The third day of the GRC 2011 Science Visualization Conference. The line up today was Ric Lowe, Curtin University; Vickie Williamson, Texas A&M; David Geelan, University of Queensland; Graham Johnson, The Scripps Research Institute; and Larry Gonick, San Francisco.
Ric Lowe (Curtin University, Perth, Australia) kicked off the morning with an excellent talk called “Dangerous Dynamics: Animated Visualization as a Two-Edged Sword”. Lowe talked about why animations can be problematic for learners. Animations have features that can be a poor fit with well-known limitations to our human capacities to process. Animations are transient, packed with information, simultaneity across display, and require the learner to integrate over time. Various solutions have been applied – cueing the learner, slowing the action, including insets, providing the learner with strategies. Lowe contends these solutions are not sufficient. He suggested more tailored approaches, specific to the animation’s content – that is, getting inside the content sufficiently to know how to break it down and steer the learner toward the significant parts of the animation and why.
We tend to present animations as a full, coherent, and operating system. Perhaps what we need to do, instead, is to facilitate the composition of a mental model inside the learner’s head. Some of the information they need to understand the model is already there, but there is other information that the learner will have to obtain from the animation. So what do they pull out of the animation and in what order? (like putting back together a dis-assembled watch, there is significance to the order in which you reassemble the pieces). partitioning the flux. We readily partition everyday experience into coherent objects, events and scenes. We are very good at recognizing and using spatial and temporal boundaries and that partitioning is a big part of our understanding of the experience.
The extraction system must be a cumulative process. It’s important to remember foveal vs. peripheral vision. The variation in clarity, detail, color across the visual field are different and you are able to notice movement in your peripheral vision but detail, color, focus less so. Try this: Extend your arms in front of your eyes with your thumbs up, and stare at them. Notice the detail. Now move your arms (thumbs up) out to the side of your body and notice the change in your ability to see your thumbs. You know they are there, but you can’t see detail. Now wiggle them. You can definitely detect the movement, event though they are in your peripheral vision.
Lowe also talked about salience and relevance in a very interesting way. If my perceptual salience and thematic relevance are misaligned, there will be learning problems. If you add dynamics to a static graphic you run this risk. The dynamic contrast may out-compete visual-spatial contrast regarding the direction of the learners’ attention. If you have a mismatch – that is, low salience and high thematic relevance – there’s a problem. And if you have dynamic elements to those low salience items, you exaggerate the problem.
Lowe cautioned us that the act of decomposing an animation – dynamic relationship cueing (progressive path), focus on event units (build up entities), single event units or groupings of them for composing – can be tricky. Learning from animation is iterative and cumulative. Feed the learner a carefully sequenced set of information. Each subset consists of one or more event unit with well-defined boundaries. He advises that we shift the emphasis to the composition process.
Here is Lowe’s original paper, Perceptual and Cognitive Challenges to Learning with Dynamic Visualizations, on which his talk was based.
David Geelan, University of Queensland, gave a talk called “Does it Work? For Whom? Visualizations in High School Classrooms.” He talked about a new book out by John Hattie called Visible Learning, that points out that almost everything works. So, what works is maybe not the right question, we should instead be asking what works best? Hmm…. I question the notion that “everything works”. Maybe most things that educators try work, for at least a short time. But this doesn’t seem like a solid conclusion to rely upon.
Geelan reported a study from his research group, comparing high school student conceptual knowledge gains (in chemistry and physics) when taught with and without visualizations (in this case, simulations with variables that can be changed by the student). His results? No statistical difference. Visualizations made no difference. He goes on to say that these visualizations are doing no harm and since they do provide other benefits (e.g. motivation, enjoyment, and engagement), we should use them, but we should be very careful about the claims we make about their potential impact.
While Geelan’s talk was provocative, I thought that the simulations he used in his study (all of which were “found on the web”) were of variable quality and the two that he showed us were, to my eye, remarkably weak. Also, the topics of the visualization (Newtonian physics and Le Chatelier’s Principle) are not principles that obviously benefit from a visualization – as one questioner pointed out, these are concepts that can typically be understood by students with a text description. His research also did not look at the context for these computer-based simulations – how did the teachers use them? DId the students do these as homework or in class? Did they work individually or in groups? Nonetheless, I think it’s enormously helpful to hear talks like this from a researcher who is not afraid to present “unattractive” data and question assumptions. It’s very hard to get negative results published and it’s important to remember that it’s important to listen to all of the data.
In the evening session we had two artists – Larry Gonick and Graham Johnson. Johnson is a classically trained medical illustrator who has now completed his PhD in biophysics. He has been working to take one of David Goodsell’s beautiful water colors (example pictured to the left) of cell interiors and turn them into 3D models. Part of his talk illustrated the packing problem behind this objective – how to efficiently pack all of these shapes, surfaces, and ingredients into these 3D mesoscale environments. He showed us some beautiful images – most of which were over my head – but they certainly gave you a good feeling for these crowded, complex cellular spaces.
When asked about where this is all going – what he sees the long-range future to be – he talked about the idea of simulating cells. Turning these static 3D models he’s creating into agents with behaviors that could interact in systems. With that, one could replicate cell functions – perturbate them, test them, and see where it goes.
Larry Gonick was a breath of fresh air. As he put it – no analysis, no research, no data, and no conclusions. Gonick is a cartoonist who has written a number of “Cartoon Guides” to various topics – chemistry, genetics, statistics (has sold the most), sex, physics, calculus to name just a few.
Allegory and analogy, says Gonick, are the cartoonist’s tool. You don’t show the thing but you show something that is like the thing, to make the point. He sings the praises of simple (obviously hand-drawn), black and white illustrations (which his all are) – they stand out and are incredibly compelling. His favorite comics online? XKCD and Hark! A Vagrant.
All comics are based on characters. He uses characters in his comic books to embody something important and serve as a guide for the reader. For instance, in his statistics comic book, he used Sherlock Holmes and James Watson characters to discuss the differences between inductive and deductive reasoning. Or he uses little guys who are performing the action, point out where to look, or helps the subject come to life.
Comics are really quite cinematic – they tell a moving story – with the added advantage of being able to change the size of the frame. Gonick says that he likes constraints (certain lengths, certain sizes, certain tools). He claims that they help him organize his thinking and get it done. He talked about the importance of narrative and explained that it’s important to have a beginning, a middle and an end when illstrating a procedure. More precisely stated, he explains that he’s always careful to add a beginning and an end to the step-by-step middle of a procedure. His science stories do suffer from anthropomorphism (Larry would say that they employ anthropomorphism) but, as he said, “I don’t worry about that so much”. What he thinks he accomplishes with his comic books is distillation – he makes selections and finds essences.