A Comprehensive Model for Human Behaviour in Industrial Environments

A Comprehensive Model for Human Behaviour in Industrial Environments A. Sneddon, P.T.W. Hudson, D. Parker, M. Lawrie, M. Vuijk & R. Bryden University of Aberdeen Leiden University Manchester University Shell International Exploration and Production Abstract The lack of an integrated approach to theories of human behaviour in the industrial environment forms an unaddressed problem for Psychology as a discipline. People behave as a totality whereas we only have local micro-models to explain specific behaviours in restricted environments. In this paper we present a comprehensive framework that makes explicit the relationships between a wide variety of existing and well-validated models in different areas of specialisation within the study of psychology. The model integrates the range from perceptual processing, through decision-making and its biases, to the modelling of selection of action and the implementation of intentions. Lower level theories, perception and decision-making are typically the preserve of Cognitive Psychology, while later elements are usually seen as part of Organisational, Social and Health Psychology.

The lack of an

integrated model makes it difficult for non-psychologists, typically those who wish to make use of the insights provided after more than 100 years of study, to understand how human behaviour is explained and might be directed in an industrial setting. The model was originally developed to understand how people behave safely, or unsafely, in hazardous industrial settings, leaning upon work investigating high reliability industries and advanced safety cultures. It appears to provide a robust framework that allows extension to areas such as effective leadership behaviours, situation awareness, real time decision-making under stress, as well as driving behaviour.

A Comprehensive Model for Human Behaviour in Industrial Environments Introduction The nature of high reliability organisations is a work environment which possesses the capacity to be very hazardous. Managing these hazards to reduce the risk of accidents dictates the way in which most jobs are performed. Supporting the individual worker there or usually engineering safeguards, management systems, competence programmes as well as rules and procedures. However, despite these preventative mechanisms accidents still tragically occur and the question remains why? Research into a number of high-hazard, high reliability industries has indicated that human factors causes can be attributed to around 80% of accidents (Hoyos, 1995).

In

particular, research into human factors causes of accidents in the oil and gas industry (e.g. Rundmo, 1995; Rundmo, Hestad & Ulleberg, 1998) has indicated that lack of care and attention is often reported as one of the main immediate human factors causes (Mearns, Flin, Fleming & Gordon, 1997). Usually behind these immediate causes are several underlying causes such as managerial decisions and other leadership failures with their roots much deeper in the organisation. One reason for the prevalence of human factors causes in incidents is that human error is inevitable, because of the fundamental imitations of human cognitive architecture. In high-reliability industries such as nuclear power, fire-fighting, aviation, and oil and gas exploration errors can have potentially catastrophic repercussions, which is why there are precautions to minimise the impact of errors. What Lawton (1998) showed was accidents happen when unintentional acts (errors) combine with intentional acts such as rule breaking. Consider, for example, the work of a drilling crew on board an offshore rig. Much of the overall drilling process is performed by heavy machinery and monitored by computers, yet human crews are still required for the manual processes of, for example, looking after the drill pipe – adding or removing drill strings, etc. an incident will usually involve a error by one person whilst someone is taking a calculated risk .

The motivation for why people behave unsafely varies, but tend to

fall into one of two categories – unsafe behaviours of which they are aware, and unsafe behaviours of which they are not aware. People may knowingly participate in behaviours that are unsafe for a number of different reasons (this will be explained in more detail later in the paper) – for example, they have developed a bad working habit which has been reinforced by other workers (or possibly even supervisors), or has gone unobserved or unchallenged, and so has developed into an acceptable practice. However, as mentioned previously, people also behave in an unsafe manner, but are unaware of the fact that they are doing so. It is therefore imperative to distinguish the reasoning behind why people are behaving in this manner (i.e. safely both knowingly and unknowingly) in order to attempt to modify the unsafe behaviour and rectify the situation. Most approaches to unsafe or at-risk behaviours have concentrated upon listing the reasons why people perform unsafe acts, such as poor attitudes to safety, laziness, corner cutting or, frequently, deliberate taking of risks by individuals. This information is typically used to justify remedial measures intended to prevent such atrisk behaviours in the workplace. Such models, however, do not explain why people might have what appear to be such poor attitudes, or trade off safe but lengthy activities with more dangerous short cuts. The approach taken in this model is quite different, starting from asking what it would take to have people acting safely, as the standard, and then ascertaining why they might deviate from normal safe behaviour, demonstrating unsafe behaviours as pathologies of normal safe behaviour patterns. The assumption made here is that people, even thrill seekers, are usually interested in their own safety, as well as that of others; this assumption is not apparent in the usual unsafe behaviour models. The Safe Behaviour Model ‘Safe behaviour’ may be considered as an end product of a number of different steps or stages, as it requires all of these to be followed successfully in order for the entirety of behaving safely and avoiding an incident to be achieved. In conjunction with Leiden and Manchester Universities, Shell Exploration and Production developed the ‘Safe Behaviour modelTM’, which outlines the four main steps necessary for safe behaviour. Figure 1 details this model with four main stages, from Sensing to Action.

