Introduction
Recent contributions to the field of Resilience Engineering (RE) have added to the continuous development of new concepts and methodologies to improve resilience at different organisational levels. Part of these contributions has focused on training for adaptive capacity of individuals and teams to cope with changes and disturbances of work, since literature recognise that working tasks (at least in complex settings) are not as stable as procedures, manuals and regulations might depict. It is becoming accepted that more is needed than training for recognition of pre-defined situations and application of corresponding procedures, that is, individuals and teams should be prepared by their training to also cope with unexpected situations. In previous volumes of RE contributions we have introduced new methods in order to address these unexpected situations (Bergström, Dahlström and Petersen, 2011; Dekker, Dahlström, van Winsen and Nyce, 2008). In this volume we will rather discuss the theoretical foundation of team training and the potential to align such a foundation with RE theory. Guided by two of the four cornerstones of RE (Hollnagel, 2011) our argument is that traditional approaches to sharp-end training should be reviewed, revised and readapted to concepts more aligned with RE thinking.
This chapter will initially discuss the traditional perspective on high-risk industry team training in order to present some of its discordance with resilience concepts. Then an alternative approach to such training is suggested. The suggested approach connects the RE abilities of responding and monitoring (Hollnagel, 2011) to the theoretical perspectives of distributed cognition and complexity. The aim of outlining a theoretical agenda is for future development of methods for training and assessment, of team performance in high-risk industries, to be more aligned with RE theory.
Traditional Training Principles
An important starting point to understand the current paradigm of traditional high-risk industry team training is that it is based on the notion of task stability. Since technical systems are assumed to be highly predictable, and humans provide a potentially important but unpredictable and fallible variability to the system, proceduralisation of work is central. Consequently, training is focused on identification and reaction, that is, a work situation is identified and the corresponding procedures for the situation are executed. In this sense, the human operator is seen as an effective repository for a large number of stimuli-response programs for pre-defined situations (Hollnagel and Woods, 2005).
The study of man as a stimuli-response repository puts the emphasis on human cognition, which is supposed to produce correct and rational decisions according to optimal information processing, based on accurate awareness of the situation, assertive communication and effective interaction between leaders and followers. The ultimate implication of this information-processing paradigm, currently and traditionally dominant in the field of Human Factors and team training, is the use of different techniques to assess cognitive processes by means of behavioural analysis (Flin, O’Connor and Crichton, 2008). The central idea is that human behaviour is a genuine representation of human cognition, since certain cognitive processes should result in certain behaviours. This has led to an increasing use of evaluation techniques based on behavioural markers, such as NOTECHS and different forms of on-the-job observations focused on error identification and categorisation, such as Line Operations Safety Audit (LOSA).
But it is not only the focus on human cognition and behaviour that constitute the traditional high-risk industry training paradigm. In recent years, an increasing importance has been given to the use of scenarios based on previous incidents. In this case, past incidents are brought into training in order to ‘fix’ a perceived problem that caused those incidents to occur. In most cases this training reinforces existing procedures or introduces additional procedures intended to prevent the same incident from reoccurring. This again assumes stability; not only of tasks but also of risks, assuming that increased system rigidity can indeed eliminate discovered risks. International Air Transport Association (IATA), for example, have recently, as a part of their training and qualification initiative, brought together several of the above-mentioned concepts under the umbrella of ‘Evidence Based Training’ (EBT). EBT emphasises a need for gathering, sharing and categorising data from past events in order to have the lessons from them incorporated in future training (Voss, 2012). Some of the first conclusions drawn from the EBT project highlight issues related to pilots’ attitude toward safety, such as problems related to non-compliance with standard operating procedures even in threatening circumstances. Also the accident with Air France 447 has triggered central stakeholders to draw similar conclusions, including The European Safety Agency (2012) and the UK CAA (2013) publishing accounts which reproduce what is considered ‘acceptable’ explanations, depriving the industry from far more difficult and complex explanations. Such explanations would allow an increased understanding of the role of the operator in the system and direct how this role could be strengthened by a focus on resilient behaviour rather than on the stimuli-response behaviour of procedural adherence.
