UCSF Study

General topics and issues.
Space
Posts: 179
Joined: Mon Jan 16, 2006 5:54 am
Location: Cincinnati, OH

UCSF Study

Post by Space » Thu May 02, 2013 7:17 pm

Ok, so I was reading about the UCSF study here: http://perfectpitch.ucsf.edu/study/ and thinking about one possible fatal flaw.

The researchers only did genetic testing on AP possessors and their AP possessing family members. Basically, it's like taking a group of people that can juggle and comparing their genes to see what is similar and what is different and finding something and assuming that because all jugglers have 'x' gene in common, then that gene must be responsible for their juggling ability. Then, failing to make the realization that MOST people have this gene, even non-jugglers.

I'm probably completely off and don't know enough about the protocols of such research to judge but what do you think about it, Chris? I really wanna understand this stuff!

Rob

aruffo
Site Admin
Posts: 1700
Joined: Tue Dec 14, 2004 12:09 pm
Location: Evanston, IL

Post by aruffo » Thu May 02, 2013 11:12 pm

I don't know too much about it myself, but there are certain problems that go right along with what you're saying. The less-convincing argument, that persisted for a very long time, was that it must be genetic because it ran in families.

An argument the geneticists advanced some years ago, which seemed to me a colossal cop-out, was that AP was genetic but had to be "activated" through learning. In other words, if you learned AP, that meant you already had it.

The newest stuff, though, I think, is a bit more plausible. As I recall, yes, they gather people whom they believe to have AP, and compare their genetic material-- but, having located commonalities, they don't assert that the common gene(s) must be responsible for AP. Rather, they observe that absolute listeners share this particular gene, and (if they're responsible) make no causal attribution; the more interesting action, which may be what they're doing, is to recognize what other phenomena have been associated with that particular gene (or genes) and speculating about relationships between the two. F'rexample, if the AP-common gene were also associated with high language ability, or autism, or.. well, or whatever, then that might give us insight about what AP is related to. In short, it wouldn't be necessary to assert that AP "is" or "isn't" genetically bestowed, because the purpose of the research would be to find clearer ideas about what other abilities AP is most nearly like.

And I can get behind that.

Space
Posts: 179
Joined: Mon Jan 16, 2006 5:54 am
Location: Cincinnati, OH

Post by Space » Fri May 03, 2013 2:04 am

Well I can certainly get behind any decent, sound, research myself, whatever the outcome. However, the issue of AP seems to come with a lot of strange baggage and it always seems like there is a 'belief' about the nature of AP that is driving the researchers. Much like Rush and Nering's studies where results seemed to be skewered in favor of Burge's methods, I feel that genetic studies are driven by folks that are convinced that it IS genetic and are looking for results that serve their agenda.

Maybe that's not what's happening, though. Maybe it's a matter of media grabbing hold of something like the UCSF study and applying it to the 'nature/nurture' debate. As soon as these genetic studies started coming out, there are suddenly a nice little list of news articles (if you google "absolute pitch genetic", a bunch come of) that have that bent of "it's all in the genes", etc.

I mean, of course it's all in the genes. Without genes, we wouldn't have auditory processing apparatus to begin with. The point, though, is there are people like this guy I couldn't seem to stop combating with that was so vehemently against the notion that AP can be developed to any capacity. He goes on to claim that I'm fooling myself, wasting my time, and I'm some horrible person for encouraging other people to put effort into AP development themselves.

His interpretation of the UCSF findings is that AP is purely 'genetic'. If you don't have the genes, you'll never have any of it and you're wasting your time trying to develop it, yatta, yatta.

I'm an idiot to let these types get under my skin. I mean, my experience with AP development has been plenty successful in terms of what the AP-like abilities I've acquired have done for my musical advancement. So I'm not gonna stop working on it simply because some random no-name dude wants to try to take the wind out of my sails.

But there's the issue of everything that you've put into your own research and the development of ETC and all of that. I mean, it's been extremely valuable to people like myself and he's basically saying that AP is genetic, it's a wrap, a done deal. So why continue researching it and particularly why continue trying to develop ways to aquire it?

This is why I think it'd be incredibly beneficial to put together a rigorous sound study of AP development that doesn't cherry-pick, takes into account all participants equally, and involves PET scans and genetic testing pre and post 'development'. It would be interesting to put participants through a rigorous AP development program that lasts something like a year or so, involves a sound method for that development, and just basically kicks arse overall.

Yes, I realize the funding for such a study would probably be enormous and it'd be hard to find willing participants that would put themselves through rigorous daily eartraining for an entire year, but maybe someday!

Rob

aruffo
Site Admin
Posts: 1700
Joined: Tue Dec 14, 2004 12:09 pm
Location: Evanston, IL

Post by aruffo » Sat May 04, 2013 10:05 am

I was surprised when I read the geneticists' research.. its language is more speculative than the public buzz. The fact of the matter is that it is not possible-- literally, not possible-- to demonstrate a causal link between AP and any kind of genetic source without performing terribly unethical procedures (e.g., deliberately breeding people with and without AP, and then keeping their offspring isolated from music).

