moving through level 4

Comments and questions about Absolute Pitch Painter
aruffo
Site Admin
Posts: 1694
Joined: Tue Dec 14, 2004 12:09 pm
Location: Evanston, IL

moving through level 4

Post by aruffo » Tue Apr 08, 2014 10:29 pm

...and boy is it tough. Not because it's difficult, but because I keep letting myself take guesses when I'm unsure. Because there is only one essential characteristic of a tone (its frequency), the only way to recognize its category with full confidence is to compare it to other category members... and I keep blundering ahead and guessing instead of waiting and comparing.

Once I know I'm wrong, of course, I know exactly which category it belongs to, and so I'm able to drag it to the correct place without learning anything. So instead I try to lose track of my mistakes so that it will surprise me all over again when I find it.

I'm going to write this soon on the Strategy & Statistics page, but here's a quick brief of how to interpret your stats screen (the pitch stats, not the level-up stats). In a nutshell: a steep dropoff at the category boundary represents a strong category. The gentler the slope, the worse you are perceiving that category.

Here are my results for Level 4 so far. I drew red lines so you can see that my perception of the blue/grey upper boundary is very weak, but my perception of boundary between brown and yellow/red is quite strong... which shouldn't surprise me, because that's the first boundary I learned in the game.

Image

It looks like generally, I'm better at lower boundaries than upper boundaries. I think this is mostly because, when I let myself guess, I get suckered by pitch height. The blue/grey I think of as "lowest", so if it doesn't sound low enough, it must be brown. The orange I think of as "highest", so when I hear a higher tone, I don't always stop to make myself perceive the green-category quality, and I wrongly label it "orange". The yellow/red is a little different, but I still get suckered by height. In the Level 3 thread, some folks mentioned that there seems to be an unusual timbral quality to C. I notice it as sort of... well, "plunky", and when I hear that quality I know that it's the red/yellow category. The upper part of the red/yellow category, though, does not show that "plunky" quality quite so obviously-- so if I don't immediately detect that plunkiness, and I can tell that the tone is nonetheless higher than the regular plunky tone, then I'll wrongly guess green without making myself stop and detect the weaker plunkiness.

The secret to the whole thing seems to be, as I think I've said before, not guessing. When I let myself guess, I get suckered by height and ambiguous tones-- I give tons of wrong answers, and I wonder if I haven't learned anything after all. When I don't guess, but put the uncertain ones to the side, I'm thrilled at how many tones are totally obvious, and I get both high scores and renewed confidence.

Angelman
Posts: 20
Joined: Thu Oct 04, 2012 10:07 pm

Post by Angelman » Thu Apr 10, 2014 12:31 am

I've been pondering the questions you posed such as "what would happen if we introduce a tone outside of the categories?" and also noticed the problem of height fooling us in the game.

I wonder if the introduction of octaves in the game would force the listener to further refine the categories. A high sounding egg would be marked wrong if it belonged in the blue category and then the listener would have to rethink his strategy.

Also, I often find that if I play in such a way that I'm in a 'zone', I lose sense of height in the pitches. I made mistakes this way because I couldn't discern the height of pitches and realized I was using height to help me paint the eggs.

aruffo
Site Admin
Posts: 1694
Joined: Tue Dec 14, 2004 12:09 pm
Location: Evanston, IL

Post by aruffo » Thu Apr 10, 2014 8:13 am

Introducing octaves will have to come at some point, to be sure. I didn't include them initially because I tried using them and I was unable to play the game competently. I might've been able to pick it up if I'd really kept at it, but it was so difficult to begin with that I didn't want to keep trying.. and if it was that discouraging to me, it surely would've been more so to others.

I've been contemplating the use of height to make judgments.. I think that'll end up as my next main-page update.

aruffo
Site Admin
Posts: 1694
Joined: Tue Dec 14, 2004 12:09 pm
Location: Evanston, IL

Post by aruffo » Thu Apr 10, 2014 4:06 pm

And it did end up as my next main-page update.

I'm idly contemplating whether it's desirable (or even possible) to find a way to thwart height judgments. Even though it is one hell of an effort to avoid them while playing the game-- I'm three dozen games into the Large size, and I still have only two scores over 2000-- the game is doing exactly what I had hoped it would do. Namely, you can not win the game if you use traditional ways of listening to tones (either height or intervals). There is no way that I can see to "cheat" at the game that doesn't guide you right into the learning process. You've got to make categorical comparisons or those off-tuned notes will get you every time.

To that end, I'm not sure it's a good idea to find a way either to block or confuse height judgments. I suspect it's more important for your brain to be convinced, by this task, that chroma judgment is superior to height judgment, even when height is obvious and easy to hear.

Axeman
Posts: 146
Joined: Wed Jul 20, 2005 4:20 pm
Location: New Zealand

Post by Axeman » Fri Apr 11, 2014 5:21 pm

Isn't the fact that we are using height judgement proof that the game is not inducing us to use chroma.

