Justin Starren (November 22,1995)
My question is: is there a literature on discovering the intrinsic representations in a field? My specific domain is radiology, but I'm looking for general literature addressing the issue, particularly with respect to categorical data.
For example, if an examination of the chest shows an infiltrate and I want to display that fact in a easily recognizable way, I could show a outline of the chest and an "infiltrate icon" in the area of the disease. But what does an "infiltrate icon" look like? Are there methods for discovering the best design for that icon, other than trial and error?
I have found a huge literature on methods for discovering "process" and "work-flow" as well as much information on how to test representation. But there seems to be very little literature on how to get the initial designs. One researcher even said "I can do it, but I can't teach you to do it."
Am I doomed? Does this literature exist? If so where? What is this sub-domain called?
Justin Starren, MD
Thread | Message Archive | Home
Werner Kuhn (November 23,1995)
Whether there are intrinsic representations is probably an open question, though i doubt that there are (whatever this would mean). however, we are rarely designing on a 'tabula rasa'. previous representations exist in many forms, analog and digital. for instance, the words and gestures that people use in doing or describing their tasks, the vocabulary and pictures used in instructions and regulations, existing tools and instruments etc.
The best and (IMHO) only way to come up with useful and usable designs is to investigate these existing symbols and to analyze what metaphors they suggest. taking existing concepts (like 'infiltrate') as source domains for interface metaphors is the approach i would suggest. i can't call it a method, as there is only very limited methodology on this available yet. But I believe there could be much more and we have to formalize some of the cognitive linguistics findings about metaphors in order to come up with a more engineering-like design approach to the problem you raise.
An attempt at this can for exammple be found in: Kuhn, W. and A.U. Frank. "A Formalization of Metaphors and Image-Schemas in User Interfaces." In Cognitive and Linguistic Aspects of Geographic Space, ed. D.M. Mark and A.U. Frank. 419-434. Dordrecht, The Netherlands: Kluwer Academic Publishers, 1991.
Thread | Message Archive | Home
James Souttar (November 23,1995)
Justin
This is perhaps only half an answer to your question. I'm not convinced myself that there is such a thing as *intrinsic representation* - most successful iconography seems to depend on the use of easily recognized generics or clever tropes (e.g. the icons used by computers to represent printers or documents). Arguably, visual (particularly iconic) representation depends as much upon shared meanings as other forms of language - and (back to Saussure) the relationship between signifier and signified is as arbritrary [for Robin's benefit, perhaps I should say *unmotivated* ;-) ].
I'll stick my neck out by saying that I don't think there is a methodology that can reliably result in creative inspiration. Typically, creative responses come during a period of relaxation immediately following a phase of intense concentration on the problem in hand - just prior to falling asleep or waking, in reverie or whilst listening to music, in the bath or whilst peeling potatoes. If the process were more controllable, there might be more good design about...
What I do think we can say is that there are some things one should specifically *avoid* in developing visual representation. The Whorf/Sappir hypothesis holds that our perceptions of the world are shaped by the language we use to describe it. A word or an icon may unwittingly send out messages to its users that could be interpreted in a mistaken or dangerous way. The classic case is Whorf's description of the explosion that resulted from a group of workers smoking close to empty gasoline containers - they had (in that circumstance) wrongly equated *empty* with *safe*. Icons may be more susceptible to this than words, due to the potency of their metaphors (we easily forget that a computer document is not actually an entity, only a convenient way of thinking about otherwise undifferentiated binary data held in a file system).
Hope this is useful
Thread | Message Archive | Home
David Sless (November 23,1995)
I don't know that there is clearcut answer to this question because much of the practical knowledge of designers on how to create new symbols or other representations is tacit. It grows out of the craft of doing. It is a knowledge that is seldom written up and is quite different from the knowing *about* things.
I tried to answer this question myself some years ago when I was doing research on symbol design. The nearest I came to an actual example in the literature was: Bedno, E. (1972) A program for Developing Visual Symbols, *Visible Languages* VI 3 355-363. But there were many problems with this approach, not least of which was that there was no attempt to establish the validity of the methodology proposed.
I tried to develop an approach to this which integrated the use of practical drawing skills with testing.
Sless, D. (1979) Image design and modification: an experimental project in transforming. Information Design Journal, 1, (2), 74-80.
This is now more widely known in the HCI literature as rapid prototyping.
