Keywords

4.1 Introduction

In his book ‘Man-Made Disasters’, the safety theorist and organisational sociologist Barry Turner remarked that ‘a way of seeing … is always also a way of not seeing’ (Baddeley and Hitch [1], p. 49). Turner was talking about what he termed the ‘decoy problem’—the fact that attention may be paid to a well-defined problem or source of danger, but this also distracts from other more potentially dangerous problems lurking in the background. In this chapter, I want to focus on a slightly different set of issues and questions centred around ‘seeing’ in safety, in particular the use of visual representations in safety tools, models and methods. In particular, the chapter focuses on two main questions:

  1. 1.

    What does the use of diagrams and other forms of visual representations by safety researchers tell us about their thinking and the underlying theory of safety they seek to promote?

  2. 2.

    How do the various types of visual representations used in safety models, methods and tools work influence their use by researchers and practitioners?

4.2 John Berger and ‘Ways of Seeing’

There is a well-established tradition within the History of Art which focuses on the influence of pictorial and other forms of visual representations on our ability to perceive, understand and interpret works of art. The art historian Ernst Gombrich for example, in his book ‘Art and Illusion’ (1960) argued for the importance of ‘cognitive schemata’ [2] in analysing works of art. Gombrich claimed that artists learn to represent the external world by learning from previous artists, and as a result representation is often achieved using stereotyped figures and methods.

More recently, the artist Bridget Riley has argued that the use of colour and black and white in her work has the power to trigger perceptual and cognitive illusions and other visual stimulations in the viewer [3]. One of the most important works of the writer and cultural theorist John Berger (1926–2017) was a book based on a television series which was screened in the UK in the early 1970s. ‘Ways of Seeing’, Berger [4] set out to criticise traditional Western cultural aesthetics and raised questions about hidden ideologies in visual images (e.g., magazine advertisements). In one episode of the programme, Berger showed the continuities between post-Renaissance European paintings of women and modern-day posters and advertisements, by juxtaposing the different images and showing how they similarly rendered women as objects. The book was partly written as a riposte to the more traditionalist view of the Western artistic and cultural canon (e.g., the work of Kenneth Clark and his book/TV series ‘Civilisation’, 1969) and the TV programmes and book criticise traditional Western cultural aesthetics by raising questions about hidden ideologies in visual images. Berger offered a Marxist alternative which shifted attention towards the cultural messages and sub-meanings which are embodied when we look at objects and art.

4.3 Visual Representations in Accident and Safety Research

Berger’s analysis of the process of seeing art and revealing some of the implicit hidden meanings in paintings and other forms of art, might serve as a useful basis with which to probe deeper into the theoretical roots and origins of many of the types of visual representations used in the world of safety. In a series of articles describing the history of safety science, Swuste et al. [5, 6] make use of a wide variety of visual materials including posters, warning signs and cartoons to illustrate the way in which theoretical and methodological approach to understanding accidents has changed over the last century. Similarly, [7] argues that safety science and occupational safety and health relies heavily on visualisation as a means to communicate safety messages and sometimes act as metaphors and boundary objects. Warning signs, for example, rely heavily on icons, pictorials and other visual materials and the models, methods and tools used by safety practitioners. It is also a testament to their dominance that some models (e.g., the Bird/Heinrich triangle) still remain popular even in the face of significant criticism within the scientific community (e.g., [8]).

4.4 Why Are Visual Representations So Popular?

One of the most obvious reasons why visual representations feature so prominently in safety can be attributed to the Chinese proverb ‘a picture is worth 10,000 words’. Because we think with the help of images, pictures, diagrams and other forms of imagery help us to reflect deeply beyond words. Visual representations may also be augmented and act as ‘cognitive aids’ through the use of metaphors [9], for example Bow-Tie models or Reason’s three bucket model of human error. When they are solving problems, human beings use both internal representations, stored in their brains, and external representations, recorded on a paper, on a blackboard, or on some other medium [10, 11]. Amongst engineers, scientists and designers visual representations are also commonly used as a means of structuring and communicating complex problems [12,13,14]. Visual representations serve as ‘boundary objects’ which are open to interpretation across the various communities and specialisms (e.g., occupational safety and health managers, risk managers) involved in safety, but also serve as a common focal point supporting cross-disciplinary communication and collaboration [7, 15]. Henderson [16] further developed the concept of ‘meta-indexicality’ in order to underline the ability of visual representations to combine many diverse levels of knowledge and to serve as a meeting ground for many types of disciplines and individuals.

