This paper is also available as PDF (suitable for printing).

Extract from Blackwell, A.F. (1998). Metaphor in Diagrams
Unpublished PhD Thesis, University of Cambridge.

Chapter 1: Introduction

These circles, or rather these spaces, for it is of no importance what figure they are of, are extremely commodious for facilitating our reflections on this subject, and for unfolding all the boasted mysteries of logic, which that art finds it so difficult to explain; whereas by means of these signs, the whole is rendered sensible to the eye.
Letters of Euler to a German Princess, tr. H. Hunter 1795, p. 454.

For 20 years, new computer software has presented information graphically as well as in textual form. The usual justification for this practice has been that the graphical form is easier to learn, understand and apply because it allows metaphorical reasoning. Consider these forthright statements from introductory textbooks on software user interface design, all published within the last two years: "Designers of systems should, where possible, use metaphors that the user will be familiar with." (Faulkner 1998, p. 89). "Metaphors are the tools we use to link highly technical, complex software with the userís everyday world." (Weinschenk, Jamar & Yeo 1997, p. 60). "Select a metaphor or analogy for the defined objects Ö real-world metaphors are most often the best choice." (Galitz 1997, p. 84). "Real world metaphors allow users to transfer knowledge about how things should look and work." (Mandel 1997, p. 69). "Metaphors make it easy to learn about unfamiliar objects." (Hill 1995, p. 22). "Metaphors help users think about the screen objects much as they would think about real world objects." (Hackos & Redish 1998, p. 355). "Very few will debate the value of a good metaphor for increasing the initial familiarity between user and computer application." (Dix et. al. 1998, p. 149).

The goal of this dissertation is to investigate the psychological evidence for these claims. This investigation is perhaps overdue. Not only are computer science students advised to use metaphor as the basis for their designs, but software companies routinely base their research efforts on this assumption (Blackwell 1996d), and the most influential personal computer companies insist on the importance of metaphor in making computers available to everyone:

You can take advantage of people's knowledge of the world around them by using metaphors to convey concepts and features of your application. Use metaphors involving concrete, familiar ideas and make the metaphors plain, so that users have a set of expectations to apply to computer environments.
"Metaphors" from Chapter 1 of the Macintosh Human Interface Guidelines. (Apple Computer, Inc. 1992).
Familiar metaphors provide a direct and intuitive interface to user tasks. By allowing users to transfer their knowledge and experience, metaphors make it easier to predict and learn the behaviors of software-based representations.
"Directness", from Windows Interface Guidelines for Software Design (Microsoft Corp. 1995).

The conclusion of the research described in this dissertation will be that the case for the importance of metaphor is greatly over-stated. This should not be interpreted as a deprecation of graphical user interfaces. Graphical user interfaces provide many advantages - the problem is simply that those advantages are misattributed as arising from the application of metaphor. A more prosaic explanation of their success can be made in terms of the benefits of "direct manipulation", which indicates potential actions via the spatial constraints of a 2-dimensional image. The concept of direct manipulation has been thoroughly described and analysed (Shneiderman 1983, Lewis 1991). It will not be discussed in any detail here, but the implication of the current investigation is that, if the expected benefits of metaphor have been exaggerated, these low-level virtues and by-products of direct manipulation are even more important than is usually acknowledged.

Overview of the Thesis

Chapter 2 considers previous work in HCI, but it also reviews theories that have been proposed to describe diagrammatic graphical representations and to describe metaphor. It then considers the manner in which diagrams and metaphors can be used as cognitive tools, before returning to the question of HCI.

Chapter 3 presents the results of three contrasting surveys, investigating how computer scientists and professional programmers regard their use of visual programming languages. Researchers developing these languages are greatly influenced by cognitive theories, including some theories of metaphor, but professional users appear to have little awareness of the potential cognitive implications of diagrammatic representations, instead emphasising more pragmatic benefits.

Chapter 4 describes two experiments which manipulated the degree of metaphor in diagrams. The metaphor was used to teach elements of a visual programming language, then of more general diagrams, to people who had never programmed computers. Their performance was compared to that of experienced computer programmers, in order to judge the effect of the metaphor on learning. The use of metaphors provided little benefit relative to that of experience.

Chapter 5 investigates which properties of visual representations assist the formation of complex abstract concepts in visuo-spatial working memory. The value of mental imagery as a design strategy for abstract problems is an underlying assumption of much of the literature on visual metaphor. Four experiments were conducted to measure productivity when the appearance of the visual representation was manipulated. Metaphorical content appeared to have little influence, and there was also little consistent evidence for significant benefits from mental imagery use.

Chapter 6 returns to the type of explanatory diagram introduced in chapter 4, and presents the results of three further experiments which manipulated both the metaphorical and visual content of the notations. Diagrams were described with and without instructional metaphors, and both memory and problem solving performance were measured. Metaphor had little effect on problem solving, and memory was improved far more by pictorial content in the diagram than by explicit metaphorical instructions.

Chapter 7 concludes that the main potential advantage arising from metaphor in diagrams is a mnemonic one, rather than support for abstract problem solving or design with mental images. Furthermore the mnemonic advantage is greater if diagram users construct their own metaphors from representational pictures, rather than receiving metaphorical explanations of abstract symbols. This finding has considerable importance for the future study of diagram use and human-computer interaction.

Continue reading chapter 2 , or return to table of contents and download information .