Figure 1. The Safe Behaviour Model Stage 1 Sense – the perception of hazard and hazardous situations The first stage that is necessary involves perception of information about the environment. A person needs to be able to see, hear or otherwise receive information that there is a hazard that requires a response. This perception may be direct, as when one sees a snake, smells a noxious product or feels an unusual vibration that may presage a disastrous event. If there is no such perception then it is, at best, difficult to react to something that does not appear to exist as it is essentially invisible. In addition sensory processes will be subject to habituation, making the visible invisible again. There is a second level of perception, that applies to what we can learn to recognise as hazardous. In some cases this is possible. For instance, a vehicle may be immediately recognisable as dangerous, because it is travelling too fast or is clearly likely to fall apart, but experienced drivers also learn to recognise other, often more critical dangers. An inexperienced driver is sometimes recognisable from the way they drive, drivers with poor attitudes (far more dangerous in terms of the traffic statistics) often show visibly recognisable behaviours such as close following. Notions of perceptual ‘tuning’ apply here, where the perceptual system becomes sensitised to configurations that have become meaningful. Remedial measures at this level would consist of sensitising perceptual systems, inducing perceptual learning, or of undoing the effects of habituation. (Gigerenzer, & Selton, 2001)

Stage 2 Know – understanding which risks deserve attention The raw ability to perceive a hazard or hazardous situation does not, on its own, provide a sufficient impetus to behaviour. While specific hazards might, conceivably, have the potential to cause substantial harm, in a great many cases the potential for harm may (appear to) be negligible and therefore no special action would be merited. What is perceived needs to be regarded as sufficiently important to warrant action and in many cases there may appear to be more pressing needs. This means that hazards have to be understood in terms of their likelihood to cause harm; the hazardous objects, event or situation has to be reframed as a risk, by adding in information about the frequency with which harm may actually be expected. This information can change the weights people assign to possible outcomes, which they can use to determine where to set their priorities. Thus, if a situation is judged to be relatively risk free, in an environment of competing priorities for action, an at-risk behaviour will be more likely to be chosen if it meets more pressing goals. As a matter of experience people are unlikely to be influenced by one or two statements that an action is dangerous if this conflicts with their personal experience over many years that it is usually not. For instance, most people’s personal experience of years of accident free driving often makes adhering to speed limits a losing competitor against their certain expectation of a later arrival. Seen as an intuitive risk judgement they multiply the potential consequences with their personal (and statistically reasonable) expectation that they, as an individual, will not have an accident, this contrasts with the certain outcome of increased journey time. The product of potential outcome and probability, the risk, drives behavioural choices. There is a substantial literature on this area, and it is clear that individual judgements of both potential outcome and perceived probability are open to significant cognitive biases (Tversky & Kahenman, 1974; Kahneman, Slovic, & Tversky, 1982). Stage 3 – Plan which action to implement Just as it is necessary to sense hazards, albeit indirectly, and essential to treat hazards once perceived as sufficiently important to merit taking some action, it is also necessary for safe behaviour that people know what they can do to manage the risks they appear to face. An individual’s behavioural repertoire may include possible behaviours to cope with identified risks, but may also be limited to a restricted set of

