A Cognitive Approach to Situation Awareness: Theory and Application

Sébastien Tremblay, Simon Banbury

• Publisher: Routledge
• Year: 2017
• Topic: Psychology
• eBook ISBN:
  9781351962650
• Subtopic: Applied Psychology
• Index: Psychology

The importance of 'situation awareness' (SA) in assessing and predicting operator competence in complex environments has become increasingly apparent in recent years. It has been widely established that SA is a contributing factor to many commercial and military accidents and incidents. Yet determining exactly what constitutes SA is a very difficult task, given the complexity of the construct itself, and the many different processes involved with its acquisition and maintenance. This volume brings together recent developments from researchers and practitioners from around the world who are studying and applying SA from a cognitive perspective. The 41 contributors represent many different theoretical perspectives, research approaches and domains of application. Each chapter has a primary emphasis around one of three main topics - theory, measurement and application and examines the considerable inter-linkage between them. To bring further coherence to the book, all of the contributors received draft manuscripts of those chapters most relevant to their own. Designed to be completely international and interdisciplinary, the authors themselves present varied perspectives from academic departments and industrial organisations from around the world, and from broad applications - with contributions from researchers in the domains of process control, sport, aviation, transportation, and command and control. The readership includes practitioners, academics and researchers within human factors, ergonomics and industrial psychology; Graduate and Undergraduate students specialising within these areas during their final year.

PART I
THEORY

Chapter 1
Defining and Modeling Situation
Awareness: A Critical Review

Robert Rousseau Sébastien Tremblay Richard Breton

Introduction

The concept of Situation Awareness (SA) is now well established in the domain of human factors studies in complex environments. In practice, a long list of examples exists to persuade someone that SA has its own reality and its own importance. Indeed, knowledge of the information relevant to efficient task performance is critical to safety and productivity in a wide variety of situations such as air-traffic controllers, jet pilots, nuclear power plant operators and military commanders (see Durso and Gronlund, 1999). However, when an attempt is made to define SA, the result is highly variable. Reviews of definitions from varied sources (e.g., Dominguez, 1994; Breton and Rousseau, 2001) provide a clear indication of the variety of viewpoints about SA. One might not be too concerned with that situation. As Pew pointed out:

The term situation awareness shares a common history with several psychological concepts such as intelligence, vigilance, attention, fatigue, stress, compatibility, or workload. During decades, all these terms were poorly defined. However, each became important because they attracted attention on critical processes or mental states that were previously unknown. Ultimately, they changed the ways to study human factors problems, and they brought new benefits. (Pew, 2000, p. 33)

Defining SA

Surveying definitions of SA reveals the variety of conceptions currently conveyed in the literature. Breton and Rousseau (2001) performed a systematic classification of 26 SA definitions. These definitions turned out to be evenly divided in two classes corresponding to the now accepted duality of SA as a State or as a Process.

State and Process-oriented Definitions

SA is often considered a ‘buzzword’ that serves as a label for a range of cognitive processes (see Prince, Salas and Brannick, 1999). Sarter and Woods (1995, p. 16) proposed that: ‘the term situation awareness should be viewed just as a label for a variety of cognitive processing activities that are critical to dynamic, event-driven, and multitasks fields of practice’. Endsley (1988, 1995) takes an opposite position reflected in the well-known definition of SA as: ‘the perception of the elements in the environment within a volume of time and space, the comprehension of their meaning, and the projection of their status in the near future.’

Endsley further claims that SA is a state of knowledge that needs to be distinguished from the processes used to achieve that state. These processes should be referred to as situation assessment. That distinction between ‘Process’ and ‘State’ definitions is of considerable importance. One of the major difficulties in working with SA is to avoid confusion between SA knowledge and the underlying cognitive processes such as perception, memory, attention or categorization that underlie the production of SA. The proposition of Endsley (1995) to distinguish between situation awareness (a state) and situation assessment (a process) has been very influential on development of measures and modeling efforts (Pew and Mavor, 1998). Unfortunately, as pointed out by Pew and Mavor, the similarity in acronym for situation awareness and situation assessment has led to some confusion. Moreover, situation assessment is included in a number of decision-making models without any reference to Situation Awareness. For instance, Klein (1997) refers to situation assessment as central to decision-making and as including appropriate goals, cue salience, expectations and identification of typical actions. Given such a definition, it is apparent that in Klein’s view, situation assessment is a very broad concept that covers situation awareness content as defined by Endsley as well as actions. Thus, simply identifying the processes supporting Situation Awareness as a Situation Assessment process is likely to bring some confusion given the way situation assessment is already defined in the broader literature on decision-making.