I wonder at the introduction of new categories. Since they cross with former categories isn't the game continually breaking down old category boundaries in our hearing (I'm thinking about the nerons or whatever at brain level) and rebuilding them at different places.
It seems a bit like teaching an infant colours by only using the three primary colours and then trying to later introduce the secondary ones.
Should the process of learning be so radically different to what a child goes through while learning music as a youngster being introduced to a set of categorical tones.
I suppose there is the scale degrees issue. If all of the categories were introduced at once it would be like going from out-of-tune piano to in-tune- piano to out-of-tune-piano ad infinitum in the real world. Isn't this similar to the experience of a young child learning to hear in AP? Also similar to the idea of a normal distribution of notes.
I suppose the issue for me as I play is that the former upper category (im at level 2) is now broken into two so most of my mistakes feel like I am identifying a note as belonging to the former category but now it belongs to a different one.

aruffo
Site Admin
Posts: 1694
Joined: Tue Dec 14, 2004 12:09 pm
Location: Evanston, IL

Post by aruffo » Fri Apr 11, 2014 8:10 pm

Isn't the fact that we are using height judgement proof that the game is not inducing us to use chroma.
Yes, it is. I'm considering that a temporary flaw. It's necessary to use chroma to succeed at the game, but the game doesn't teach you to listen for chroma. So for now, it's necessary to play APA and learn to hear chroma from that. It is definitely a failing of the game that I have to really, really convince myself to not guess or else I can't win-- the game mechanics should make it easy to follow the best path to learning-- but as we play and discuss our experiences, this leads toward the design of ETC v7.1.

The learning process must be different from how a child learns, or else we could all learn AP fairly easily. The category splitting seems, at present, to be the most reasonable approach, because it accommodates the binary decision-making that our minds tend to do-- starting with broad judgments and then refining.

[Are you using an earlier version of the game? The Small and Medium games should start you out with only the two new categories.]

Angelman
Posts: 20
Joined: Thu Oct 04, 2012 10:07 pm

Post by Angelman » Fri Apr 11, 2014 8:39 pm

aruffo wrote: It is definitely a failing of the game that I have to really, really convince myself to not guess or else I can't win.
I actually thought this was on purpose, if you can never win through guessing, wouldn't that be motivation to pay closer attention? For example, when I don't reach the target score, I realize I have to slow down and take the time to listen to each egg carefully.

aruffo
Site Admin
Posts: 1694
Joined: Tue Dec 14, 2004 12:09 pm
Location: Evanston, IL

Post by aruffo » Fri Apr 11, 2014 9:15 pm

Well, yes and no. The yes part is that what you say is absolutely true-- that making an error encourages a player to slow down, listen carefully, and compare. The no part is that the comparison isn't an automatic part of the game. In Absolute Pitch Avenue, you are forced to compare sounds, because that's the way the game is structured. In Absolute Pitch Painter, comparison is a strategy that you can learn to use. A better game would be one that deliberately and unobtrusively (even invisibly) guides you into the winning strategy.

I also had another thought in response to your noting that APP doesn't teach chroma listening. This is exactly what I was hoping that APP would achieve: to show us what does and doesn't work. I did not know, prior to designing the game, how-- or why-- chroma and categorical perception worked together. Now, thanks to APP, I do know. What to do about it, I haven't yet figured out, but it's a pretty big conceptual leap that I'm quite pleased to have made.

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

Post by abminor » Sat Apr 12, 2014 5:14 pm

I'm only at level 2 now so I can't express a definitive opinion but subjectively I would say I use height judgments as part of my strategy but not only. I mean because I'm dealing with only 3 categories I feel I mainly become sensitive to the middle category. If a tone doesn't fit the category I usually tend to paint it according to its height relative to the middle category. Of course this strategy will become useless in upper levels as the number of categories is gonna increase.

As I play through the game, the middle category becomes "engrained" in my short term memory so I can use it as a reference to sort all the tones. I'm don't feel like I have it in my long term memory yet. I hope it can happen with time. Yet I was surprised by the level up statistics because one day elapsed since i played the game and yet, they showed a tendency toward the ideal slope chris mentioned (except at the new boundary between the two new categories). So maybe i developing long term memory of those categories without realizing it.

About the scale degrees judgment, I would say it's possible I unconsciously use but if I do I must have a precise reference tone in my head to avoid being thrown off by the constant tuning variations. What I can say is that when I try to use it explicitly I tend to make mistakes so I avoid it now.

I have altered my playing strategy to eliminate guessing as chris suggested because starting at the large mode, I began failing too many games, so now I delay my answer for tones I'm unsure of and I place them between two buckets. I have not failed a game since I use that strategy so it seems helpful.

I must admit that because categories are now so broad, I have trouble assigning one color to each of them. They can have several colors really (as the buckets show). I, maybe, naively think that it will become easier for me when the tone span of categories will be be narrower.