actions. What a person confronted with a risk regarded as significant has to do is to decide what behaviours are appropriate to manage that risk. They will have to make a plan of action to be performed. Failures at this level constitute the Rule-Based mistakes identified by Reason (1990), when possible choices of action compete, or Knowledge-Based mistakes when they have no specific behaviours in their repertoire and have to generate a fresh plan of action to solve the problem, being confronted with a clear risk. Wagenaar and Hudson (1997) distinguished two types of planning failure. In the first case an individual, faced with a known problem, may only have a single possible plan of action, which is a source of problems if the plan is inappropriate (“To a man with a hammer, everything looks like a nail”). Alternatively people may generate more than one plan but fail to think the consequences through adequately. Certainly, people under time stress will be more likely to select the first plan that looks as if it might offer a solution (The politician’s syllogism from the popular series Yes Minister goes: ”I must do something, this is something, so I must do this”). Stage 4 – Act and maintain the chosen behaviour The final stage of the Safe Behaviour model involves actually implementing the chosen behaviour, together with the requirement to continue to perform the chosen action in the face of already identified risks. This can prove problematical if there are pre-existing competing behaviours, possibly including non-behaviour (do nothing), because those alternative choices, which are typically seen as at-risk behaviours, are well established. This is the level at which most Behaviour-Based Safety Methods (Geller, 2001; Daniels & Rosen, 2004) are aimed, when people can see the hazards, know the risks, have the necessary plans of action in their repertoires but still perform other, usually riskier, behaviours. This final stage can prove to be problematical even when all the other stages have been passed successfully. One reason may be that frequent reward of at-risk behaviours, without negative consequences, leads to the formation of skill-based behaviour (Rasmussen, 1983) which can by-pass the Know and Plan stages, becoming a learned reflex triggered by environmental conditions, and only monitored occasionally. This analysis suggests that people act with at-risk behaviours because

they are no longer driven by rational thought, but by direct response to the environment within which they find themselves. This is the same conclusion that was reached earlier for different reasons (Wagenaar, Hudson & Reason, 1990) in the development of the Organisational Accident Model (Reason, 1997). The model as presented is based upon a logical analysis of what it might take to behave safely in the face of hazards. Is there any empirical evidence that such an approach forms a good and fruitful basis for ensuring people behave safely? Wagenaar and Groeneweg (1987) analysed 100 major maritime accidents, examining 57 of those in terms of the notion that people cause accidents by taking deliberate risks, similar to the alternative models described above. Their results are shown in Table I, where they came to the conclusion that at most 1% of all the accidents studied were primarily caused by deliberate and explicit risk-taking behaviour. While their categories are not immediately related to ours here, it is clear that 27% of problems were at the Sense level, that finding options – plans of possible action - was a significant factor (15%) and that calculations of risk or of successful outcomes of plans were also substantial issues (36%) that prevented people from doing what, in hindsight, would have appeared to be the ‘right’ thing. These data suggest strongly that, in real life scenarios, people confronting hazards can fail at any point along the route. Table I. Where decisions about risky actions go wrong in 57 accidents at sea, reported by the Raad voor de Scheepvaart (Netherlands Maritime Accident Board) in 1984 and 1985 (some accidents are represented in more than 1 line: from Wagenaar & Groeneweg, 1987) ------------------------------------------------------------------------------------------------------Decision phase

Number of accidents scenarios in which the decision went wrong Absolute number

(%)

_____________________________________________________________________ 1

Receipt of information

17

21

2

Detection of problem

21

27

3

Finding options

12

15

4

Evaluation of consequences

19

36

5

Conscious acceptance of risk

1

1

All of these stages are necessary for a person, faced with a hazardous situation, to perform the appropriate actions. The scientific understanding is not drawn from a single part of Psychology, as the earliest stages reflects Perceptual Psychology, then the inspiration is from work on biases in Cognitive Psychology and Economics, followed by Social and Health Psychology when it comes to the implementation of intentions. At the same time many of the issues covered are also to be found in the area of Organisational Psychology, although there is usually a very different literature in such analyses of the problems encountered in industrial safety, such as we are discussing here. One point that should be stressed is the complexity of the intellectual problem faced by ordinary people in hazardous situations, especially when they are already so well defended that a safe outcome is usually guaranteed. Shipping operations are typically simple, and the safety measures rudimentary. Oil and gas exploration and production, in contrast, is usually complex and highly defended with a wide range of overlapping safety measures. Wagenaar and Hudson (1997) showed how many unsafe acts were required to have a major accident in the two different industries; the results are shown in Table II. While it took 2-3 unsafe acts (mean = 3.4) to lead to a major shipping accident, Table II. The number of unsafe acts necessary to cause an accident. Number of

