Publications

Journal article | Peer reviewed

The good, the bad, and the lost: a meta-analysis of rail signal passed at danger incident investigations using AcciMaps and systems thinking ↗

by Anjum Naweed, Teal Evans, Gemma Read, Paul Salmon and Brian Thoroman

2026
Safety Science

Signal Passed at Danger (SPAD) incidents, though infrequent, remain a serious threat to rail safety. While often investigated as discrete operational events, SPADs are symptomatic of deeper system conditions. Investigations tend to follow linear logics of human error and focus corrective actions on individual behaviour—especially that of the train driver. However, such approaches may obscure rather than reveal meaningful system-wide learning. This meta-analysis used the AcciMap technique to analyse and aggregate findings from SPAD investigation reports (N = 49), offering a unique holistic perspective. A total of 859 contributory factors were identified (∼17 per SPAD), with most (∼62 %) reported to occur at the level of physical processes and actor activities. The largest aggregate contributory factor being attributed to SPADs was driver cognitive (dis)engagement (63 factors). Thematic aggregation revealed four themes: (1) A signal lost in the noise, relating to information overload in the rail environment and inevitable separation of human attention from signals; (2) Lost in translation, highlighting interchangeable and indistinct use of labels and concepts employed during investigation; (3) Lost in the brake down, reflecting a dominant narrative of driver (in)action and wrong-doing; and (4) Lost in the loop, pointing to surface-level corrective actions that repeated what was expected rather than redesign ineffective controls. Findings are discussed by reflecting on the question: “what might SPADs be trying to teach us?” By shifting the analytical lens from individual to system, this study challenges dominant narratives of driver blame and considers whether current investigations genuinely reflect systems thinking or merely perpetuate an illusion of learning.

---

Journal article | Peer reviewed

Naturalistic decision-making of football coaches ↗

by Scott McLean, Isaiah Elstak and Paul Salmon

29 December 2025
International Journal of Sports Science & Coaching

Effective coach decision-making is important for achieving success in sport, with the decisions made by coaches during matches having a major influence on match outcome. However, to date there has been minimal research investigating the cognitive decision-making processes of coaches during dynamic match play. The current study aimed to investigate the factors that influence the decision-making process of high-level football coaches during dynamic moments in football matches. We applied the critical decision method to assess the decision-making process of 11 high-level football coaches across two football match scenarios 1.a Send off, and 2. Chasing the game. Data was thematically analysed and coded to the perceptual cycle model which is a model of decision-making whereby information available in the world activates decision-maker schemata resulting in an action in a cyclical nature. The results indicate that decision-making processes differ based upon situational constraints. In the Send off scenario, participants relied on mental models and schemata rather than information taken from the match to select a course of action. In contrast, in the Chasing the game scenario, participants tended to use information from the match to assess the situation prior to selecting a course of action. The findings have implications for coach education programs, which are discussed.

---

Journal article | Peer reviewed

Applying a systems thinking lens to child sexual abuse in sport: an analysis of investigative report findings and recommendations ↗

by Colin Solomon, Scott McLean, Karl Dodd and Paul M. Salmon

2025
Child Abuse & Neglect

Background Sporting organisations and governing bodies are facing increased pressure to prevent child sexual abuse (CSA) in sport. This has led to an increase in investigative reports into CSA that include recommendations on how sporting organisations could improve child safeguarding. Current peer reviewed literature on the prevention of CSA in sport, indicates that the majority of research has been on interventions at the levels of the victim and perpetrator, rather than on broader components of the sports system. However, it is not clear whether this is the case in investigative reports. Objective The aim of this study was to analyse investigative reports into CSA in five Australian sports (Swimming, Cricket, Gymnastics, Football, and Tennis), to evaluate the extent to which a systems thinking approach was adopted to understand the broader systemic factors enabling CSA in sport. Method Factors enabling CSA detailed in the reports, as well as their accompanying recommendations, were mapped to a systems thinking-based framework. The identified enabling factors and recommendations were then evaluated to determine the extent to which a whole of systems focus had been adopted in the investigative reports. Results In total, 30 enabling factors to CSA were identified, with the majority focused at the higher levels of the sports system (e.g., Governance, Policy, Reporting/Handling issues etc.). This contrasts with the peer reviewed literature. Conclusions The findings indicate that the identified enabling factors to CSA align with a systems thinking approach, whereas the recommendations to safeguarding partially adhere to the tenets of system thinking.

