Cognitive Daily

Last week’s Casual Fridays study elicited quite a few confused reactions. Listeners heard short music clips and then were asked to indicate which color they associated with each piece from a list of 13 colors.

So were we able to come up with consistent results? Yes. For some of the clips, our 377 listeners were more likely to associate particular colors with that clip. Consider this clip:

Here are the survey responses:

Significantly more listeners picked “yellow” than any other color, and half of all listeners picked yellow or pink. Practically no one chose beige, brown, or any dark color. And listeners weren’t simply assigning the same colors to all the clips: this was the only clip for which yellow was the most popular response, and significantly more listeners associated pink with this clip than any of the other clips.

The excerpts in the study came from two sources: the piano selections have been used in several studies led by Isabelle Peretz, and the drum selections are used in Greta’s lab. Both sets of excerpts have been rated for emotional content. The selection you heard above, for example, was rated significantly more “happy” than any of several other emotions, including sad, angry, and fearful. We chose the drum clips because they matched the emotions in the piano selections. Greta’s lab has found that listeners also rate this drum piece as “happy”:

This raises the question: Is the emotional content of a piece of music responsible for the colors we associate with it? After all, many people do associate emotions with particular colors. Let’s take a look at the results for the two “happy” excerpts:

There’s much less variation in the responses for the drum excerpt — it almost seems as if the respondents chose randomly. In fact they did still choose dark blue significantly more frequently than any other color. But as you can see, the responses were quite different than for the happy piano excerpt: Almost no piano listeners picked dark blue, brown, or dark purple — the three most popular choices for the drum excerpt.

But perhaps other emotions were more similar between the two instruments. Here are the results for fearful music:

Things aren’t looking much better: while brown is the most popular choice for the drum piece, red and dark orange top the list for piano. How about sad excerpts?

Nope. It seems that listeners have almost completely different associations for the drum pieces. In fact, there is one important difference between the drum clips and the piano clips, summarized nicely with this small graph:

Drum clips were much more likely to be associated with beige or brown, while piano clips were associated more frequently with all the other colors.

We might be tempted to say that emotion has little to do with the colors we associate with a music clip, and the tonal qualities of the particular instrument are more important, but there is a potential problem with the data. Because the clips were rated for emotion in two different labs, the emotions portrayed in each clip might not match up exactly. Take a look at this graph:

The top two colors for the happy drum clip match exactly with the sad piano clip (although the response for dark purple on the drums isn’t significantly different from brown). These clips sound very different to me, as you can see here:

Happy drums:

Sad piano

But maybe they evoke similar levels of certain aspects of emotion, and this is why the pattern of their results looks similar.

In any case, we can certainly say from these results that there is a strong relationship between music and color — and that both the particular instruments chosen and the emotion conveyed by the music are also related to the particular colors we associate with music.

Comments

#1 sprinkles
February 8, 2008

I’m sort of curious as to how much of this color association is a product of the environment. It is sort of hard to keep an open mind when, at least personally, I can think of many instances in which people associate specific emotions with certain colors as a fact. Although, I think that I have seen studies claiming red evokes anger/fear universally in humans.
#2 dotti
February 8, 2008

here’s a brief explanation of timbre – the “color” of music, which goes a little way in explaining the difference between drums and piano. piano has a specific pitch, which is related to its timbre. drums don’t have specific pitch, making their frequencies more variable, maybe leading people to have a sense that they’re muddy, and therefore brown or beige.

http://cnx.org/content/m11059/latest/

i took the test without thinking analytically and came up with many of the most frequent responses. i remember feeling like the piano pieces were much easier to link to a color than the drum pieces.
#3 6EQUJ5
February 8, 2008

I took your test last Friday (?) and noticed the pattern of my responses. If it sounded like a movie sound track, the only color I associate with movies is blue — as when they are shooting day for night. The rest of it sounded like music played live on a stage, and the color I associate there is yellow — from the yellow tinge of incandescent lights.

I don’t think I associate color with any emotions, only with places. As in blue, the color of the desert sky in daylight; green, the color of the Appalachian woods in the summer; or black and white, the color of the same woods in winter.
#4 Joe Shelby
February 8, 2008

Well, I was annoyed that you didn’t include “white” or any other variation of grey. Percussion for me is colourless. Still solid and opaque, but generally a shade of white/grey/silver. It’s not a reflection of “happy/sad” – i saw the same color with all the percussion pieces except the one that involved the tamas.
#5 Stephen Downes
February 8, 2008

Yeah, but does it work the other way around. Do certain colours evoke certain types of music?
#6 mickgrierson
February 8, 2008

When I saw this on my rss feed I almost exploded with anger. However, it looks like you got some interesting results.

I would propose that trying to find connections between “music’ and “colours’ is a study based on erroneous distinctions. In a way, it’s as if you’re drawing a comparison between a note and a symphony, or a colour and a portrait. You’re just not measuring like against like. Timbre and texture are related – very closely. So, a rough sound like a snare roll, and a piece of sandpaper may have a similar texture. But these are not notes and colours. They are complex patterns that have structural similarity. It is n surprise that we perceive them as relative.

However. Your results make me think of the importance of luminance, which is a much more significant and fundamental visual element than colour perception, and could related in an odd way to some metaphorical interpretation of harmonic content against time. God knows how. Hmmm. All in all, texture is a better measure.

