Why Perfect Pitch isn't always perfect

Blogging on Peer-Reviewed Research

There are lots of people who, with training, can identify musical notes when they know the starting point -- when they hear a song starting with "C," they can name the rest of the notes in the song. But much rarer is the ability to identify musical notes without any context. This is what people are talking about when they talk about "perfect pitch" or "absolute pitch."

Let's do a quick test to get a rough sense of how many CogDaily readers have absolute pitch. Listen to this note:

Now, what note is it?

Obviously these results won't be perfect, but they should give us a general idea. I'll give the answer below so you can see how many people got it right.

But what is the nature of absolute pitch? Do people with "absolute pitch" ever make mistakes? Does the ability change as we age? A team led by E. Alexandra Athos recently published the results of the largest-ever study of absolute pitch. They collected data from over 2,000 individuals, 981 of whom were defined as having absolute pitch. They're still collecting data online, and you can participate -- even if you don't have perfect pitch. So what did they find?

First of all, there was a dramatic difference in ability between people who have absolute pitch and those who don't. People were asked to name 36 random notes generated by a computer (the "pure tone"), and 36 random piano notes. This graph shows their scoring:


You might expect scores on piano tones to be the same as pure tones. In fact people tended to score better on piano, which is why most of the points on this plot appear above the diagonal line which represents equal scores on both tests. The gray box in the lower left represents where you'd expect 95 percent of the scores would fall if the results were purely due to chance -- and there's a large group of people scoring in this range. The other large group is in the upper right -- indicating nearly perfect scores on both tests. For the purposes of this study, anyone scoring to the right of the gray vertical line representing a score of 24.5 on the pure tone test was counted as having absolute pitch.

But with so many people participating in the test, the researchers could also see how scores corresponded to age. This graph shows the average scores for each age range among those passing the test:


The scores declined with age. Note too that 60- to 70-year-olds didn't even average 24 correct. Wasn't 24 supposed to be the cutoff for "Absolute Pitch"? Actually the researchers gave credit of 0.75 points for an answer that was one semitone (a sharp or a flat) off. This allowance helped them make an additional observation: when these answers were scored slightly differently (-0.5 for an answer one semitone flat [low], +0.5 for an answer one semitone sharp [high]), a pattern emerged:


Rarely were mistakes made in the flat direction, and responses became progressively biased towards "sharp" as respondents got older.

What's more, there was a pattern involving which specific notes were incorrectly identified. Take a look at this graph of deviation from correct responses in pure tones:


G# was the note most often misidentified as sharper. A# was the only note misidentified as flatter. The note that's one semitone sharper than G# is A. The note that's one semitone flatter than A# is also A. Since respondents with Absolute Pitch are rarely off by more than a semitone, this means that A is guessed more often than any other note.

A also happens to be the note that orchestras use to tune before concerts. But not every orchestra uses the same value of A -- some are slightly off of the standard value of 440 Hz for tuning a piano. Some ancient music orchestras tune A at 415 Hz -- equal to the standard piano value for G#.

Athos' team speculates that people with absolute pitch have accommodated to a wide variety of A-values as a result. Over time, this has led to a perceptual magnet effect -- the same effect we discussed a few days ago on Cognitive Daily. The perceptual magnet causes people with Absolute Pitch begin to accept G# values as equivalent to A, just like it causes people to hear the Swedish "y" sound as an English "ee".

But the perceptual magnet doesn't explain the gradual increase in "sharp" responses over time. The authors believe that this change may be due to a physical problem with the ear over time. It's likely that everyone's ears start sensing pitches as "sharper" over time -- they just don't notice it because they don't have Absolute Pitch.

So, how did you do on our little one-note test? The note in the recording is a G-sharp (we did an initial survey where the note was an E. Now we're collecting more data).

Athos, E.A., Levinson, B., Kistler, A., Zemansky, J., Bostrom, A., Freimer, N., & Gitschier, J. (2007). Dichotomy and perceptual distortions in absolute pitch ability. PNAS, 104(37), 14795-14800.

More like this


i only knew it was an E because i'd been playing this game called absolute pitch blaster for a long time. I've practically memorized the notes C and G so I used those for reference.

i'm surprised so many people are getting the poll correct. (hell, i was surprised _i_ got the poll correct!)

Yes, it's interesting. I think I might do an analysis of this tomorrow morning. Obviously a shortcoming is that we're only doing one note. Also, since it's just a third away from C, it's likely not as "random" a note as some I could have picked. I sort of wish I had chosen a "black key," since even people with Absolute Pitch have more trouble with these notes.

Also I did a piano note rather than a pure tone, which also makes the task easier.

If you play an instrument with an E string, that's an easy note to recognize.

Wow, there is some interesting data and good theorizing there. I don't have a good sense of absolute pitch, but my relative pitch is pretty good and I got it right. Chatham makes a good point about musicians and E strings. ...maybe the test pitch should have been a G for more exciting commenting potential.

I tell you what. I'm going to change the note and put up a second poll. Then we'll have two data points to compare.