I went on to do some work on symbols with Standards Australia, but that was mainly concerned with introducing sensible testing methods to stop people introducing silly symbols. I have not done much work in that area for some time. Unfortunately there is still a sad lack of good work in this area. At a recent conference in Holland (Public Graphics 1994) there were a number of papers on symbols, but they were all on issues to do with testing rather than developing symbols. I found this very disappointing. Someone really needs to develope an integrated methodology in this area. So the answer to your question may be that we still need to do some work in this area. I hope others give a more positive response.
Thread | Message Archive | Home
Justin,
I suggest you hire a graphic designer with experience doing iconography or logos. They have years of experience, knowledge and skills of taking information and visualizing it graphically. Hours of reading can't replace true experience. As an MD, I'm sure you understand!
What an 'infiltrate icon' looks like will depend on the viewer and the knowledge they have about the subject. Try to understand what infiltrate means to your users and test imagery to authenticate that understanding. Also, I recommend you couple an image with text for the most efficient communication.
Good luck.
Thread | Message Archive | Home
Patricia Chalk (November 23,1995)
The sub-domain is called cartography in some circles. In some ways you and I deal with some of the same sort of representational problems for spatial data - in your case the space is some part of the the human body, in mine it is some subsection of the world. I am a cartographer! On a daily basis I design patterns or icons to represent some phenomena or other occuring over the earth's surface or among it's peoples. Perhaps you should align yourself with a cartographer who has skills in this area to assist with the design of the symbology you need.
Regards.
Thread | Message Archive | Home
Hi all,
I would like to add to this thread the work in psychology in category theory. Rosch, Mervis, Lakoff have very interesting things to say about basic, subordinate and superordinate categories of objects that impact representation as 'natural' or intrinsic...
Thread | Message Archive | Home
Betty Murphy (November 29,1995)
To Justin Starren:
I would strongly recommend William Horton's "The Icon Book," published by John Wiley (1995). It has excellent introductory chapters, including Chapter 3, Representing Ideas Graphically; Chapter 4, Showing Relationships, and Chapter 5, Designing an Iconic Language. Later chapters deal with Drawing Icons, Color in Icons, and Testing Icons, among other topics.
Beyond this, as suggested by Brad Weed, consulting a graphic designer would be a good idea. And combining text labels with your icons is also strongly recommended in the literature on human-computer interaction, which is voluminous.
Best wishes.
Thread | Message Archive | Home
Justin Starren (November 29,1995)
Thanks for all of the excellent responses. It is clear that others have faced similar problems. Werner Kuhn's, James Souttar's and David Sless's responses speak very directly to my problem. The "Icon Book" sounds like a must have resource.
Myke Gluck's observations on psychology are also interesting. As I am not familiar with the psychology literature, are then any particularly central papers, or is there a literature database to help find them?
Patricia Chalk's parallel to cartography is very interesting. I would wonder how cartographers approach the need for new symbols. For example, if I need a map that shows widget factories and foobar factories, do cartographers have a standard method for creating those symbols? Or do they learn a little about widgets and foobars and make a guess? If there are standard methods, I'd be very interested in pointers to the literature.
I agree with Brad Weed that graphic design skills are valuable, and generally underrepresented in software projects. His suggestion to "try to understand" the field and then "test imagery to authenticate that understanding" is exactly the type of "generate and test" cycle I'm trying to address.
If one imagines all possible representations for a concept spread out as a plane and then plots the "goodness" of those representations as the vertical scale, one will get a surface with hills and valley. Many representations will be simply bad and appear as local minima (or "valleys"), but some will be good and appear as local maxima (or "hills"). The goal is to find the global maximum (or highest hill) on the graph.
My concern is that a "generate and test" cycle will be equivalent to a simple hill climbing algorithm and be very good at finding local maxima, but may not find the global maximum. Of course, there is probably no way to be know what the *true* best representation is, (if one exists), but starting with a better initial representation should get one closer to the global maximum.
A skilled graphic designer can make any design better (at least any design I could come up with). A graphic designer may also be very good at comming up with initial designs for the general public, but I suspect that getting optimal designs for expert domains may require such intense knowledge of the domain that experts are the only valid candidates for generating the initial designs. (Of course, we could train a graphic designer as a physician in every subspecialty, or visa versa, but that would get very expensive.) Thus, I'm looking for a methodology for to extract from the experts representations that could then be refined by conventional rapid prototyping and graphic design principles. Some of the articles mentioned seem to address this need.
Thanks again for all the input. I'm off to the library.
Thread | Message Archive | Home
Peter Simlinger (November 29,1995)
Reacting to Justin Starren's question concerning "Discovering intrinsic representations":
To show an infiltrate one might not need an "infiltrate icon". Depending on the target audience and the communication purpose of the message various means of visual presentation can be imagined.
