I've kept my eye on the synesthesia literature a bit, but I'm not an expert in it. Thus post was a fun excuse to update myself on some of the grapheme-color literature, but there are lots of other kinds of synesthesia too! Brit Brogaard and Jennifer Matey, among others, have recently been exploring the implications for our understanding of consciousness in general.

  5. I have grapheme-color synesthesia! A is red too. Just saying, synesthesia is something developed from birth. There is no physical experiences that make y yellow. When we are young, we don't know that "yellow" is a word until we learn it. Most synesthetes are born with synesthesia, so there is no way that theory is true.