4.5 The Evolution of Two Safety Models (Swiss Cheese Model and Accimaps)

Probably the most well-known model in safety is James Reason’s Swiss Cheese Model. Reason developed at least two different versions of the model over the course of the 1980s and 1990s. What is interesting about the earlier version of the SCM is that it retains an element of the types of box models of cognition common in psychology in the 1960s and 1970s (e.g., [17] model of working memory). Similarly, the box model or flow chart is also reminiscent of the various types of error taxonomies which were developed in the 1980s (e.g., Reason’s 1988 GEMS model, Rasmussen’s 1983 SRK framework), as well as the fault trees commonly used to assess risk in the nuclear and other high-risk industries. By the time we reach the late 1990s, the model has become a metaphor which supports a more systems-oriented way of thinking about latent and active pathways (‘holes’) across the various levels within the system (‘slices’) and the role they play in causing human error.

A final example of how safety models shift and evolve over time and what this tells us about implicit theories of safety is [18] Accimap model. An earlier version shown in [19] was a hand-drawn model of the abstraction hierarchy, elements of which formed the basis of Cognitive Work Analysis, Cognitive Systems Engineering and the Risk Management Framework (Rasmussen [18]). The RMF also served as the basis with which to develop the first Accimaps, a method for analysing accidents which has proved popular and spawned many variations and ‘remixes’ [19].

4.6 Augmenting and Extending How We Use and Evaluate Visual Representations in Safety

Many of the visual representations mentioned in this chapter are used as methods for analysing and sometimes investigating accidents. One of the most common ways of facilitating comparison and evaluation between the methods is to look at their scientific properties (e.g., validity, reliability, coverage of systems thinking components). This proves often to be problematic [20] and from the point of view of safety practitioners, often misses the point. Safety investigators and risk assessors, for example, often cite the usability of the method as more important than scientific concerns. Being able to learn the method quickly and how resource-intensive, it is are also important considerations. One way forward then may be to shift the focus of evaluations of safety models and methods away from strict scientific criteria and more towards ease of use and usability in general. Green [21], for example, describes a set of cognitive dimensions of notations which are designed to provide a lightweight approach to analyse the quality of a design, rather than an in-depth, detailed description. They provide a common vocabulary for discussing many factors in visual notation, user interfaces or the design of programming languages. The dimension ‘role expressiveness’, for example, is defined as ‘how obvious is the role of each component of the notation in the solution as a whole?’ and an associated question is ‘When reading the notation, is it easy to tell what each part is for?’ Are there some parts that are particularly difficult to interpret? Likewise, the ‘viscosity’ of a notation might be assessed through questions such as ‘how readily can required parts of the notation be identified, accessed and made visible?’. Finally, the ‘abstraction gradient’ dimension [22] might be applied in order to assess the degree to which methods or models allow further refinement and elaborations (‘abstractions’) of causal components. The sort of trade-offs involved in different methods as might take the form of comparing for example, FRAM [23] (possibly high on ‘role expressiveness’, possibly low on ‘viscosity’) with other methods (e.g., Accimap—possibly scoring high on both dimensions). In other words, these types of what might be called ‘low fidelity’ criteria might prove helpful in further improving the visual representations used in safety models, as well as assessing their fitness for purpose within the practitioner community. It might also be one way in which we break out of what is sometimes seen as rather dry and sterile debates surrounding which method is better for accident analysis as compared to another [24].

4.7 Some Conclusions: Ways of Seeing Safety

The diversity of visual representations which are used in safety science is evidence enough that we have evolved multiple ways of ‘seeing safety’ in the sense that John Berger suggested. Models and methods provide us with one of the many lenses with which we can look at safety and afford a way of what Barry [1] described as ‘turning the kaleidoscope’. They also raise a number of challenges for the future, not least how to deal with ever-increasing complexity in terms of systems, technology and organisations. How we trade off complexity against simplicity in our attempts to model safety in these systems will continue to be an important question for the foreseeable future. Some might argue that some of our present models and methods (e.g., FRAM, STAMP) have gone too far and widened an already large gap between safety researchers and practitioners [25], whilst at the same time casting the net too far in terms of how we might design solutions which might prevent accidents or attempt to link organisational factors to casual factors contributing to accidents [26, 27]. How we narrow some of these gaps, as well as wider questions about how visual representations structure and shape our views on safety will continue to occupy our attention in the coming years.