The intellectual origins of the "new view" in the field of cognitive system engineering

Starting from the 1970s, Jens Rasmussen, Erik Hollnagel, and David Woods worked on human-machine design, human reliability assessment, and accident investigation within the context of safety-critical systems. They developed perspectives on cognition, human error, causality, and accident and safety models.

Jens Rasmussen, in particular, emphasized the contextualization of errors and introduced the concept of self-adaptive properties of individuals in performing tasks. He viewed human error as a result of intrinsic learning properties and saw accidents as the migration of adaptations beyond the boundaries of safe performance. Rasmussen also proposed a vertical sociotechnical view called Accimap to represent accidents graphically.

James Reason developed his own approach to error, cognition, causality, and safety and accident models. He took a taxonomic approach to human error, categorizing them into slips, lapses, mistakes, and violations. Reason's safety and accident models were based on the defense-in-depth model and he introduced concepts such as latent/active failures and sharp-end/blunt-end.

Hollnagel and Woods, in the 1980s, developed their perspectives on cognition, human error, and safety and accident models within the framework of cognitive system engineering. They proposed the concept of Joint Cognitive Systems, defining a cognitive system as a functional coupling between humans and machines. They rejected the mechanistic view of machines applied to humans and criticized the use of the information processing model in cognitive psychology and engineering. They also raised doubts about the concept of human error and emphasized the contextual and control nature of cognition.

Hollnagel focused on developing methodologies for human reliability assessment and accident investigation, introducing CREAM and later FRAM. FRAM conceptualizes performance variability leading to unexpected outcomes, similar to Rasmussen's defence-in-depth fallacy.

Woods studied cognition in context and advocated for distributed cognition, viewing human error as a mismatch in the triad of screens-tools, people, and the world. They contrasted their approach with the conventional view of human error and highlighted the distributed nature of cognition, problematizing causal attribution, and emphasizing goal conflicts and operational trade-offs in real-life situations. They recommended practical design options for expert systems and interfaces, cautioned against tabulating errors as a preventive strategy, and emphasized the understanding of people's contribution at the sharp-end in the context of latent causes and the sharp-end/blunt-end distinction.

In the early 2000s, Sidney Dekker expanded on the distinction between the "new view" and the "old view" of human error, which was previously established by Hollnagel and Woods. He summarized the points of the new view in a more accessible way, contrasting it with the conventional view in a table format. Dekker's contributions made the distinction easier to understand and accessible to investigators. Dekker wrote a series of critical essays in the 2000s, challenging Reason's ideas on human error: He rejected Reason's conceptualization and terminology, arguing that it carried negative connotations and failed to consider the subjectivity and constructed nature of science. Dekker also criticized Reason's Swiss Cheese model, arguing that it was too linear, mechanistic, and simplistic to account for the complexity of accidents. He relied on philosophical and sociological ideas to support his critical stance, targeting the human factors and system safety community as a whole. Dekker also challenged other established notions in aviation human factors and system safety, such as situation awareness, function allocation, and procedures. He revisited the concept of just culture and developed an opposition between the mainstream represented by Reason's work and the alternative school based on CSE, emphasizing the potential of the latter. Dekker's radical stance led to a divide between the two perspectives, and while his ideas were not acknowledged by Reason in later publications, they had a significant impact on the field.

As we move forward in history from the mid-2000s to 2020, some ambiguities arise in understanding the evolving "new view" of human error. The concept of Resilience Engineering (RE) and related concepts like Safety II, Safety Differently, and the Theory of Graceful Extensibility were introduced.

In the mid-2000s, Woods and Hollnagel edited a book on Resilience Engineering, which encompassed the developments of the CSE authors in the 1980s and 1990s. The term "resilience" emphasized adaptive capabilities and the ability to bounce back from stresses and shocks. Hollnagel emphasized what goes right with the notions of ETTO (Efficiency-Thoroughness Trade-Off) and Safety I/Safety II. Woods developed his Theory of Graceful Extensibility inspired by complex adaptive systems and resilience. Dekker pursued a critical approach, targeting bureaucracy and advocating for Safety Differently with a political inspiration of anarchism. These three perspectives brought a greater differentiation within the "new view" and evolved from the original formulation by Dekker in the early 2000s. This creates some ambiguity and confusion, especially when readers are not familiar with the nuances and developments in these conceptual approaches. Additionally, Dekker's "new view" itself has three separate versions, reflecting an eclectic empirical and conceptual background.

The "new view" initially formulated by Dekker was based on Woods and colleagues' distinction between conventional and new approaches to human error research in the 1980s and early 1990s.  Dekker's program became more radical with an explicit rejection of Reason as a representation of the mainstream. This created a second version of the "new view" that differed from Woods' perspective. Dekker drew from various backgrounds, including epistemology, philosophy, and sociology, using concepts such as constructivism, radical empiricism, postmodernism, and system and complexity philosophy. He borrowed from the aviation and maritime industries, which are safety-critical, to support his rejection of Reason. Later, with Safety Differently, Dekker's intellectual stance became more political, influenced by anarchism. His criticism expanded beyond Reason to include behavioral-based safety, Heinrich, and Vision Zero. This shift led to a third version of the "new view," which focused on occupational safety in different industries and had an increasingly critical tone. These three versions represent different iterations of the "new view" that are rarely discussed or described.

To avoid conflating these programs, the term CSE and RE school is preferred over the "new view". The CSE and RE school has had significant influence and success in recent years, as demonstrated by the academic interest, conferences, workshops, and publications in the field. The healthcare safety research community and practitioners in fields like air traffic control have shown receptiveness to the ideas and concepts promoted by CSE and RE authors. By positioning their ideas in relation to Reason's influential models of error and safety, the CSE and RE authors have successfully engaged practitioners and prompted critical reflection on the limitations of existing models. Reason's language of unsafe acts and failure, criticized by Dekker, often does not align with the complexity of real-life situations and organizations, leading to the need for alternative approaches and terminology proposed by CSE and RE authors. These ideas emerged at a time when traditional safety frameworks based on compliance, incident reporting, and indicator-based management were being challenged, which further contributed to the enthusiastic reception of CSE and RE concepts. The active translation and adaptation of these ideas for specific practitioner audiences, such as through the HOP approach, have played a role in popularizing the "new view."

One of the main critiques raised by some is the lack of independent research to support the claims and effectiveness of the CSE and RE ideas in transforming safety practices. Questions regarding the differentiation from traditional approaches and the extent to which old approaches should be rejected or retained remain unanswered. While the limitations of cognition as solely in the head have been acknowledged, the more radical critiques of established notions like situation awareness, function allocation, errors, and the Swiss Cheese model have been received more cautiously. Nancy Leveson challenges Hollnagel's Safety II approach on several aspects, including the conceptualization of complexity and causality. Andrew Hopkins is critical of Dekker's anarchist view of safety and emphasizes the importance of organizational structure and rule management.

Ref.

Le Coze, J.C. (2022), The ‘new view’ of human error. Origins, ambiguities, successes and critiques, in: Safety Science, Volume 154, October 2022, 105853

Full article here.