---

Editorial

Football clubs and their ‘hive mind’: distributed situation awareness (DSA) and its role in injury management ↗

by Mitch Naughton, Drust Barry, Michael Brownlow, Scott Mclean, Paul Salmon and Tom Page

2025
British Journal of Sports Medicine

Correspondence to Dr Scott Mclean; smclean@usc.edu.au Injury rates in elite football remain persistently high, despite significant investments in sports science, medical infrastructure and performance technology.1 This longstanding problem highlights potential limitations of the systems, processes and data currently involved in understanding and responding to the complexity associated with football injuries. Emerging qualitative investigations, particularly into the communication of performance teams, demonstrate important considerations for injury risk reduction frameworks.2 Shifting attention towards how awareness is shared, connected and acted on across the complex, interdisciplinary systems (ie, performance, medical, coaching, management) that define modern football clubs requires a pro-active approach.3 One concept offering explanatory and applied utility in this context is distributed situation awareness (DSA).4 5 DSA recognises that no single individual actor holds all relevant awareness and that effective decision-making requires the integration of disparate, yet complementary, awareness across a system of actors.5 This editorial introduces DSA as a critical concept to improve injury prevention and management within elite football clubs. What is distributed situation awareness? DSA has been extensively applied in numerous safety critical domains such as aviation, healthcare and defence, where teams, organisations and even entire sociotechnical systems must make time-sensitive, high-stakes decisions in complex and dynamic environments.5 6 In such contexts, the emphasis of the DSA model is not on an individual’s ability to perceive and interpret information in isolation, but on the system’s capacity to develop and maintain a coherent understanding of its operational environment.6 This involves the connection of mental models, timely communication, role clarity and compatible use of information and technologies. Previous yet scarce applications of EAST in sport have demonstrated its utility in modelling DSA and identifying key nodes (ie, tasks, actors), information bottlenecks and pointing to opportunities to optimise DSA and performance.14 Improving player health and performance in the Premier League and elite football will not come solely from more data, better algorithms or increased specialisation.

---

Magazine article

Nobody wants to talk about AI safety. Instead they cling to 5 comforting myths ↗

by Paul Salmon

2025
The Conversation

This week, France hosted an AI Action Summit in Paris to discuss burning questions around artificial intelligence (AI), such as how people can trust AI technologies and how the world can govern them. Sixty countries, including France, China, India, Japan, Australia and Canada, signed a declaration for “inclusive and sustainable” AI. The United Kingdom and United States notably refused to sign, with the UK saying the statement failed to address global governance and national security adequately, and US Vice President JD Vance criticising Europe’s “excessive regulation” of AI.

Grants

31 October 2025

A computational decision support system for doping prevention

World Anti-Doping Agency (Canada, Montreal) - WADA
Grant no. 0980030126.

Paul Salmon

---

3 February 2025 - 29 August 2025

Level crossing camera PoC human factors evaluation

Queensland Department of Transport and Main Roads (Australia, Brisbane) - TMR
Grant no. 0980030383.

Paul Salmon, Zohre Abedi and Gemma Read

---

1 August 2025

The Understanding and Preventing Led Outdoor Data Systems (UPLOADS) project - 2025

Sport and Recreation Victoria
Grant no. 0980031089.

Paul Salmon

---

15 September 2022 - 30 June 2025

A Sunshine Coast Hospital and Health Service system digital twin decision support model

University of the Sunshine Coast (Australia, Sunshine Coast) - UniSC
Grant no. 0980027688.

Gemma Read, Scott McLean and Paul Salmon

---

6 June 2025

Level Crossing Behavioural Assessment - Kilgour Street, Geelong

Victorian Department of Transport and Planning
Grant no. 0980030760.

