A blog post about what a well-designed graphical abstract can do. More than a text…
It always intrigues me when a figure looks really complicated. And I start to wonder: does it have to be that confusing? How could I fix this? Either with better design, or even the underlying concept needs a major makeover.
A tweet from science magazine showed the visual comparison of two models of body weight regulation. I tried to compare the two and that was hard. It shouldn’t be, which is why I read the publication and tried to fix the figure. In this blog post, I am sharing three things that I would change: color, arrangement and visualization of the flux, plus two minor design changes.
This is how the original figure looks like:
Color is one of the most powerful ways to catch people’s attention. Not all colors are equally drawing our attention to them, but a bright red certainly is the one color that does. If we take a look at the figure as a whole, the red parts pop out. Let’s take a closer look at what is red and why.
Red was used to color the food, the tissues and the repression arrows. To show repression in red is conventional and given the shear number of connecting lines, plus that they are grouped in three types of lines (repression, stimulation, flux) it certainly makes sense to use different visual cues for the different types of lines. Thereby it is possibly to visually separate repression from stimulation and flux. It becomes very easy for the reader to focus on each of these groups and quickly tell which of those groups a certain line belongs to. Black is flux, blue is stimulation, red is repression. Got it. But wait, that is true – except for – all those other red things.
There is too much red in here to make it work. While the tissues and food have a slightly more orange hue than the repression lines, they are too similar to be visually distinct. My suggestion would be to reserve red for the repression lines and use more subtle colors for the food and the tissues. This avoids blending and makes each category visually distinct.
Order of Reading
Looking at the figure at a whole, we quickly grasp that it is divided into two parts. Following the conventional order of reading, we start at the left part in the upper left corner. That means what feels like a natural starting point is “Energy expenditure”. This is not the intended starting point. The intended starting point is actually “Food environment”. My suggestion is to flip the layout vertically for each of the two models. This leads to “Food environment” being the first thing people read.
Let’s take a quick look at the message first: The carbohydrate insulin model is describing a feed forward loop, starting with high carbohydrate intake, leading to fuel being distributed to adipose tissue by postprandial insulin. As fuel is now missing in other tissues, these are sending a cellular starvation signal, leading to higher food intake. Plus, the missing fuels are leading to lower energy expenditure, further enhancing the imbalance of energy intake and energy expenditure.
Flux seemed to be key. Other tissues are fuel deprived because the flux of fuels is directed to adipose tissue. This is not visible at all! We need to make our way through multiple arrows that show the direction of the flux, while other little arrows going up or down and labels give us some hints about the magnitude of that flux. My suggestion: use a sankey diagram. It shows both the magnitude and the direction of flux.
Speaking of visualizing it: Another small change I made is at the food composition. Carbohydrate intake is greater than fat and protein intake. This should be reflected visually. Here, it is only represented by a small arrow next to the label “Carbs”. Increasing the size of the Carb icon (or the frequency) compared to the other two icons will give a visual cue that there is more carb than fat or protein.
Minor design changes
When looking at the initial figure, I thought that I would prefer the blue arrows being more distinct from the black lines. This is why I intensified the blue color.
The makeover figure uses visual layering (more about that in a future post – it is one of my favourite design tools, I certainly will talk about that again). To explain it just in the context of this figure: the original looks messy because a lot of lines are crossing each other. While in the left part of the figure the makeover avoids this altogether (by re-arrangement), on the right part we can see the same crossings as in the original. But they are a lot less disturbing, because the bright blue lines seem to float on top of the grey area representing the flux. The blue lines and the grey area are on different visual layers – this is visual layering.
Diving deeper – the (hidden) meaning
This is the end of how far you can go with better design choices on what I call the cosmetic level. The cosmetic level is what you can do without changing what is displayed, but rather how it is displayed. For this figure, I feel it needs a major makeover, rethinking what needs to be shown to get the key messages across.
What initially was drawing me to this figure was the level of confusion that arose when trying to make sense of it. Two models are compared. When contrasting two things, it should be easy for the reader to tell what the major difference is. However, I stared at the figure and while discovering minor differences, I just couldn’t tell what the point is. That is a bad sign. Figures should help in explaining scientific ideas, not concealing them.
Eventually, I was grabbing my highlighter, starting to highlight the differences I could spot. What are the differences? What is the one key message with that comparison between the two models?