30 thoughts on “The tone problem in psychology

    • Interesting. 25/32, where <=19 would be tone deaf. Basically, anything 1/16th or smaller sounded the same to me. This is consistent with my self-perception as having rather minimal musical abilities.

      • It doesn’t anything to do with musical ability. The smallest tone step in western instrument is a half tone.

        I heard two 64ths clearly and guessed right on at least two, and guessed wrong on three, so there must have been eight, so you nailed all the 32nds and one fo the 64ths, which is more than good enough tone awareness to be successful as a musician.

        29 for me.

        • 27 for me. (Includes one try when I pressed the wrong key from the one I intended to press. Probably related to why I never learned to play the piano well, but did fine with flute and vocal lessons.)

        • “Includes one try when I pressed the wrong key”…

          I was surprised I didn’t to that! Usually on that kind of thing I’m all tense. This time I closed my eyes and cleared my mind and let my subconscious take over.

          Would be interesting to know how the average person performs on the 64ths.

        • Yeah I did the eye closed thing too. I didn’t want to get in my own head. But I guess since they were showing us the answers this was supposed to be a continuous learning thing?

          > It doesn’t anything to do with musical ability.

          I guess that’s probably part of what this study is gonna try to figure out. Presumably we can’t learn much from the 32 binary responses of any individual.

          > The smallest tone step in western instrument is a half tone.

          Sure but to be accurate to a half tone you’d want to be able to keep your tuning error less than that. Have you played any instruments? That seems like a musician sorta thing to know :D.

        • I play a little guitar, very little, and I have trouble getting into precise tune by ear. I also had trouble with the 1/64th tones, and some of the falling 1/32nd tones. But 1/8th tone and 1/16th tones were more or less ok.

          I took this test rather quickly the first time, and got 16th percentile, and then took it more carefully and got 50 something percentile…

          I think they could use a little more statistical nuance in their scoring, it seems like the percentiles drop off really rapidly, so that a single mistake could make the difference between say 16th and maybe 35th or whatnot. I don’t think they’re using speed in their reported score/statistics at all. It seems like they should, as well as age, male/female, and they should be starting on more than one initial tone, some people will just not be particularly accurate in certain frequency ranges… this is always A4 = 440Hz to begin with.

        • A logistic regression of accurate classification against up/down, male/female, age, speed, and headphone use would give a more useful relative ranking score (ie. your score vs expected accuracy range for you averaged across the population)

        • Ben said,

          “> The smallest tone step in western instrument is a half tone.

          Sure but to be accurate to a half tone you’d want to be able to keep your tuning error less than that.”

          I used to play the flute (haven’t for a long time), and I recall something called “beats” that came up in tuning:
          From Wikipedia: “In acoustics, a beat is an interference pattern between two sounds of slightly different frequencies, perceived as a periodic variation in volume whose rate is the difference of the two frequencies. With tuning instruments that can produce sustained tones, beats can be readily recognized.”

          I think that using beats we got to much less than a half tone of difference. I’m pretty sure the beats from a half tone would be noticeable.

        • @Martha, oh, good point. Like it’s probably way easier to compare two tones if they’re played at the same time vs. different times. That makes sense.

          @Daniel yeah I was trying for speed too.

        • “Sure but to be accurate to a half tone you’d want to be able to keep your tuning error less than that. Have you played any instruments?”

          If you can’t hear the 64th it doesn’t matter if your tuning is out by that much. You just can’t hear it. It only becomes an issue if it’s progressive – that is each successive string is out another 32nd more in the same direction or something like that.

          I’ve played guitar for a long time. I learned how to tune by ear because there were no tuners (except pitchforks) when I learned to play. Now I have one but I can’t find it right now, so I’ve been tuning by ear the last few weeks. As Martha said, when you strike two tones that are close in pitch, the interference of the two slightly different wave lengths creates a “wa wa wa wa” sound – a secondary vibration I guess – that decreases in frequency and disappears as the strings are brought into tune. That’s helpful. The sound is sharper when using harmnoics (lightly touching the string over certain frets and striking it)

          For me the main problem with tuning by ear is that as the string vibration dampens due to the strings getting dirty, both the interference and the preciseness of the tone are harder to hear, especially in the base strings, and I wind up working the tuning a lot. Also I stretch the strings a lot which knocks them out of tune.

        • “I’m pretty sure the beats from a half tone would be noticeable.”

          Yes actually a half tone separation is common between chord notes, so Amaj7 would have G# and A. The effect is roughly similar to a minor chord, somewhat of a melancholy sound.

          The “half-tone hammer” is also a common “trick” in rock music: play the same chord but move one finger from the 3rd tone in the scale/chord to the 4th tone and back, which is a half step, let’s see: the intro to Jack and Diane comes to mind.

        • I guessed the first 1/64 note test correctly (wasn’t prepared for the test being that hard!) but once I figured out the game, I got them all correct.

          (I played violin from 4 to 20, fingerpicking guitar from 24 on, and jazz guitar for the last 10 years.)

