Author Topic: CW/ACW bias study results  (Read 20262 times)

Rosanella

  • Newbie
  • *
  • Posts: 3
    • View Profile
Re: CW/ACW bias study results
« Reply #30 on: March 15, 2008, 02:57:07 pm »
Hello Everyone  :-)

I am new to this forums and have analysed 2280 images so far (probably not as many as others have done) , following the news of GalaxyZoo as it appeared on Astronomy magazines, Astronomy related forums and various websites where it made the headlines. I've been studying Astronomy for not very long, though my passion for the Sciences began quite a few years back, as a very young girl.

As I begun analysing galaxies (naively, I should say, as a complete amateur), it soon became apparent that I could not quite understand why a clockwise galaxy looked anti-clockwise, and viceversa; but it became even more apparent that it wasn't just me encountering this little glitch in the way the brain perceives motion through pictures, so as to generate quite a topic of discussion. By the look of it, I've become part of a psychological study in the identification of images. Well, the field of Science does involve a great deal of creativity, I'm well aware of that and it is probably one of the aspects that draws me to it 8), but this has gone a little too far laterally considering that a simple, yet more thorough explanation could have been added in the tutorial section of Galaxy Zoo with a small imaginative exercise. Thus saving us some time in working out what button to click, and yourselves with the bias results, as I suspect the results have shown quite a few discrepancies.
I can understand that researchers need to assess the accuracy of the analysis in various misterious, or not so misteriuos ways ;) ;  yet, I think that Part 2A ... Clockwise and Anti-clockwise Spiral Galaxies should read along these lines:

N.B. Note how the arms of both kinds of spiral galaxies point the opposite
way to their motion.  If this is confusing you, just imagine your hand holding the
galaxy at the center and rotating it. 
Let's practice with the image of the galaxy spinnig in a clockwise direction. Place
your hand a couple of centimeters from the screen and imagine holding the galaxy with
your fingers at the center. Now, imagine rotating it and notice your hand spinning clockwise with the spiral arms being dragged behind going the opposite direction.
Repeat the exercise with the image of the galaxy spinning anticlockwise. Again,
place your hand a couple of centimeters from the screen and imagine holding the
galaxy with your fingers at the center, rotating it. Notice
you hand spinning anti-clockwise with the spiral arms being dragged behind
going the opposite direction.
Basically, don't focus your attention on the direction of the spiral arms, but
keep focusing at the center of the spin.

                                                                        ********
Last but not least, taking into consideration that analysing galaxies is a very repetitive task ---one can simply get carried away by clicking on the nearest button the mouse cursor is resting onto (I've clicked on the elliptical button more often than I should have, at times  :-[  ), I strongly suggest that the buttons are placed in a different order, in the Galaxy Analysis, with the elliptical button not being larger than the others so as not to focus too much on it (are the Psychologists reading this? :P ).

So, these are my thoughts (and you're free to disagree with them) and hope that the next generation of Galaxy Zoo is going to be re-designed with a few thoughts in mind that take into consideration the simple facts of life: am I turning this thing clockwise or otherwise?   ::) , and then we can all move on.

Thank you for taking time to read this.
« Last Edit: March 17, 2008, 09:22:00 am by Rosanella »
Rosanella
(...don't change my name. It has colour and it has sound. It is what it is.)
If we do not try, we will not know. (Ayya Khema) 8)
[/color]
[/font]

Half65

  • Hero Member
  • *****
  • Posts: 10288
  • "Eagle Nebula"
    • View Profile
Re: CW/ACW bias study results
« Reply #31 on: March 15, 2008, 06:41:31 pm »
Welcome to the zoo.

Rosanella

  • Newbie
  • *
  • Posts: 3
    • View Profile
Re: CW/ACW bias study results
« Reply #32 on: March 15, 2008, 10:01:56 pm »
« Last Edit: March 17, 2008, 09:22:15 am by Rosanella »
Rosanella
(...don't change my name. It has colour and it has sound. It is what it is.)
If we do not try, we will not know. (Ayya Khema) 8)
[/color]
[/font]

elizabeth

  • Hero Member
  • *****
  • Posts: 33783
    • View Profile
Re: CW/ACW bias study results
« Reply #33 on: March 16, 2008, 05:00:34 am »
 ;D ;D ;D Welcome to the zoo. I do hope you enjoy yourself here. ;D

njtm

  • Hero Member
  • *****
  • Posts: 1731
  • The Wonder Down Under :)
    • View Profile
Re: CW/ACW bias study results
« Reply #34 on: March 16, 2008, 10:16:41 am »
Hi Rosanella. :)
"it soon became apparent that I could not quite understand why a clockwise galaxy looked anti-clockwise, and viceversa"
That bothered me when I started analysing, but I got used to it. :) :) :)

Rosanella

  • Newbie
  • *
  • Posts: 3
    • View Profile
Re: CW/ACW bias study results
« Reply #35 on: March 17, 2008, 09:21:09 am »
;D ;D ;D Welcome to the zoo. I do hope you enjoy yourself here. ;D

I hope so too.
Cheers!
Rosanella
(...don't change my name. It has colour and it has sound. It is what it is.)
If we do not try, we will not know. (Ayya Khema) 8)
[/color]
[/font]

mallory weiss

  • Sr. Member
  • ****
  • Posts: 369
    • View Profile
Re: CW/ACW bias study results
« Reply #36 on: May 14, 2008, 06:20:39 pm »
Uh-oh...
Zookeepers alert!

