October 4, 2004 11:32 AM PDT

Navigating PCs with pictures, not words

Related Stories

Microsoft aims for search on its own terms

November 24, 2003

The mouse that roared

September 18, 2003
Do you remember faces better than names? A team of university researchers is developing software that applies the same phenomenon to PCs.

VisualID is an experimental program that seeks to help individuals more rapidly find their word files or spreadsheets by tagging them with graphical--and largely arbitrary--icons.

A snowflake with pop artist Peter Max overtones, for instance, might represent a document listing the members of an organization, while a dachshund-with-a-hat doodle might become the visual mnemonic for the household expense file.

News.context

What's new:
An experimental software program called VisualID labels files with images that can be easily remembered.

Bottom line:
The idea behind VisualID is that humans can remember pictures better than words--and can navigate faster and easier with distinct, individual images than with generic icons.

More stories on this topic

The principle behind VisualID is that humans can remember pictures better than words, and can navigate faster and easier with distinct, individual images than with generic icons (like the consistent, omnipresent "W" that tags Microsoft Word documents).

"Our visual brains 'suck up' the appearance of everything we see. It doesn't require effort or even conscious thought. We just do it, all the time," J.P. Lewis, a researcher at the University of Southern California and the principal author of the project, wrote in an e-mail interview. Researchers at USC and at the Massachusetts Institute of Technology are cooperating on the project.

The quest for better desktop navigation is as old as the computer industry itself. Mice, graphical Oases and directories have all helped simplify PC controls. More recently, Google, Apple Computer, Microsoft and others have announced plans to release search tools that will scour hard drives for files, although Microsoft recently delayed its project.

The appeal of VisualID is that it takes advantage of the familiarity of the graphical icons concept and the hardwiring of human psychology. The program won't replace text file names--just enhance them.

"Visuals always help," said Jerry Tarasofksky, CEO of iPerceptions, which analyzes Web sites for large companies. "When you are looking for shoes, an icon is so much clearer than 'shoes.'"

One of the more intriguing findings with VisualID has been that the icons themselves don't have to be meaningful for people to remember them. A squiggly sea monster-like icon can just as easily represent a file outlining monthly sales figures or a stored URL on beach vacations.

It turns out that "scenery" of this sort is not used by the brain for data visualization but instead enables visual search and memory, according to Lewis. The icons of VisualID can relate to each other. The program fashions similar icons for files with similar names, and icons can be grouped under a thematic meta-icon. But, overall, no intrinsic connection between content of the file and doodle exists.

"We cannot always guess the appearance of a book due to arrive in the mail from knowledge of its subject matter, but its somewhat arbitrary appearance is almost instantly learned and remembered once we see it," Lewis writes. "I think the realization that relating the visuals to content is not necessary is one of our bigger contributions."

Doodle away, maestro
VisualID icons, called doodles, expand on the icon vocabulary in software today. Applications on PCs typically sport icons--like the script triangle that signifies Adobe Acrobat files--but the icon stays the same for each document or file created with that application. VisualID gives each file its own distinct symbol automatically.

While the shape of an icon is random, the program creates mutations of an original icon if a new file shares a similar name. The icons differ from each other by shape, but color may be added as a future identifier.

Determining what makes a good icon remains a challenge. The VisualID icons are deliberately complex to avoid confusion and overlapping imagery, but also try avoid excessive complexity. The group adopted the Kolmogorov definition of complexity, which says that an object is complex to the extent that each part of it is unpredictable given the other parts.

"A humanized version of the definition would be: How many words would it take someone to describe a particular icon to someone else over the phone with enough detail that it could be picked out from among other icons of the same general sort?" Lewis said.

For instance, a triangle is easily identifiable, but there aren't enough simple shapes like this for the thousands of files inside the average PC. A circle containing random dots is clearly distinct and individual, but nearly impossible to remember with precision.

A look at results
Surprisingly, people can keep many squiggly figures in mind. "In one study, people were able to form much better than chance recognition memory for a set of 100 of these icons after seeing them once for a few seconds each," Lewis wrote. "The brain is finite, so there is some limit to how many icons can be learned, depending on how long a person spends working with the icons...I would guess that this number is in the several hundreds to several thousands range."

Fortunately, not all of the figures have to be kept at the forefront of the mind. Because most people access just a few files during a given period--and often never access many files after a relevant period again--only a few files need be actively recognized at any given time.

It's too soon to tell whether the VisualID work will influence the future development of software. But early tests indicate that the system works. In one experiment, VisualID users took 25.2 seconds on average to conduct four file searches versus 30.5 seconds for those seeking with generic icons. In another experiment, subjects recalled VisualID tags 37 percent of the time a day later, as opposed to 24 percent of the time with genetic markers.

Subjects contacted six weeks after the study were able to still correctly identify previously seen icons (as opposed to icons they weren't shown in the experiment) 80 percent of the time, according to researchers.

5 comments

Join the conversation!
Add your comment
where's the link?
um...there's no link in the story or below to the company making VisualID. any ideas?

jake
Posted by (2 comments )
Reply Link Flag
link
search for "visualids", it is at these places

<a class="jive-link-external" href="http://www.idiom.com/~zilla/Work/VisualIDs/visualids.html" target="_newWindow">http://www.idiom.com/~zilla/Work/VisualIDs/visualids.html</a>
<a class="jive-link-external" href="http://www.idiom.com/~zilla/Work/VisualIDs/visualids.pdf" target="_newWindow">http://www.idiom.com/~zilla/Work/VisualIDs/visualids.pdf</a>
<a class="jive-link-external" href="http://www.idiom.com/~zilla/Work/VisualIDs/visualids_faq.html" target="_newWindow">http://www.idiom.com/~zilla/Work/VisualIDs/visualids_faq.html</a>
Posted by (1 comment )
Link Flag
I would rather just read a list of titles.
They want us to remember which file belongs to which icon? What a bunch of nonsense.
Posted by Seething Ganglia (12 comments )
Reply Link Flag
bunch of nonsense
<a class="jive-link-external" href="http://www.analogstereo.com/isuzu_oasis_owners_manual.htm" target="_newWindow">http://www.analogstereo.com/isuzu_oasis_owners_manual.htm</a>
Posted by George Cole (314 comments )
Link Flag
we need good old fashioned search
these icons are nifty but why cant we just get a simple search of the hardrive to work in under 15 minutes. google has no icons and it works find to search the entire internet.
Posted by confucious_says (7 comments )
Reply Link Flag
 

Join the conversation

Add your comment

The posting of advertisements, profanity, or personal attacks is prohibited. Click here to review our Terms of Use.

What's Hot

Discussions

Shared

RSS Feeds

Add headlines from CNET News to your homepage or feedreader.