Last changed 11 Mar 2005 ............... Length about 4,000 words (25,000 bytes).
(This document started on 11 Mar 2005.) This is a WWW document maintained by Steve Draper, installed at You may copy it. How to refer to it.

Web site logical path: [] [~steve] [patera] [this page]

Using educational theory and advanced technology in colour education

M. Patera, S.W. Draper and M. McWhirr
Digital Design Studio, Glasgow School of Art
Glasgow, G41 5BW, U.K. /
Dept. of Psychology, University of Glasgow
Glasgow, G12 8QQ, U.K.

Contents (click to jump to a section)


This paper begins with an educational problem: the defects in how colour is often taught. It then uses educational theory to suggest an analysis of the problem in terms of a failure to connect learners' personal experiences of colour to the theories being taught. It then describes the development of a promising method of learning colour theory. Observations from a comparative study between this activity-centered learning method and learning from text are then discussed. Finally we outline the advantages and disadvantages of using advanced 3D technology in color education, a preliminary version of which we also tested.


This paper describes what was learned from the first stage of a project to develop an improved method of teaching colour theory which will eventually involve the application of advanced computer technology. We have developed an initial version of the method, and run a preliminary trial of it on learners, comparing the new method with learning from an excerpt from textbooks.

The overall strategy we are following for the successful application of technology to improving the learning and teaching of some topic, in this case the theory of the colour space comprising hue, saturation and brightness, is in outline this:

  1. First identify some existing problem. Simply replacing teaching that works adequately is not likely to lead to worthwhile improvements.
  2. Analyse the existing problem, using educational theory.
  3. Develop a solution using human tutoring, firstly because that allows rapid changes and improvements, secondly because it allows detailed observation of the learning and learners, and thirdly because the so-called "Wizard of Oz" technique of having a human simulate what will eventually be done by computer is an established rapid prototyping technique in Human Computer Interaction.
  4. Develop the technological solution: convert the human teaching into software.
  5. Experimental testing of alternative solutions e.g. existing textbooks versus the human tutoring versus the technological version.

The pre-existing educational problem

Colour is frequently not taught as a separate course but integrated into other subjects [1]. Occasionally it is not taught at all, even in programmes where it is important to students' subsequent professional practice e.g. art and design and architecture.

Too often a course may consist mainly of lectures and notes presenting highly abstract concepts, such as hue, value, saturation and colour-order systems. Didactic lectures on Colour Theory, with no visual teaching aids or practical activities, can lead to shallow learning, where students do not connect the theoretical concepts to real life situations. Black and white textbooks can be ineffective for building connections between symbols and visual recollection of the colours. In such cases, learners find it difficult to master the concepts, and hard to find much motivation to learn.

On the other hand, some courses on colour are based on practical activities, excluding entirely the theoretical aspect. When practical assignments are not linked to any theory, they can also lead to limited learning that does not transfer or generalise well. Students tend to concentrate more on how to mix paints accurately and to cut out colour samples from magazines than actually reflecting on the purpose of the exercise.

We selected the small topic of the hue, saturation, and brightness colour space as our first target for several reasons. We needed something conveniently small for our initial work, partly in order to try out our overall strategy. Secondly, several studies [2] have shown that 3D virtual environments are particularly appropriate for teaching complex and abstract concepts and since colour space has an inherently three dimensional nature, this could offer a significant advantage. Finally, as we shall see, it turns out that surprisingly few people even of those who have worked with colour are in fact familiar with this conceptualisation.

