Understanding Safety / General
Decision Making is the central focus of Human Factors. A good decision will save the day whilst a bad decision can lead to catastrophe. In order to better understand decision making, we need to explore the Human Information Processing system, memory, workload management and situational awareness. We should remember that when discussing Human Factors, contrary to engineering where we have a causation model, i.e same inputs will always give the same outputs, human processing is more probabilistic.
Information processing has 3 distinct stages. Like any information processing, the human information process is comprised of input, processing and output. Input is our senses – vision, smell, taste, hearing, feeling (touch). Most of the inputs we receive come from vision and hearing. These are further called the iconic and echoic memory respectively. Echoic memory lasts longer, to a maximum of 4 seconds. That is, if we hear a sound without processing it, we will forget it after 4s. Iconic is even less!
Christopher Wickens created a model of the Human Information Processing to illustrate the flow from input to output (decision).
Receptors – attention mechanism – perception – working memory – long term memory – motor programmes – response
Receptors are the stimuli, the senses. But our senses pick thousand of information per moment. We therefore need to filter them through the attention mechanism. Attention mechanism is a precursor to all other cognitive functions and it filters data in order to distinguish between important ones and less significant ones, thus allocation the resources accordingly. Human Information Processing has a bottleneck point, in this exact process, as we are unable to process all the information simultaneously.
Attention is selectable, shiftable and divisible, based on practise and level of arousal. Attention has four distinct areas: selective, focused, divided, sustained. Some argue that the information is not filtered out entirely (blocked), but rather attenuated. The cocktail party effect is a phenomenon which demonstrates this, where being in a noisy room engaged in a conversation at one side of the room with specific people, you suddenly hear your name at the other side of the room and you shift immediately your attention. This is very profound for pilots when they pick their own callsign in a busy radio environment.
Selective attention is what helps filter the received information and process them accordingly. Perception is all bout how we deal the received information. We can ignore, respond to, consider or remember the info, or activate other process. Sensation precedes perception, as the information is not yet processed. The contradiction and difference between sensation and perception is what leads to illusions. Hence the information will pass on the Working Memory where the central Decision Making is taking place. The effectiveness of the Working memory is dictated by the level of the brain arousal. Through the process 3 supportive function will assist the flow.
Motor Programmes: These are autonomous physical responses that feed directly to a response. They are associated with learnt, developed and practised skills (e.g. driving a car). To establish a Motor Programme one must go through 3 stages. Cognitive stage where as the name suggest you have to think about it (knowledge based), then through the associative stage where reactions come after making the association with learnt or past knowledge and finally the autonomous stage where no processing is required by the brain. As motor programmes pass the working memory straight to a response, by acquiring more of these, we increase the Human Information Processing by releasing more resources from Working Memory.
Short term memory stores information just as long as they are needed and then can be forgotten. It is relatively restricted (7 items +/- 2) or up to 20 (if using techniques like association and chunking).
Long term memory if the hard drive of the brain. It is argued that once information passes into LTM it is retained for ever. In order to pass information from STM to LTM we need to meaningfully and mentally process them using techniques (most common one is by repetition). LTM is divided into semantic, procedural and episodic memory. As the names imply, information is stored in LTM because of semantics (facts, principles, rules), procedures (skills, habits) or an episode (happy, dramatic, scary, shocking). LTM has an important function the Human Information Processing, which is retrieval. What we refer to as forgetting is basically a failure of the retrieval function. Retrieval from LTM can greatly altered our Decision Making.
The interesting part of the Human Information Processing is that although of limited capacity, it has the ability to combine inputs and stored information to produce novel solutions! HIP is not a passive process. Even though we don’t consciously think of the process, it remains an active and dynamic one.
Due to the bottleneck structure and the limited capacity of the Working Memory, DM can be affected by the workload level (especially mental workload). Mental workload can de defined as the cost in cognitive resources for the performance of a task. The increase of automation has relieved the pilot from Motor Programmes tasks and has shifted at the same time the resources to the increased mental workload as the systems need to be managed. This places more emphasis on the cognitive skills. For pilots, six basic workload functions have been defined (by FAA): flight path control, collision avoidance, navigation, communication, operation and monitoring of systems and finally the command of decisions.
Mental workload can be measured in 4 different approaches. Analytical, behavioural, physiological and subjective. However, none of the approaches can satisfy all the criteria for correct evaluation. This is on the reasons we are having trouble in fully defining and explaining mental workload.
Mental workload is similar or even more demanding than physical workload. We observe an increase in energy expenditure, by increase of heart rate and change of pupils of eyes It is common to see physical exhaustion from mental workload. This issue is even more elevated when considering the dramatic increase of mental workload during a flight, as during the cruise the pilot is in a phase of mental underload and during the approach and landing has to sometimes transit directly to mental overload. Mental underflow is NOT the same as low workload. In low workload there is a limited use of capacity, whilst in underload we observe a shrinking of the attention capacity!
Decision Making can be also affected by Situational Awareness. Situational Awareness is a form of Human Information Processing as it involves a continuous gathering of information on environment and surroundings which integrate with previous knowledge. This will then form a coherent picture, to help perceive the current situation and anticipate future events. Situational Awareness lives the Working Memory and is the precursor to all Decision Making. Situational Awareness is a function of available situations (facts & data), perceived situations, expected situations and the inferred situations. Practically S.A. is a 3-step process: perception of elements, comprehension of current situation and projection of the future status.
It is important to remember that in the aeroplane there is a team of at least 2 pilots and a number of cabin crew. Therefore S.A. of each individual might be different. We can have situations of shared, common amongst everyone, S.A., or simply overlapping S.A. where only some elements are shared by the whole team. Open communication and information sharing is probably the easiest way to merge the individual S.A. into a shared one. A team S.A. can be also a distributed S.A. which differs from a shared one, in that each team member is more S.A. in his/her own responsibilities of the flight but these different requirements and purposes are compatible with the other members.
Decision Making is therefore a systematic approach to mental process in order to reach an action/ response. In critical situations D.M. is very tricky as apart from time pressure and risk, one is also faced with uncertainty. It is therefore, due to our limited cognitive capacity, the a simply satisfactory decision is sought rather the optimum one.
Expert naturalistic D.M. is not about generating a list of possible alternatives and arrange of options and then compare them to evaluate the likelihood of an outcome. It focusses on understanding the situation. Then based on familiarity a viable solution is chosen. If at any point the solution seems unsuitable then you move to the next plausible option. So you don’t create multiple options from scratch but sequentially form a plan of action. The process becomes easier by diving this plan into smaller sub-tasks. A significant problem arises when the problem is not well-defined leaving too much room for individual interpretations and too many plausible options.
Many models have been created to aid the pilots make a structured and well examined decision when under pressure. It is interesting to see the closed loop of most models (i.e. the checking/ evaluating of the plan as it unfolds)
SHOR: Stimuli – Hypothesis – Options – Response
FORDEC: Facts – Options – Risks – Decision – Execute- Check
DESIDE: Detect – Estimate – Set (Objectives) – Identify – Do – Evaluate
DODAR: Diagnose – Options – Decide – Assign - Review
SHOR might be more suitable during time critical phases whilst DODAR & FORDEC for knowledge-based decisions.
DM should never be evaluated on the outcome (hindsight) but on the stages and the process followed.
Other decision models for various tasks:
Eisenhower matrix – how to work more efficiently
Consequences model – on making prompt decisions
Conflict Resolution model –
Supermemo model – Increasing LTM
Pareto Principle – 20/80 rule
Black Swan Model
Black Box Model – Faith Vs Knowledge decision
Drexler – Sibbet – Team performance model