100

unsafe acts

shipping accidents

21 accidents in oil & gas exploration & production

_________________________________________________________________ 0

4

0

1

3

0

2

46

1

3

22

2

4

14

2

5

5

1

6

2

1

7

2

1

>7

2

13

more than 50% of the oil and gas accidents required in excess of 7 contemporaneous unsafe acts (mean = 8.0). In the oil industry, therefore, an individual unsafe act will almost always not lead to an undesirable outcome, as an unforeseeable combination is usually required for a major accident actually to occur. This is the message of Reason’s (1990; 1997) Swiss Cheese model. Under these conditions people are likely to downplay the potential negative consequences, actually reflecting their personal history of managing the risks effectively, so that 20 years of personal experience that a hazard has not become a problem for an individual weighs more heavily than someone else’s advice that it is very dangerous. It is, of course, very different, seen from the vantage point of a corporate centre, when millions of man hours are aggregated, as opposed to the viewpoint of an individual who will, in a well defended high risk industry, probably never personally experience a major accident such as a fatality. The view from the centre is that an incident that occurs once in several million working hours is actually quite frequent when the total world-wide workforce may accumulate a million man hours in a single day. Nevertheless any one individual only works about 80,000 hours in a lifetime, so it takes 13 people to accumulate in their lifetimes the million hours the company achieves in a day. In short an individual may never actually experience a serious accident at first or second hand, but even a very safe but large company will be confronted with the negative consequences of unsafe behaviours on a daily, or at least weekly, basis. Interventions - Using the model to generate safer behaviours A scientific model can be constructed simply to explain a particular phenomenon. The Safe Behaviour model makes the different stages explicit and allows us to understand why people might behave in ways observers, often blessed with hindsight, would find odd and even dangerous. This model was, however, constructed primarily to provide a framework for intervention, to prevent people from harming themselves and others. It underlies two intervention tools in the suite of nine Hearts and Minds tools, Working Safely (Shell, 2003) and Driving for Excellence (Shell, 2004), designed to eliminate industrial accidents in two specific areas, unsafe working practices and on the road. These two tools operate by inducing biases in favour of safe behaviour. To support individuals the model stresses that each of the stages can be supported by processes of looking, telling and listening involving (i) learning to look at a hazardous world and

recognise danger, (ii) learning how to tell people that what they are doing is unsafe, and (iii) teaching people to accept such information without being offended or feeling patronised. Another set of processes have to support the continuation of each stage, so that people do not habituate perceptually, continue to rate risks highly despite their actual personal daily experience, maintain and develop their repertoire of possible actions and, finally, continue to behave in ways appropriate to the hazard. As it is in everyone’s interest to ensure that people are supported in their efforts to act safely, it is vital to ensure that the personal pay-offs for safe behaviour outweigh those for unsafe behaviours. The next section examines how a particular problem identified in many incident investigations, a lack of situation awareness, can be understood in the context of the Safe Behaviour model. Situation Awareness Numerous theories present in the domain of psychology all play their role in particular stages of the model. However, one particular psychological construct that maps neatly on the Safe Behaviour model is of situation awareness. In order for humans to function effectively in daily life without mishap, they must have at least some sort of rudimentary understanding of their surroundings in order to act accordingly and successfully avoid any potential accidents. It is obvious that one critical factor in preventing accidents should be maintaining an adequate understanding of the current situation. This is needed in order to perceive the conditions of the environment, and judge the consequences of any actions taken in relation to the safety of the work, in order to avoid adverse events. By having full and correct understanding of the situation, the potential risk involved in an action can more effectively be gauged and in turn minimised, thus reducing the risk of an accident.

However, if the

understanding of the situation is impaired, then the ability to predict the outcomes of actions is more flawed as it is based on less accurate information, and due to this the risks of an accident occurring are increased. This is supported by Orasanu, Dismukes & Fischer (1993) who found that pilots who had experienced an accident had a tendency to interpret cues wrongly, and also in many cases underestimated the risk that was associated with a particular challenge, thus over-estimating their perceived

ability to deal with the situation.