[update] Okay, I've changed it. Here are the results from the first poll (correct answer was E):

1. A (16)
2. A-sharp (B-flat) (13)
3. B (10)
4. C (18)
5. C-sharp (D-flat) (5)
6. D (11)
7. D-sharp (E-flat) (8)
8. E (31)
9. F (16)
10. F-sharp (G-flat) (3)
11. G (14)
12. G-sharp (A-flat) (2)
13. I don't know (18)

Hah. I'm on the second batch; the only reason I knew it was G# was because it's the opening note of Rachmaninoff's Prelude in C# Minor, which I have engraved on my brain because my father played it beautifully on the piano. Interesting poll!

I'd agree with poster 7; recognize it in the same way but had to mentally move the note up an octave or two so that it sounds like the opening note on the right hand for Beethoven's Moonlight Sonata, or, Chopin's Fantasie Impromptu.

By Tony Jeremiah (not verified) on 10 Oct 2007 #permalink

Along these lines is the "air ball chant" phenomenon. I'm no academic, only a Dave Barry fan, but I thought all readers should know about this one. Mr. Barry reported on Heaton's study which found that when a basketball player's shot hits nothing but air, fans spontaneously chant, in perfect unison, "air" on the note "F" and then down to "D" for "ball." How cool is that?


i wonder if seniors whose pitches drifted by half a step were consistent, each, in the direction it drifted.

my projecting synaesthetic (pitch to syllable) gradually went down in pitch along with my solitary flute as i spent years avoiding listening to music, my non-projecting pitch-color syn. failing to intervene, until my mental a was half a step below the 440-hz a. finally turning on the radio/buying the full bach, disturbed (distressed) by the "wrong" key, but didn't realize the problem was mine until your test (clicked f, twice, by accident, in different browsers). just spent an hour rehabilitating with pieces i know by heart and much better at guessing pitches in random mp3s off the web.

if enough of your readers echo my story, i'd ask for a casual friday with a different pitch.

i should note that my relative pitch is terrible; i used to be bitter that my teachers hadn't trained it, and my singing, letting my (former) absolute pitch take care of the exercises.

on a separate note, i do feel (consistent with studies, i think) that my synaesthesia degraded with age as i gradually stopped needing it in situations which tested learning. i am almost sure that pitch>color used to be projective at ~3-5 y.o., at least.

on a separate yet, projective=curse/violence (note names, even wrong, outshout thought like overheard conversation does; wrong note names out-argued the colors and outshouted the right names and the projected fingerings which I tried to force today for the first ~15 min after failing your test)

what i've just written makes me think, but not intuit, the following: 1. my syllables may be linked to projected fingerings, not pitches; 2. or my pitch drift may be related to the result that people in vertically inverting glasses adjusted (and saw the right-side-up image) after some time.

By egoblasted (not verified) on 10 Oct 2007 #permalink

just read the comments; unf. can't retake since i've read the right answer to the new pitch (piano, no? for strings, (lower) freq will affect id speed (negatively)).

re lisa's comment: i've recently read somewhere on the internet that (vague follows as i can't google back the study) vowels (stressed?) fall on the c-scale in (a subset of?) native speakers (of what. pardon the figuratively? amnesiac); also,

"The striking difference between the CCOM and ESM groups found here supports the conjecture that, if given the opportunity, infants can acquire absolute pitch as a feature of speech, which can then carry over to music." via http://www.aip.org/148th/deutsch.html

By egoblasted (not verified) on 10 Oct 2007 #permalink

An interesting tidbit: I am a musician, and tune to A. I used to get the A by hitting a little tuning fork on my head. Eventually, I tried tapping a pencil on my head. It worked. I, like Pavlov's dogs, heard an A. Of course, Schubert and others have muscle memory (propreoception) such that fingering a desk as though it were a piano reproduces the sound inside the person's head, at at least allows them to know the pitches. There is a definite correlation between imagining timbre (texture of the sound), pitch, and rhythm, to being able to reproduce these on the instrument. I teach this and find it a continuously useful basis for evaluating my own state of consciousness. Cheers! Peter Metcalf, Musical-Presence.com

By peter b metcalf (not verified) on 10 Oct 2007 #permalink

Whoa. I actually got it. It just 'sounded' like a black key, and 'felt' like G#.

Now I'm really starting to wonder about myself. My relative pitch has always been pretty good, but...

I was only off by a half-step. I thought it was G.

Can you have perfect pitch without having memorized the non-musical language (A, A-sharp, B-flat, B, etc..) that goes with it?

Another way of asking the same question: does having perfect pitch essentially require *naming* the pitch?

I guess those piano practices do pay off.
I guessed it to be G#. I tried a "hypothetical" (it was correct) C note and dropped step by step to G, which sounded slightly lower than the note above. Usually, I tune using the C note because it's most familiar.

does having perfect pitch essentially require *naming* the pitch?

Every time I've heard it defined, it's always involved naming the note. The ability very likely emerges from a combination of genetics and early musical training. Athos et al. speculate that because of the extremely bimodal distribution of responses to the test, the genetic component is probably a single gene.