Similarly, defining SA as the relevant knowledge of which an operator is aware is not without problems. According to Smith and Hancock (1995), SA cannot be equated with the momentary knowledge of which an operator is aware. Clearly, SA cannot simply be equated to any verbal report of the content of consciousness about a situation. Even more fundamentally, the question remains as to whether SA should be considered as meaning conscious mental content, precluding all forms of tacit knowledge (see Banbury, Andre and Croft, 2001; Croft, Banbury, Butler and Berry, Chapter 5, this volume).

As pointed out by Durso and Gronlund (1999) that difficulty could very likely stem from the existence of two basic approaches to the question. One approach, the Operator-focused approach, is concerned with the properties, basically mechanisms, of the operator as they determine SA. The second approach, the Situation-focused approach, views SA as determined by the environment or situation in which the operator is at work. The interest of that point of view is that it resets the issue of SA definition around the two basic factors on which SA depends. A State-oriented definition could then be associated with a Situation-focused approach and characterized as driven by the properties of the Situation aspect of the SA concept. On the other hand, a Process-oriented definition, associated with an Operator-focused approach, would be centered on the properties of the operator or agent. The distinction between Operator and Situation focused approaches helps to reframe the classical state-process distinction.

Operator-focused Approach

An Operator-focused point of view is more concerned with the set of cognitive processes supporting the production of the mental representation corresponding to the SA state. These processes are by definition a general property of the human agent. That approach follows an information-processing framework that considers a mental representation of the world to be based on a processing with specific functions (e.g., Marr, 1982; Dawson, 1998). That framework requires an explicit description of the processes involved in providing humans with cognition. As an example, Dominguez (1994) includes a set of processes on which SA depends: information extraction, information integration, mental picture formation, and projection and anticipation. In a way, it could be argued that Endsley (1988) also describes three cognitive processes or functions: perception, comprehension and projection, even though she defines SA as a State. Most of these Process-based definitions do not rely on available generally accepted human information processing (HIP) models like the one described in Wickens and Hollands (2000). The HIP models identify a network of processes like perception, attention and memory.

Situation-focused Approach

A Situation-focused point of view is clearly concerned with the mapping of the relevant information in the situation onto a mental representation of that information within the agent. State-oriented definitions limit the description of processes involved in SA. In fact, they are on line with the fundamental concept of direct perception derived from the work of Gibson (1979). Direct perception is based on a number of principles two of them being of interest for SA: (i) all the information necessary for perception is contained in the environment; (ii) perception is immediate and spontaneous. Based on these principles one can assume that in order to understand perception, the priority must be on understanding the environment without concern for possible underlying processes.

The Situation-focused approach provides a more factual basis on which to define SA as a State. The Situation can be defined in terms of events, objects, systems, other persons and their mutual interactions. However, this is complicated by the fact that a Situation is domain-dependent and that elements of the situation that should be within SA in aircraft control will be different from the ones that are required for an anesthesiologist.

Pew (2000, p. 34) provides an example of what could be included in a generic definition of Situation, even though it is somewhat specific to aircraft piloting. He defines a situation as ‘a set of environmental conditions and system states with which the participant is interacting that can be characterized uniquely by a set of information, knowledge, and response options.’ Pew then proposes that SA should integrate, when applicable, five aspects of the situation:

• the surrounding environment;
• the mission’s goals;
• the system;
• the available physical and human resources;
• the crew.

In that view, ‘situation’ takes a very large meaning. It includes task and mission features, as well as the other human agents in the significant environment. It is much more than the awareness of the distribution in space of objects within a contextual environment. Actually, Smith and Hancock (1995) view SA as a concept that is centered on task goals. They explicitly include task goals, criteria of performance and cues in the environment or situation.

From Defining to Modeling

In the context of developing the concept of SA, one is then left with a double-edged sword problem. If SA is a state, it is essential to give a precise definition of the knowledge that defines the state. There should be a certain mapping between a situation schema and a knowledge schema. If one is to improve SA, the elements of the situation critical for SA should be specified, and the SA content definition should follow from these elements. On the other hand, if SA depends on a set of processes that are not an intrinsic part of SA as a state but on which SA depends, it becomes important to specify which processes are essential to SA. SA improvement, for instance, will depend upon changes in the operation of these processes.