Consequently, the question is: what is this

mindfulness that we all appear (to a greater or lesser extent) to possess which allows us to have successful passage through our day, avoiding most potential accidents? How is it that people garner and use information from their constantly changing environments to keep operations safe? In short, the answer to these questions is ‘Situation Awareness’ (SA). Definitions of situation awareness vary greatly, but the most cited definition is provided by Endsley (1988, p97), who explains it as “...the perception of the elements in the environment within a volume of space and time, the comprehension of their meaning, and the projection of their status in the near future” . In simple terms, SA is the ability to successfully pay attention to and monitor the environment, and essentially ‘think ahead of the game’ to evaluate the risk of accidents and ensure a safe working environment. Sarter & Woods (1991, p50) state that situation awareness “…is based on the integration of knowledge resulting from recurrent situation assessments”, i.e. by continually appraising the situation and incorporating facts from it, situation awareness is derived.

These so-called ‘situation assessments’ comprise of three

levels: perception, comprehension/information integration, and projection (Endsley, 1995). Vuijk (2004) mentions that when people are unaware of their unsafe working practice, that there is a problem with interpretation of information, or indeed the senses themselves, whereas if the issue regards bad habits, then the stage requiring attention is the planning/acting step. Let us consider how SA can help to explain why people behave unsafely. Referring back to the Safe Behaviour model, the three levels of SA clearly correspond to the sense, know and plan stages: ‘sense’ equating to ‘perception’, ‘know’ referring to ‘comprehension’, and ‘plan’ being associated with ‘projection’. The parallels between these stages and the SA levels are perhaps more clearly understood by examining them in more detail. Level 1 SA: Perception

This is the basal constituent of SA: in order for

SA to be achieved, objects and information in the surrounding environment (i.e. stimuli) must be noticed. The stimuli must be mentally processed to determine to which ones attention should be paid in a work task, and the surroundings should be continually monitored to attend to changing stimuli. This is very closely aligned to

the Safe Behaviour stage of ‘sense’ which regards the perception of the hazard. This stage does not involve merely seeing the hazard with our eyes, but we should also employ our other senses, for example smell (for the perception of dangerous gases in the atmosphere), touch (to feeling unusual vibrations) and our hearing (to avoid heavy machinery which we cannot see). Attentional processing (La Berge, 1995; Schiffrin & Schneider, 1977) is intrinsically linked to the theory of SA, but attention is bound by the limits of the working memory (Baddeley & Hitch, 1974). The fact that attention is limited can be a problem, as a person is unable to attend closely to every single detail of his/her environment. This is an important issue to consider in safety critical environments, given that it is often the case whereby factors outside the central focus of attention are those that have the potential to be harmful. Consequently, critical elements may be missed in the observation/perception stage, leading to an incorrect mental model (the representations of objects, people and tasks that people hold in their minds of the understanding of the various roles and relevance of the items concerned) being formed. Possession of a poor or incorrect mental model can increase accident risk as there is no ‘template’ to guide actions. This is further supported by Klimoski and Mohammed (1994, p405) who state that mental models “…provide a conceptual framework for describing, explaining, and predicting future system states”, and so if this was to be impaired, then consequently the resultant SA would similarly be impaired. In summary, sense and perception relate to the same psychological construct, as they both involve paying attention to the environment, and ‘picking up’ on the information relevant to the task. The situation should be monitored in order to keep track of and perceive any changes to the situation (should they occur), as this will affect the workers’ SA. Level 2 SA: Comprehension

This involves the combination, interpretation,

storage and retention of the aforementioned information (Endsley, 2000b) to form a picture of the situation whereby the significance of objects/events are understood, essentially deriving meaning from the elements perceived.

The degree of

comprehension that is achieved will vary from person to person, and Endsley (1995) maintains that the level attained is an indication of the skill and expertise held by the operator. Applying this level specifically to the Safe Behaviour model stage of

‘know’, it means understanding how the information that you’ve perceived (be it, for example, visual information from seeing something, or information from other senses such as smelling gas) relates to you and what you are doing, and how the information can help to influence your actions to behaving in a safe manner. Success at the ‘know’ stage can only occur of the meaning assigned to the information is correct. For example, the information can correctly be perceived (i.e. success has been achieved at the ‘sense’ stage), but its importance or meaning to the overall situation or goal can be misunderstood (Endsley, 1999). This is often the case when a situation is encountered for the first time, and a worker’s expectation of the state of affairs is not known. For example, a worker may see a reading on a dial in the control room of an offshore platform, thus has perceived the information, but does not understand how it connects with the task he is performing, and so has not achieved the correct comprehension level relevant to his work. Also, it may be that certain objects or events can be classed as hazards in one situation, but harmless in another, and so this also requires correct comprehension (or knowing) in order to make the distinction between them and act accordingly. Level 3 SA: Projection

The final level of SA is projection, which occurs

as a result of the combination of levels one and two.