Still interested in how to test this without requiring we name the pitch. I am pretty sure I know what a perfect C sounds like in any octave and could tell you whether what you played was a C or not. In the case where it is not a C, I could certainly tell you what it is by describing "how I got there," even if I don't have to consciously count the steps up/down from the familiar C pitch. For instance if it is G sharp, I wouldn't leap to calling it that. I'd call it "that pitch that is five and a half [arbitrary unit that you may call a step]s up from C", or "four down from C."

The extent of my formal musical training was singing scales in grade school choir practice and that is the only reason I remember C, we always started with it... We didn't read sheet music or do any exercises that required naming or transcribing a note to its proper letter-label-place on the staff, based on what pitches we heard.

That's more of a rote memory categorical naming-thing to me, even though what all normal people will hear and how we know it's discretely not some other pitch is fundamentally the same experience as naming it properly according to some formal system someone made up.

I know I sound angry, but I'm really not. It's just that when I lack the formal training or vocabulary to participate, I feel left out :)


I'm not sure we could do it right here, but I believe there have been some attempts to do what you're describing. I've seen abstracts to studies that attempt to study absolute pitch in untrained individuals. Technically it's not the same thing as being able to identify notes, but it covers the same principle. The idea is to ask people whether they've heard a note before -- then play some notes in tune and some out of tune. But it still involves a training phase and a testing phase, something that wouldn't be very easy to accomplish on a blog.

I may have gotten this wrong, so if other readers know more, feel free to correct me.

Very interesting critiques posed by Stephen (#17) and Katy (#20). It speaks to the same critique present in language and mathematical literacy, and literacy in general-- that differing levels of literacy exist in any area. The particular questions posed by both seems relevant to the observation that some autistics are capable of playing back an entire musical score (e.g., on piano) after hearing it only once and without having any formal training associated with learning musical language/notation. The analogy for language, is that of a person with oral but not written literacy; but again, this indicates differing levels of literacy for a particular area.

So the question is how do such individuals demonstrate perfect pitch without learning musical notation (one type of music literacy)? Answer: They are literate as it concerns knowing how notes sound on a particular instrument. So from this particular example, another way to test absolute pitch without understanding formal musical notation, would be to have participants answer by pressing a note on a keyboard or some other visually presented instrument that a particular musician might play. However, this would be very difficult to do if a person only has had voice training. And in this case, the only appropriate way to communicate possible absolute pitch would be to indicate this knowledge by clicking on a visually presented musical sheet containing the relevant notes.

If one cannot indicate musical knowledge either through indicating the note being tested via a visually presented instrument, or, through a musical sheet (staff) with notes on it, or, by understanding specific musical notation, then it's clear that an assessment of absolute pitch would be biased against persons having literally, no musical literacy.

So it might be worth doing this online poll again by asking for answers in a different way as suggested above, if you're really curious to see if different levels of musical literacy makes a difference in estimations of absolute pitch.

By Tony Jeremiah (not verified) on 11 Oct 2007 #permalink

I have perfect pitch-I was diagnosed with it at the age of four. My two older sisters just started taking piano lessons and I knew the names of all the notes without even having to look at them. One sister cut the notes out of the John Schaum green book and taped them to the notes. I just looked at the pieces of paper, listened to the notes and memorized them. I also walked up to the piano after their lessons and played every song they played totally by ear. I think any 3 or 4 year old could probably train himself to do that if he really wanted to. A parent should put the names of the notes on the middle keys and have the child listen and look. I liken that to learning how to read.

Also, I can tell you 4 and 5 note chords just by listening. Or if someone asks me to sing a particular note, I can sing it on cue. My niece is the same way-BUT, she can't play any instrument by ear, like I can. She needs music in front of her

I was an augmented 4th off. This explains a lot....

Interesting. I misidentified as A rather than G#, but I'm accustomed to playing recorders tuned to A427. A415 is perhaps more accurate historically, but still a little dull to my taste. Besides, I can get up to concert pitch from A427 by overblowing a trifle; from A415 I have to transpose.

I suppose the conclusion is that my absolute pitch is not good enough to resolve quarter-tones reliably. :)

By Another Kevin (not verified) on 05 Dec 2007 #permalink

hey im only 13 years old and i knew it was g#-well im playing the piano-so doas it mean i have absolute pitch?

Hey Im only 13 and I knew it was G#
Does it means I have absolute pitch?

I have absolute pitch for notes played on string instruments. I played the violin and piano from kindergarten. When I was in high school, I started playing the French horn. This instrument is not in concert pitch, so when you read a note on a musical score and play an 'F' on the horn, the sound that comes out of the machine is the B-flat below. This tripped me up horribly, because I knew perfectly well that I wasn't making an 'F', but I couldn't get a the sound I expected to produce out of the instrument with the fingering we were supposed to use. I finally untrained my absolute pitch for the French horn, but the result is that if I hear a note played on the cello, viola, violin or piano anywhere in the two octaves that form the natural range of a violin, I know exactly what note it is, while if it's played on a trumpet, horn, flute, clarinet, or similar instrument, I don't have a clue. I suspect that, with the wind instruments, I'm getting two answers, one in concert pitch and one in F, and I can't disambiguate them, since both answers are correct, depending on context.