The debate remains as to whether a definition of SA should be limited to content or should include the processes or functions linked to the awareness of the situation. Should SA include (or not) what some authors refer to as situation assessment? It is not possible to provide an answer to that question from the strict analysis of SA definitions. However, many authors have expanded their definition of SA by developing models of SA. These models are reviewed in the next section.

Modeling SA

In its most generic form, a model is a representation or description, more or less complete, of an object or a process. The representation of an object refers to a statement, formula or image that depicts or describes that object. A model describing a process will include one, or a series of operations, that accomplish an end. Models are useful to support reflection about the effectiveness of action or of a particular device with which a human interacts. They provide a logic that lends itself to measurement or experimentation about a problem related to the model. Many different types of models have been developed that address specific aspects of performance, particular processes or goal-directed actions. For the sake of reviewing, models of SA will be divided in two groups: descriptive models and prescriptive (computational, numerical or simulation-based) models (Zacharias, Miao, Illgen, Yara and Siouris, 1996).

Descriptive Models

Most SA models are descriptive. Descriptive models of SA are driven by an effort to describe the actual SA process. A good descriptive model will reflect the operator’s, or most often the pilot’s, decision-making processes. It is very flexible and can take into account most constraints that are typical in an operational setting. However, the descriptive models are difficult to represent in formal prescriptive terms. They will provide rather general predictions. Their main strength resides in the analytic power they carry. They provide a systematic description of what SA is and how it is produced.

A systematic analysis of SA descriptive models is provided in Endsley, Holder, Leibrecht, Garland, Wampler and Matthews (2000). They provide a list of eight models presented as ‘… models of how people achieve SA in complex domains …’ (Endsley et al., p. 34). Amongst these models, Endsley’s model (1988, 1995, 2000a) clearly stands as the reference for most work done on SA. It is the prototypical descriptive model for SA. A number of other models focus on specific aspects of SA but remain within the constraints of Endsley’s model (e.g., Endsley and Jones, 1997; McGuinness and Foy, 2000). Given the tremendous influence of Endsley’s SA model on the evolution of the concept and its acceptance by field experts, we will briefly describe the model and then address the basic aspects of Endsley’s model.

Endsley (1988, 1995) presents an SA model with two main parts: a core SA model and a set of various factors affecting SA. The first part, the Core SA model represents the processes directly responsible for SA. The second and much more elaborate part describes in detail the various factors affecting SA grouped into four broad classes: external world, task and environmental factors, individual factors, and a set of domain factors. These factors also include contributions of all components of current HIP models like goals, active schemas, past experience, attentional processes and memory. The core SA model is what is usually referred to as the SA model. The core model of SA presented in Endsley (1995) is the basis for much of the current modeling of core SA. It is a three-level system that can be described as follows.

Level one, Perception of the Elements in the Environment, is the first step in achieving SA. It provides information about the status of the relevant elements in the environment. The interaction with long-term memory knowledge makes possible the inclusion of classification of information into understood representations. The subset is under attentional selection based on task requirements. The elements are structured into meaningful events situated in time and space. This content is active in working memory thereby providing a basis for its awareness.

Level two, Comprehension of the Current Situation, is a synthesis of disjointed elements of Level one. It provides an organized picture of the elements with a comprehension of the significance of objects and events. Schemata or mental models stored in long-term memory are the basis for level two SA. Level two SA is then defined as a situational model depicting the current state of the mental model.

Level three, Projection of Future Status, is achieved through knowledge of the status and the dynamics of the elements and comprehension of the situation. It enables predictions about the states of the environment in the near future.

Endsley’s model is a complex model in which State and Process aspects are meshed. Endsley et al. (2000) describe an extended version of Endsley’s (1995) model, adapted to infantry operations. It is, by far, the most extensive SA model currently available including an impressive set of cognitive and non-cognitive contributing factors. While this makes the model all-encompassing in terms of factors affecting SA, it does not push the modeling of core SA any further.

Process Aspects of Descriptive SA

The essential aspect of the Endsley family of SA models about process issues is that SA is achieved through a hierarchical linear processing system. This sets SA within a well-known approach on descriptive modeli...

Table of contents

• Cover Page
• Title Page
• Copyright Page
• Contents
• List of Contributors
• Preface
• PART I Theory
• PART II Application
• Author Index
• Subject Index