Designing for the Grand Illusion
(SIGCHI Bulletin November/December 2001)
[For an example of change blindness and links to research mentioned in this article, please go to //www.syntagm.co.uk/design/articles/cb.htm.]
Psychologists have identified a slightly worrying failure of visual perception known as change blindness. It occurs when the movement normally associated with change is temporarily masked. In the real world this might be caused by blinking or a brief focus of attention elsewhere. While this manages to sound fairly innocuous, the effects of change blindness can be quite startling. Imagine being stopped by a stranger and asked for directions. Part way through the exchange there is a distraction, but the conversation continues to its conclusion. Do you think that you would notice that the stranger was really two fairly dissimilar people who had switched places? In an experiment by Simons and Levin, half of the participants failed to notice the swap.
Accounting for change blindness has led to some interesting theories of how visual memory works (or fails, depending on your point of view). One of the proposed theories is that passive "seeing" is an illusion. We believe that we see everything in front of us, but the reality is that we only take in those things that we give deliberate attention to. In other words we lack a passive visual memory. Hence the theory that Kevin O'Regan refers to as the grand illusion.
What impact, if any, does change blindness have on interaction design? In the desktop world of high bandwidth and instant page changes, probably none or very little. But even with relatively high speed networking, the web presents a different story. The description of a typical change blindness experiment in the diagram below (from Ron Rensink's web site) could equally well describe a typical web browser interaction. A user sees a page, clicks a link and a new page appears after an intervening blank field. (Note that neither longer initial exposure to the image nor the color of the blank field has any significant impact on the outcome.)
Just how difficult it is to notice what has changed between the two pages is demonstrated by the change blindness examples on the web page associated with this article and the other sites referred to.
Change blindness presents us with no problems where consecutive pages are unrelated. However, it completely dashes our hopes that minor changes between pages, such as error messages or search reports, will be noticed. For example, I still come across search pages that start with an "items found" count of zero, which is replaced by the search result count once the search is completed (with a separate link to the search results). In some cases, this is the only change to the page. Needless to say, users are frequently baffled by their inability to notice any difference between the "before" and "after" images.
We need a more reliable mechanism for dealing with change in such cases. Most well-designed search engines have obvious differences between the search and results pages. However, things are not so straightforward in form validation and confirmation sequences. In form validation, the original page is sometimes redisplayed with only minor changes indicating where errors have occurred. The use of a different color will help combat change blindness, but large and obvious differences will be much more effective.
In fact, this is one occasion in which limited use of animation might actually be beneficial, because what is missing is the conventional visual cue of motion that we associate with change. Small, blinking change markers (especially in a contrasting color) will allow users to compensate for the change blindness introduced by the awkward delays of the web.
[For an example of change blindness and links to research mentioned in this article, please go to //www.syntagm.co.uk/design/articles/cb.htm.]
Bibliography
Daniel J. Simons, ed (2000), Change Blindness and Visual Memory, (Psychology Press) Harvard University, Boston, MA (also at Amazon.co.uk)