What the hell, it’s casual fridays after all.

peace

M
#7 katy lawley
February 9, 2008

I wonder how your results would look if test-takers had selected from jpg swatches rather than textual (hence indirect) representations of colors.
#8 mdreyer
February 9, 2008

I wonder if it made any difference that you gave actual sounds but a written word for color. a niggling point, but– if you had given color swatches, maybe there would be less or no verbal mediation, and once you go there, I think things get complicated with what you know and remember verbally.
#9 joel
February 9, 2008

I know that I would generally have identified color associations most strongly with tambre. An orchestration that blends tabre and harmony in certain ways can evoke color very strongly even with a single note and no tempo. But when I realized that there were only two tambres in play, I had to amplify minor differences in the color I felt to avoid giving them all the same color result.
#10 Tony Jeremiah
February 10, 2008

I suspect the explanation for this may be related to three basic types of information contained in the energy that is detected by our sensory organs: frequency, amplitude, complexity. For light energy, frequency differences are perceived as color, amplitude differences are perceived as brightness, and complexity is perceived saturation. For sound energy, frequency differences are perceived as notes, amplitude differences are perceived as loudness, and complexity is perceived as timbre. The music/color correspondence might be the result of an implicit matching of the low-to-high range of information contained in sound and light energy.

As an example, the famous ROYGBIV acronym represents visible spectrum colors proceeding from low to high frequency light. CDEFGABC could represent notes on a piano proceeding from low to high frequency. So in this scenario, it’s possible that red could correspond with the note C on a piano since they both represent low frequencies on their respective scale; violet could correspond with note B since they both represent high frequencies on their respective scale. Since the clips are playing a variety of notes, the music/color correspondence is probably more connected with similarities in the complexity dimension (i.e.,a similar mixing of frequency, amplitude, and duration of particular sensory information).
Explaining how this is all connected to emotion is slightly more complex, since it involves a connection between brain/body temperature and emotions.

There’s some theory that suggests that high brain temperature is associated with negative emotions (e.g., sadness), and lower brain temperatures are associated with positive emotions (e.g., happiness). And, we know that when we detect the temperature of something, it’s due to the rate at which particles in that something is vibrating–low frequency vibrations are detected as cold; high frequency vibrations are detected as warm or hotter.

So if we connect this with the low-mid-high range correspondence between sensations, that would mean that positive emotions would be connected to colors such as red and orange as well as lower notes on an instrument; negative emotions would be connected to colors such as indigo and violet.

At least for the piano, that looks to be the case since it appears that a happy piano is associated with relatively lower spectrum colors (yellow in particular), and a sad piano is associated with relatively higher spectrum colors (blue and purple in particular). I’m not entirely sure what to make of the fear (negative emotion) piano corresponding with lower frequency light (red) though. Although, fear is known to correspond with a reduction in body temperature.

The difference between the drum and piano might have something to do with the greater sound complexity that seems apparent with the drum clips (i.e., more overtones) than the piano clips (i.e., more pure tones).
#11 Heidi
February 10, 2008

The order of the colors is different for each musical clip. Were these orders randomly generated for each user, or did everyone see the colors in the same order? I ask because I did notice that occasionally I would think about the colors at the top of the list first, b/c I would see them in my peripheral vision as I listened to the music. Do you think this effect could have affected the overall results? It might be interesting to have a quiz were the music and the response choices were on different screens.

Some of the colors I picked because of emotional quality, but others I chose because of other associations I have with the music. For example, I associate jazzy music with blues and purples of album covers and dark clubs and some drum music with browns and oranges or other colors of Africa or South America.
#12 Dave Munger
February 10, 2008

Heidi: The colors were randomly generated for each individual.

Tony: Your explanation makes some sense. I’m still thinking that emotion has something to do with it. To really get at the answer we’d have to run a much better controlled study, which simultaneously asked listeners to rate the emotional valences of each clip.
#13 Tony Jeremiah
February 10, 2008

Dave,

Indeed it would be an interesting follow up study. Also, from my understanding of the reported research, it looks like there are actually three different phenomenological relationships that would need to be tested:

(1) Emotion/Color (indicated by the study mentioned in the post)

(2) Emotion/Music (indicated by your suggested follow up study)

(3) Music/Color (basically my explanation for the current results)

The third relationship is separate from the other two because it is more of a reference to purely cognitive processes (i.e., the brain associating two perceptual processes), whereas the remaining phenomena implicate cognitive-affective associations.

I think the most direct test for my explanation would require running a spectral analysis on each sound clip (which would quantify the complexity of the sounds from each clip), and correlate that with the colors selected for each clip.
#14 John Carter
February 10, 2008

Could you post a link to the raw data?
#15 Alp
February 10, 2008

I was wondering if the brown/beige color associations to drum could be the result of actual (most likely) color of the drums..
#16 Tony Jeremiah
February 10, 2008

Alp,

An interesting hypothesis for the trend concerning the greater choice of beige/brown for the drum relative to the piano. However, that would still leave two things left unexplained: (1) why there isn’t an overriding preference for dark colors for the piano as well (a good portion of pianos tend to be black and brown); and (2) when you look at the happy drum/sad piano comparison, dark blue and dark purple seems to be as strongly associated with the drum as the beige/brown bias.