Chapter 5. A study of web developers’ mental models of web accessibility
5.2.1. What are mental models?
Mental models are commonly understood to be internal cognitive frameworks that enable us to understand how something works in the real world. They have been found to be an alluring concept across many fields of research, including: philosophy,
education, psychology, cognitive science, and HCI. Researchers have appealed to the concept of mental models in areas as varied as the human mistakes that led to the Chernobyl nuclear disaster (Johnson-Laird, 1999) and people’s understanding of home central heating systems (Kempton, 1986). Conant and Ashby (1970) consider the development of these internal representations of external situations to be a
“compulsory” component of cognition. The idea that the brain models its environment is, they claim, unquestionable. Similarly, Rouse and Morris (1986) contend that without resorting to the concept of mental models, many aspects of human behaviour are difficult to understand. Payne, Squibb and Howes (1990) consider mental models to be “one of the most important ideas in the psychology of human-computer interaction” (p. 417).
The concept of mental models has endured for a very long time. German physicist, Heinrich Rudolf Hertz (1899) proposed that humans form “internal images or symbols of external objects” that underlie our thought processes and allow us to simulate the external world. Around the same time, American philosopher, Charles Sanders Peirce (as cited in Johnson-Laird, 2004) claimed that humans form internal “diagrams” on which we perform “mental experiments” to reason about the world around us. The picture theory of language proposed by Austrian philosopher, Ludwig Wittgenstein (as cited in Johnson-Laird, 2004), proposed that humans use language to form “pictures of facts” corresponding to reality. Similarly, French philosopher, Georges-Henri Luquet (1927, translated 2001) argued that children construct “internal models” through interaction with the external world. The term “mental model” was not actually coined, however, until 1943 when Scottish philosopher and psychologist, Kenneth Craik used it to refer to the “small-scale models of external reality” that humans construct to explain, to reason, and to anticipate future events (Craik, 1943).
Despite the enduring popularity and apparent necessity of mental models, there is still a surprising lack of consensus among researchers about what constitutes a mental model. This is apparent from many reviews of the mental models literature (e.g. Rouse & Morris, 1986; Wilson & Rutherford, 1989; Moray, 1999; Payne, 2007; Turner & Sobolewska, 2009), which are prefaced with warnings about terminological and methodological inconsistencies. This confusion over the term is notably characterised by Rips (1986) as “mental muddles.” Richardson and Ball (2009) attribute the source of the confusion to the plethora of terminology afflicting the literature on mental models.
They highlight how mental models are referred to, often interchangeably, by a range of terms, including: mental representations, internal representations, mental images, conceptualisations, mental simulations and many more. They also comment that “the lack of unified terminology is confusing, especially when the same type of model is described with different vocabulary, by different authors, from different disciplines and from different perspectives” (Richardson and Ball, 2009, p. 337). Payne (1992), on the other hand, considers the confusion over mental models to have arisen from different authors using a single common term to represent very different ideas. Payne (2003) attributes the source of much of this confusion to the 1983 publication of two very different books with the same title: Mental Models.
In one of these books (Johnson-Laird, 1983), mental models represent the cognitive processes underpinning people’s thinking, reasoning and problem solving. Mental models describe the mental mechanisms that allow people to process information, make inferences and deduce conclusions. For example, given the premises “Person A is taller than Person B” and “Person C is shorter than Person B”, Johnson-Laird alleges that we form an analogous mental representation of the three people allowing us to deduce the conclusion that Person A is taller than Person C. Similarly, given the premises “The scissors are to the left of the stone” and “The paper is to the left of the scissors”, we allegedly form and consult a mental model of the spatial arrangement. The construct of such models allows us to make deductive inferences (e.g. that the paper is to the left of the stone) that were not explicitly mentioned in the original premises. These
representations are believed to be transient and dynamic and manipulated ‘on the fly’ within people’s working memory (Wilson & Rutherford, 1989). This interpretation of mental models, as the dynamic cognitive structures that allow people to process information and reason about the world around them, has also been extremely influential and inspired a substantial body of research.
An alternative interpretation of mental models is presented in the other book published in the same year (Gentner & Stevens, 1983), in which mental models represent the collection of common-sense knowledge, beliefs and theories that people formulate about the world. Drawing on a disparate array of domain-specific examples, including motion, mechanics, ocean navigation, electricity, calculators and computer systems, Gentner and Stevens characterise such mental models as “naive representations” that humans construct to guide their thinking and behaviour. For example, many people
adopt the relatively simple analogy of water flowing through pipes to understand the more complex concept of electrical current “flowing” through wires. Similarly, many people liken the behaviour of an atom to that of the solar system with electrons orbiting the nucleus in the same way planets orbit the sun. Such analogies may not always be accurate (and in many instances may be completely wrong) but they provide people with a working knowledge on which to base their actions. These static and relatively unchanging representations (often referred to as conceptual models) are believed to be constructed within people’s long term memory (Welford, 1961). This interpretation of mental models, as domain-specific knowledge that allows people to reason and make inferences about that particular domain, has also been very influential, particularly in the field of HCI, where mental models have been used to represent people’s
conceptualisations of computer systems and devices.
Despite contrasting interpretations of mental models and the plethora of terminology used to label and describe them, the field of mental models may not be as conflicted and uncertain as it appears. Johnson-Laird (1989) suggests that different approaches to mental models merely represent different aspects of the same cognitive phenomenon, claiming that “theorists are talking about the same beast” (p. 490). This is echoed in Richardson and Ball’s (2009) comprehensive review of the mental models literature, which begins by reaffirming the seemingly contrasting interpretations of mental models proposed by Johnson-Laird (1983) and Gentner and Stevens (1983). The review
concludes, however, that the two interpretations are, in fact, complementary. Together, they form a cognitive framework in which dynamic representations constructed in working memory during a particular activity are populated and informed by static representations of knowledge and theories residing in long term memory. In other words, people hold knowledge, theories and beliefs about the world around them and use this to support the mental mechanisms that allow them to process information, make inferences and deduce conclusions. Thus, the two types of cognitive
representation can both be considered mental models, working together to underpin all higher-level cognitive activity, including thinking, reasoning, and problem solving.