También disponible en Español

Inf@Vis!

The digital magazine of InfoVis.net

An image is worth 1000 words
by Juan C. Dürsteler [message nº 18]

Not always. Quoting Colin Ware, Director of the Data Visualization Research Lab, 'Natural language is the most elaborate and universal symbol system that we have'. We learn it from our childhood and we use it constantly. For this reason visualisation is necessarily a hybrid of images and words. How and when to use words or images is one of the most important decisions the visualisation designer has to take.

Let's think of the outstanding boom of mobile telephony, that allow us to speak everywhere, whenever we want. Nowadays it's difficult to imagine a more ubiquitous and easier to use communications system.

The power of language relies on the fact that with only one word we can evoke images, sensations or complete experiences lived previously.

Verbal information and visual information excite different zones of the brain. The language is essentially sequential and dynamic. To read or to explain something unknown beforehand requires a certain amount of time. Contrasting with this, relatively large sections of static images and diagrams can be understood at once. 

The basic description of the platypus occupies 85 words in my encyclopedia and, frankly, without the photo it would be very difficult to clarify what type of animal is this. Nevertheless, for a person that already has seen or knows what a platypus is, the mention of the word is enough to link a complete set of experiences related with this animal.

How can we use, advantageously, both (verbal and visual) systems in order to improve the understanding?. In chapter 9 of the interesting and very commendable book of Colin Ware 'Information Visualization: Perception for Design' some of the most recent research on this topic are revised. The key findings that are there presented are summarised in the following lines:

Static Images:

  • Images better than text
    • Images are better to show structural relations, like links between entities and groups of entities.
    • The tasks that involve localisation information are better represented by images.
    • Visual information is generally better remembered than verbal information. This applies except for abstract images.
    • Images are better to provide detail and appearance. Some studies suggest that first we comprehend the general structure and shape of an object and then the details.
  • Text better than images
    • Text is better than graphics for abstract concepts like freedom or efficiency.
    • Procedural information, like computer algorithms, are better represented using text (flow diagrams have long since been abandoned for pseudocode). Static images are not effective by themselves to present complex, non-spatial, instructions. Nevertheless, there are exceptions like planning using Gantt Diagrams.
    • Text is better to show information specifying conditions under which something should be done or should not be done.

Animated images.

The work of different researchers shows that some events can be better represented using movement and animation.

  • Animation allow to represent causality in an effective way. With an appropriate animation a causal relationship can be perceived in a direct way and unambiguously.
  • The acts that express communication or flux are better presented through animations. As an example, a communication event can be described with a symbol (the message) moving from the source to the receptor.
  • Reorganization and restructuration phenomena are well described with animated images whenever the complexity is low. A structure can be gradually transformed using the animation to show explicitly the phases of the reorganisation.
  • In Software visualisation, a sequence of movements in the data structures of an algorithm can be presented through animation to improve the understanding of the underlying mechanism. A good example of this, also mentioned in the book, is the movie 'Sorting Out Sort' that represents the movements of the data in the memory for several sorting algorithms. You can see the animation clicking on the figure that appears following the previous link.
  • Animation can be helpful in showing sequences of complex spatial actions. An outstanding example is that of the assembly instructions that come in the CD-ROM of the Lego Mindstorms toys. They allow to see step by step with clever animations how to mount the parts of the robots properly. A great job.

To know when to use text and when to use images is by no means a simple task. The contributions of research are still in an early stage and many are subject to debate. In a hybrid visualisation scheme, the problem of how to link text and images properly arises. But this will be a matter for the next issue.

Incidentally, this message has been written using a speech recognition system…

Links of this issue:

http://www.ccom.unh.edu/vislab/CWBio.html  
http://www.ccom.unh.edu/vislab/index.html  
http://www.infovis.net/printRec.php?rec=llibre&lang=2#InfoVisWare  
http://swt-www.informatik.uni-hamburg.de/~1friedri/sv/jvlc/JVLC-fig2.html  
http://mindstorms.lego.com  
© Copyright InfoVis.net 2000-2014