Analysis of the problem using educational theory

Educational theory suggests that a better balance between theory and practice should result in a learning experience that is both more effective and more enjoyable. The most frequently mentioned educational concept is constructivism [3]. According to this theory, people do not just receive or reproduce, but actively (re)construct, new knowledge and understanding founded on what they already know and believe. This is also expressed in Laurillard's model [4] of the teaching and learning process, which asserts that all well-designed teaching will attend equally to both the public, abstract, conceptual aspects and to the private, concrete and experiential aspects of a topic, and also to creating connections between them. In fact it is desirable to connect new ideas being taught to three things. Firstly, to the prior concepts (whether right or wrong) the learner may have embraced e.g. that the colours of the spectrum form a linear sequence and not a circle of hues, or that black and white are colours and thus should be included in the spectrum. Secondly, to their prior experiences of colour and using colour, especially since people experience colour from a very early age: it is not sensible to try to teach even a simplified theory that cannot deal with facts that the learner already knows and which to them are salient and familiar e.g. that brown and pink can seem independent colours in their own right, not simply intermediate shades between other primary colours. Thirdly to their present experiences: although some learning occurs without any actions being performed, it is also clear that important learning occurs through attempting tasks. Since colour is a perceptual experience, this is both important and relatively easy to arrange.

Educators need to build on these personal conceptions and experiences in ways that assist learners to achieve a deeper understanding of the subject. Thus in general, the design of learning materials should be preceded by an investigation into what the target learners typically think, rightly or wrongly, about the topic. It should also include practical activities where possible, partly because people learn by exercising the ideas they are trying to learn, but even more in order to link the theory to practical experience.

Tutoring method

We therefore developed a teaching method for our chosen topic of Runge sphere colour space, that consisted of a set of exercises in arranging colour samples guided by a set of prompts by a human tutor in the spirit of Socratic dialogue. This evolved into our design for effective teaching, satisfying the theoretical criteria discussed above, and containing a balance of theory (the abstract structure of the space in the form of the Runge sphere) and practical experience (arranging and placing colour samples in relation to each other). However it also was valuable as a research method, because it gave prolonged opportunities for observing how participants arranged colours, the reasons they gave for their actions, and what this revealed about the different ways people think of colour. We report on this in the next section.

The method we used was to sit a participant down in front of a table, and take them through a sequence of tasks. A typical task would be to give them a pile of colour squares (for instance 10 samples ranging from white through pink to fully saturated red), and a request (for instance "please arrange these in any way you think right"), and after they completed it, to ask them why they chose the arrangement they did. They would then be asked a few other probe questions: e.g. where they would now insert one or two more squares (handed to them) in their current arrangements. The sequence of tasks was generally:

We also implemented an alternative digital way of presenting this, using 3D digital models built in the Maya software package on a desktop computer to replace the material cardboard squares and sphere (but still with a human tutor presenting the tasks and asking probe questions), and tested this too on a set of participants.

The study

We compared learning using the cardboard equipment, the computer models, and from a text (combined from passages from several textbooks). (We also compared performance on immediate post-tests i.e. at the end of the learning phase, with performance on delayed post-tests i.e. about a week later.) We used participants, about 50 in total, who were mostly but not all in their early twenties, from a variety of backgrounds. Some had art and design backgrounds, others had not. All those given the computer models had current experience on using the same computer modelling software, so that familiarity with the user interface conventions would not be an issue.


While the participants were performing the tasks, notes, photographs and some video were taken in order to record the process.

The first strong impression is that those performing the material version of the learning tasks found it enjoyable, and indeed absorbing. Even those who took over an hour to complete them all, were surprised at how the time had passed and never showed any signs of fatigue or boredom. These participants often used the word "play", while the ones who performed the digital version considered it more like a problem-solving task. None of this could be said of those with the textbook version.

Every person arranged the coloured tiles in a different way. As regards spatial relations, the patterns the participants formed varied from straight lines, circles and triangles to squares, spirals, stars, and zigzag diagonal patterns. From the viewpoint of the concept we were trying to teach, we had the tacit view that for the black-white, red-black, and red-white sets a straight line was best, while for the hues a circle was better, and combinations of these sets would require two or three dimensions. Particularly in early trials, the fact that our samples were themselves square in shape (whereas we could have made them circular, say) often seemed to influence people to use square or rectangular layouts. They also tended to reason in terms of the number of tiles in the current problem e.g. if there were nine, then a three by three square layout might be attractive. Increasing the number of samples and using probe questions that required them to add an extra tile or two to a completed layout tended to reduce this type of reasoning (which is demanded by some tests of spatial intelligence).

