submitted for partial completion of HCI 450 – Lawler
DePaul University Summer II 2001
Jeffrey Absher xxx-xx-xxxx
The World Wide Web and the concept of hyperlinks, intertwined with the availability of cheap displays, and the relative ease of untrained authors to create and publish documents has lead to inefficiencies in task completion when the tasks rely on information gathering from a webpage. Simultaneously with the public’s changing style of reading for information gathering (from newspapers to web-centric) and the public’s increased expectation for instant delivery of data, the evolution of data presentation formats has brought about new distractions that slow the seek for data. When a system is viewed as both humans and machines working together to perform a task, the capabilities of machines to display data in millions of colors and the aesthetic desires of document authors to present a pleasing or attractive document with many hues hinder the speed of the human-machine system for gathering and seeking data.
As more information migrates to the medium of hypertext, more tasks will depend on accurate location of data within that hypertext. Strictly speaking, hypertext is not a sufficient term to describe the display capabilities of a webpage; but in this document it shall refer to webpages’ text-display capabilities. While automated scans and search engines are effective manners to cull hypertext documents for target strings in some cases, they are not useful for all searches for information or data. It is conceivable that in the future natural language queries will suffice for all information searches, but today many searches still end up with a user scanning a page for a target. The same display capabilities that made webpages an enjoyable medium and created the demand for the mass migration of information to hypertext format are those that slow humans’ abilities to locate data efficiently within documents at a perceptual level. If one imagines a time-sensitive task that relies on the split-second location of a text fragment within a document to determine action, one can start to see that multicolored hypertext may not be the best medium for those documents, for example:
A doctor needs to know the drug to administer to a patient having a stroke. As the doctor visually scans a prose hypertext page seeking the drug name, he is distracted by multiple hues within the text. This slowed scan leads to lost seconds in proper treatment determination of the patient.
Because the disciplines of Human Factors and Human Computer Interaction deal with aspects of making the machines and systems fit the humans, they should investigate whether machine-systems should take into account that extraneous color in text slows scanning for targets.
The concept of using color to highlight important text is not new. “In the days of text-only terminals, IBM had a terrific advertisement for what was then the novel idea of color text displays: An abnormal value jumped out of the clutter immediately when shown in a distinctive color.”[3] The developers of CERAS refer to this as a pop-out function.[8] CERAS is a tool for automatic coloration or highlighting of words in a document. Also some researchers are attempting to develop a program that implements rulesets to appropriately color text within a user interface. This tool is called CRAFT, the Color Rule and Font Tool.[6]
Winn [9] summarizes that pop-out is a preattentive process over which humans have little or no voluntary control. This roughly describes the perceptual stage in the information processing approach to cognition. Winn also argues that color is attentively processed as well by pointing out that the feature integration theory promotes the view that color may or may not be dealt with on a preattentive basis and the features that humans attend to preattentively vary. Trummell, [7] taking a gestalt view of color in typography, labels color detection as a percept rather than a concept, but does not go further to hypothesize on any determinable functional (relating to pre-cognitive target location) effect of a color percept within a document. Instead he points out some of the post-cognitive effects of color such as emotional impact and “mood and feeling.” While these effects may be correctly documented, if they are cognitively attended to rather than perceived then any delay in processing or interference with searching would occur at that stage rather than at the perceptual stage.
Color has varying functions in a document. Apple Computer [1] has an explicit viewpoint that color should not be included in the initial design of a user interface. Quoting from their guidelines to developers, “You want the user’s attention focused on the content of the application, rather than distracted by color…” Though the context of this quote is referring to an interface rather than an electronic document, it shows that they feel that extraneous color is a distraction from a task. The two preattentive purposes of color in a document according to Winn [9] are direction of attention and delimiting of shapes and area.
Winn’s article is one of the few that tries to soften the anti-color sentiment of most researchers. Both he and Horton [4] agree that “communicators should understand color, respect its limitations, and use it for specific, functional purposes.” Though Horton does not separate purposes into preattentive and attentive as strongly as Winn does; the positive purposes of color utilization that he lists that can be classified as having a large preattentive factor are “Color Speeds search” and “Color focuses attention.” The negative effect that he points out is that “Color Distracts Viewers.”
Though none of the articles explicitly state it, they all seem to align themselves with a viewpoint that there exists a threshold of the number of hues in a particular document over which addition of hues becomes a distraction rather than an enhancement. Jacobson and Bender [6] illustrate this and state it in their report:
“As more color is added, attention is drawn away from the gray patches and a new sense of order emerges: a progression of hues. However, with the addition of still more color, a point is reached where a consistent message can no longer be viewed in the image. What began as an ordered progression has turned into a cacophony of colors.” Reference their image at http://www.research.ibm.com/journal/sj/mit/sectiond/jacob1.gif
While many researchers have determined their own lists of guidelines for color usage, the general consensus of the summarizing articles comes up with these basic posits:
R. Carter has conducted an experiment on visual search time within text for a target.[2] Target color and color relationships between the target and the remainder of the display are his independent variables. He defines two concepts that are necessary to understand his research. TCS is the target class size. It is the number of objects that are indiscriminable from the target object in their hue characteristic (all of the “words” that are the same color as the target.) He also defines display density as the “total number of items in the search area.” In addition to these variables, Carter manipulated other variables to minimize error and provide control groups.
