come in number nine

[aruffo]

Now on the Large games for Level 9. Let's see the Level 8 (Jumbo) and Level 9 (Medium) discrimination scores back-to-back. The new categories are the red and red/yellow (rellow?) split.




The reason for looking at these back-to-back is that the Small and Medium games of Level 9 are, of course, only red and rellow tones. Therefore I would likely expect that the discrimination results would remain essentially the same for all but the new categories. I haven't run any stats tests to determine whether these results are significantly different from each other in any respect, but a glance suggests that the higher categories got a little better and the lower categories a little worse (blue remains suspect due to the same end effects).

I decided to again let myself relax on the Large games, which again made a royal hash of my identification stats, but gave me some interesting insight. Now that I'm working with more than seven categories, when I just let myself relax, my mind will quickly and automatically default to a height strategy. And if I just sort-of pay attention, I will rely on height for many of the tones.

So I decided I would tell myself that I couldn't remember all 10 categories currently in play. Rather, I would let myself recognize seven of them: blue, grey, brown, red, yellow, green, and blorange. And then, whenever I heard grey, red, or green, I would make a further decision as to whether the tone might be grue, rellow, or orange, respectively. And this strategy netted me the highest score of my six games so far.

[aruffo]

Well, that's level 9 completed. And here's what the identification stats look like:



..which is an appropriate summation of my experience. Border tones are the major culprit, as they should be, and certain border tones fool me every single time (the one between grey and grue is particularly nasty, and I keep second-guessing myself on the border between brown and red). When I hear either rellow or yellow I can't always tell which one I'm hearing until I find the other one. I know that rellow has a certain "bounce" to it that yellow doesn't, but I frequently need to hear the sound without the bounce before I can confirm that I am in fact hearing the bounce in the first sound.

Even so, with the eggception of rellow and yellow, I'm pleased at how confidently I can identify the non-border tones. The only thing that makes the game difficult is carelessness with the border tones; if I play carelessly, then I seem to stumble on border tones with just enough frequency to keep myself at about 4x, and the multiplier needs to stay at 7x or higher to reach a "passing" score. So I still have to play a strategy of not guessing, which on the one hand is frustrating because I have to be so careful and patient, but on the other hand is encouraging because I can identify so many without guessing.

Here are levels 8 and 9 for discrimination:




I thought perhaps the results I've been getting for the past couple of levels have been because I made the task too hard-- tones were only 12¢ away from each other, and the just-noticeable difference can be as high as 15¢-- so I widened the differences here to 18¢. The results of the wider differences, I think, are somewhat more representative of how well I hear each category, because the obviously weak ones are rellow and yellow, which are indeed my worst. I'm not entirely sure about red and orange, though, and blue and blorange may yet be end effects.

On to Level 10!

[EvilBlade]

Nice

I was wondering, do you have any new theory about the possible "magnet vs. anchor" situation?

I'm still on level 5 at the jumbo games, and after some initial difficulties on the Large games, I'm proceeding through them pretty easily with good confidence in the tones identification. But, as usual, my level up stats show the opposite situation of the ideal categorical perception, as they look like this for most categories.

Level 5


Also, for example, the brown category had a different graph in the previous levels, as it became "inverted" like this only in the last two level up procedures, for medium and large.

I also want to say that experienced some spountaneous positive identification notes in music, especially of C and E, but I'll wait to get a clearer idea of what it is like before saying anything else.

[aruffo]

No new theory, just yet, except that the experiment I solicited in another thread is essentially complete (I may or may not need to recruit a few more relative listeners, but the results won't change even if I do) and it shows that, with certain sounds, the "ideal" results may be reversed.

So although I would expect the center-middle-cross "ideal" categorical perception to be of increasing sensitivity, a large part of the reason I'm gathering data now is to find out whether that should or should not be the actual case here.

At the very least, it's interesting that your middle tones-- the ones with, presumably, the weakest cues to category membership-- are mostly the ones you find easiest to discriminate.

[Nikolaus]

hey Aruffo,

after reading your own writing, i put two and two together and decided to reclassify my own perception as a form of absolute pitch.

let me explain: in the timbre of my own instrument, and in the central octaves, i've developed broad categories that stretch across certain semitone combinations. for instance, C# and D, or Eb and E. this being the case, i'd confuse D for C#, but never for an Eb. (that'd be like seeing brown, and thinking that was yellow! yes they're that different!)

these broad categories fit all of the definitions of absolute pitch, and would survive all of the tests. so no matter what key or tonal combination, i'd never confuse them. ever. another interesting thing is that i don't need to compare them to know what they are. so for instance if i hear the broad category of Ab and A (red), i just know, and i don't need to compare it to the F#-G category (green), or the Eb-E category (yellow).

another interesting thing is that these broader categories have lighter and darker versions of themselves in the other octaves. so right now all of my observations pertain to the fourth octave in this particular timbre, but the Ab-A category definitely has a darker "red" version of itself in the third octave, as does the F#-G category (a much darker green)

in tests, this would show up in the numbers as constant confusion of certain semitone combinations. so change the timbre on me, and this pattern would disappear. i'd be making semitone errors for sure, but they'd be due to height again. (maybe... or just broader categories that haven't developed yet).

