Our sensory memory helps us recognize important information in the world around us at lightning speed. We do this by scanning for pre-attentive features. In a mini-series of three blogs, I'll explain how to make use of these powerful tools in data visualizations. With today: how to use shape effectively.

A brief introduction

Just a brief introduction: pre-attentive features are characteristics(attributes) of things around us that we register in our brain before we are really aware of them(pre-attentive). We do this in our sensory memory(sensory memory). Now, most people have five senses (sight, smell, taste hearing and touch), but we limit ourselves here to sight and thus pre-attentive visual attributes. 

We can divide pre-attentive visual properties into roughly three main categories: shape, position and color. Of course, you can very well combine different categories with each other, but which feature has the most effect at which moment depends on what information you want to convey. Today we look at the different ways to use shape effectively.

For what visualization would you use a Penrose triangle?


Shape is a multifaceted property of visual objects since it is not only related to external features, but because it is also related to the environment. However, these properties are discussed in the blog on the third category of pre-attentive properties: position. In the external world, we have three spatial dimensions in which a shape can exist, but since data visualizations are usually displayed on a screen, this does not apply here. Nevertheless, there are several ways to employ shape to guide the eye:

  • Length
  • Width
  • Size/size
  • Form

Below I explain for each of these four points how best to use them, and also what not to do. 


Length, or the difference therein, is one of the most easily distinguishable characteristics to the human eye. Even a small difference is still very easy to perceive. It is not for nothing that the bar chart is one of the best-known forms of data visualization. Length is good for comparing quantitative values. The prerequisite is that the objects are on the same axis or line, and preferably also sorted. Otherwise the lines are more difficult to compare, as you can clearly see in the visualization at the bottom of this page.


When do you talk about width, when about length? You can argue about that. In this case, by width I mean the emphatically shortest side of a mark, regardless of orientation. Width is also used for quantitative values, but less often than length. This is because, by (my) definition, the differences in width are always smaller than the differences in length, making this form not so easy to deploy. 

But when the number of marks is small and the exact differences between lines are not important, width can be used just fine. A popular example of this is the Sankey diagram(by Ian Baldwin), in which the marks branch like a river several times. In the example below, this happens only once, but there are also examples with further branching. Finally, width can also be used textually by highlighting totals, for example.


Size in visualization practice often refers to the area of a mark. With that, size is a two-dimensional unit, where length and width are usually equal (squares, circles or stars). So it is obvious that size is best employed for quantitative values. Yet precisely because of this, different objects are not very easy to compare, since our idea of an object's size is strongly influenced by its surroundings. It is easier for the human eye to see differences across a single axis or dimension than across two. 

Size can play an important role at a higher level, though. In the design of a dashboard, large headers and visualizations stand out more than small ones. For example, the titles and headers in a blog attract your attention more strongly than the individual text. 

Both center circles are the same size


This blog is about form in the broad sense of the word. But this paragraph is about a specific shapes, shapes in English, rather than the more general form. Examples include a square, circle, triangle or star, but in Tableau it is possible to use your own shapes or even import images. Shapes are mainly used to group or distinguish categorical or qualitative values.

Here it is important especially not to use too many shapes. As with many forms of visualization, on average we can store up to seven, but sometimes even fewer, individual objects in our short-term memory. Above that we lose the overview and quickly make mistakes. If possible, using shapes or images with a direct visual link to the value they depict helps, as in the example in my Tableau Public-story below.

A visual reminder

The above tips are mainly intended to help you design your visualizations as effectively (and thus simply) as possible. But of course, effective and simple is not always beautiful, fun or funny. So definitely feel free to unleash your creative mind to create a great infographic. But if you're looking for some well-meaning advice, you can always grab the story below. It visually summarizes the most important tips from this blog.

This is the third of a series of blogs on the how, what and why of data visualizations, and second in a mini-series on pre-attentive properties, in addition to color and position. More blogs (with links) will follow in the rest of this month. Click here for the complete listing of all my blogs, or check out my profile at The Information Lab or on Tableau Public.

Check out our other blogs on Tableau, Alteryx, and Snowflake.

Work together with one of our consultants and get the most out of your data. 

Contact us, and we'll help you right away.