The researchers that came before Carter proved a few facts that he took into account in his experiment. There is a linear relationship between search time and TCS for a specific range of TCS (The more objects colored the same as the target, the longer it takes to find the target.) Higher display density with the same sized TCS slows the search. A stronger color difference between the TCS and the background items speeds the search. Placing TCS members in a pattern also sped the search. See article [2] for relevant references.
Carter performs three experiments. In the first he allows four colors to be used in the non-target set of objects, and one color to be used in the target class. The subjects are told the color of the target and the first two of three characters in the three-character target string; subjects are to determine the third character. The determination of the third character indicates successful location of the target. Display density and TCS are varied. Carter’s results extend the specific range of TCS in the TCS-search-time linear relationship (noted above.) Carter’s third experiment relates to searching for targets within a pattern and is beyond the scope of this document.
In Carter’s second experiment, he
replicates the conditions of the first except that the non-target colors were
not chosen from a set of four hues, but were either the same hue as the target
class at a different intensity, or a single complementary hue. In his results
he proves that “dissimilar background items [have] the same effect as no
background items.” Comparing data between the two experiments, one sees that
having a single target object as the only instance of the target hue among many
background objects of different hues facilitates search compared to the
background items all being the same hue. Having many objects of the target hue
among many objects of varying hues slows search compared to many objects of the
target hue among many objects of a single non-target hue. Restating that, if
your object is the only RED object on a multihued page, you will find it
quickly compared to only saturated RED on an unsaturated RED page, but if your
object is one of many RED objects on a multihued page, you will take longer to
find it than if it is one of many RED objects on an unsaturated RED page.
Brockmann [1] cites an additional study that agrees with
Carters, stating that “polychromatic color …[leads] to distraction of eye
motions, increase in time required to read and [a] decrease in performance.”
Carter’s study is an excellent
example of research. He reviews the pertinent earlier research, describes his
assumptions and procedures, controls for confounds, clearly states his
findings, includes the relevant statistical calculations, and suggests further
research directions.
Color is a powerful tool in a document on multiple levels. The pre-attentive, cognitive, and emotional effects of colors contribute to the manner that a document is read and comprehended. Focusing on the higher-level cognitive or emotional effects can cause unwanted pre-attentive side effects that slow the reading and scanning of a document. Like many other aspects of life, the quality that initially attracts humans to a document (many colors) is sometimes incompatible with efficient interaction with that document. Authors may enjoy using color, but they should gear their color choices and strategies toward facilitation of the document’s likely use. A document’s context may require that a color scheme be designed for preattentive speed, attentive efficiency, aesthetic value, or none of the above. The available research points to color as a useful tool in document design to attract attention, but if three hues are better than two, four may not be even better.
[1] R. J. Brockmann, “The Unbearable Distraction of Color,” IEEE Transactions on Professional Communication., vol 34, pp.153-159, 1991.
[2] R. C. Carter, “Visual Search With
Color,” Journal of Experimental Psychology: Human Perception and
Performance., vol 8, pp. 127-136, 1982. (original research)
[3] P. Coffee, “Convergence Points Dressed up to Distraction,” eWEEK Commentary., 04-June-2001., http://www.zdnet.com/eweek/stories/general/0,11011,2766207,00.html (10-Aug-2001)
[4] W. Horton, “Overcoming Chromophobia: A Guide to the Confident and Appropriate Use of Color,” IEEE Transactions on Professional Communication., vol 34, pp.160-173, 1991.
[5] L. Hunter, “Text Nouveau: Visible Structure in Text Presentation,” http://www.info.kochi-tech.ac.jp/lawrie/tnsecret.html (10-Aug-2001) - questionable link, cached copy available at Google Cache (13-Aug-2001)
[6] N. Jacobson and W. Bender, “Color as a Determined Communication,” IBM Systems Journal Vol. 35, No. 3&4, 1996 - MIT Media Lab; also available at http://www.research.ibm.com/journal/sj/mit/sectiond/jacobson.html (10-Aug-2001)
[7] P. Trummell, “Shape Concept: Color Percept...Graphics, Geometry, and Gestalt,” IEEE Transactions on Professional Communication., vol 34, pp.174-179, 1991.
[8] T. Uchida and H. Tanaka, “An Automatic Document Coloring and Browsing System,” http://www.mtl.t.u-tokyo.ac.jp/Research/paper/1997/E97-conference-tomo-2.ps.gz. also available at http://citeseer.nj.nec.com/cachedpage/98260/1 (10-Aug-2001)
[9] W. Winn, “Color in Document Design,” IEEE Transactions on Professional Communication., vol 34, pp.180-185, 1991.