anyway... what brought me back to absolute pitch studies has not only been your activity as of late, but also because about one month ago i listened to a piece of music in this timbre i mentioned, and i was blown away by how colorful it sounded to me. i mean, this is both practical (when refined by relative pitch), but even more than that, aesthetically pleasing. like i'm hearing color or something! but change the timbre, and everything's dull again, and i have to resort to relative pitch.

anyway... i never made any conscious effort, other than the ear training i was doing, to develop these categories - they just kinda emerged on their own as i did the drills.

having said that, and to be fair, some of the categories have become even tinier. so for instance, the "random cadence, random tone" drills forced me to start hearing the differences in the B-C category, and it broke up. same goes with the C#-D category, as well as the Ab-A category. Eb-E i can get right on a good day, and i can certainly hear the difference, but that reflects more in the numbers than in my perception. so far as i am concerned, even one error out one hundred can expose the weakness of my perception - that's how strong it is for the other categories that have developed, and that is my standard. and then there are those renegade pitches, that always just had categories of their own - Bb, for instance, and F. and sure i could attribute this to their unique sounds, but it's also because the neighbors on either side of each just sounded too different to cause confusion. so for instance, with F, the Eb-E set was just too yellow, and the F#-G set too green. as for those categories that did break up further, i see them like brothers - in one sense the same, but also different. and it took a lot of time, as well as comparing, to finally begin reliably distinguishing them. like you said (i think)... C# just seems to have this "oomph!" to it that D does not have, but only by comparing them are you able to hear it - perhaps you continually remember the more obvious quality that makes them similar, but forget the more subtle one that makes them different.

your listening strategies and experience with APP reminded me so much of my own experience with the "random cadence, random tone" drill that i had to mention it. for instance, it has also proven too difficult (at first) to generalize the sound of a note across different octaves, and different timbres. - each new variation of that note needs to be tackled one at a time, at least until one's perception is developed further. what do i mean by this? one timbre at a time, and add on notes only a few at a time.

i see all of this as good news, because even though my perception is unrefined, it is not only practical, but aesthetically pleasing to the ears. and when combined with my relative pitch, a great boost to my perception.

and i know that we might have a tendency to get discouraged because we are dependent upon the sound of a timbre to make accurate identifications, but so are a lot of other true APers. let us not forget that!

[aruffo]

A fascinating description. I'm particularly intrigued by the combinations you mention-- now that I'm dealing with 11 categories, they're joined for me in the same way as they are for you. Ab and A, C# and D, Eb and E. (I suspect that the reason F# and G are not joined is because they're at opposite ends of the height spectrum right now.) When I make a snap judgment, without careful consideration and comparison, I'm as likely to misjudge one for the other.

I wonder how your suggestion that we remember the obvious similarity but forget the subtle difference would fit in with the unavoidable magic number seven. I knew, even before I designed APP, that it is not feasible to expect that we could actually learn 12 pitches, any more than we are capable of learning more than seven colors. Being able to identify more than seven of anything is a multi-step process of broad to narrow categorization. With twelve pitches in a single octave, it's a two-step process:
- decision one: semitone combination or renegade?
..if it's a combination, then
- decision two: which of the two?

Right now I'm pretty good at nailing the lower six categories-- blue, grue, grey, brown, red, and rellow (which is G through C#)-- which takes up six of my seven. Unsurprisingly, therefore, when it comes to everything above C#, I readily and easily lapse into a height strategy, which is of course a failing strategy.



It appears as though my decision process, when I'm not being careful, is this:
- decision one: renegade or above-rellow?
...and if it's above rellow, then
- decision two: where does this go into the upper range?

When I am being careful, it's
- decision one: do I know this tone?
- decision two: could I be mistaking it for another tone?

At that point, what you're describing about the obvious/subtle qualities kicks in. With decision one, I can usually make a clear judgment of any of the lower six categories, or at least recognize whether a tone belongs to C#/D, Eb/E, or F#/F. Then, once I find another one to compare, I can make a confident decision. Curiously, I can also often make a confident decision by finding another identical tone; the second one seems to confirm the first somehow.

And, when I'm using this strategy, I am still using seven divisions-- I become more aware of having to make a two-step decision for Ab/A, making the divisions 1. blue, 2. grue/gray, 3. brown, 4. red, 5. rellow/yellow, 6. grellow/green, and 7. orange/blorange.

It's difficult to prevent myself from using the former strategy (lower categories, higher range) in the Large games, though, when there are this many categories. Because of the randomly-weighted way in which the game chooses pitches, there are far fewer border tones than in the Jumbo games. This means that, for the six lower categories, I can usually make accurate snap judgments.. but then, when I allow this, it leaves only the one mental division for the entire upper range.

Still, I'm pleased that I cannot win the game if I persist in this strategy, so gradually I'll persuade my mind to use the second strategy and eventually succeed. It's still way too early to tell if this game is "working" to deliver absolute pitch ability, but I do appear to still be learning and progressing, so I'm keeping at it. (Once I get through Level 11 I will have to do another major change to allow new octaves.)