Although many of these models and techniques are simple to understand and convenient in how they ascribe safety-related problems in the operational environment to the behaviours of the individuals operating in such an environment, our argument is that this simplicity represents a considerable trade-off, that is, there is a price to pay. While these models and techniques normatively assume task stability, they effectively exclude and hide the complexities of numerous highly contextual factors in the work situation. If these factors would instead be identified they could contribute to operator understanding of the inherent risks in their work processes and enhance their resilience in handling the variability in day-to-day situations as well as their handling of unusual and unexpected situations.
The Resilience Approach
Hollnagel (2011) explains that the approach of RE is focused on an in-depth understanding of an organisation’s ability to ‘adjust its functioning prior to, during or following changes and disturbances, so that it can sustain required operations under both expected and unexpected conditions’ (p. xxxvi). This broader definition is then operationalised as the study of ‘the abilities to respond to the actual, to monitor the critical, to anticipate the potential and to learn from the factual’ (p. xxxvii). The first question in order to develop a framework for a resilience approach to sharp-end training becomes which of these four abilities that can at all be addressed in training. In this chapter we will focus on outlining a conceptual agenda for how team training can be used to enhance the abilities to respond to the actual and monitor the critical, with the abilities to anticipate and learn seen as potential secondary effects of such training. Underlying this conceptual agenda will be the perspectives provided by cognition as a phenomenon distributed among the actors engaged in a specific context (Hutchins, 1995a) as well as complexity theory. Together these two perspectives embrace the context-specific nature of monitoring and response as well as how both these abilities are emergent properties of multiple relations and interactions rather than the products of reliable behaviour at the level of the individual actor.
Complexity theory embraces the corollary that a system cannot be fully described nor fully controlled (Cilliers, 2005), which is fundamentally important for understanding safety-critical work from an RE perspective. Complexity is what emerges when a system, put together by physically separated actors, elements and artefacts (for example, different wards at a hospital or different aircraft on approach to an airport), shifts from loose to tight coupling and from high autonomy to high interdependence in a short span of time, for example, as when different hospital wards normally functioning relatively autonomously become highly interdependent in response to an escalating situation (Bergström, 2012; Dekker, 2005). Likewise, instead of training only for task stability and ‘correct behaviour’, the RE approach needs to embrace that variability is not only normal, but also necessary for the ability to dynamically adapt to unexpected situations. This does not mean that all procedures are detrimental to safe work, but that understanding of procedures and their application in a variety of predicted situations, as well as skills for handling of unexpected situations, must be the focus of training.
From the approach of analysing cognition not as individual-centred but rather as a distributed trait of a situation, the focus shifts from the human as an information processor to the actual work in which the human engages, together with technological systems and other humans (Woods, 2003). In this sense, the unit of analysis is changed from the mind of an individual to the joint system of humans and artefacts engaged in a particular working situation (Hutchins, 1995b).
Thus, from the perspective of RE the understanding of the ability to respond and monitor lies neither in observing human behaviour, nor in deconstructing human work from engineering-centred task analysis models, motivationally based models or concepts such as workload management, situation awareness and decision-making. Instead the interest of analysis lies in the complexities facing the sharp-end operators in their day-to-day work and how to improve their adaptive capacity in order to promote safety.
Important Principles for Training Response and Monitoring
In our respective organisations we are incorporating the notion of resilience into training programs as well as in methods to assess team performance. Bergström is involved in developing a training program for multi-professional team training of healthcare staff.
Henriqson and Bergström are together developing methods for team assessment based on the theoretical fundaments outlined here, and Dahlström has, together with Bergström, worked for several years with alternative approaches for training team coordination in escalating situations. Dahlström is in recent years closer than ever to ‘reality’, being the Human Factors Manager at Emirates Airlines Flight Operations Training. With the context of our work outlined, the last part of the paper will be focused on describing the main principles of the resilience approach to team training and team performance assessment.