So if someone asserts an unequivocal claim based on the genetics research, the only fact there is that they don't understand (and probably haven't read) the actual research.

Conversely, though, there's no hard evidence on the other side, so there's no way to persuade anyone-- it's just he said-she said, and opinion wins the day. So it's frustrating, but better to put one's energies toward finding a way to get that hard data.

[Which reminds me that I recently discovered that Realbasic has a built-in normal-distribution number generator-- ETC v7, when it appears, will have to be for mobile devices, but Realbasic can be a good way to test the hypothesis..]

zacxpacx
Posts: 145
Joined: Tue Oct 23, 2012 3:51 pm
Location: United States

Post by zacxpacx » Sat May 04, 2013 2:32 pm

aruffo wrote:Conversely, though, there's no hard evidence on the other side, so there's no way to persuade anyone-- it's just he said-she said, and opinion wins the day. So it's frustrating, but better to put one's energies toward finding a way to get that hard data.

[Which reminds me that I recently discovered that Realbasic has a built-in normal-distribution number generator-- ETC v7, when it appears, will have to be for mobile devices, but Realbasic can be a good way to test the hypothesis..]
Hey Chris, I just wanted a little clarification. Is it your intent to find hard evidence to argue the nurture side of perfect pitch? Since we can agree it's mostly opinion at this point, I'd like to know what your opinion is. I know you've said you think it can be taught to adults in the past.

Also, another thing to think about for all of us: at what point will we deem it a reasonable "next step" to take MRI scans of adults who have undergone training? Once categorical perception is developed? Once one of us is able to ID pitches as well as someone with "natural" AP?

I guess once we're able to ID pitches as well as anyone with AP, it's obvious that taking an MRI would be logical to compare if what we've learned is the same as AP. So the real question is whether or not there's a time before then that taking a brain scan would be useful.

zacxpacx
Posts: 145
Joined: Tue Oct 23, 2012 3:51 pm
Location: United States

Post by zacxpacx » Sat May 04, 2013 2:34 pm

Oh, Space, I know you think it's reasonable to take an MRI test as of now, since you're able to ID notes well.

You might want to try a categorical perception test for pitch first. I think the results of that would be interesting to see.

Space
Posts: 179
Joined: Mon Jan 16, 2006 5:54 am
Location: Cincinnati, OH

Post by Space » Sat May 04, 2013 7:07 pm

How does one go about that? Test for categorical perception, that is?

Space
Posts: 179
Joined: Mon Jan 16, 2006 5:54 am
Location: Cincinnati, OH

Post by Space » Sun May 05, 2013 1:51 am

Also, what difference does it make to test for categorical perception first? The point of the MRI/PET scans is to see what happens when I identify pitches 'as is', the way I do it now. I'm not trying to 'make sure' that I am identifying pitches like a 'natural' before testing. The point is to ID pitches the way I already do it and see what my brain is doing.

In fact, it would be far more interesting to find that there are some pitches where the neural activity is spot on an AP'er from Zatorre's scans, and some where purely RP is involved and some where maybe both are involved.

Hell, maybe there's even a synesthetic element in there somewhere! Not necessarily for color, but shape. I tend to experience the pitches as 4-dimensional shapes and I have graphene-synesthesia, so I experience the note-names themselves as being colored (not the actual pitches, though).

aruffo
Site Admin
Posts: 1700
Joined: Tue Dec 14, 2004 12:09 pm
Location: Evanston, IL

Post by aruffo » Sun May 05, 2013 6:06 pm

I'm sure you're both right, for the reasons you give.

The way to test for categorical perception would be to identify tones that were out of tune. You'd see a consistent accuracy right up to the categorical border, and then a sudden switchover at the division between categories.

A couple alternatives that would be indicative of something other than categorical perception:

Accurately identifying only those tones which are precisely in tune.

Accurate identification that decreases in a slope toward the categorical boundary.

abminor
Posts: 87
Joined: Wed May 23, 2007 7:55 am

Post by abminor » Sun May 05, 2013 11:45 pm

Just a question here,

Do you think there could be a training for AP based on the categorical perception test you mention ?

For example:

1) Guess if an out of tune pitch is belonging to the category of on target pitch. Give yes or no answer. The target tone should be played at regular interval for reference. The interval should increase so that user doesn't rely on short term memory, but memorize the tone instead. Ideally at the end of test, it should not be played anymore.

2) Add one neighbor pitch to the first target tone and test now for 2 categories. Idealy only the first target tone should be memorized now so only play the second target at decreasing interval. The answer should be now which tone category the out of tune tone belongs to: tone 1, tone 2, none ?

etc... adding neighbor tone after neighbor tone.

This idea has probably been talked about in the forum already.

Space
Posts: 179
Joined: Mon Jan 16, 2006 5:54 am
Location: Cincinnati, OH

Post by Space » Mon May 06, 2013 1:44 am

"then a sudden switchover at the division between categories."

Not to say that varying degrees of accuracy with this denote a lack of categorical perception per se but I think with most AP possessors there wouldn't be a 'sudden' switchover. Probably more of a grey area where considerable mistakes are made.