One of the reasons for developing the digital version was to provide the participants with the option to utilize the third dimension, which a flat tabletop might prevent them from considering. They were told at the beginning of the test that they were free to use all the views available in the software -- orthographic or perspective -- and switch between them. Yet, most "digital" participants performed the task using the top view and did not consider using the third dimension. Even when they were prompted to view their arrangement from a different perspective, most of them switched back again to two dimensions.

Watching participants' movement of tiles as they assembled a layout often made it clear how important placing samples side by side is for comparing small differences in colour. This can be done extremely rapidly and conveniently with cardboard tiles on a large tabletop: it seems an important feature to support in any computer implementation.

As regards the properties they were expressing in their spatial layouts, some participants were organising the colours according to their intensity or hue, while others were dividing them as cool and warm, and many said they were arranging them by lightness, even in cases where the samples they were working on had been intended by us to show equal brightness, although limitations in the colour printing used to generate the tiles meant this was imperfect. Only a couple of them argued that colours should be randomly scattered and one claimed that colours are like music notes.

In the early development of our tutoring method, it emerged that the number of samples in a set was a significant issue. From the point of view of rapidly sorting samples on a given dimension, a small number seemed most convenient. However for getting learners to see a set as a smooth sequence of a single varying property, a powerful intervention or question was to offer another intermediate value and ask where it could be inserted into an arrangement. Introducing more and more fine distinctions seems important in dislodging people from the opposite tendency of seeing colour in terms of a small handful of landmark primary colours with no particular relationship to each other. This is in fact a form of a very general educational tactic that can be important in quite different areas, is sometimes known as "bridging" [5], and which consists essentially of suggesting the learner consider an intermediate case midway between two cases they regard as poles apart and unrelated, in order to see a connection between them.

The great variety of ways of arranging the squares, both spatially and in terms of the colour properties expressed, was true for both those from an Art school background and from a computing background. Initially we expected the educational and vocational background of the subjects to play a significant role on their performance. Almost half of the subjects had studied art and design, whether that was painting, sculpture, graphic design or product design. One might assume, since artists and designers apply colour extensively in their works, in the form of pigments or computer graphics, that they would have had a mental model of colour relations in their minds. Surprisingly enough, their performance did not differ from that of the other participants. It shows however that this kind of approach is effective at eliciting from each learner the properties of colour they are already aware of, and which ideally they need to relate to any abstract theory or organising principle. This remains true even for conceptualisations, such as the Runge sphere, that only express some properties of colour. In this case, a full appreciation of the theory should include a definite realisation of the properties it does not capture e.g. the cool/hot dimension, or the emotional connotations of different colours, as well as of the satisfactory integration of the properties it does address.

The majority of the participants did not have a clear mental model of how all possible colours could be related before the 3D sphere model was presented to them; however after arranging the tiles in each task, most of them provided good reasons to support their schemes. When the 3D sphere model was presented to them, most of them agreed that it related colours in a coherent way and tried to revise their conceptions. A literal interpretation of the notion of constructivism, and of Socratic dialogue, is that each learner should construct the target concept for themselves, perhaps under the impact of experiences and questions arranged by the teacher. This was what we aimed for, but in most cases did not achieve in this version of tutoring method. However our impression from our observations is that this may not matter. The tasks together with our tutorial questions served first to get learners thinking of their own notions of colour, whether explicit or implicit, and to address the question of how to organise colours. They grasped, both practically and intellectually, that this is a problem that is not trivial, but which perhaps could be solved. This puts them in a position to appreciate what is good about the Runge sphere when it is offered to them, and their ability to place new test colours within it seems to show that they grasped what they were seeing. In contrast texts describing the Runge sphere do not say why it is good, nor what the problem it solves actually is. They do not prompt readers to discuss whether a cylinder would or would not be just as good as a sphere, why the axis must be black to white, not unsaturated to fully saturated, or where brown would fit in the sphere and what its neighbouring colours might be.