This stage is extremely

important, as it means possessing the ability to use information from the environment to predict possible future states and events (Endsley, 1988, 1995; Sarter & Woods, 1991). Having the ability to correctly forecast possible future circumstances is vital, as time is made available to dispel potential discords and formulate a suitable action course to meet goals (Endsley, 1995, 2000; Stanton et al, 2001). Endsley (2000b, p7) states that “…experienced operators rely heavily on future projections” as this “allows for timely decision making”, adding that “It is the mark of a skilled expert”. Just as level 1 SA (perception) mirrors the Safe Behaviour model’s stage of sense, and level 2 SA (comprehension) corresponds with stage 2 in the model (know), the third level of SA (projection) maps perfectly onto the third stage of the Safe Behaviour model – plan. In accordance, whilst projection involves using the information you have assimilated during the previous stages of SA to prepare an appropriate plan of action, ‘plan’ entails doing likewise.

By understanding how the information is

associated (both directly and indirectly) with the work being performed, then workers can use the data to plan not only for their immediate work, but also for future work,

and can develop appropriate strategies and working patterns. In addition, workers can use the information to aid in the planning of colleagues’ work as well. In other words, as long as the information is shared, we can warn others about the hazards and dangers associated with particular objects, situations or processes of which other workers were possibly not aware. The fourth stage of the model is act and maintain. This stage is essentially about keeping doing the safe behaviours that were planned in the previous stages in order to manage the hazards in the workplace. This can be considered in conjunction with process occurring throughout the process of achieving safe behaviour – Look, speak and listen (sharing information with others, and listening to the information they have which can help your situation) and maintaining all the stages.. When taken as a whole, it overall sums up the modus operandi of situation awareness, as the understanding of a situation is being repeatedly updated through continual situation assessments (Endsley, 2000b).

These situation assessments are the cognitive

processes of gaining and interpreting new knowledge, be it knowledge gained from the worker own senses, or information shared from colleagues. The new knowledge can then be applied to feed into the representation that workers have of the current situation, continually updating their awareness. This feedback loop is essential to maintain accurate SA, as new information must be incorporated, as workplaces in safety critical environments rarely remain static. The importance of good quality SA in maintaining a safe working environment can be seen when we look at the human element root causes of accidents in safety critical environments. As stated previously, lack of care and attention has been found as one of the main reported human factors causes of accidents in the oil and gas industry (Mearns, Flin, Fleming & Gordon, 1997; Flin, Mearns, O’Connor & Bryden, 2000). Since attention is central to SA, this may provide a focus area for addressing some of the main immediate causes of incidents, this should then be followed by a deeper investigation of the reasons “why” as discussed so far in the paper, in order to address the underlying causes.

Errors in SA can occur at any of the three levels: At the base level of SA (perception, or ‘sense’ in the Safety Behaviour model), errors can easily occur, most notably arising from the reality that some element of the environment has been discerned incorrectly, or perhaps even not at all. This may be the fault of the individual concerned (i.e. he/she simply did not pay enough attention, and so did not notice the event/action/object), or it may have been a characteristic of the environment (e.g. an object was obscuring the particular feature that was missed from view) (Endsley, 1995; 1999). While this means that there will be enhanced SA for those elements of the environment to which attention was allocated, the resulting SA for the entire situation will be poorer, as the representation that the worker holds will not have incorporated information from elements which received little/no attention. Prior expectations about how a particular situation should be can also have an impact here. It may be that workers are so used to working in a particular environment, doing a particular task, that the work is so routine that they become complacent, feeling that they already know exactly how things should ‘turn out’. This can easily lead to information being missed, as full attention is not being paid to the situation. Errors in Level 2 SA involve the failure of certain aspects of the situation to be correctly understood in relation to the targeted goals, i.e. there is an error at the ‘know’ stage. For example, the information can be perceived accurately by those concerned, but its importance or meaning to the overall goals desired is not adequately appreciated (Endsley, 1999). This is often the case when a situation is encountered for the first time, and an expectation of the state of affairs is not known. The worker may therefore perceive all of the information available, but not recognise the relevance of it, so not integrate it properly to his/her awareness. The final stage at which an error to SA can occur is at the level of projection (plan). At this level, future occurrences need to be anticipated in order for the best and most effective course of action to achieve the goals, deal with hazards, and maintain a safe environment to be decided upon. The most common cause for failure of SA at this level is simply due to the fact that some people appear to be very poor at mental projection (Klein, 1989b; Endsley, 1999). The prediction of future events and actions is a task that places a great deal of demand on both the attentional and memory

