[author: Dr. Udo Schultheis]

Introduction

Humans as we are, “Homo sapiens,” can be traced back around 300,000 years. Our large and complex brains work continuously to develop tools, devices, and complex technologies. We build teams, start families, live in social structures and organizations, and interact apart, with, and also against each other in our environment. As humans we do a lot of great things; however, we are not perfect. Some of our actions can take us in unexpected direction or toward an outcome we did not plan; we can make mistakes and errors. Homo sapiens have an enormous power to learn. We can analyze what we did or did not do, learn from each other, and do better the next time.

There are a lot of definitions for “human factors” (HF), and myriad opinions, publications, and other data sources. Every definition differentiates from others depending on the viewpoint of the author. At the end, HF is nothing more than an empirical “takeaway” from an endlessly cumulated body of knowledge. Empirics tell us “how it is” and not how something ought to be. Our individual and social behavior is specific to us humans and influences the functioning of simple or highly technological “systems” in the environment. Human behavior “is what it is.”

Understanding Human Factors

Modern HF is an applied science based on psychology with all its concepts and constructs stemming from behavioral and cognitive psychology (e.g., the law of effect[1], conditioning, attention, perception, decision-making, workload, stress, situational awareness, and more), psychophysics (the relationship between physical and mental qualities), psychophysiology (e.g., emotions), and more.

It is important to understand, in HF, what is meant by a “system.” In short, a “system” represents the physical, cognitive, and organizational artifacts that people interact with. More specifically, a system can be a technology, a medical device, a software; furthermore, a person, a team, an organization; a procedure, a policy, a guideline; or a physical environment. As used in HF, interactions between people and a system are “tasks” (for example, driving a vehicle).

HF can have numerous applications. It can help with advancing technology by determining how the technology can fit the needs of people controlling it (e.g., pilots flying an airplane), using it (the passengers of an airplane using a seat and an in-flight entertainment device), or designing it (for human well-being and organizational success). Furthermore, HF can train people in teams, organizations, etc., to work efficiently together, help to identify their abilities and limitations under normal, abnormal, and in emergency situations (e.g., medical personnel, operators of complex technologies, etc.) and help to prevent incidents and accidents.

Each of the above psychological concepts underlying HF—human attention, memory, experience, and others—is highly complex. One single concept determining human behavior is extremely scarce and almost impossible to find. Instead, at almost all times several concepts are mingling to produce a specific human behavior. This will be illustrated more specifically below.

What is Cognition

The concept of cognition can be divided into three separate but related tasks—noticing, interpreting, and deciding. Many people assume that “seeing” is an objective, complete, and passive process that simply happens when a viewer looks at a scene such as that which is produced with a camera. This would mean that seeing is just the light (i.e., electromagnetic radiation) entering the viewer's eyes just as light enters the lens of a camera. This assumption, however, reflects a fundamental misunderstanding of the nature of human seeing. Key points of “seeing” are:

1. Light from a source goes directly from the source to the human eye.
2. The eye’s light sensitive layer, the retina, is stimulated.
3. Through chemical processes on the retina the light is converted into electrical impulses.
4. The electrical impulses are propagated and travel up through specific pathways to the human brain.
5. As a result, “seeing” occurs, i.e., a human consciously perceives something.
6. Most importantly, we don’t “see” the retinal image which is immediately converted to nerve impulses. After all, there is no picture in the head and no inner viewing screen.

Cognition is the concept behind seeing which interprets the electrical impulses and tells the viewer about the world. Most interestingly, the human viewer is unaware of this extremely fast transition from noticing to interpreting. For us, it seems interpretation happens simultaneously with noticing. If it would not work in this way, we would not be able to walk or run and definitely would not be able to drive a vehicle, as scenery would change too fast, and the interpretation of what is going on would always happen in the past. Instead, even better, our brain can process an interpretation even if the sensory input is low, novel, ambiguous, or caught on the periphery of the retina. A possible gap which can occur between noticing and interpretation is filled by the brain with expectations and/or experiences from prior interpretations. For instance, when driving a car, an object on the side of the road can initially be interpreted by the driver through the above-mentioned process as a mailbox. However, with more “noticing”, and resulting “interpretation”, the earlier “mailbox” can all of the sudden become a pedestrian, and this pedestrian’s movement is interpreted as having the intention to cross the road.