If this game fails to teach me absolute pitch I will have at least considerably improve my sensitivity to pitch variations as i can see the upper score in my level up stats has considerably increased :-) It s something I always was bad at because I always tended to consider the tones more similar than different or the commonality between them seemed more remarkable than the differences let's say.

Axeman
Posts: 146
Joined: Wed Jul 20, 2005 4:20 pm
Location: New Zealand

Post by Axeman » Sat Apr 12, 2014 7:15 pm

What the game is showing is the blue and grey bucket, then two more one red and yellow the other green and orange. This is on the small level. have i got the right version Chris?

As to the splitting of categories - do they splitt in the middle to the former category boundaries or accross them?

aruffo
Site Admin
Posts: 1694
Joined: Tue Dec 14, 2004 12:09 pm
Location: Evanston, IL

Post by aruffo » Sat Apr 12, 2014 8:51 pm

The Small and Medium games will show all the buckets, but will present you with eggs that use only the two newest categories.

The new categories split down the middle of the old ones.

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

Post by abminor » Sat Apr 12, 2014 11:14 pm

I 'm wondering. Shouldn't the game be more tolerant regarding tones located at the boundary of two categories ?

I mean if a sudden drop of accuracy at the boundary if considered normal for categorical perception, I think missing thoses tones should have little impact of the score. One should not fail because of those otherwise maybe we gonna end up training a different skill that categorical perception (more like a boundary perception).

aruffo
Site Admin
Posts: 1694
Joined: Tue Dec 14, 2004 12:09 pm
Location: Evanston, IL

Post by aruffo » Sun Apr 13, 2014 12:52 am

I suspect the answer to your question would be made clear by changing the emphasis from one word to another. The important phenomenon at the category boundary is not a sudden drop, but a sudden drop. As you move through category A, everything sounds like A, and then BAM you're suddenly hearing category B.

The only reason there's a drop at all is because items that are right at the border have qualities that could be legitimately construed as either A or B. Items on the A side have more A-quality than B-quality, which is why they are classified as A (and vice-versa for B). When the qualities are almost the same in strength, it's difficult to tell which one has more... but it is possible, and the ability to recognize which quality is stronger is a hallmark of categorical perception.

To put it another way, categorical perception is not categorical perception unless you have a sharp sense of the boundary. The game's intolerance of the boundary errors can be incredibly challenging-- I only just this moment qualified to level up to Jumbo on Level 4, and let me tell you I was sweating bullets toward the end of it-- but it forces you to make those fine distinctions and figure out where the border is. If the game were more tolerant of border errors, then it wouldn't be necessary for your mind to really scrutinize those border tones and figure out what aspects of the tone truly represent its category.

aruffo
Site Admin
Posts: 1694
Joined: Tue Dec 14, 2004 12:09 pm
Location: Evanston, IL

Post by aruffo » Tue Apr 15, 2014 1:14 am

Geez. The mechanics of how the game scores were not something I spent a great deal of time on-- I just came up with a scheme that seemed fair and appropriate to the task-- but wow does it seem to be on target.

I just cleared 4,000 points on the Jumbo game. Going into the final 40 eggs or so, I had a huge run of 10x/11x/12x eggs, so I let myself relax a bit.. and all of a sudden WHAM BONK SPLAT I was back down to 2x, and was no longer certain of victory! I had to pay ever-so-close attention to the sounds because one more slip-up could mean missing the mark. And I didn't... but it was a darn close shave.

So it's a big plus, I say, that the game's scoring system keeps me paying attention right to the end.

Axeman
Posts: 146
Joined: Wed Jul 20, 2005 4:20 pm
Location: New Zealand

Scoring system

Post by Axeman » Tue Apr 15, 2014 11:47 pm

I know you've posted somewhere before RE the scoring system for APA Chris. Is it the same for Painter? I seem to remember something about an accumulative increase in score for the number of correct answers in a row. Is that right? If that is true it will explain some of the low scores I get at times.

I tend to agree with the comment above about the near-category scores since at times in certain games if the tones that are being generated randomly happen to ask you to often distinguish between the tones around the fuzzy category boundary then you are being asked to perform a harder task than for games that have more mid category tones.

I see your point about that being the whole point though. I tend to have developed a strategy that retains the tone nearest to the boundary in my mind as I listen to other notes and then any that are ambiguous I end up comparing to this memorized note. This seems to be a RP technique but is often the only way to get through a level. And yes it is very frustrating. What if the game had a bonus round where there was a normal distribution of tones generated over the near-category boundary much like old video games that would get you to position your space ship to exact firing points during bonus rounds (I am thinking of the old game called Galaga) thus developing a more targeted focus for short bursts.
When I am playing Painter I get a kind of chroma sensation that serves me well enough for most notes until I get to the near boundary notes. The top category has a sort of twang to it and the bottom some other quality. This tendency is at least encouraging.

Post Reply