Paul Salmon and Gemma Read

---

26 August 2021 - 28 April 2025

Optimising international doping control systems: Cycling and Football

World Anti-Doping Agency (Canada, Montreal) - WADA
Grant no. 0980026843.

Paul Salmon, Kelly Murphy and Scott McLean

---

9 January 2024 - 31 March 2025

Toward a reconfigurable command control experimental testbed at UniSC

University of the Sunshine Coast (Australia, Sunshine Coast) - UniSC
Grant no. 0980028753.

Melissa Nursey-Bray and Paul Salmon

---

12 March 2025

Level Crossing Behavioural Assessment - Anderson Street, Yarraville

Victorian Department of Transport and Planning
Grant no. 0980030445.

Paul Salmon and Gemma Read

---

11 March 2025

Level Crossing Behavioural Assessments - Donnybrook Road, Donnybrook & 2025 Site Comparison

Victorian Department of Transport and Planning
Grant no. 0980030446.

Paul Salmon and Gemma Read

---

13 January 2025

Implementation of the National Driver Distraction Roadmap: Incorporating Human Factors in Transport Systems

Austroads (Australia)
Grant no. 0980030045.

Gemma Read, Zohre Abedi and Paul Salmon

Supervision

Doctoral Thesis Supervision - Completed

Anticipating System Safety Performance: System-Level Safety Leading Indicators ↗

Students: Elizabeth Grey

Associated Researchers: Paul Salmon and Gemma Read

1 May 2014 - 15 May 2025

Proactive safety management uses safety performance indicators or safety leading indicators to provide information on current and future system states. These indicators are designed to provide early warning of potential future events. However, leading indicator systems have tended not to be aligned with contemporary systems-thinking models of safety and accident causation. To ‘future-proof’ sociotechnical systems (STSs) against accidents, leading indicators are required that provide insight into system states at the systems level. This thesis aimed to develop a systems thinking-based safety leading indicator scheme. A railway projects environment was selected as the application domain, given the rail industry has experienced system-scale accidents over time, is relatively accessible from a research perspective and is an interesting expression of a complex sociotechnical system. Previous work synthesising contemporary accident causation models produced 15 system accident tenets (SATs) that were adopted as the underpinning theoretical approach to this thesis.

---

Doctoral Thesis Supervision - Completed

Using Distributed Situation Awareness to Understand and Enhance Multi-Agency Emergency Response to Natural Disasters ↗

Students: Alison O'Brien

Associated Researchers: Paul Salmon and Gemma Read

1 February 2018 - 1 March 2025

Natural disaster events such as fires, floods and cyclones are increasing in rate and intensity worldwide. Apart from the devastating human and economic costs, there are other substantial short and long-term social, health, psychological, cultural, and environmental impacts. Effective disaster management systems rely on multi-agency emergency response in the immediate aftermath of a disaster. Situation Awareness (SA) has commonly been identified as a contributory factor in sub-optimal responses; yet, from a systems thinking perspective it is poorly understood and it is unclear how it can be optimised. Distributed Situation Awareness (DSA) is a contemporary theory that allows the examination of complex multi-agency systems and the identification of factors that promote effective emergency response measures. The theory posits that SA emerges from interactions across an entire system and challenges the notion that it is SA solely an individual's to ‘lose’. The theory is underpinned by a set of tenets that have not been explicitly tested in multi-agency emergency response. This research aimed to explore the application of DSA in a civilian multi-agency emergency response context and to test the tenets of DSA within this domain. Additionally, the research sought to understand how DSA can be optimised to improve disaster management outcomes. A literature review established that while the importance of SA was well accepted, it was under-investigated from a systems perspective. To address this gap three studies were conducted applying the Event Analysis of Systemic Teamwork (EAST) framework, firstly to a retrospective case study, and secondly to data collected during a live training exercise. The third study, a prospective risk assessment, applied EAST Broken Links (EAST-BL) to a generic response model. EAST works on the premise that complex collaborative systems can be understood through a network of networks (Salmon & Plant, 2022), this approach entails the examination of three distinct networks, task, social and information. A task network describes the order and interdependences between tasks, the social network describes the organisation and communications among system members and the information network describes the information that is used and communicated during task execution (i.e., DSA). By adopting different perspectives, these studies enable a comprehensive evaluation of DSA and the testing of its tenets across various multi-agency response scenarios. The findings suggested that failures in information transfer between tasks led to more severe consequences compared to failures between agents. The influential role of social media in developing and maintaining DSA during emergency response was also evident. While some of the DSA tenets required modification the studies confirmed the validity of others. The findings are discussed in terms of their theoretical, methodological, and practical implications.