For example, take a pitch that is right in the middle of two categories, and say the categories are B and C. I don't know what the figure would be but say the grey area is 20 cents from the middle of either side, so there would be a 40 cent wide grey area where pitches are consistently mislabeled as either B for C or C for B.

Oh and I was just reminded of this little bit from the UCSF study:

"absolute pitch possessors tend to err on G# far more than any other tone, an error that occurs only on pure tones. Most often, G# pure tones are misidentified as "A" tones. We hypothesize that this phenomenon reflects the use of A as the universal tuning pitch in Western music. Since the actual frequency of A used in tuning varies widely, from A415 in early music to A446 in some orchestras, we suspect that absolute pitch possessors accommodate a wide range of frequencies in their naming of A. We further speculate that this accommodation is not used in piano tones, since pianos are generally tuned to A440. This phenomenon is reminiscent of a property referred to as "perceptual magnet" in language acquisition."

Also, I know this has come up before and it doesn't really suit what we're talking about precisely here, but I strongly feel there is a link between AP and the harmonic overtone series. I don't feel that AP'ers simply categorize fundamental frequencies. However speculative this may be, I believe that the 'chroma' that they are perceiving is based on a full overtone spectrum for each 'pitch' so that when a pitch is sounded and identified, they are responding to not just one but a sort of 'family' of specific 'nodes' along the basilar membrane. Each 'family' of nodes is unique to that 'pitch' and gives the subject a group of neural links to categorize.

This would explain a number of various 'levels' of AP perception that we see among various possessors, for example:

- AP'ers have a harder time identifying pure tones

- some people possess AP for only one instrument or group of similar instruments (in fact, there was a guy, Glenn Drodge who managed to develop AP for HIS specific brand of electric piano and was hopeless in trying to carry this over to other timbres, even other pianos!)

- Some people possess what might be called 'key' AP or 'piece' AP, which is the ability to identify keys but a lesser or non-existant ability to identify specific pitches

My feeling (again, speculative but I believe carries some genuine merit) is that it's a matter of memory 'tags' and being able to streamline these memory tags to the lowest number necessary to form the category.

For someone with 'key' AP, they require a much greater amount of information to form the categorical judgement of 'key'. A single pitch doesn't carry enough information to stimulate the neurons dedicated to that category to make a confident judgement of pitch.

For someone with 'full blown' AP, it seems though pure tones can be 'tricky', they still do ok (hell, when I took the UCSF test, I scored 35.5 out of 36, so my pure tone score had to be pretty decent).

Of course, how this really works in terms of brain structuring and neural networks, who the heck knows. Maybe the full-blown AP'er simply has more neurons dedicated to each category, therefore requiring less information to 'trip' the accurate judgement alarm.

I truly believe, though, that there's a relationship between all forms of absolute pitch identification, from 'key' to 'full-blown' (I know my terms for these things are so very scientific ). :P

Absolute judgement of pitch falls along a spectrum and there is not just 'one' kind of 'one' level of the ability. Afterall, the UCSF researchers did have to break the AP'ers up into 4 categories based on scores. So clearly there was a spectrum of performance even among those considered genuine AP possessors themselves.

Space
Posts: 179
Joined: Mon Jan 16, 2006 5:54 am
Location: Cincinnati, OH

Post by Space » Mon May 06, 2013 1:49 am

Oh and as far as testing for categorical perception using out of tune notes, Prolobe is supposed to have advanced levels (called 'ear manglers) that involved judgements of quarters and thirds of pitches, but apparently they aren't operable.

Oh, and Clint hasn't been heard from in ages. The site was down for years, then he got it back up again and it's still there, but HE seems to have disappeared!

Space
Posts: 179
Joined: Mon Jan 16, 2006 5:54 am
Location: Cincinnati, OH

Post by Space » Mon May 06, 2013 3:01 am

Man, sorry bout all of these long-winded posts but I forgot about another thing.

Chris, you mentioned ETCv7 and the random number generator. Is this to incorporate randomly selected 'out-of-tune' pitches?

Rob

zacxpacx
Posts: 145
Joined: Tue Oct 23, 2012 3:51 pm
Location: United States

Post by zacxpacx » Mon May 06, 2013 7:47 pm

Space wrote:Not to say that varying degrees of accuracy with this denote a lack of categorical perception per se but I think with most AP possessors there wouldn't be a 'sudden' switchover. Probably more of a grey area where considerable mistakes are made.

For example, take a pitch that is right in the middle of two categories, and say the categories are B and C. I don't know what the figure would be but say the grey area is 20 cents from the middle of either side, so there would be a 40 cent wide grey area where pitches are consistently mislabeled as either B for C or C for B.
No need to speculate there. I'd take a look at Chris's research another time, Space. He may have posted some stuff since you last looked at it however many years ago. There have been studies done for categorical perception of pitch on APers already.

Space
Posts: 179
Joined: Mon Jan 16, 2006 5:54 am
Location: Cincinnati, OH

Post by Space » Mon May 06, 2013 10:04 pm

Nice!

Post Reply