100 Things You Should Know About People: #18 — What People Look At On a Picture Or Screen Depends On What You Say To Them

Yarbus Eye Tracking Picture
Yarbus Eye Tracking Picture

Eye tracking is a technology that allows you to see and record what a person is looking at, and for how long. One way it is used is to study web sites to see where people are looking on a web page, where they look first, second, etc. It’s a pretty interesting technology, one of the benefits being that you don’t have to rely on what people SAY they are looking at, but can collect the data directly. Like any technology, however, it’s not perfect, and one of the problems with eye tracking is that you can’t just give people a web site to look at and then assume that where they look is what they are “really interested” in.

We underestimate the effect our instructions have on where someone looks. Look at the picture at the beginning of this post. In research by Yarbus, people were shown this picture, and then given different instructions of what to think about while looking at the picture. Below are the eye gaze patterns matched with the instructions that people were given:



Proceed with caution. To me this data says: a) If you are using eye tracking as a technique to evaluate how people are using your website then you must be very careful about the instructions you give, and you must make sure you are giving everyone the exact same instructions. b) You can’t assume that just because people look at one spot on your website when they first see it that they will always look there. It might depend on what they were coming back to do. c) It’s nice to have a measure that doesn’t rely on what the user says or how they think they are reacting, but even these “objective” measures aren’t as objective as we think!

This research is from way back, but I believe it is still relevant. I haven’t found any recent replication of it yet. If you know of any please do pass on the reference:

Yarbus, A. L. (1967). Eye Movements and Vision (B. Haigh, trans.), New York: Plenum


Did you find this post interesting? If you did, please consider adding your comment, subscribing to the blog via RSS or email, signing up for the Brain Lady newsletter, and/or sharing the post.

Eyetracking Studies -- 7 Traps to Avoid
Quick Review of Dan Zarrella's Social Media Marketing Book

21 Replies to “100 Things You Should Know About People: #18 — What People Look At On a Picture Or Screen Depends On What You Say To Them”

  1. There are two points I would like to raise on this point, firstly in the example the person is looking at an image, not a website and therefore by definition the viewers behaviour will be different – when looking at an image they are passive unless given a task and will just browse the picture, normally looking at faces, mouths and following cues such as pointing fingers, roads that lead into the distance and so on. When you give them a task such as ‘how many people are there’ or ‘judge their ages’ then of course their behaviour will change as they try to answer that question. When doing web testing with eye tracking the participant is in an active state as they will have been given a task prior to going onto the web page (unless it is a free roam exercise which usually offer little insight and we do not recommend this type of task) so they will of course be trying to complete their task (make a purchase, log in to an account, search for ‘x’ or whatever) and of course their behaviour will be driven by the task and their interaction and gaze within the page will be appropriate to the task in hand. So if we run three different tasks on a web page then we will have three different interactions and gaze journeys for each participant and we analyse them accordingly so they are relevant to search strategy, menu naivgation and so on and therefore the data is relevant, and invaluable, to assessing the sites usability against the benchmark or task applied. It would be a very bad usability practioner that gives someone a website and tells them to do whatever they want and then try to draw some conclusions….

  2. Thanks for commenting Jon. But I disagree with you some. I read many usability studies with eye tracking where they ask someone to open a URL, and then they take measurement on where the person is looking before they have even given them a task to do. And part of my point is that you have to be very careful about the specific instructions. So if you say to one participant, “See if you can find a present to give your sister for her birthday.” And to another participant you say, “Your sister is having a birthday. See if there something at the xyz site that you would like to get her for a present.” Those aren’t worded exactly the same way. We don’t now what or how much an effect it might have on the pattern. A picture is not the same as a web page, but we don’t really know how different. So my point is to be careful what you say and what you assume.

  3. Agreed, you must be very careful with how you set up a usability test, particularly with eye tracking involved.

    It is very risky unleashing people on an eye tracker who have limited understanding of the scientific method.

  4. Susan, the comment you make about the instruction given to the participant is valid – but a trained usability practicioner should be able to create the correct scenario and instrcutions to give accurate and useful data, and the eye tracking information gleaned from this will be constant again if the moderator is constant! As James says, you need to be careful who you let loose with technology…. as for how different interactions are with images as opposed to web pages there are numerous studies, white papers, case studies and academic papers regarding this type of variation – and also things showing the differences between different protocols such as think aloud as opposed to retrospective think aloud and so on… Scribd and Slideshare are full of this type of resource, we have some on there also (search under Acuity ETS or Tobii). Look forward to more of your posts…

  5. Jon,

    Yes, a trained person should be able to avoid the pitfalls. The original post above was to point out some of those pitfalls … I think there are often people interested in eye tracking that may not have the full background.

    Thanks for your comments and the resource suggestions.

  6. I’ve recently started a blog, the information you provide on this site has helped me tremendously. Thank you for all of your time

  7. User tasks in a web page clearly influences the way they look at them.
    We cannot assume users see all elements we show at them.
    In your post, I don’t understant ‘Intructions’ in that context.

  8. The other thing I’ve never understood about eye-tracking studies is that they discount the designer’s ability to direct the eye. For example – If the page only contains a red circle in the middle, I guarantee you that no user is going to scan that page in an F-pattern.

    Or am I crazy?

  9. Visual Cognition 2009 released a book all on behavioral findings of eye movements and Yarbus’ 1967 work is featured prominently in quite a few of them. One article was specifically targetting why Yarbus’ study was important and what recent research has to add to the topic.

  10. It’s ironic that I’m studying eye tracking data as an online teacher who will not be face to face with her students! ; ) However, the use of images and memory with students is wholly relevant.

  11. This is very interesting. I enjoyed examine the artwork, as well as, reading the article about eye tracking. Recently CNN had a story on this. This is a great way to engage students in reading and visual literacy.

Leave a Reply

Your email address will not be published.