---

Thesis Supervision - Completed

Using Systems Human Factors and Ergonomics to Identify and Mitigate the Risks Associated with Future Invasive Brain-Computer Interfaces ↗

Students: Brandon King

Associated Researchers: Paul Salmon and Gemma Read

2021 - 2024

Brain-computer interfaces (BCIs) are an emerging technology that can be implanted within the brain to read and decode brain signals and control electronic technologies. BCI technologies are advancing rapidly, and their utility has been demonstrated in relation to therapeutic applications, such as facilitating communications for people with completely locked-in syndrome or restoring motor functions. In the near future, BCIs could also be used widely for human enhancement or entertainment purposes. Such an invasive technology will inevitably introduce risks not just to users but to the entire sociotechnical BCI system; however, the extent to which these risks have been considered is unclear. This thesis is a direct response to this gap and aims to apply systems Human Factors and Ergonomics (HFE) methods to identify the system-wide risks of BCIs and identify potential controls that could be implemented throughout the BCI lifecycle. The ultimate goal of this research is to informthe safe and ethical development and introduction of advanced and efficacious BCIs. The research undertaken in this thesis includes a systematic literature review of BCI risks, a comparison and evaluation of systems HFE methods that could be applied to address the thesis aims, three studies modelling the BCI system lifecycle and prospectively identifying risks and controls, and the development and evaluation of a set of risk controls for the BCI system throughout its lifecycle. The systematic review (Chapter 2) found a range of risks have been discussed; however, a key gap was that no formal risk assessment methods had been applied to identify risks associated with BCIs. Next, a set of systems HFE methods were evaluated and selected according to their ability to identify the risks of BCIs and potential risk controls (Chapter 3 and 4). The Work Domain Analysis – Broken Nodes (WDA-BN; Chapter 5) and the Networked Hazard Analysis and Risk Management System (Net-HARMS; Chapter 6) prospective risk assessment methods were then applied to model a future envisioned BCI system lifecycle and identify potential risks and controls. A computational modelling approach, Agent-Based Modelling, was then applied to dynamically simulate the emergence of risks over time following the envisioned introduction of BCIs within a simulated society. This analysis was used to identify the societal conditions likely to lead to the most favourable and unfavourable outcomes for BCI users and society (Chapter 7). Risks identified through the application of the three adopted methods ranged from physical health risks of BCI implantation and degradation, shortfalls in BCI user and stakeholder training, lack of BCI user and past user support, and uncontrolled malicious hacking. The risk controls identified across the three studies were then aggregated, synthesised, desktop tested, and refined into a set of 10 high-level risk control strategies before being evaluated by BCI subject matter experts (Chapter 8). Overall, the findings of the thesis provide various complementary models of an envisioned BCI system lifecycle and a detailed overview of the individual, organisational, and societal risks that could emerge when BCIs are introduced. Key contributions of this thesis include the development and application of a novel future risks many model process framework, the proposal of a prototype model of BCI risk sources (Chapter 9), and the development of a practical set of high-level risk control strategies for BCI stakeholders. Future research should build on this by further testing the high-level risk control strategies and evaluating their application, and testing the reliability and validity of the future risks many model approach and the modified HFE methods.

---

Doctoral Thesis Supervision - Completed

Applying Systems Thinking to Understand and Prevent the Risks Associated with Hazardous Manual Tasks

Students: Peter McCormack

Associated Researchers: Gemma Read and Paul Salmon

2019 - 2023

---

Doctoral Thesis Supervision - Completed

Testing and Applying Driving Research Paradigms to Understand Visual Attention in Learner Drivers

Students: Rachael Wynne

Associated Researchers: Gemma Read and Paul Salmon

2016 - 2022