Your eyes do funny things when you look at a web page. Eye tracking technology shows us that they tend to follow an F-shaped pattern of eye movement across the screen. But new research could bring this into question.

The new research shows us this may be down to the "instructions" we receive when we open a web page. Back in the 1950s and 1960s a Russian psychologist, Jarbus, showed that our eyes move in different ways according to the motivation we have. For instance, if we go into a room of people and we simply want to get to know as many people as possible, our eyes move all over the place. But if we are told to try and remember the clothing everyone is wearing, our eyes move in a completely different way, his research found.

The new study, from the University of Iowa, shows a similar finding. It found that when asking people to look at a scene, their eye movement depends upon their motivation. In particular, this new research showed that if we are asked to search for something, we don't revisit areas of the scene that we have already looked at.

It appears that if we are not searching, we do revisit areas of the image we have already looked at. For decades psychologists have worked on the assumption of the "inhibition of return" - the notion that we are prevented from returning to look at things we have already looked at in a bid to save effort. However, this new study suggests the opposite - as, indeed, did some of the earlier work from Jarbus. Together they show us that when we are asked to search for something we tend to do the opposite of what we would normally do. Usually, we look all over the place, returning to areas we have looked at already. When we are asked to search, we only visit areas once.

This has significant implications for search engines as well as ordinary web pages. The mere fact it is called a search engine means we are likely to look at each result only once; we will not return to that area of the page. However, if we were to return to a result we have already looked at we may well find it is better or more appropriate than search results we subsequently look at. In other words, the bias to avoid returning to areas of the screen we have already looked at while in "search mode" is potentially preventing us from actually finding the information we may want.

For SEO and web page optimisation the aim is to get you into the Top 3 of a search engine's listings so that you appear in the right place of that F-shape pattern. That's great - but this new eye tracking study shows that people will not return to your top listing if they have ventured further down the page. In other words, in some instances, being top works against you.

Furthermore, a two-year-old study of Google shows that a "blended" results page (where the page shows videos and images as well as the traditional list of web sites) produces an altogether different pattern of eye movement. It seems that when we don't see the page as something we are supposed to be "searching" we use our eyes differently.

This has significant implications for web site owners and web designers. It all means that if you present your page as though the reader needs to "search" for what they want, their eyes will behave in one way. Whereas if you don't imply they should search for things, their eyes will move differently and (importantly) will return to areas of the screen they had previously looked at. This means they may well find items of interest that would otherwise have been discarded if their mind was in "search mode".

So, does your web site suggest it is something that people can "search" for what they want? Or do you just give it to them? Implying that people can look around your site could mean they view less of it. Giving them what they want and loads of other interesting things they could look at as well can change their eye movements, bring them into contact with more of your site. Perhaps we need to revisit the whole notion of having a search facility as a prominent feature on a web site, since it can change the motivation of your readers, making them less likely to look at your web pages completely.