I've been self-correcting for the possible bias the whole time... without even knowing about its existence.
Should I stop doing that?
"At least one child was removed because its image was too similar to a supposedly different child." - SDSS

zookeeperChris

  • Admin
  • Hero Member
  • *****
  • Posts: 723
    • View Profile
Re: CW/ACW bias study results
« Reply #37 on: May 14, 2008, 06:42:04 pm »
It's okay; take your best guess at each one. One of the interesting things is that the bias seems clearly apparent to classifiers; for example, I was sure I was clicking more acw than cw galaxies, and was surprised the total was only a few percent different. One for the psychologists, I think.

mallory weiss

  • Sr. Member
  • ****
  • Posts: 369
    • View Profile
Re: CW/ACW bias study results
« Reply #38 on: May 14, 2008, 08:00:12 pm »
It's okay; take your best guess at each one. One of the interesting things is that the bias seems clearly apparent to classifiers; for example, I was sure I was clicking more acw than cw galaxies, and was surprised the total was only a few percent different. One for the psychologists, I think.

Yes, I always did the best guess... but with spirals... it's a whole different story...

I meant to confess for a while, but felt kind of embarrassed... :-[

Like (I'm pretty sure)  everybody else, I have split-second, immediate reactions to the images displayed: "definitely elliptical,"  "fuzzy blob," "definitely a star/artifact..."
but here's one for the spirals:
"definitely a spiral with discernible direction of rotation... which (the direction) I will determine in just a second" - these are reactions, not conscious thoughts/analyses just yet
Note there's no "definitely CW/ACW" - because it takes another half-second for the conscious analysis to kick in...

The embarrassing part is that I've been always (from the very start) taking 5-10 times longer than usual classifying CW/ACW spirals if the immediate reaction was "spiral with discernible direction of rotation" because I always felt it's important to not report the opposite direction... I stare at the image for a while and sometimes even (more embarrassing still! :-[) move the cursor on the screen in circles to confirm the direction. I caught myself several times wanting to report the wrong direction and then correcting myself after staring at the image a bit longer... I always thought it's only me, and no one else takes any extra time to double-check the direction. Also, I've never used SDSS links to verify the direction of rotation.

So... in the nutshell:
1. I think it takes a much-slower conscious, analyzing part of the brain to tell the actual direction of rotation, whereas most other information (pattern recognition) can be classified by the lightening-fast "reactive" part (supposedly primary visual cortex*) which just might favor ACW... ???

2. I hope I wasn't going against the grain of the whole bias study by self-correcting from the very start...  :o

_______________________________
* here's some evidence...
According to this source (http://webvision.med.utah.edu/VisualCortex.html#response%20timing) Primary visual cortex (V1) isn't very skilled in discerning the direction/orientation:
"While the vast majority of V1 cells show some degree of orientation selectivity, only approximately 25-35% of V1 cells are strongly directionally selective" whereas the mediotemporal area (MT) has many Meynert cells which are "highly direction selective" ... 


I wonder if the reason for bias is that many users click the classifying buttons before the visual signal arrives at their MT area... ::) ;D

Edit: fixed 1 typo  :)
« Last Edit: May 15, 2008, 12:50:08 am by mallory weiss »
"At least one child was removed because its image was too similar to a supposedly different child." - SDSS

ElisabethB

  • Hero Member
  • *****
  • Posts: 13200
    • View Profile
Re: CW/ACW bias study results
« Reply #39 on: May 14, 2008, 08:06:53 pm »
Hey Mallory,
You are really not alone ! I too make the appropriate motions on the screen more often than not.
E.

mallory weiss

  • Sr. Member
  • ****
  • Posts: 369
    • View Profile
Re: CW/ACW bias study results
« Reply #40 on: May 14, 2008, 09:13:27 pm »
Hey Mallory,
You are really not alone ! I too make the appropriate motions on the screen more often than not.
E.
...and sometimes I do it with my index finger in the air in front of the screen... imagine the sight :o
Thanks, I feel better! ;D

Chris probably classified my post as TLTR and promptly left ??? ;D
"At least one child was removed because its image was too similar to a supposedly different child." - SDSS

Alice

  • Global Moderator
  • Hero Member
  • ****
  • Posts: 31782
    • View Profile
Re: CW/ACW bias study results
« Reply #41 on: May 14, 2008, 11:21:02 pm »
Don't worry, he's not like that and I still move the mouse in circles round galaxies too! ;D

zookeeperChris

  • Admin
  • Hero Member
  • *****
  • Posts: 723
    • View Profile
Re: CW/ACW bias study results
« Reply #42 on: May 15, 2008, 07:24:57 am »
I'm by far and away the least consistent of all of the zookeepers on this, so I sympathise.

Chris

veggy

  • Full Member
  • ***
  • Posts: 100
    • View Profile
Re: CW/ACW bias study results
« Reply #43 on: May 15, 2008, 08:37:13 am »
Me too. Also fatigue creeps in and you suddenly find you no longer know whats ACW and whats CW.  Then you have to work it out slowly for yourself. And yes I too found myself calling them ACWs unless I watched myself carefully.  I did make lots of mistake by going too fast but sadly couldn't return to correct the faults because that system didn't allow such a GoBack facility.

mallory weiss

  • Sr. Member
  • ****
  • Posts: 369
    • View Profile
Re: CW/ACW bias study results
« Reply #44 on: May 16, 2008, 03:31:17 am »
I'm by far and away the least consistent of all of the zookeepers on this, so I sympathise.

Chris
OK, Chris, so this one's for you... hope it'll haunt you in your dreams... ;D



Do you see the dancer spinning clockwise or anti-clockwise?

(hijacked from http://www.news.com.au/heraldsun/story/0,21985,22556281-661,00.html)
"At least one child was removed because its image was too similar to a supposedly different child." - SDSS