Eye tracking research; PaperSense 2.

This is a non-academic publication, based on my personal reflections on a highly interesting research project that I have been partly involved in.

In the previous article, No. 1, some background information was given about the research project itself. If you want to read more about it, follow this link.

This article presents three new pictures with the analysis and data that emerged under the leadership of including Siv Lindberg and Annika Lindström at Innventia in Stockholm.
The pictures used for the test were provided by IKEA and are copyright Inter IKEA Systems B.V.

Again interesting and revealing heat maps show where the test persons focused their gaze during the test:
– Examine whether there is a difference of attention when an image is presented on paper or a screen,
– Investigate how much influence the presence of a person in the picture has to the attention given to the picture.

Again, the results are strong. When the results were compiled and calculated the first time after the test, the differences were so extraordinary compared to what one is used to seeing in research, that it was assumed that the test data or calculations were wrong. So all was trashed and recompiled from the beginning.
But the large differences were of the same magnitude.
– When the image appears on paper, it gets up to twice the amount of attention, sometimes more, compared to when the image is shown on a screen.
– When the image contains people, it catches up to three times as much attention.

Are there any obvious signs and differences in the pictures, which can be attributed to the impact of image composition?
The answer to that question is both yes and no.
When an image is more colourful and cleaner in its composition, it gets higher rashes, while a cluttered, dark and indistinct image gets punished with a lower amount of attention. The presence of people in the image often raises the amount of attention, in some few cases although people have limited impact.

Eye-tracking can provide interesting conclusions about the composition of the picture and its implications on the viewer.
But an effect following this type of experiment is that the information and task you give to the test persons affects how they look in the pictures. Even if one gives general and neutral instructions, the situation is still a test. It might give rise to a kind of loyalty in which the test person look a little different than when you’re sitting at home flipping through pages, in print or on the Web.
So we can assume that during an eye tracking test, people may look a little extra, maybe to try to discover something that they think the experimenter wants them to find. Probably test persons are more alert and attentive than in a relaxed situation.

Some pictures from the Paper Sense, Part II.
Now follows three pictures of the test suite. These images are the three pictures that ended last among the twelve in the suite, concerning the total value of attention in the images.

Picture 1, kitchen 1 – with people. Lowest ranking, position 12 of 12.
PaperSense - pict 01. With people.
All original pictures copyright Inter IKEA Systems B.V.

This image is somewhat cluttered in terms of composition. Although it has a central perspective, which often is described as to lead the eyes into a picture. And the eyes will be directed towards the bright windows in the background where two people are clearly depicted. A child can be seen in the sofa with red pillow in the foreground.
This picture actually gets the lowest attention value, regarding the total sum of all the twelve tested images.
The version printed on paper gets value 543, while the screen version get the value 261, which is down to 48%.

Kitchen 1 – kitchen without people.
PaperSense - pict 01. No people.

The kitchen is now without people and is slightly quieter and less cluttered. The heat maps show that the eye wandered around a bit more in the picture.
The version printed on paper gets value 481, the screen version gets value 195, a decrease of 89% and 75% respectively of the values that ​​was observed when people were present in the picture.
The total value of this picture is 1480, which results in the lowest amount of all tested images.
An explanation for the low amount, despite the central perspective, can be that the picture has a complicated structure and lacks clear focus points that the eyes can find interesting.
I think the picture is built according to recognised rules of composition, but the image itself lacks charisma, and does not have much to tell.
Good composition can hardly compensate for weak charisma.

Picture 03, kitchen 2, with dog and people. Place 10 of 12.
PaperSense - pict 03. With people.

Once again a kitchen, now with an active person and a dog. The image is built around a central perspective and is compared to the previous image cleaner and simpler in structure. The picture gives a warmer impression with more color contrast and clearer focus points for the eyes to attach to.
Yet this image gets rather low attention value. Compared with the previous image, with the lowest value of all the images, this image is only third in place from the bottom.
The version printed on paper gets value 721, the screen version gets value 356. The picture displayed on a screen only gets about 50% attention compared to the image viewed on paper.

Picture 03, kitchen 2 – without people.
PaperSense - pict 03. No people.

The kitchen now without person and dog.
The heat maps now show that the viewers look more around in the kitchen. One can see that the shiny dog bowl attracted some attention, even though it is in a corner.
The attention values now fall to about half when there are no people in the picture (the dog suddenly was classified as people).
The version printed on paper gets value 378, the screen version gets value 176, a decrease of 52% and 50% respectively compared to when there were the people in the picture.
As well in this version the attention drops below 50% when there are no people in the picture.
The total value of this picture is 1631, resulting in just about 10% more attention than the previous kitchen image.

Picture 10, a product image, partly with people. Place 11 of 12.
PaperSense - pict 10. With people.

The following image has more character of a product picture, where neither the couch or the person is shown in full. The composition is straight and simple, with rising lines that leads to a person sitting on the armrest. The colours and the environment are unspectacular apart from the pillow.
The heat maps reaffirm that a person attracts the most attention, even though we can not see the face of the person.
The screen version gets value 197 and the version printed on paper gets value 628. The picture displayed on screen gets only about 31% attention compared to the image viewed on paper.

Picture 10, without people.
PaperSense - pict 10. No people.

The picture without a person, a product picture with a colourless sofa or armchair.
The screen version which had people in it got value 197, while the version on screen without people got value 272, a sudden increase, plus 38%.
Still, in the absence of people, the attention is reduced to about 50% compared to when we have people partly visible in the picture.

It is understandable that the first dark kitchen and the product image with sofa/ chair are low in popularity.
But I am confused by the second kitchen with the dog, which received such low values. We have the values 1480 – 1606 – 1631.

So, we can see where the eyes have looked in the pictures and how much they have looked, but we can not look into the mind of the test persons and find out what they thought about the pictures during the test.

Leave a Reply

Your email address will not be published. Required fields are marked *