systems of the brain, and so those with a poor mental projection will inevitably have more trouble in this task. It may be that workers do not think ahead of their tasks and only concern themselves with what is happening currently, thus do not project at all, or alternatively do attempt to forecast situations, but the schemes that they develop do not account for all possible eventualities. Previous analyses of the causes of accidents have shed some light on the role of SA in incidents. Recent accident analyses by Jones and Endsley (1995, 1996) support the notion that SA is a major issue in aviation accidents. They investigated aviation incidents, analysing data contained within the Aviation Safety Reporting System (ASRS) database from January 1986 to May 1992, searching for the term ‘situational awareness’, and found 143 incidents that met the criteria for the study (i.e. contained enough information, cited problems with SA, and were submitted to the database by either the pilot or controller involved).

It was discovered that of the accidents

attributed to situation awareness error, 76.3% were categorised as perception errors, 20.3% were comprehension errors, and 3.4% were projection errors. Of the perception error, the majority were found to be due to the operator not attending to all the required information, and thus did not perceive (or sense) the information in the first instance. A recent study by Sneddon, Mearns, Flin and Bryden (2004) also found a very similar pattern in oil and gas exploration and production. They investigated a multinational oil and gas exploration company’s accident database of drilling incidents (all well operations, onshore and offshore) occurring within a ten-month period, and found that where SA was the root cause (135 incidents), the incidents largely were due to problems with perception. In parallel with Jones and Endsley’s research, 66.7% of incidents were due to errors in perception (sense), and 20% were found to be triggered by problems with comprehension (know). This research shows that the vast majority of incidents appear to have their cause rooted in errors occurring at the perception and comprehension levels, or the sense and know stages. These results corroborate Wagenaar and Groeneweg’s (1987) study reported in Table I, but there may be a difference related to the differences found in Table II. Workers in the oil and gas industry may be much better trained to start with, having a better knowledge and planning ability, so that their failures load much more heavily on the initial sense component, as found by Sneddon et al (2004). It therefore appears that

people are essentially not even picking up on the information in the first instance, or perhaps doing so but then misunderstanding it’s importance, or immediately disregarding it, thus not taking it into account in their work practices. In order to increase safe behaviour, the areas of sense and know must be focused upon to facilitate increasing perceptions and understanding of the hazard in the first instance. Conclusion While the SA framework provides a fundamental base for the Safe behaviour model, it is not the only theory that has been drawn upon in order to understand why people behave safely or unsafely. Many other theories and models found in psychology play an important role in explaining human behaviour in the workplace, nevertheless the model described here represents a much more unified approach, synthesising knowledge garnered from the whole gamut of Psychology. There appears to be some empirical evidence that the model is valid for more than just theoretical reasons, as well as making it clear what kind of interventions can be used to increase safe behaviours in hazardous but well defended systems.

References Baddeley, A. D., & Hitch, G. J. (1974). Working memory. In G. H. Bower (Ed.), The psychology of learning and motivation (Vol. 8). London: Academic Press. Daniels, A. C., & Rosen, T., A. (2004) Performance management: Changing Behaviour that drives organizational effectiveness. Performance management publishers Endsley, M. R. (1988). Situation Awareness Global Assessment Technique (SAGAT). Paper presented at the Proceedings of the National Aerospace and Electronics Conference (NAECON), New York. Endsley, M. R. (1995). Toward a theory of situation awareness in dynamic systems. Human Factors, 37(1), 32-64. Endsley, M. R. (1999). Situation awareness and human error: Designing to support human performance, Proceedings of the High Consequence Systems Surety Conference. Albuquerque, NM: Sandia National Laboratory. Endsley, M. R. (2000a). Situation models: An avenue to the modeling of mental models, Proceedings of the 14th Triennial Congress of the International Ergonomics Association and the 44th Annual Meeting of the Human Factors and Ergonomics Society. Santa Monica, CA: HFES. Endsley, M. R. (2000b). Theoretical underpinnings of situation awareness: A critical review. In M. R. Endsley & D. J. Garland (Eds.), Situation Awareness Analysis and Measurement (pp. 3-33). Mahwah, NJ: Lawrence Erlbaum Associates. Flin, R., Mearns, K., O'Connor, P., & Bryden, R. (2000). Measuring safety climate: Identifying the common features. Safety Science, 34, 177-192. Geller, S, E. (2001) The psychology of safety handbook CRC Press Gigerenzer, G. & Selton, R. (2001) Perceptual tuning learning Bounded reality: The adaptive toolbox. Cambridge Massachusetts, MIT press Hoyos, C., (1995). Occupational safety: Progress in understanding basic aspects of safe and unsafe behaviour. Applied Psychology: An International Review, 44, 233-250 Jones, D. G., & Endsley, M. R. (1995). Investigation of situation awareness errors. Paper presented at the Eighth International Symposium on Aviation Psychology, Columbus, OH. Jones, D. G., & Endsley, M. R. (1996). Sources of situation awareness errors in aviation. Aviation, Space and Environmental Medicine, 67(6), 507-512. Klimoski, R., & Mohammed, S. (1994). Team mental model: Construct or metaphor? Journal of Management, 20(2), 403-437.