Responding to the Actual
Just as concepts such as success, safety or risk, the performance of teams needs to be understood as an emergent property of a multitude of interactions and relations, embracing that ‘there are no fundamental differences between performance that leads to failures and performance that leads to successes’ (Hollnagel, 2011, p. xxxv). This should be seen as a contrasting view to the idea that team performance can be ensured by best-practice guidelines or by gathering increasingly more data on their past performance. This tenet also questions the idea that concepts such as ‘correct behaviour’ can be imposed on the system. To move beyond containing or limiting the emergence of variance in performance, by means of stronger emphasis on following standard operating procedures or replacing unreliable humans with reliable technology, the resilience approach asks us to establish practices for dealing with variability and uncertainty. In this sense, in order to analyse the capacity of individuals and teams to perform work, or to analyse the design of technology, it is necessary to focus on the phenomena emerging from the interactions of joint cognitive work (rather than on delimited individual cognition and action).
The interactive and dynamic context of normal work can offer guidance for action at the same time as it will be affected by the action carried out. In this case, there may be no precedence in regards to specific modes of interaction, except for those required for immediate needs of the event (for example, for inflight collision avoidance). For example, Henriqson, Saurin and Bergström (2010) identified that local representations, which condition the broader coordination context in a cockpit, are results from interactions between internal representations (interpretative structures unique to the individual) and external representations (elements in the action context, such as symbols, numbers, data and shapes), which are in themselves always partial and incomplete. This means that rather than only providing finalised action programs (that is, procedures) for events, it is imperative to ensure operator understanding of the multitude of available representations and interactions and develop the skills to apply these to a variety of situations.
Just as Nyssen (2011), we emphasise that the joint cognitive systems-notion of coordination forms an important part of the framework in understanding organisational response. In the study mentioned above (Henriqson, Saurin and Bergström, 2010) we describe how coordination can be interpreted as a situated and distributed cognitive phenomenon in the cockpit of commercial aircraft. The study provides an integration of the perspectives of joint cognitive systems theory with four coordination requirements described in the literature: common ground, interpretability, directability and synchrony (Klien et al., 2005). We have also used these coordination requirements in studies aiming at interpreting joint cognitive work in escalating situations, forming an early suggestion for a method to assess team performance based on joint cognitive systems theory (Bergström, Dahlström, Henriqson and Dekker, 2010).
When developing programs for team training the importance of the multi-professional approach is stressed. In a multi-professional setting the participants can together identify the situations in which the relationships and interactions between them, rather than their respective reliability as safe components, establish safety (or risk) as an emergent property of the system. This multi-professional dialogue is seen not only as a vital activity for enhancing the ability of the organisation to enhance future response to changing conditions and monitor its current performance, but also for widening the learning-loops in the organisation and anticipate the characteristics of future interactions.
Furthermore, joint cognitive systems theory has adopted a cybernetics approach to define control by its circularities of feedback and feed-forward. This approach combines the cybernetic notion of regulation (Ashby, 1959), the Perceptual Cycle of Neisser (1976) and Hutchins’ ideas of distributed cognition (1995a, 1995b) to provide a functionalist approach of control. In this sense, control ‘happens’ during the interaction between ‘human-task-artefact’ and is goal-oriented and influenced by the context in which the situated activity takes place. From the resilience perspective outlined here we can also see control as an emergent property of the systems’ ability to respond. We have been involved in developing methods for team performance assessment through operationalisation of Hollnagel’s Contextual Control Model (Hollnagel and Woods, 2005) in order to map how the interpreted level of control shifts during a scenario by including both the participants’ own reflections and the observer’s interpretations (Palmqvist, Bergström and Henriqson, 2011). Noteworthy is that a contextual model of control is, in contrast to behavioural markers or error categorisations, not normative, meaning that no level of control is seen as more appropriate than another but instead as highly contingent upon the situation and context. This is a promising way to go beyond the behavioural markers or error classifications of the information-processing paradigm.
Monitoring the Critical
Even a widely accepted information processing notion such as ‘situational awareness’ becomes problematic from the resilience perspective. Complex systems do not allow for complete or any other finalised and ‘correct’ descriptions. Instead of using hindsight to accuse operators for not having had optimal situation awareness and trying to improve this by training of ‘correct monitoring’, the focus of training needs to shift to ensuring that different and competing productions of meaning (based on different experiences and viewpoints) are available when working in safety critical environments. This can allow a parallel development of best practice for routine situations as well as skills development for unusual and unexpected situations.