However, if it must be an icon, one should know that ISO / TC145 standardizes pictograms of any kind (ISO 3000). But if it comes to signs in general and pictograms in particular with some affinity to radiology DIN might have been more productive (DIN 30 600). The "symbols" shown in these standards are considered to be expert symbols which must be learnt (TC145 as well as DIN do not do any testing to find out whether the pictograms proposed for standardization in either ISO 3000 od DIN 36 600 are understood without previous learning).
ISO / TC145 / SC1 has elaborated and is in the process of elaborating standards resp. technical reports dealing with the development, application, and testing of public information symbols. Still, these documents would not explain how graphic images are conceived and designed up to such a standard that they are correctly interpreted (so that their image content triggers off the intended associatons, thus qualifying to be considered conventional signs = symbols). On the work of ISO and DIN every national standards institute should be able to give further information.
To my knowledge until now there is no ISO or DIN body focussing on pictograms (icons) for the display on screens.
Be aware that usually good visualizations are created by designers with specific education and experience. Good interpretations of e.g. the nature of infiltrations and recommendations how to treat them might equally well require previous expert training and experience.
Of course, you might occasionally come across some stimulating "how to do" literature. "Designing Pictorial Symbols", which actually relates to trademark design, is such a book which I think could inspire creative thinking. However, trademarks and icons are difficult to compare.
Nigel Holmes with Rose DeNeve: Designing Pictorial Symbols, Watson-Guptill Publications, New York, 1985, $ 27.50
Thread | Message Archive | Home
Deirdre Morrissey (November 27,1996)
I am a fourth year psychology student at Trinity College Dublin. For my final year thesis I am investigating why the care labelling symbols of clothes are so difficult to understand. I believe that existing care labelling symbols are ineffective because they are badly designed.
Context:
I wish to focus on the development, design and evaluation of visual information products such as graphic symbols, user instructions and warnings in a context of sharply increasing incidences of novel and quite complex products for public information and awareness. I wish to know other peoples views on how applied behavioural psychologists and designers can learn from each other and work together to create effectively designed products.
Why don't we always obey warnings? Warnings should be matched in their urgency to the urgency of the situation they warn about. In my design I intend to investigate if an array of negative care labelling symbols will have a more positive effect that an array of positive symbols, i.e. the idea that consumers will pay more attention to "do not" symbols than to "do" symbols.
In an undergraduate study carried out in this college last year, it was found that 65% of her subjects misinterpreted the "do not iron" symbol incorrectly and 40% interpreted the triangular bleach symbol as the symbol for ironing because they stated that the triangle facing them reminded them of an iron in a standing upright position.
The majority of these symbols do not follow basic design priciples and also lack a very necessary marketing strategy. I already have information from several editions of the Information Design Journal edited by Paul Stiff (Vol.1, no.3, 1979; Vol.6, no.1, 1990; Vol.6, no.2, 1990; Vol.7, no.3, 1994).
Any information regarding public graphics, warning design and warnings and risk communication would be greatly appreciated and also any recent work by Bob Dewar, who has written extensively in the area. Also I would love to hear more from David Sless and his work on symbol design and any advice that he could give me.
Thread | Message Archive | Home
Marian Petre (December 3,1996)
Deirdre,
You might want to contact Yvonne Rogers at Sussex University, whose doctoral research was on icons in computer interfaces. I believe the title was something like "Pictorial Representations of Abstract Concepts in Relation to Human-Computer Interaction". I suspect that care labels bear some analogy to interface icons.
Thread | Message Archive | Home
Kirk Doggett (December 3,1996)
Deirdre:
Here's a book that may be very helpful to you:
The Icon Book, (Visual symbols for Computer Systems and Documentation), by William Horton, John Wiley & Sons, Inc. publisher, 1994, ISBN 0-471-59900-X (400 pages)
John Wiley & Sons, Inc., Professional, Reference, & Trade Group, 605 Third Avenue, New York, NY 10158-0012
"This book collects advice, design principles, rules of thumb, thousands of examples, and just a dash of theory into a guide for icon designers. It draws on my own experience leading icon design efforts and on the work of many of my consulting clients. This book strives to report and show those good design practices that most of us discover only by costly trial and error..."
-from Preface, p. iv.
Though targeted at computer interface designers, the information contained in this book can apply to *any* icon design, not just for computer screens.
I hope this is helpful.
Thread | Message Archive | Home