Wow. 2,481 clicks after only 2 days. That's a lot!

Philipp, its does give some interesting zones. 4 circles represent more or less the center of the image ..

This was a great idea. I thought I was so smart clicking on the watch (coins?) on the table, but so did a lot of other people!
Matt

I clicked the cards near the middle of the image.

Intriguing. It sheds light to where people are the most likely to have their mouse when they visit a webpage... I think. I'll have to muse over this one.

Interesting experiment. I imagine there must be similar stuff within the labs at MS, Apple, Yahoo, Google etc, and especially amongst the game developers.

Were you just curious, or is it research toward a specific purpose?

And why did you post the comparison with areas of brightness? Of course there's some overlap, but the main click targets seem to be hands, faces, cleavage(!) and artefacts – even when those items aren't bright, and ignoring some bright but otherwise uninteresting areas. It looks more like the activity of humans than magpies. Maybe you should try with a landscape, and an abstract?

(BTW, since you've already got the logo attached to more than one poster, why not fiddle with things to make this into a real page :- http://blogoscoped.com/forum/posts-by-googlers.html ?)

> Were you just curious, or is it research
> toward a specific purpose?

Just curious :)

> And why did you post the comparison
> with areas of brightness?

I think the artist uses light to focus attention, as people often are attracted to the brightest parts of an image... the lady in the middle seems to be the center of attraction, her face and cleavage being some of the consistently brightest part in the composition. But I didn't want to propose a conclusion, I just wanted to offer this image as footnote for reference so that we can make our own conclusions. I think the eyes and objects held in hands are also "hot spots." I think in any given composition humans will give special attention to faces...

By the way, I wouldn't be surprised if the results of an eyetracking study on this painting would come to similar results.

Interesting results. (FWIW, I clicked the cards in the middle of the image too.)

I had a dream last night that this was just part of a long online game. You had to click where you thought was the most clicked part of the painting. You got three chances and once you got it right you had to guess again on another painting. And it went on and on and on and on and... :-)

I had that idea this morning...

Philipp, there are indeed already all kind of studies with eyetracking devices.

On this page you can find some interesting information about how a user watches a webpage, its ads (e.g. textads are more attractive than graphic ads), navigation, etc.

http://poynterextra.org/eyetrack2004/main.htm

There is plenty of stuff about eye movement and eye tracking on the web. But I'm not an expert in it...

Heh. This is an interesting idea. Not only could it be a game, it would also generate lots of these heat maps. The game would need to be "initilaized" with say 100 clicks each to get some starting data, but that could be done with some help. (A variation of the game could be... "click where almost no one clicked before.")

That reminds me of something else.

Let's say your objective is to meet your friend. You don't have the day or time or location. You only know that your friend gets the same objective to meet you, with the same unknown location and time. Can you too meet each other? Where do you meet? At what time? What are the reasonings here?

I believe I read about this in The Foucault's Pendulum. Here the objective was for a secret society to meet each other... but they were able to choose any location in the world.

I recall reading that Mickey Mouse was given white gloves so it would be easier to follow his actions. Similarly, Shigeru Miyamoto has said Mario's outfit and mustache were designed so that you could distinguish the parts of his body despite that his sprite was tiny.

where was this posted? i read all the posts and still couldn't find the original post.

Do you mean this post Zoolander?
http://blogoscoped.com/archive/2006-07-14.html#n35

tits = clicks. internet truth 101

Not much people clicked on the ring of lady!

Hi Philipp,

First of all, thanks for the study.

I think you should carry on this experiment with some more commentary based on scientific researchs.

Here are just two simple pages about virtual perception:

http://www.ocf.berkeley.edu/~wwu/psychology/attraction.shtml
http://perceptionweb.co.uk/perc1200/subjix.html

Hope to see the "results" :)

Out of curiosity, I ran the picture through some well known feature-detectors developed by the computer vision research community.

The first is Harris Corners – it responds well to corner-like patches
in the image.These seems to be a fair degree of correspondence between this and where people clicked.

http://www.robots.ox.ac.uk/~mjc/HarrisCorners.jpg

The second is Phase Symmetry – it is a measure of local symmetry, and
responds well to lines and blobs. Doesn't seem to predict clicks so
well.

http://www.robots.ox.ac.uk/~mjc/PhaseSymmetry.jpg

Anyway, just though it was interesting. If you want to have a play around yourself, all of the code I used is public and available here:
http://www.csse.uwa.edu.au/~pk/Research/MatlabFns/index.html

Shoot, I myself just clicked in a random area in the black background, as I was certain those spots wouldn't get very much click lovin' ... and as I can see, there were a handful of others who did the same thing. Could the background be taken as a whole?

Also, consider this: I clicked based on where I thought others would click. I knew this was an experiment and what the inputs were. For something more realistic, you'd probably want people to not consider where others would click, and perhaps ask them to click on the first thing they focus on as opposed to "anywhere" ... and, probably not even ask period, but rather see how many people click where without you asking ...