Estimated reading time: 6 minutes

Summary

Consider already established practice when choosing colour, particularly as these are applied in your own field.
Be aware that colour is sometimes used very differently within different fields in the life sciences.
Physical conditions might play a role in your choice of colour as well.
The cultural significance of colour is a bias but can also be useful in thinking about which colour to choose.

If and when you use colour as part of your visualisation process, you will have a few key decisions to make. While there is no right or wrong and no definitive rules to follow, keeping a few principles in mind might be helpful.

Within a specific scientific field, there may already exist an established practice that you will need to take into account. A consensus may have arisen regarding colour usage, often because it increases understanding and allows for more effective communication.

Another thing to be aware of is that the meaning of colour may vary among disciplines; colour is sometimes used to send diametrically opposed signals depending on the field of study.

In other instances, physical circumstances or the cultural significance of colour might guide your choices, as well as our general pre-understanding of the meaning of different colours.

Also note that there may be specific journal policies regarding for example colour choice, image processing and colour manipulation when preparing images for publication.

Considering already established practice

One example of an established practice is the common division in anatomical textbooks between arteries and veins, an example of which can be seen below. Here, red is used to designate arteries; blue is used to designate veins.

Illustration of veins and arteries shown in blue and red, respectively.

Illustration of blood vessels depicted as blue on the venous side and red on the arterial side. Image by Karolinska Institutet University Library. Creative Commons BY SA

Oxygenated arterial blood is naturally a lighter red than darker, venous blood, which tends to be almost purple. In textbooks, this difference is exaggerated for educational purposes. Within medicine and biology, exaggerating the colours in this way has become its own distinct visual language.

Physical conditions also matter

Models of biomolecules

Models of biomolecules, such as neurotransmitters, proteins, DNA, etc., often make use of what is known as the CPK scale, a colouring convention aimed at differentiating between different substances. Practical molecular models, where atoms and bonds are represented by balls and sticks, generally follow the CPK colour convention. Many colours are furthermore rooted in some manifestation of the substance itself; for instance, carbon (C) is black; the colorless gas hydrogen (H) is white; sulfur (S) is expressed by a deep yellow, and so on. Here, physical circumstances and conditions are to some extent taken into account when creating a visualisation.

In the example below, for instance, the amino acid proline has five black carbon atoms, nine white hydrogen atoms, one blue nitrogen atom and two red oxygen atoms. Each is coloured according to the CPK scale.

Plastic ball-and-stick molecular model with coloured atoms, and colour key below image.

Source: Illustration by Karolinska Institutet University Library. Photo by Peter Murray-Rust, via Wikimedia Commons. Creative Commons BY SA

Fluorescence

Physical conditions also play a role when fluorescence is used, i.e., where the ability of certain molecules to emit light in an excited state is used for visualisation. For example, the technique can be used to "tag" a specific protein using a molecule that can emit light at a certain wavelength. The wavelength of the visible light emitted by the molecule then determines the colour used in the visualisation.

Examples of this would be nicotinamide adenine dinucleotide (NAD+), which emits a blue light, or green fluorescent protein (GFP), which emits light at the wavelength perceived as green.

Colour spectrum showing emission peaks for NAD+ and GFP.

Emission peaks for NAD+ and GFP marked on the colour spectrum. The light emitted by the two molecules when they return to a non-excited state is strongest at 447 nm and 509 nm, and is read as blue and green fluorescent light, respectively.

Image by Karolinska Institutet University Library, Creative Commons BY SA.

Nanoscale

Any cell or molecular biologist producing a visualised representation of their data will also need to consider the nanoscale, i.e., what exists below the wavelength range of visible light. Here, there are no colours. This means that all images produced using an electron microscope are monochrome. To the extent that such images nonetheless use colour, they have been coloured afterwards.

Närstudie av TSH-receptor, monokrom rendering av 3D-modell

Close-up study of a TSH receptor. The monochrome expression emphasizes the absence of colour at the nanoscale.

Source: www.scivis.it. Creative Commons BY NC SA.

If we consult Mackinlay's ranking, we see that texture, i.e. surface structure or pattern, is almost as effective as colour in separating categories from each other. However, the use of valence, or a weaker or stronger version of the same colour, which is almost inevitable in monochrome work, is much less effective.

What colours are used, how they are used, and for what purpose might vary, but if you are working at the nanoscale, know that the use of colour has to be a conscious choice.

Cultural significance

Some colour choices feel so natural that we hardly think about them. This is likely because they are rooted in preconceived cultural notions about what colours mean. An example might be a thermometer with both plus and minus on the scale, or an image such as the one below, where red signals hot temperatures and blue cold temperatures.

Blå och röda ränder av olika valör som illustrerar temperaturskillnad.

An example of how blue and red are used to express differences in temperature.
Source: Wikimedia.

Creative Commons BY SA

Preconceptions like these can also be described as biases, and play a significant role in how we interpret colour. This is especially true when colour is the main conveyor of information, such as when colour is used to encode quantitative data, say in choropleth maps or colour charts. We tend to interpret a deeper colour as a higher frequency or quantity. In an article published in the Journal of Vision (2020), Sibrel et. al refer to this as the dark-is-more bias.

There are of course exceptions, cases where consideration of the "dark-is-more principle" instead makes it more difficult to interpret the visualisation effectively. For example, when the darkest shade approaches black to the extent that it risks being perceived as empty. Alternately, when dark dominates the image surface, while the much smaller light areas are in such strong contrast that they attract the viewer's gaze even though this is where the data is thinnest. This is sometimes referred to as hotspot-is-more bias.

Compare the difference in the choropleth maps below.

Två kartor där färg indikerar kvantitet, på den ena betyder mörk färg mer och på den andra är ljust mer

Choropleth maps of daily rates of confirmed cases of COVID-19 in the Republic of Ireland and Northern Ireland, based on data from 2020-12-31. Source: Angus, Colin (2021): Covid Plots and Analysis. The University of Sheffield. Dataset. https://doi.org/10.15131/shef.data.12328226

Creative Commons BY

The two choropleth maps above both use the same colour palette to encode quantity, but in the right map the palette has been inverted.

In the left map, dark colour is used to indicate a higher number of daily cases per 100,000 inhabitants, while light yellow represents a lower number of daily cases. This makes the central parts of Northern Ireland, which were hard-hit with COVID, look almost like a black, empty hole.
In the right map, however, the colour scale is the opposite. In this map, dark colour is used to indicate a lower number of daily cases per 100,000 inhabitants, while light yellow represents a higher number of daily cases. Although a viewer's gaze may initially be drawn to the fiery bright yellow area in Northern Ireland, there is a risk that the dark, larger areas on the rest of the island might be incorrectly perceived as more hard-hit (due to what we mentioned earlier: we humans tend to interpret a darker colour as a higher quantity).

Always consider which palette is most appropriate for your particular data set. The trade-off should be made on a case-by-case basis. In cases where multiple colour charts are used, be consistent in your choice of colour.