What is Reasoning

This leads us to another complex cognitive concept, decision-making or “reasoning.” Reasoning operates on the prior perceived meaning of something, a situation. It helps us to understand what is happening, predict the future, and determine our course of action., i.e., in the mailbox/pedestrian example, the driver might decide to apply the brakes to avoid a collision with the pedestrian.

A possible different outcome of the mailbox/pedestrian example can be that the driver circumnavigates the pedestrian or accelerates to pass the pedestrian faster. Whatever the driver decides in this situation, it makes sense for him at this time, and, retrospectively, if nothing undesirable happened, it was successful.

When stopping for a red light several minutes later at the next road intersection, the mailbox/pedestrian episode is vanished from the driver’s remembrance. It did not make its way from the short-term memory (aka working memory) to deeper structures of the brain called long-term memory. It does not play a role for the driver anymore. This fortunate outcome (being unimportant and, therefore, not “stored” in our brain) would have changed drastically if the driver would have had a collision with the pedestrian. The experienced collision would have been further processed by the brain and passed through the “bottleneck” between the short-term and long-term memory. In the reconstruction of the collision, human factors would be involved to find an explanation for what happened. Maybe the driver was changing the radio station at the time prior to the collision (when he initially recognized the pedestrian as a mailbox) with the effect of having a delayed response to the actual pedestrian situation, or the pedestrian was texting on his phone, was distracted, and was trying to cross the street without paying attention to the approaching vehicle. Explanations can be manifold; it’s up to the human factors expert to reveal the details as they really were at the time of the event.

Fallacies and What Is Not Human Factors

So far so good. We could stop here, but some important information would be missing. Knowing “what something is” is only one side of the coin. The other side reveals “what it is not.”

In the mailbox/pedestrian example, matching the context-specific particulars (the driver focusing on the radio searching for a different music channel, or the pedestrian focusing on text messaging on the cellphone) of a collision with the mentioned human factors explanation seems straightforward.

However, the praxis of human factors, especially when experts are called in for delivering explanations, can be full of human experts’ fallacies. The HF expert is human, too, after all.

Hindsight Bias

The most common fallacy is an extremely persistent one; it is known as “hindsight bias”, mistaking or mixing our own reality with the one that was surrounding people at the time of an incident or accident. Retrospectively, we always know more about an incident or an accident than the people who have been in the situation. Hindsight bias allows experts to easily pinpoint what people missed and, even more so, what they should not have missed and what they should have done instead. However, it needs to be understood that taking the retrospective position is not human factors.

Moreover, when trying to trace a sequence of events back to the outcome, which is already known to the retrospective viewer, we will always identify some joints and fork junctions where it is retrospectively obvious that people involved in an accident had an opportunity to revise their judgment of a situation but failed to do so. The driver in our mailbox/pedestrian situation should have focused on the road and not on the radio. He was also given the choice of making an avoidant steering movement before hitting the pedestrian but perhaps did not take it. We are talking here about “counterfactuals,” easy “binary” choices contributing powerfully to the hindsight bias. These kinds of judgments unfortunately make their way into many accident analyses.

Counterfactuals

Counterfactuals often require an (HF) explanation too (“Why did the pedestrian not stop while texting?”). Failing to use the right choice before an accident makes it easy to judge someone by comparing the choice(s) he made to a set of standards, rules, procedures, and other human artifacts (i.e. “Why did my football team not take the obvious chances that would have led to a win? It was so easy…”). The practice of judging people for not doing what they could have or should have done does not explain anything and is not human factors.

One of the most interesting findings within human factors is that in most incidents and accidents people were doing exactly that what they would usually do. Moreover, they were doing this repeatedly and successfully. People do what makes sense to them given the perceived situational indications, organizational norms, and social or other pressures at the time.

Mishaps, incidents, or accidents, in the majority of cases, are not the result of strange or even erratic individual behavior. Instead, they are the result of influences on continuously ongoing human reasoning at a given time in a specific space. Furthermore, human errors are rarely about the people who committed them; rather, they are coupled to a series of combinations of circumstances, tools, and tasks (as mentioned above, “tasks” are interactions between people and a system or several systems).

