in science imagery

by Ingrid Kallick, NCSA Graphic Designer, USEO; Graphic by Ingrid Kallick

NOTE: This AAAS panel was organized by Ingrid Kallick and David Curtis of NCSA and Bonnie Caroll, Secretariat-Director of CENDI (Commerce, Energy, NASA, National Library of Medicine and Defense Information management group). For further discussion of this topic, see Ingrid Kallick's article, "Science Icons: The Visualization of Scientific Truths," in a forthcoming issue of Leonardo (volume 27, number 4, 1994).

What makes imagery important to science? What makes scientific imagery truthful and accurate? Can scientific imagery be objectively evaluated? If so, by what criteria?

These and other questions were the focus of a symposium at the annual meeting of the American Association for the Advancement of Science (AAAS), San Francisco, in mid-February 1994. A panel of visualization specialists included Donna Cox from NCSA, computer graphics artist and UIUC professor of art and design; Karen Kaye of the NASA Science and Technology Information Program, specialist in scientific image database management; Gregory McRae of MIT, chemical engineering professor who has used visualization extensively to inform emission control policymaking; Jon Palfreman from WGBH (Boston's Public Broadcasting System outlet), executive producer for NOVA and numerous documentaries utilizing scientific graphics; and David Curtis of NCSA, creative director of scientific communications and executive producer of NCSA video programming. Curtis introduced the symposium.

Insight through images

Images have long been a part of the communication and practice of science, but the technologies that drive scientific visualization are in development. Once relegated to the static pages of textbooks and academic journals, scientific images can now be set in motion. Broadcast to homes, shown in the conference rooms of policymakers, traveling across the Internet, and being animated on computer screens, scientific images continue to expand their role as visual interlocutor between research scientists, the media, and policymakers.

"This panel seeks to address a specific need, namely that in an image-hungry society like ours, a keener awareness of the elements of 'truthfulness' or 'validity' of scientific imagery and visualization is desirable for a greater understanding of science--and particularly its impact on society," said Curtis in his introduction.

Animation and video media have transformed scientific images into powerful storytellers. Yet how reliable are their stories? Conceptual illustration and data-driven visualizations are often used interchangeably. While data- driven images are usually considered to be more valid, the relative validity of one form over the other is not always clear.

lllustrations are subjective pictures of scientific concepts, yet the supposed objectivity of data-driven visualizations from numerical simulations may not be so different. In visualizations there is no one-to-one correspondence between data and representation, and the process of numerical simulation introduces subjective elements. In a recent article in Science [1], the difficulty of evaluating numerical simulations was discussed. Similar concerns have prompted the DoE to begin work on standards for evaluating numerical simulations of groundwater pollution. [2]

The debate about standards for visual representation persists along with questions about numerical simulations. This was illustrated by an animation from the Cornell Theory Center by Animator/Producer Wayne Lytle called The Dangers of Glitziness and Other Visualization Faux Pas, using fictitious software named "Viz-o-Matic." [3] The video, shown in the Electronic Theater at SIGGRAPH 93, documents the enhancement and subsequent "glitz buffer overload" of a sparsely data-driven visualization trying to parade as a data-driven, thoughtfully rendered presentation. (See Figure 1, below, courtesy of Wayne Lytle, CTC.) The video was a humorous response to discussions at an April 1992 meeting of the MetaCenter visualization community at SDSC.

Insight gained

Though some might question any use of "glitzy" graphics, McRae argued for using persuasive and sophisticated--if not exactly glitzy--visualizations. "Imagery and images are extremely important in the policy process. . . . In order to be able to interpret whether or not an image may be telling you the truth requires a level of understanding and education."

McRae says that to inform a decision-making body of basic scientific concepts, it is imperative to have a presentable visualization. With examples of his work on monitoring air pollution in Los Angeles and Mexico City, he sought to demonstrate that a carefully crafted and readable visualization will do just that. (See Figure 2, below.) For McRae's purposes, the risk of possible misinterpretation of visualizations was secondary to the benefits of the audience understanding the science. As McRae reminded the AAAS audience," 'The purpose of computing is insight', not numbers." [4]

Insight in question

More recently, concern has grown in the visualization community and the science media over the use of scientific imagery, its interpretation, and the relation of the data to its interpretation. Computer manipulation of images, in the general sense, has been a subject of interest in the popular press. Appearance is important to the interpretation of a scientific image, but not solely. Context, narration, sound, authority and viewpoint of presentation, editing, and reliability of and amount of underlying data contribute to the meaning of a scientific image; all can give very different interpretations to the same design.

Palfreman emphasized the metaphorical qualities of imagery in his presentation, stating, "Like literary metaphors, the visual metaphor aids communication by suggesting how a new unfamiliar thing--for example, the life cycle of the HIV virus--can be thought of in more familiar terms." Science is presented as a story or a metaphor. In light of this, scientific imagery must be evaluated as information, as well as design, demanding a broader view of both.

When they are removed from the scientific research, poorly labeled, or shown in a sequential video without context, scientific images tell little about the procedures that generated them. Intentionally or unintentionally, the use and reuse of such images may mislead viewers. Texts in scientific journals are formally peer reviewed, but scientific imagery is not.

Cox emphasized that peer review may be the necessary step in improving the credibility of scientific visualization as a vehicle for insight into data. "There are currently real problems with the. . . untestability of visualization techniques. . . . We do not have peer review of the algorithms, data sampling techniques, standards for spatial/symbolic mapping, or the methods for representing the data. This condition places scientific visualization in a precarious position."

Cox said that usually the intended audience and message determine the form of a visualization. It is imperative to have this information when evaluating a visualization or other scientific imagery, and it should be passed on when images are placed in a new context.

New contexts, new media

The computational environment of visualization is currently experiencing an explosion of online capabilities.

Kaye presented solutions for documenting, accessing, and scrutinizing scientific imagery online, including a NASA Home Page and image database using NCSA Mosaic with (currently restricted) access to the underlying visualizations.

NCSA's numerical relativity exhibit and digital library are also being expanded to include detailed information about NCSA scientific imagery.

Another example given was Al Globus's "Annotated Scientific Visualization URL Bibliography," which provides useful information about what visualization resources exist online. Globus is a research assistant at NASA Ames Research Center.

Critical insight

The consensus of the panel was that online hypermedia holds great promise for developing presentations that combine narrative, sequential videos with a rich, in-depth database. (An example is the CD-ROM presentation accompanying the book by Robert Wolff and Larry Yaeger that is reviewed.) While information design issues continue to be important, context, use, and intent must be considered when evaluating scientific images.

Finally, it was determined that education and training of producers and consumers of scientific imagery will enhance critical thinking about scientific images.

References
1. N. Oreskes et al, "Verification, Validation and Confirmation of Numerical Models in the Earth Sciences." Science 263: 641-646 (February 4, 1994).
2. Ibid.
3. "Theory Center Visualization." Forefronts 9:1 (Summer 1993).
4. Quote from Numerical Methods for Scientists and Engineers by R. W. Hamming (New York: Mc Graw-Hill, 1962). Note: The 1973 edition is prefaced with this variation of the quote: "The purpose of computing is not really in sight [sic]."

access / Summer 1994 / NCSA