On the other hand informal questions to the participants showed that they had not connected the technical terms "hue", "saturation" and "brightness" to the dimensions. This is not surprising since these are never mentioned in the version of the tutoring method tested so far, while they are of course mentioned in textbook treatments. However this should be an easy improvement to make. It is also necessary if a full balance between practice and theory is to be achieved.

Digital versus material versions

As noted above, we trialled both a computer based ("digital") and a material (cardboard) version of our tutoring method. The preliminary data and the researchers' observations do not suggest any great difference in effectiveness or enjoyableness between these two implementations. However, the digital version proved more flexible since it could be updated and altered much more rapidly and easily by a programmer than the cardboard materials. It is also portable and can run on most computer systems. The physical model needs a large carrier bag to carry around, and it requires plenty of space, usually a big tabletop (about 1.5 by 1 metres), larger than most desks. After a number of tests, the cardboard began to bend and crush unlike the digital version, which of course does not wear out.

In future we will explore making the digital version more autonomous i.e. explore automating the human tutor. This will require a) making the user interface readily usable by novices (in the trials, we used only those already familiar with the controls of the software package): this is probably fairly easy. b) Building in the tutor's prompts. Since these were deliberately designed to be a standard set, this should be unproblematic. c) Automating the tutor's choice of which prompt to select. This will require the software to recognise features of the user's arrangement for each task (e.g. whether the squares were positioned into a straight line or a circle). This is more difficult. However since we have the impression that learners are not misled by inappropriate prompts, but rather benefit from those questions which are telling, it is probably not hard to do this well enough to stimulate learning.

Participants reported enjoying "playing" with the cardboard squares because they could touch them. The tactile sense and the feeling of texture are important to some people. The computer screen is a single-sensory medium thus it does not offer this experience, but more advanced technology can. Tactile feedback can be provided through hardware and software mechanisms incorporated within special gloves, which allow the user to interact with the virtual objects. These systems have been used for science education, medicine and engineering with some very positive outcomes. Colour education may also be able to benefit from the use of such advanced 3D visualisation. Our aim is to take this study further and create a virtual 3D environment for teaching and learning colour theory.


Preliminary inspection of results from these early trials suggest that there is no great difference between digital and material implementations of our tutoring method, and so that further development of digital versions is worthwhile to gain the advantages of robustness and ease of distribution. Participants seemed to retain their knowledge longer from our tutoring methods, than did those who were given a passage from a textbook on the same theory. This is consistent with our analysis based on educational theory. However still more benefit may be obtainable if we are able to improve our tutoring method further in the light of our observations. The challenge is to refine it so that participants gain a still fuller appreciation of what the Runge sphere concept does well (offering a way of organising all possible colours into three dimensions), and what aspects of their ways of thinking it does not directly accommodate, even though they are valid, such as express divisions of colours into "warm" and "cold".


  1. Bergstrom, B. (2001) Creative Colour Education AIC Colour 01, Rochester, USA.

  2. Winn, W. (1997) The Impact of Three-Dimensional Immersive Virtual Environments on Modern Pedagogy HITL Technical Report R-97-15, Human Interface Technology Laboratory, University of Washington, Seattle, WA, USA.

  3. Larochelle, M., Bednarz, N., Garrison, J. (1998) Constructivism and education Cambridge University Press, Cambridge.

  4. Laurillard, D. (1993) Rethinking university teaching: a framework for the effective use of educational technology Routledge, London.

  5. Brown, D., and J. Clement. 1991. Classroom teaching experiments in mechanics. In Research in Physics Learning: Theoretical Issues and Empirical Studies, R. Duit, F. Goldberg, and H. Niedderer, eds. San Diego, Calif.: San Diego State University.

Web site logical path: [] [~steve] [patera] [this page]
[Top of this page]