Information-Processing Method vs the Context-Oriented Approach

Some experts follow a traditional information-processing approach for explaining specific human performance difficulties. It needs to be understood that this is a fine line and can be a shift from using the proven context-orientated approach to a concept/construct/theory-dependent approach. The problem with that is that most people they work with (like lawyers, judges, juries) did not study psychology and don’t know what it means when an expert says that a driver of a car lost “situational awareness”, or the attention of a pedestrian was limited by “inattentional blindness.” Sometimes, especially highly educated experts use such terms to intimidate non-experts. However, this can easily slip into so-called “pseudoscience,” which is drawing supposedly logical conclusions but unfortunately leaving the strong analytical connection between those conclusions and the actual data behind. Besides the fact that we do not know what was really going on in the mind of the driver or the pedestrian introduced in our mailbox/pedestrian example, it does not explain anything, and this is also not human factors in the understanding of this paper.

To clearly understand, a short example described by Dekker (2005) might help:

After an aircraft crash, the accident investigators always analyze available data from the aircraft’s black box (i.e., flight parameter, like speeds, altitudes, and many more) and the cockpit voice recorder. When investigating a fatal crash of a commercial airplane the investigators (Aeronautica Civil, 1996) noticed that one pilot in the cockpit asked the other one something like, “What is going on?” and “What shall we do?” The conclusion of the investigators was “that deficient situational awareness was evident” and that “the Crew Resource Management (CRM) of the crew was deficient” (CRM is a training for commercial pilots for identifying their own limitations and using their resources as a crew).

A closer look, and the intended lesson in this paper, help reveal that none of these conclusions explained anything. They are nothing but “finger food” or “fragments,” meaningless outside the context that produced them. One piece may not have anything to do with the other piece. The question of ‘’What shall we do?” from one pilot to the other, for instance, could have been the initial question for a decision for one of several identified options. Therefore, the “labels,” “situational awareness,” and “CRM” are just theoretical models, enabling a quick and easy concordance between several experts with different backgrounds, but not the cause of an accident. The investigators in the described case did nothing more than use a deceptively mundane phenomenon like “getting lost” and “being confused” for the causation of the accident. Such conclusions do not explain anything. Even worse, they are unverifiable and are not human factors.

Conclusion

The information presented in this paper focused on explaining human factors when reconstructing behavioral sequences. It was introduced that human factors is an applied science with numerous applications. As a science, human factors is based on a continuously growing body of knowledge and encompasses concepts from psychology, mainly behavioral and cognitive psychology. Every single concept describes a highly complex process finding its expression in human behavior. The complexity was primarily shown via the processes of human vision and decision-making. A simple example, a mailbox “turning into a pedestrian,” was used to demonstrate that human factors deals with explaining a situation, an incident, or accident, i.e., “how it happened.”

In a second look, the human factors “coin” was viewed from the other side, introducing the fallacies and “traps” when applying human factors in accident reconstruction. When looking retrospectively we have the benefits of knowing the outcome of an incident or accident. However, this benefit must not lead us to taking the position of pinpointing what people missed or should have done in an accident. Human expertise is not about judging by comparing different choices someone should have made.

Despite human factors being based on numerous scientific concepts, these concepts alone cannot be parroted as an explanation for a human error or an unfortunate outcome of a situation. This would be a concept-orientated approach with the danger of not being understood by an audience and can enter the realm of being unverifiable and nonanalytic.

This paper pointed out that human factors work takes a straight context-orientated approach. The human factors expert explains how something happened in the context of the time and space of a situation. It shows that human behavior “is what it is.”.

References

1. Aeronautica Civil (1996). Aircraft accident report: controlled flight into terrain, American Airlines flight 965, Boeing 757–223, N651AA near Cali, Colombia, December 20, 1995. Bogota, Colombia: Aeronautica Civil.
2. Boring, E. G. (1933). The law of effect. Science, 77 (1995), 307-307.
3. Dekker, S.W.A. (2005). Reconstructing Situated Performance in Human Error Investigations. In Noy, Y.I., & Karkowski, W. (2006), Handbook of Human Factors Litigation. CRC Press, Boca Raton.