Ultimately, this is a shift towards the notion of diversity, which is fundamental for complexity theory (Cilliers, 1998). Complex systems are resilient when they are diverse, which also needs to be embraced by the resilience interpretation of the ability to monitor. Diversity implies that different practitioners deploy different repertoires for responding to what, from their respective perspective, is seen as evidence as well as to each others’ constructions of such evidence (Dekker, 2011). This argument was also made in a recent study analysing the joint interactions of the multiple professionals (especially midwives) involved in the process of labour. The study raised questions regarding the idea of understanding performance by reference to best practice guidelines, emphasising that:
Patient safety efforts, then, might recognise, celebrate, and enhance the positive aspects of diversity that guarantee the emergence of resilience in complex situations. Such efforts can be made in activities of inter-professional team training where medical staff representatives are given the opportunity to identify complex as well as complicated situations in their work to achieve more efficient and effective patient-centred care (Dekker, Bergström, Amer-Wåhlin and Cilliers, 2012).
From the resilience perspective the notion of monitoring needs to be raised to a level of a joint cognitive effort, facilitated by a focus on strategies to monitor and bridge the current gaps in the system (Cook, Render and Woods, 2000). Again, when developing concepts for multi-professional team training in complex settings it is important to gather the actors that at certain points might become tightly coupled and highly interdependent (Bergström, Dekker, Nyce and Amer-Wåhlin, 2012). Together they need to be given the opportunity to build common ground, aided by facilitators, and enrich their different perspective with those of others in order to create prerequisites for successful response at the organisation’s most complex moments.
Summing Up the Argument
Resilience Engineering theory embraces how successes and failures are the results of the same kind of processes in dynamic and goal-conflicted environments. This important principle challenges traditional practices that assume a task-stable environment. Consequently, in efforts to enhance organisational resilience, there is a need for the establishment of a theoretical agenda to team performance and training which is more aligned with RE theory. This chapter discusses some thoughts for what to include in such an agenda.
In order to be aligned with the central notions of RE theory we argue that concepts to enhance the organisational abilities of responding to the actual and monitoring the critical need to be rooted in complexity theory and the theory of distributed cognition. The two perspectives complement each other. The perspective of distributed cognition helps to shift the unit of analysis from the individual actor to the coordination activities of the joint cognitive system. These activities can be understood and analysed based on certain requirements for successful coordination and/or the level of control emerging from the coordination activities. With the concept of emergence being essential to complexity theory, here is also where this perspective becomes important. Not only does complexity theory suggest that the unit of analysis needs to be the interactions and relations between actors rather than the behaviour of the individual actor, but it also helps outline the agenda for how to monitor the critical by the tenet that the complex system is resilient when it is diverse. Consequently the agenda for team training needs to be one of emphasising and enhancing diversity rather than, as is the risk of traditional approaches to team training, uniformity and rigidity.
The multi-professional discussion is seen as important to enhance organisational diversity and here the third and fourth cornerstones of RE as outlined by Hollnagel (2011) come in: the notion of anticipating and the notion of learning. It is in this multi-professional discussion that actors are able to learn about each other’s perspectives in order for them to establish the common ground which might make them able to anticipate the current and future actions of the joint cognitive system in which the actors engage. From this perspective the four activities are not disconnected or isolated when it comes to team training, but rather all four are tightly intertwined activities and can benefit from such multi-professional learning.
Commentary
Safety management, whether resilient or not, obviously depends on the competence and experience of the people working in the system. This competence and experience is to a large extent based on the training that is provided. This chapter shows how the fundamental premise of Resilience Engineering (that tasks are not stable and that variability and adjustments therefore are necessary) has consequences for training. The chapter presents a discussion of the requirements for a theoretical agenda for team performance and training that is aligned with the principles of Resilience Engineering, especially the abilities to respond and to monitor. Developing such an agenda is an important step from requirements to specifications.