          I went on to the site’s other tests, and found they have an axe to grind. Their axe is that since every society everywhere has and likes music, there must be some universals in there, and that those universals are dance, love, healing, and soothing babies.

          I get about zero on those tests, which are based on horrifically bad recordings of “world music”, which to them means aboriginal music. There’s lots of Middle Eastern, Klezmer, Japanese, Indonesian, South American “world” music that I like. None of that was there.

          Reading the explanation of the clips, I really don’t think it’s possible to tell what most such songs are about without understanding the words. But they do.

          They also have a “We’ll find you songs that you like” test*, and the musical style choices don’t include any of the music I listen to: folk, blues, or jazz.

          * Also based on those bad recordings and their ideas of what music is about.

          Go figure…

        • Yes to all that stuff about their website and Axe to grind… the recordings they use in the “find songs you like” test are pretty bad, and I LIKE world music a lot (Andean flutes, Gnawa trance music, singing the Koran, Shakuhachi, etc)

          I spent all this time rating their horrible recordings, then they offered me a couple of “genres” to choose from, I chose Latin because I love Latin Jazz (Ray Barretto, Chucho Valdez) but all they had was like Julio Iglesias or some such thing… yech. I felt cheated.

    • Zad, take it again and see what happens. See Daniel Lakeland’s report above, 16th percentile first time and 50th another. Obviously for most of us there is a substantial amount of random guessing — the 1/64ths were coin tosses for me, and probably the 1/32nds were too — so the score can fluctuate. I would guess you’ll regress towards the mean but still be worse than median. But also maybe one gets better with practice, even just with a couple of repetitions.

    • That twitter thread was a turn-off for me. It wasn’t so brief that I got all the way through it. Instead I googled the title of the paper. It’s certainly been a PR success; within days of the study’s release it’s been covered in relatively glowing articles in Scientific American, Atlantic, and other places, with very limited carping from those pissed-off ethnomusicologists.

      I haven’t read the article, although I did dip into it. That all human societies have music is no surprise, although its hard to believe this study has proved that conclusively (there must be outliers somewhere) or that earlier studies hadn’t gone a long way toward establishing that. That listeners can discern at rates much better than chance the difference between lullabies and dance songs in other cultures is interesting, but I wouldn’t conclude anything from that about what’s in our DNA. I’d think there are lots of activities associated with putting babies to sleep or with celebration that will be discernible across cultures (hand-clapping, for example). That listeners aren’t so good at distinguishing love songs from other types of songs is amusing, but not so surprising once I think about it (what popped into my head as a love song was MC5’s “My love is like a rambling rose”).

      The “A new science of music” section at the end of the article offers what I take to be their main conclusions. Among them, “music varies more within than between cultures.” Wouldn’t this depend a lot on what factors you’re measuring to define the variation? Their search for universality in music might bias them toward measures that average out across cultures. And their final finding: “melodies and rhythms are balanced between monotony and chaos.” I would hope so.

      • David said,
        “That listeners can discern at rates much better than chance the difference between lullabies and dance songs in other cultures is interesting, but I wouldn’t conclude anything from that about what’s in our DNA. I’d think there are lots of activities associated with putting babies to sleep or with celebration that will be discernible across cultures (hand-clapping, for example)”

        Duh. Lullabies are almost by definition slow and soothing. Some dance songs might be that, but dance songs are more likely to be lively — and lively is the opposite of what is likely to help get a baby to sleep.

        • +1 to that “Duh”, of course.

          Note that the non-lullaby categories aren’t categories. You can dance for love, dance to heal, dance to make rain. You can sing a love song to someone who’s in pain to heal that pain. And they left out singing in anger and rage and sadness.

          NSFW audio: https://www.youtube.com/watch?v=W8lypsvs3ls

          And, since Morris Dancing appeared in a TV rerun we watched after dinner, a joke from the Boston/Cambridge area folk music scene. “Why do Morris Dancers wear bells on their legs? So they can irritate the visually challenged as well.”

  1. I got to this point in the Science article and just had to stop:

    “To specify the coherence of these clusters [formality, arousal, religiosity] formally rather than just visually, we asked what proportion of song events are closer to the centroid of their own type’s location than to any other type (Text S2.1.6). Overall, 64.7% of the songs were located closest to the centroid of their own type; under a null hypothesis that song type is unrelated to location, simulated by randomly shuffling the song labels, only 23.2% would do so (P < 0.001 according to a permutation test). This result was statistically significant for three of the
    four song types (dance, 66.2%; healing, 74.0%; love, 63.6%; Ps < 0.001) although not for lullabies (39.7%, P = 0.92)."

    At least they resisted the urge to go to two decimal places!

    • “Overall, 64.7% of the songs were located closest to the centroid of their own type”

      And how was the centroid of a type defined? Presumably by looking just at songs that had been classified as being in the same type. So the “result” is unsurprising.

Leave a Reply to Martha (Smith) Cancel reply

Your email address will not be published. Required fields are marked *