How scientists fail where artists succeed

In 1543 Andreas Vesalius published De humani corporis fabrica which is one of the earliest and most beautifully illustrated anatomy text books ever written. The detail put into each illustration is tremendously exquisite, and each panel stands alone as a piece of art. The effort and care put into those drawings was probably as much for artistic reasons as for educational ones, but the artwork was eminently useful for spreading anatomical knowledge beyond the dissection tables. Three hundred years later, Henry Gray would publish the first edition of Gray's Anatomy, which would become the foundation of anatomical education for the next 150 years and inspire every anatomical textbook that followed it. Visualizations have gotten more sophisticated and in today's classroom, we supplement our textbooks with physical and digital anatomical models, but they are all fundamentally refinements of the humble anatomical drawing.

Sketching anatomical structures is a useful method of studying because it forces a student to look at shapes and forms in a complete and visually critical method, but most students don't engage in an artistic examination of human anatomical forms because of their lack of artistic training. The extent of a student's engagement with any educational visualization is often just a shortcut to understanding the material without having to read extensively. It is pretty obvious that visualizations are useful tools for learning, just imagine how much harder it would be to understand how a car's differential worked by reading a narrative paragraph about the rotation of spider gears on two simultaneous axes versus how much easier it would be to just watch an animation with an accompanying verbal explanation. You don't need to look any farther than at how popular instructional YouTube videos are to appreciate how much more readily people will learn from visual information than from written information. At this point, you may be thinking to yourself, "but science textbooks have all kinds of illustrations in them, and I always turned right to the pictures to help me learn.", and you'd be right to think that... partially. Science textbooks are replete with images, but they're usually cartoonish and unrealistic diagrams that oversimplify a complex system. If you do a Google image search for "Animal cell", you'll get a myriad of derivative cartoons of a blobby sack of water cut open with each organelle carefully labeled. So what's wrong with that? Well for one, the cells in your body (and in all multicellular animals) are packed closely together, so they actually have more geometric shapes with flat sides, there are no blobby oval-shaped cells in your body. But maybe more importantly, the nice clean image of over-sized organelles neatly arranged so that everything is on display equally is just not how cells are. The diagram is a visual laundry list of all the parts of a cell, but not a representation of what cells are really like. If you ask a biology student how many mitochondria are in a cell, they might think back to that cartoon in their textbook that had two or three. Would that cartoon be informative for their estimate of red blood cells that don't have any mitochondria, or their estimate of a liver cell that can have a thousand or more mitochondria? When a student learns from a cartoonish oversimplification, they also pick up unintended elements of the diagram as factually true about that system, like the number of mitochondria. The diagram was intended to just show the student that mitochondria were an organelle within the cell, but the student learned several things that were true of the diagram to also be true of cells. Perhaps for the very reason that it is so easy to learn from visual sources, we should be careful with what we draw.

Another problem with visualizations is that they fail to teach variability, and sometimes worse than that, they teach students an 'idealized' form that all examples should look like. Most of the skeletons that are used in anatomy classrooms in America are now high quality plastic replicas of just a handful of individual skeletons. The skeleton in my own classroom is a 'typical' adult male (although he's only 5'3") and he doesn't show any sign of a frontal suture in his skull (a suture is where two skull bones knit together permanently). Everyone was born with two frontal bones that fuse together so tightly that the seam is mostly invisible, but there is a substantial subset of the population who retain a visible and distinct frontal suture throughout their lives. Inexperienced radiographers have mistaken persistent frontal sutures as a skull fractures [1]. If their training included an emphasis on the diversity that anatomical forms can take, that type of mistake might be less common. Maybe instead of a single image, what we should use is entire galleries of images, each one of a different view of a different subject. Let's also consider a more mundane example: undergraduate testing. We can imagine a student learning a phylogeny of crustaceans by opening their textbook and studying a picture of a representative of each clade. If they see the same pictures on the exam, they'll be more likely to get the answer correct than if they see different pictures of the same species. In this hypothetical, the student is learning to identify the image itself not the subject of the image. We might intend to teach the student about the elements that differ between the subjects, certainly we'll mention that lobsters have two pairs of antennae whereas crabs only have one, and that all you need to do to differentiate them is count the antennae, but the students might see a much easier heuristic, the lobster is facing to the right in the picture and the crab is facing to the left. When it comes time to take the exam... well I think you get the point.

For the moment, lets ride with my thesis that science communicators have faults in how they communicate with visual aides. After all, they've merely co-opted visual communication as a tool to inform others about scientific subjects. So who would do it better? Who should the scientists learn from? A graphic designer might be able to understand and translate the elements of a subject into the elements of a visualization much more effectively than us. Imagine that crab/lobster fiasco again only this time the student learns from two icons that make the contrast of antenna count the most bold and obvious visual feature, that student might be able to use those icons as tool to decipher an actual photograph on an exam. Picasso knew how to translate the elements of an actual bull into a visualization of a bull, even without making a painting of a bull. I would love to see how Picasso would've illustrated an animal cell. Magicians can expertly draw our attention to a single card or the tip of their magic wand, not through enchanting spells, but because their actual trade is the manipulation of our attention. If we want to know how to get students to look at the important parts of a graph on a PowerPoint slideshow, maybe we should be asking magicians for their advice. Sculptors know what it really takes to craft an enduring image without mistakes. If we are going to put drawings in textbooks that will be circulated globally for decades, maybe we should ask a few sculptors how to make sure they're perfect. If we want to get students to 'feel' at a deep and intuitive level, the immense size of a blue whale, maybe we should be asking the artist who create giant CGI monsters for movies. If we want to know how to show students a picture of a sea turtle and have their heart race because they image that animal having decades of adventures circling the globe, maybe we should ask a portrait photographer to show us that story hidden in it's eyes. Let's look around and take note of some of the amazing things that people show us, at the brilliant stories that are told visually by the masters of their crafts, then let's ask them to show us how we can tell our stories a little bit better.

[1] https://doi.org/10.1016/S1130-1473(07)70288-9