Kahneman, D., Slovic, P., & Tversky, A. (1982) Judgment under uncertainty: Heuristics and biases. New York. Cambridge University Press, La Berge, D. (1995). Attentional processing: The brain's art of mindfulness. Cambridge, MA: Harvard University Press. Lawton. R. (1998) Not working to rule: Understanding procedural violations at work. Safety Science, 28 77-95 Mearns, K., Flin, R., Fleming, M., Gordon, R. (1997). Human and organisational factors in offshore safety ( OTH 543). Suffolk: HSE Books. Orasanu, J., Dismukes, R. K., & Fischer, U. (1993). Decision errors in the cockpit. Paper presented at the Proceedings of the Human Factors and Ergonomics Society 37th Annual Meeting, Santa Monica, CA. Rasmussen, J. (1983). Skills, rules and knowledge: Signals, signs and symbols, and other distinctions in human performance models. IEEE Transactions and Systems, Man and Cybernetics, 3, 257-268. Rundmo, T. (1995). Perceived risk, safety status, and job stress among injured and non-injured employees on offshore petroleum installations. Journal of Safety Research, 26(2), 87-97. Rundmo, T., Hestad, H., & Ulleberg, P. (1998). Organisational factors, safety attitudes and workload among offshore oil personnel. Safety Science, 29, 75-87. Sarter, N. B., & Woods, D. D. (1991). Situation awareness: A critical but ill-defined phenomenon. The International Journal of Aviation Psychology, 1(1), 45-57. Shell International Exploration and Production (2003) Working Safely. Energy Institute. London. http://www.energyinst.org.uk/heartsandminds/worksafe.cfm

Shell International Exploration and Production (2005) Driving for Excellence. Energy Institute. London http://www.energyinst.org.uk/heartsandminds/driving.cfm

Shiffrin, R. M., & Schneider, W. (1977). Controlled and automatic human information processing: II. Perceptual learning, automatic attending, and a general theory. Psychological Review, 84, 127-190. Sneddon, A., Mearns, K., Flin, R., & Bryden, R. (2004). Safety and situation awareness in offshore crews, Proceedings of The Seventh SPE International Conference on Health, Safety, and Environment in Oil and Gas Exploration and Production. Richardson, TX: SPE. Stanton, N. A., Chambers, P.R.G., J. Piggott. (2001). Situational awareness and safety. Safety Science, 39(3), 189-204.

Tversky, A., & Kahenman, D. (1974) Judgement under uncertainty: Heuristics and biases. Science, 185, 1124-1131 Vuijk, M. (2004) Safe behaviour at work. Leiden University. Wagenaar, W.A., & Groeneweg, J. (1987). Accidents at sea. Multiple causes and impossible consequences. Journal of Man-Machine Studies, 27, 587-598. Wagenaar, W.A. & Hudson, P.T.W. (1998) Industrial Safety. In P.J.D.Drenth, H.Thierry & C.J.de Wolff (Eds.) Handbook of Work and Organizational Psychology. Psychology Press, Hove, England. pp 65-88 Wagenaar, W.A., Hudson, P.T.W., & Reason, J.T. (1990). Cognitive failures and accidents. Applied Cognitive Psychology, 4, 273-294