Colour perception principles


The most important application of colour in visualization is to indicate different categories.  For example, in this trend chart we use colour to distinguish between the modes actual, budget, prior year, current forecast and the cumulative variances actual vs budget and current forecast vs budget.




What factors do we need to take into account when designing a colour scheme for different categories?

In his book “Visual Thinking for Design”, Dr Colin Ware distinguishes two criteria: visual distinctness (to support visual search operations) and learnability (so that particular colours come to stand for particular entities).

Effective visual distinctness requires consideration both of the other colours used in the set and the colour of the background, such that a strong pop-out effect can be created.   When designing a visualization with a colour coded background, it is best to choose a subdued, low saturation colour for the background because we are much more sensitive to colour differences between larger areas.

Many studies have been conducted to answer the question “How many colours can be used reliably in a symbol set?” and the results have varied between 6 and 12 depending on how the experiments were conducted.  The strict limit for a complex design is 12 – more than this cannot be distinguished with complete reliability.

When learnability is important, Ware advises using the unique hues first (red, green, yellow, blue), followed by other colours that have relatively consistent names: pink, brown, orange, grey and purple.

9 colours

However, there are other factors to take into account.  About 8% of men and 1% of women are red-green colourblind and have difficulty distinguishing green from red.  It is therefore sometimes advisable to avoid situations in which red and green are frequently shown together and must be perceived as very distinct from each other.

Also, colours are often used symbolically (e.g. in western culture, red to represent danger, green to represent safety), but a particular choice of colours may have a different meaning in a different country.  For example, in China, good fortune and renewal is represented by red, rather than green.

Very importantly, colour design is subtle and can be a source of pleasure in its beauty or irritation in its ugliness.  Perhaps the greatest exponent of colour was the artist Henri Matisse.  I had the pleasure of seeing an exhibition of his famous “cut-outs” at Tate Modern and was struck by the way he achieved beauty through colour.

This quote from an interview conducted with Matisse in 1941 sheds some light on his method.  “I finally came to consider colours as forces, to be assembled as inspiration dictates. Colours can be transformed by relation; a black becomes red-black when you put it next to a rather cold colour like Prussian blue, blue-black if you put it alongside a colour that has an extremely hot basis: orange, for example. From that point on, I began working with a palette especially composed for each painting while I was working on it, which meant I could eliminate one of the primordial colours, like a red or a yellow or a blue, from my painting.”

Stephen Few suggests using soft colours as the standard colour palette for visualization which encourages the reader to explore the data calmly with an open mind.  Bright, fully saturated colours should be reserved for emphasis and used sparingly.

Examples of colours that he recommends are provided below:

Light colours (for large data-encoding objects, e.g, bars, boxes)

Light palette

Medium colours (for small data-encoding objects, e.g. data points, lines)

Medium Palette

Emphasis colours to highlight particular items

Emphasis colours

Leave a Reply